You are on page 1of 367


August 17-21, University of Leicester, UK


Selected Contributions

Selected Contributions

GIREP-EPEC & PHEC 2009 International Conference

August 17-21, University of Leicester, UK

Derek Raine, Cheryl Hurkett & Laurence Rogers

Selected Contributions
GIREP-EPEC & PHEC 2009 International Conference
August 17-21, University of Leicester, UK

Groupe International de Recherche sur l’Enseignement de la Physique (GIREP)
Department of Physics and Astronomy, University of Leicester
Physics Higher Education Conference (PHEC)
European Physical Society (EPS)
Physics Innovations Centre for Excellence in Teaching and Learning (piCETL)


Institute of Physics (IOP)
Higher Education Academy, Physical Sciences Centre (University of Hull, UK)
The Open University (OU)

Lulu / The Centre for Interdisciplinary Science, University of Leicester.
All chapters ©2010 Named Authors
This book is released under a Creative Commons licence for noncommercial non-derivative
use. for more details see
ISBN: 978-1-4461-6219-4

• Ton Ellermeijer (Co-Chair) University of Amsterdam, Netherlands

• Robert Lambourne (Co-Chair) The Open University, Milton Keynes, UK
• Ian Lawrence Institute of Physics, UK
• Laurence Rogers University of Leicester, UK
• Elena Sassi University of Naples, Italy
• Urbaan Titulaer Johannes Kepler University, Linz, Austria

LOCAL ORGANISING COMMITTEE (University of Leicester, UK)

• Derek Raine • Laurence Rogers

• Cheryl Hurkett • Lisa Brandt
• Stuart Lyon


• Ton Ellermeijer (President) University of Amsterdam, Netherlands

• Michele D’Anna (Vice-president) Alta Scuola Pedagogica, Switzerland
• Ian Lawrence (Vice-president) Institute of Physics, UK
• Gorazd Planinsic (Secretary) University of Ljubljana, Slovenia
• Rosa Maria Sperandeo-Mineo (Treasurer) Universita di Palermo, Italy



• Robert Lambourne (Chair) The Open University, UK

• Christian Ucke (Secretary) Technical University Munich, Germany
• Els de Wolf NIKHEF/University of Amsterdam, Netherlands
• Costas Constatinou University of Cyprus
• Hendrik Ferdinande University of Ghent, Belgium
• Gorazd Planinsic University of Ljubljana, Slovenia
• Elena Sassi University of Naples, Italy
• Urbaan Titulaer Johannes Kepler University, Linz, Austria
• Laurence Viennot University of Paris 7 - Denis Diderot, France

HIGHER EDUCATION ACADEMY, Physical Sciences Centre (University of Hull, UK)


• Tina Overton • Dick Bacon

• Liz Pickering • Ruth Mewis
• Tracey Maddon

The 2009 GIREP-EPEC Conference was held in the University of Leicester, UK from 17th
August to 21st August. The meeting also incorporated the annual UK Physics Higher Edu-
cation Conference (PHEC). The plenary programme was organised by the European Physical
Society and GIREP. Registration for the conference was administered by the Higher Education
Academy Physical Sciences Centre. The Local Organising Committee was responsible for the
conference administration and the programme of invited papers. The meeting received spon-
sorship from the EPS, GIREP, the Physical Sciences Centre and the Physics Innovations Centre
for Excellence in Teaching and Learning (πCETL).

The conference attracted around 130 participants. There were 10 invited talks, 93 oral presen-
tations and workshops and 27 poster presentations. All presenters were invited to contribute to
the CD of the meeting which was distributed to all registered participants. A selected number
of presenters (including those with posters) were invited, on the basis of the recommendation
of chairs of sessions, to submit papers for peer review for these proceedings. Each paper was
refereed independently by two referees. Those accepted are published here.

Derek Raine
Laurence Rogers
Cheryl Hurkett

Cover image: The Astronomical Clock at the University of Leicester, designed by Allen Mills
and Ralph Jefferson. Colour photograph kindly supplied by Alun Salt.


Pollen Project - a community approach to science education . . . (Blagotinsek, A.G.) . 1

Modelling and collaboration as cognitive tools in physics education . . . (Corni et al.) . 5
Microscopic and macroscopic aspects of student knowledge in electric conduction in
metals . . . (Fazio et al.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
How music and science can present the world (and its quanta) . . . (Frans et al.) . . . . 35
Year 8 students’ understanding of astronomy as a representational issue: Insights
from a classroom video study . . . (Hubber) . . . . . . . . . . . . . . . . . . . . 45
We Are Teaching Students - What Do They Know? . . . (Kranjc et al.) . . . . . . . . . 65
Frameworks for talking about energy - mutually exclusive? . . . (Logman, P. et al.) . . 76
Design research methodology in a physics learning material project at NDU . (Rissanen) 91
Problem based education: Modelling atmospheric changes with soap bubbles . . . (Antonov,
P. et al.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Towards a Learning Path for Computer Modelling . . . (van Buuren et al.) . . . . . . . 110
Experiments with birefringent layers and coloured polarized light from an LCD mon-
itor . . . (Faletič, S. et al.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Experiments and Models for Physics Learning in Primary School . . . (Mariani et al.) . 137
Multilayered simple experiments: an approach with increasing cognitive demands
. . . (Planinšič et al.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
Experiments with microwaves and wood . . . (Ziherl, S. et al.) . . . . . . . . . . . . . 156
“And the winner is . . . ” Using clickers and student-generated content to promote
engagement through class competitions . . . (Aliotta, M. et al.) . . . . . . . . . . 166
Attitudes Towards Science Study: CLASS at the University of Edinburgh . . . (Bates
et al.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
How Do We Identify ‘At Risk’ Students Upon Entry To University? . . . (Birch, M. et
al.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
The Impact of Low Attendance on the Success and Progression of Level 1 Physics
Undergraduates: Analysis and Intervention . . . (Casey, M.M. et al.) . . . . . . . 196
Using e-assessment to support distance learners of science . . . (Jordan, S. et al.) . . . 202
Constructing physics glossaries to support student learning . . . (Lambourne) . . . . . 217
3-D Immersive Screen Experiments . . . (Lucas) . . . . . . . . . . . . . . . . . . . . 224
Visual Images as Hooks for Problem-Based Learning . . . (Raine et al.) . . . . . . . . 240

Making Meaning with Maths in Physics: A Semantic Analysis . . . (Redish et al.) . . . 244
Testing the Development of Student Conceptual Understanding of Quantum Mechan-
ics . . . (Roberston et al.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
Research-based Curricular Changes that Improve Student Attitudes and Problem-
Solving Abilities . . . . . . . . . (Teodorescu, R. et al.) . . . . . . . . . . . . . . . . 274
Analysis of Lao university students’ collaborative activities in a pendulum experiment
. . . (Vilaythong et al.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
Alexander Graham Bell & the Assassination of President Garfield: Teaching the
Physics of Early Attempts at Medical Imaging . . . (Zollman et al.) . . . . . . . 301
Digital photography for scaffolding project-based-learning . . . (Langley, D. et al.) . . 308
Pedagogic coaching of schools and teachers in Flanders (Belgium) . . . (Peeters) . . . 324
Professional Development for Middle and High School Teachers in Nanoscale Sci-
ence and Technology: Models from the United States and Finland . . . (Sederberg
et al.) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333
Firefly - A New Contest in Science for Primary School . . . (Sokolowska) . . . . . . . 352


Ana Gostincar Blagotinsek

University of Ljubljana, Faculty of Education, SI-1000 Ljubljana, Slovenia.

Pollen is a European research and development project supported by the EU within the 6th
Framework Program, Science in Society. It is the EU’s reaction to the fact that the public
image of science and the interest for science studies among young people are very low. But
progress and our standard of living can be maintained only if we have enough educated
young people to sustain the progress in science, too.
Research shows that the attitude towards science is not formed at the time when young
people choose their careers, but much earlier (OECD, 2008). So science education should
not be fostered only in lower secondary schools and later, but also in primary classes and
Pollen Project activities stimulate and support science teaching and learning in primary
schools and kindergartens by providing tools, training and coaching for the teachers. Al-
though the Pollen project was concluded in June 2009, its outcomes will be disseminated on
a large scale within a new one, called Fibonacci, starting in January 2010. Physics teachers
can play an important role in promoting good practices in science teaching.

Scientific literacy is essential for every individual in modern society, not only to contribute to society and
participate in community as a citizen, but also in private, every-day life. But young people’s interest in
science studies and careers has been alarmingly low for some time now, and different measures, taken to
improve the situation, have had only modest effects. Unless some more effective actions are taken, the
European long term future is threatened (Rocard, 2007).
The origins of the declining interest for science studies are various: the image of the possibilities that
careers in science offer is not what the personal expectations of young people are, the public image of
science is poor, and the family environment is often not supportive. But classroom practice, which is not
stimulating and cultivating an interest for science, could be the main reason (OECD, 2008).
Educational experts mostly agree that pedagogical practices, based on active learning, a hands-on ap-
proach and inquiry based science education (IBSE) are most effective, but unfortunately they are not
implemented in the classroom practice of most European countries.
To make things worse, primary teachers are no exception in general negative trends regarding science

The Pollen Project united the efforts of twelve European countries (Belgium, Estonia, France, Germany,
Great Britain, Hungary, Italy, Netherlands, Portugal, Slovenia, Spain, and Sweden) to develop and pro-
mote IBSE in primary science education. It started in January 2006 by establishing twelve seed cities,
one in each country. A seed city is an educational territory, supporting science education through the
commitment of the whole community. The educational process starts in the classroom, but is not con-
fined to it. Cooperation with the local community, science experts, public institutions, industry, parents
and so on is encouraged and used to enrich the learning process and raise awareness of the relevance and
importance of science - both among pupils and community as a whole. To address the young generation,
especially girls, the relevance of science for improving life and solving problems in everyday life has to
be stressed. To achieve that, and change the public attitude, which significantly affects young people’s
choices, community involvement is crucial.
IBSE is at the core of the pedagogical approach supported by Pollen. Planning investigations, experi-
menting, arguing and debating results are given special attention and help pupils promote their under-
standing of science, as well as their scientific reasoning, language and communicating skills. Research
in countries with a longer tradition in IBSE showed that this approach increases pupils’ achievements not
only with science, but also with (mother) languages and maths. And, also very important in the current
situation, IBSE has proven its efficacy in increasing pupils’ interest and teachers’ motivation for science
(Rocard, 2007).
To stimulate and support science teaching and learning in primary schools, teachers are offered material,
methodological and pedagogical resources and tools, as well as support for the daily work at school,
compatible with the framework of the local curriculum.
More information about Pollen project and outcomes are accessible free of charge at www.pollen-


Activities within TEMPUS project and the resulting curricula changes in 1990s established a hands-on
approach to science education in Slovenian schools, but IBSE was not introduced yet. It was unknown to
Slovenian teachers, and consequently, pupils were not familiar with it, which was clear from their scores
in TIMSS and PISA.
As teachers have a key role in the renewal of science education, special attention was given to the teach-
ers’ preparation, as implementing new teaching methods requires additional teachers’ education (Duschl,
2006). Preparing teachers for the new teaching approach was done in three stages: Firstly, suitable parts
of the curricula were chosen, and didactic materials for teachers and classroom use were prepared. Both
scientific content and didactic issues were addressed. Adequate experimental equipment was carefully
selected next. Workshops were then organized for teachers, who carried out and tested all the activities
that they later conducted with pupils in the classroom. When they choose to do it, they can rent the same
equipment they used, which is provided in 15 sets. This enables pupils’ experimentation in pairs even
in the largest classrooms, and one set for the teacher’s use. After each use the boxes are returned to the
keeper and refurbished.
Kindergarten teachers were also invited to the workshops, and several topics prepared especially for
pre-school children.

The project activities were carried out over three years. During this time, 196 primary teachers and 175
kindergarten teachers attended the workshops. In the last year of the project alone, an estimated number
of 1700 pupils were engaged in IBSE.
Both pupils’ and teachers’ attitudes towards science were evaluated at the beginning and after two years
into the project. A significant increase was measured in attitude and confidence regarding science. At-
titudes towards other subjects, included in the questionnaire, did not change in the same time (Jarvis,
The possibility to rent the equipment was the most appreciated innovation, which teachers feel really
enables them to conduct active science lessons.


The Pollen Project ended in June 2009. As was anticipated, activities in only one city in each of the
twelve countries can not sustain the desired changes in pedagogical practice. To promote the results and
good practice of the Pollen Project (and another, Sinus Transfer, which is devoted to innovative teaching
of math) further, the EC approved a new project, proposed by partners from 23 European countries within
the FP 7 programme, called Fibonacci.
As the OECD study clearly established, the attitude towards science is formed at the age of 11, or even
before (OECD, 2008), so we all should take some measures to improve the teaching of science in the
early years. Science in grades K-5 (this is the case in Slovenia, in other countries the situation being
similar) consists of biology, chemistry and physics. Primary teachers teach them all, but fear and avoid
the physics part most.
Primary teachers were encouraged to join the Pollen Project as a team; at least two from the same primary
school should join simultaneously to provide support to each other. In some cases, a biology, chemistry
or physics teacher from the same primary school also joined the project. (Lessons for grades 1 to 9 are
usually organized in the same building in Slovenia, but organization of classes is different and separated
for grades 1-5 and 6-9.) Although this is not usual, primary and “specialist” teachers began to work as
a team within the school and support each other. “Specialist” teachers were able to help primary ones
with content knowledge. This kind of help was especially welcome from physics teachers, because the
physics part of primary science is perceived as the most problematic by primary teachers. When Pollen
equipment was in the school, all teachers involved in the project discussed problems they encountered
and shared experience, which is not done otherwise. As a result, primary teachers reported that their
knowledge and self-confidence improved. Consequently, their attitude towards science and teaching
science improved too, which was exactly the aim of the project.
But working in team is beneficial for both: primary teachers got the necessary theoretical (and, if nec-
essary material) support, and physics teachers became aware of how much physics is being learned by
pupils before they actually enter the physics classroom. If it is done properly, it nourishes the interest of
young people and builds a solid foundation, upon which the physics teacher can build.
Fibonacci is our common opportunity to improve early science and the physics community must be aware
it has to contribute too. We are all aware of the importance of promoting science in the community and
of joining our efforts to the benefit of it. I call for physics (and other) teachers to work together as a com-
munity and help each other too. It will be to the benefit of science, physics and the general educational
process. Visit for more information on the activities of the project.

Duschl, R.A. (ed.), Schwingruber, H.A. (ed.), Shouse, A.W. (ed.) (2006) Taking Science to School:
Learning and Teaching Science in Grades K-8. The National Academies Press.
Jarvis, T. (2009) Promoting creative science cross-curricular work through an in-service programme,
School Science Review, vol.90, p. 39-46
Organization for Economic Co-operation and Development (OECD) (2008) Encouraging Student Inter-
est in Science and Technology studies.

Rocard, M. et. al. (2007) Science Education Now: A renewed Pedagogy for the Future of Europe:


Federico Corni[1] , Enrico Giliberti[2], Marisa Michelini[3], Lorenzo Santi[3] & Alberto Stefanel[3] .
[1] Physics Department, University of Modena and Reggio Emilia, Via G.Campi 213/A, I-4110 Modena,
[2] Department of Social, Cognitive and Quantitative Sciences, University of Modena and Reggio Emilia,

Via A. Allegri 9, I-42100 Reggio Emilia, Italy

[3] Physics Department, University of Udine, Via delle Scienze 208, I-33100 Udine, Italy


The Green Paper on teacher education in Europe (Buchberger, 2000) highlights the crucial role
of designing learning situations in which students can find opportunities to develop structures of
meaning, knowledge and activities for a didactical reconstruction of the disciplinary contents,
integrated with pedagogical competencies, methodologies and teaching practices.
In the case of Sciences, and of Physics in particular, the methodological characteristics of the
disciplines lead also the didactical methodologies. The epistemic role of models in physics
(Hestenes, 1996) suggests that models can be included in teaching and learning methodologies
and cannot be omitted in a cultural base in science education of teachers of all levels. Models
being cognitive tools and not simple formal descriptions of phenomena, they cannot be trans-
ferred either as ready-to-use examples or taught as rules to apply. Teachers should be prepared
to offer specific experiences of model building starting from experimental activities (Michelini,
A special effort should be made in the training of primary school teachers. They, in fact, suffer
from weak scientific cultural backgrounds, and the integration of their disciplinary and peda-
gogical competences are often left to their professional experience in the field.
This integration can be achieved by presenting scientific experiments as opportunities for per-
sonal involvement, identifying the variables, the relations and the dynamical structures of phe-
Modelling, fundamental in the scientific field, is fundamental in the process of teacher devel-
opment too, as an instrument to acquire epistemological knowledge and as a “language” to
approach, analyze, describe ad interpret phenomena.
Previous research (Corni, 2006) showed that some elementary recurring basic model structures
represent “syntactic elements” of a language oriented to modelling. As an example, in the evo-
lution toward equilibrium of different systems it is possible to identify the flow of an extensive
quantity which is conserved and whose flux intensity depends on the difference in the levels of
a corresponding intensive quantity experiencing a resistance.

A formative pathway was designed, with specific activities and monitoring tools, based on mod-
elling and collaborative work, addressed to students of Primary Education of the Universities
of Modena and Reggio Emilia and of Udine (Italy). Students had to design and perform three
different experiments and to build the corresponding models with the VnR software. The three
experiments consisted of the evolution toward equilibrium in three different contexts: the level
of water in two communicating vessels, the potential in two parallel capacitors and the temper-
ature of water in two containers placed one inside the other.
In the research presented here, the key role of group investigation in a cooperative learning
environment (Sharan, 1992) was implemented in a community of prospective primary teachers.
Our proposal consisted in a research-based learning pathway based on modelling and starting
from experiments. The research aims to investigate how the modelling activity integrated with
individual and group experimental analysis of selected situations produces learning of basic
physics concepts. Moreover we detail the role of the VnR modelling tool, chosen because of its
specific qualitative approach, specially suitable for primary school teachers.

The questions that, in particular, guide this work are the following:

RQ1. Are cooperative work and discussion helpful in selecting the relevant variables of a phe-
nomenon, in recognizing correlations between them, both on the descriptive and on the
interpretative level?

RQ2. Is modelling activity, with VnR in particular, useful in helping students in selecting vari-
ables, in recognizing model structures corresponding to physical concepts, and in the use
of the feedback?

RQ3. How do the different contexts (hydraulic, thermal, electric) influence the above questions?
Are there any significant differences between contexts from a didactical point of view?

The 76 students from 2 university sites were grouped in two sample sets:

• Sample-set 1 (SS1), consisting of 64 second year university students from the Science
Teaching for Primary School Educators Degree in Modena-Reggio Emilia University.
During the activities these students were grouped in 14 groups of 3-4 students each.

• Sample-set 2 (SS2), consisting of 10 third year university students from the same class
in Udine University. During the activities these students were grouped in 5 groups of 2
students each.

The two sample groups were considered separately for most of the data analysis, due to their
numeric difference as well as their training backgrounds.

To build formal thinking and to create a competence in the use of formal representation and
relationships we choose an integrated approach based on:

1. Identification of the relevant variables, both individually and in groups

2. Experimental activity with data collecting and representation, both traditional and with
on-line sensors

3. Modelling activity, using the modelling software VnR (Variables and Relationships) char-
acterised by a qualitative approach

Each group of students had to perform three different experiments, not necessarily in order.
According to the chosen methodology, we made use of three Worksheets and VnR software.
The three experiments were chosen in order to compare similar phenomena in three different
contexts, see Table 1.

# Experiment Extensive Conjugated Reservoir Capacitance Resistor Resistance

quantity intensive
i Communi- Water Water level Cylinder Cross- Pipe Hydraulic
cating volume section resistance†
vessels of the
ii Parallel Charge Potential Capacitor Capacitance Resistor Resistance
iii Baun-marie Heat Temperature Water mass Thermal Container Thermal
capacity wall insulation‡

Table 1: Similarities between the three experiments. † Comprising the effects of the pipe
impedance and of the fluid viscosity. ‡ Inverse of thermal conductivity.


In the first step students had to plan the experiments to study the evolution toward equilibrium
in the three different contexts: hydraulic, electric and thermal.
Worksheet A, used in the first activity (first individually then in groups), contains:

• Task: study the evolution with time of the process specifying the different initial condi-
tions and the adopted parameters

• Materials

• Measurement instruments

• Planning hints

A) Initial analysis of the task

B) Designing of one or more procedures
C) Feasibility study with the available resources
D) Expected results
E) Job subdivision within group members
F) Execution of the experiment
G) Log Writing
H) Data report
I) Results discussion and conclusions


In the second step students in groups performed the experimental activity, collecting measure-
ments of the relevant variables and writing the log of the experimental work. Worksheet B
guided the students in these activities. It is not presented here as it is considered not useful for
data analysis.


In the third step students were asked to build the model of the phenomenon. We choose to
make use of the modelling software VnR due to its qualitative approach, in which the design of
models is performed using icons. Column-like icons represent the variables and can be either
positive-only or positive-and-negative. The links between variables represent relations, and can
be either a static relation (sum, product, and their inverses) or dynamic (rate of change). Figure
1 reports some basic structures with the corresponding mathematical relations.
Students followed the Worksheet C, containing some questions useful to build the model: A)
Which are the relevant quantities?; B) What are their relations?; C) Are there other quantities
that can be considered and how do they affect the previous ones?; D) How can you explain it
with a drawing or with words? How can you represent it using graphs, icons, etc.?
In the final part of the Worksheet C, students had to list all the variables and to specify their
relations in a table (Table 2).
At the end of the Worksheet C, students had to draw on paper the expected model, before
building it with the VnR software.

Figure 1: Basic relations in VnR.


While filling Worksheet C students were expected to be able to find all the relevant variables
and to include in the model the following basic concepts:

• Capacitance

• Difference of potential

• Resistance

Figure 2 shows a possible representation with VnR of the model of the three phenomena. The
model is the same for the three processes, according to the parallelism among the basic concepts.
In the final discussion teacher pointed out the main similarities and difference among the exper-
iments and the models built by the students.

Is it a free quantity or is it dependant on other physical quantities?
Description of the physical quantity

Corresponding variable symbol

Free Dependant

It varies during It is fixed at the Variables on Relation with the

the experiment beginning of the which it depends listed variables
(variable) experiment, then (list the symbols) (use the symbol)
it does not change

Table 2: Worksheet C, table for definition of the variables.

Data related with the formative activity come from various sources, in particular from the
recording of the group discussions, from the log of the experimental work and from the written
answers to the worksheets used to guide the students during the experimental and modelling
In order to investigate the role of the modelling activity we analyse and compare the group
answers to Worksheets A and C and the group models made with VnR.


The analysis of the role of the teamwork in the identification of the relevant quantities we
compare the individual Worksheets A with the corresponding group ones.
Table 3 reports the physical quantities evidenced or mentioned by students in Worksheet A,
according to the following criteria: A) Overall number of quantities; B) initial and final states of
the phenomena; C) descriptive or kinematical variables, reporting the “movement” or evolution
in time; D) interpretative variables (charge and current, energy and heat, mass and flow); E)
process quantities (resistance, capacitance, conductibility).
Data are separated according to Sample-set (Modena and Reggio Emilia, and Udine), Indi-
vidual and group data, Experiment (Communicating vessels, Thermal interaction, and Parallel
The average number of quantities for each worksheet is 2.5. Group SS1 shows very little vari-
ation between individuals and groups. Group SS2 shows a larger variation, both increasing
(especially in the case of Communicating vessels and of Parallel capacitors) and decreasing the

Number of Physical Quantities
Number of Initial/final Kinematic Interpretative Process
quantities values Variables Variables Quantities
Communicating vessels N1 N2 <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2>

Individual 26 5 2,5 1,2 2,0 1,1 0,3 5,4 0,3 0,0 1,1 2,0
Groups 6 5 3,0 5,6 2,3 1,1 0,5 4,6 1,0 0,2 1,2 1,4

Thermal interaction between water N1 N2 <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2>
Individual 22 4 2,4 7,8 1,3 1,3 1,0 0,8 0,3 0,3 0,9 0,0

Groups 8 5 1,9 4,0 1,4 1,3 0,6 0,6 0,3 0,0 0,8 1,2

Capacitors connected via a resis- N1 N2 <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2>
Individual 2 10 3,0 2,7 1,0 0,0 1,0 0,0 0,0 0,9 1,5 0,7
Groups 3 4 2,7 5,8 1,3 0,0 1,0 0,5 0,0 0,8 1,0 1,5

Mean indiviual 50 19 2,5 3,4 1,6 0,6 0,6 1,6 0,3 0,5 1,0 0,9
Mean group 17 14 2,4 5,1 1,7 0,9 0,6 2,0 0,5 0,3 0,9 1,4
Table 3: Number of physical quantities, worksheet A.
Figure 2: Possible model structure for the three experiments.

number of variables (Thermal interaction). SS2 shows a larger number of variables, both indi-
vidual and group, in the case of Thermal interaction, probably due to a previous study of the
thermal context.
SS1 reports a number of variables larger than SS2, both in initial and final states.
At least one descriptive or kinematical variable is reported; in the case of SS2 and Communicat-
ing vessels it is more marked, probably due to the presence of water, more evident to perception.
Interpretative variables (charge and current, energy and heat, mass and flow appear in a smaller
number of cases compared with the descriptive ones (< 30%).
At least one process quantity is reported, with few more variables considered by the groups of
Only small differences among contexts are recorded.
Table 4 shows the numbers and the percentages of individuals and groups mentioning the four
relevant quantities (time, potential, difference of potential, current). The average of descriptive
or kinematical variables (time, potential) increases after group activity. The peer cooperation
facilitates the identification of the variables involved in the processes, helping groups to consider
and combine the suggestions of the group members. This increase is not evident in the case of
interpretative variables (difference of potential) and flow variables (current). This is probably
due to the well-known difficulty of distinguishing between a variable and its variation.
Table 5 reports, for the three experiments, the comparison between the corresponding sections
of Worksheets A and C, in particular counting the number of physical quantities reported.

Time Potential Difference Current Charge/ Capacity/ Resistance/
of Heat/ Thermal Thermal
potential Water capacity insulation/
volume Impedance
N1 N2 %n1 %n2 %n1 %n2 %n1 %n2 %n1 %n2 %n1 %n2 %n1 %n2 %n1 %n2
Mean 50 19 32 26 64 54 11 21 20 21 16 46 80 42 22 54

Mean 17 14 41 71 65 81 14 16 18 37 12 74 76 68 18 39

Table 4: Percentage of relevant quantities, individual and group.

Number of Initial/final Kinematic Interpretative Process
quantities values Variables Variables Quantities
Communicating vessels N1 N2 <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2>

Worksheet A 6 5 3,0 5,6 2,3 1,1 0,5 4,6 1,0 0,2 1,2 1,4
Worksheet C 12 4 5,6 8,5 0,3 0,2 1,4 2,0 0,5 1,5 0,8 2,5

Thermal interaction between water N1 N2 <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2>
Worksheet A 8 5 1,9 4,0 1,4 1,3 0,6 0,6 0,3 0,0 0,8 1,2
Worksheet C 12 4 6,0 5,4 0,6 0,5 1,8 2,0 0,7 0,2 0,9 2,6

Capacitors connected via a N1 N2 <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2> <N1> <N2>
Worksheet A 3 4 2,7 5,8 1,3 0,0 1,0 0,5 0,0 0,8 1,0 1,5
Worksheet C 10 3 5,9 7,3 0,4 0,5 1,8 1,8 0,6 1,5 1,2 2,8

Worksheet A 17 14 2,4 5,1 1,7 0,9 0,6 2,0 0,5 0,3 0,9 1,4
Worksheet C 34 11 5,8 7,0 0,4 0,4 1,7 1,9 0,6 1,0 1,0 2,6
Table 5: Number of physical quantities; comparison between worksheet A and worksheet C.
The comparison of the number of variables reported in Worksheet A with the number of vari-
ables reported in Worksheet C permits the observation that in each experiment the number of
variables increases. The kinematical and the interpretative variables are reduced to the mini-
mum, preserving only the variables necessary to have an adequate description of the phenom-
ena. In the worksheet A students tend to consider all the variables suggested by the members
of the group, but in worksheet C they are required to select only the relevant quantities, having
to design the VnR model. Additionally a small growth in the number of process variables is


Comparing VnR models and the corresponding models built on paper on worksheet C only
few changes are noticeable. In particular the number of variables is systematically reduced,
presumably due to the need to build a “working” model, performing a dynamical evolution,
rather than achieving the right connections between the variables. Moreover, most of the models
are referred to partial and specific aspects of the experiments, and the contribution of only
some variables is considered. We classify the models according to the following categories: A)
models with a linear structure connecting variables in the correct order but without feedback; B)
models with feedback but without functional correlation; C) uncorrelated models; D) models in
which a transfer of structure from models built previously is recognisable.
In Figure 3 some examples of each of the four categories are reported: A) Model with flux
determined by the difference of potential; B) Model without feedback (uncorrelated); C) model
with difference of potential, flux, feedback, resistance, capacity; D) Model with transfer. Mod-
els B), C) and D) were built by the same group of students and show the evolution in the ability
of model building and the transfer of structure from one context to another.
Models distribution, classified according to the categories listed, are reported in figure 4. Cat-
egory D, showing transfer between models, is well represented. Models of type A, with linear
structure, showing no feedback, are essentially the first models built by each group, and the
models of category D are the last, built during step 2 and 3 of the activities.


From data reported in Table 1 it emerges that the role of collaborative group activity favors the
recognition of the variables, enriching their number and focussing on the relevant quantities
among the others. The peer collaboration contributed to the identification of descriptive model
elements, but did not help significantly the interpretation of the observed phenomena (RQ1). On
the other hand, the experimental activity, integrated with model design activity, enhanced the
recognition and definition of variables as reported in Table 3, showing the comparison between
worksheet A, related to the experimental activity, and worksheet C, related to the modelling
activity. From Tables 1, 2 and 3 it emerges that students are inclined to build models making
use of descriptive variables instead of interpretative ones (RQ2).
The VnR modelling tool can be considered a useful instrument to help the formalization of
mathematical relations in students with poor mathematical background.


Table 6: Examples of VnR models built by students.

Specifically, the recognition of the role of flux and feedback, the construction of descriptive
kinematical models and the recognition of model structures and their transfer from one context
to another proved to have an important role in the construction of functional relations, as showed
by the classification of models in figure 4. A lower effectiveness emerged concerning the tran-
sition from the descriptive plane to the interpretative one. VnR helped in the recognition of the
relevant variables and of their basic connections. However such connections are more likely
useful to ensure the functionality of the model, than connected to an effective organization of
concepts (RQ2).
Concerning the role of different experimental contexts (hydraulic, thermal, and electric), it
emerges, in particular, that their order of presentation is not relevant. Rather, the approach
to different contexts having formally analogical models leads to the recognition of the various
elementary component structures (RQ3).

Buchberger, F., Campos, B.P., Kallos, D. & Stephenson, J. (eds), (2000), Green Paper on teacher educa-
tion in Europe, TNTEE, Umea, Sweden.
Corni, F., Giliberti, E., Michelini, M. & Stefanel, A. (2006), The basic models in the experimental work

Figure 3: Distribution of the model categories.

of student-teachers of primary school Modelling in Physics and Physics Education: Proc. of the GIREP
Int. Conf. (Amsterdam, The Netherlands, August 2006)
Hestenes, D. (1996), Modelling methodology for physics teachers, Proc. of the Int. Conf. on Under-
graduate Physics Education (College Park, August 1996)
Michelini, M. (ed), (2004), Quality development of teacher education and training, GIREP 2003 book of
selected paper, Forum Edizioni.
Sharan, Y., & Sharan, S. (1992), Expanding Cooperative Learning Through Group Investigation, New
York: Teachers College Press.


A. Bonura, C. Fazio, I. Guastella & R.M. Sperandeo-Mineo

UoP_PERG (University of Palermo, Physics Education Research Group), Dipartimento di Fisica e
Tecnologie Relative, Universitá di Palermo, Italia.

This paper presents some preliminary results of research aimed at portraying in-
dividual models of description and explanation of electrical conduction in metals
shown by prospective physics teachers, and at studying how a specifically designed
pedagogical environment can support an adequate evolution of their knowledge and
disciplinary competencies. A teaching and learning sequence (TLS) on the subject
matter, making use of custom information technology tools, aimed at facilitating
and scaffolding learning, is described. Some relevant points of the pilot phases
of the research, like the analysis of prospective teachers’ spontaneous models of
electrical conduction, the identification of significant correlations between micro-
scopic and macroscopic aspects of knowledge, and the study of the improvement
that a pedagogical environment based on experimental and modelling activities can
give to prospective teachers’ model building abilities are discussed, in the light of
specific research questions.

Modelling matter is today part of college curricula, as well as at high school levels, given the
need to facilitate students understanding of commonly used technological devices. It can also be
considered a good way to give students a meaningful understanding of physics as a modelling
endeavour and of the quantum behaviour at microscopic level that reflects on macroscopic,
measurable properties.
In particular, some research studies (Osborne, 1983; Shipstone, 1985; Borges & Gilbert, 1999)
have focused on problems related to the teaching and learning of macroscopic aspects of elec-
tric conduction, and others have faced these problems by taking into account the microscopic
aspects of such phenomena (Eylon & Ganiel, 1990; Stocklmayer & Treagust, 1996, Wittman,
This paper describes some preliminary results of a research project aimed at designing, experi-
menting and evaluating Teaching Learning Sequences (TLSs) focused on the use of experiments
and microscopic models of metals, in order to describe and explain some properties observed at
macroscopic level. The TLS described here is based on measurements of electric properties of

conductors and on increasingly refined models of electric conduction, following an approach in-
spired by their historical development. The models are tested and confronted with experimental
results by means of “virtual experiments”, built by using the NetLogo simulation environment
(Netlogo, 2009).
The TLS is built on the ground of the model of ‘Educational Reconstruction’ (Kattmann et al.,
1995; 1997), that links analysis of the physics content, research on teaching and learning pro-
cesses, and development of prototype sequences for instruction to be experimented (the teaching
experiment). In particular, the design stage of a TLS hypothesizes detailed learning pathways
on the basis of a twofold analysis of the content structure and the cognitive requirements. The
first analysis involves concepts, ways of reasoning and processes of thinking and the second one
the results of empirical studies of students’ conceptions and representations. This framework
has been adopted in many research studies concerning the classroom experimentation of new
teaching approaches (Duit & Komorek, 1997, 2004; Linn & Songer, 1996). The main assump-
tion of such approaches is that there is not a single content structure in a particular content area,
but different content structures can be performed, according to the specific aims the authors
hold, explicitly or implicitly. Moreover, the process of interpretation used by each student is
influenced by concepts and models he or she already holds. These two issues, students’ models
and ways of reasoning as well as statements and procedures of the scientific knowledge, are
therefore accepted to be of the same relevance and treated as resources for physics education.
In this way, the physics content to be taught is reconstructed in order to realise the main goal:
to allow students to gain a fruitful knowledge of the physical world.
The hypothesis on which our research is based is that pedagogical approaches underlining con-
nections between macroscopic and microscopic aspects of physics knowledge can facilitate
students’ development of adequate competencies in analysing and interpreting natural phenom-
ena, especially in the field of modern physics. In particular, we think that the development of
adequate microscopic models can facilitate the development of a capacity for extracting infor-
mation to interpret macroscopic phenomena (Thacker et al., 1999). Conversely, a correct and
autonomous elaboration of information obtained from macroscopic data can favour the devel-
opment of adequate models of microscopic processes (Thacker, 2003).
More detail on the complete experimental and modelling approach can be found in Bonura et
al., 2007; Bonura et al., 2008; Bonura, 2009. Here, we report some preliminary results of a
pilot research aimed at analysing microscopic models of the electrical conduction in metals of
a sample of prospective physics teachers (from now on called Student Teachers, or STs) and
at identifying significant correlations between microscopic and macroscopic aspects of their
knowledge. We then describe the guidelines of the TLS, designed in order to face the difficulties
evidenced by STs in their individual modelling process and some preliminary results on the STs’
improvements in ability of microscopic model building, as well as in their use in explaining
observed macroscopic phenomena.



Our sample is composed by 21 Student Teachers attending the Graduate Program for Physics
Teacher Education at University of Palermo. All STs, graduates in Mathematics and Engi-

neering, attended, during their previous University studies, at least one course of Introductory
Physics involving the study of electric conduction in metals.
The research questions of the study reported here are the following:

1. What kind of individual models of electric conduction in metals are shown by our ST
2. How are these models used in order to explain or predict macroscopic phenomena of
electric conduction?
3. Are the initial models, and their use, modified by a teaching and learning sequence mak-
ing use of a specific learning environment based on experiments and modelling of exper-
imental results?

The pilot testing of the TLS has been developed in four class periods (the first of one hour length
and the other three involving 3 hours of work). The first class period has been used to administer
a pre-test, and the other three involved discussions about pre-test results, interactive lectures
(for less than the 30% of the class period) and active working sessions, where STs performed
experiments and simulations, discussing their results among them and with the instructor. A
researcher, other than the TLS instructor, was present during all the class periods, taking note
of all the pedagogical activities performed and the STs’ discussions (also by using an audio
recorder for finer grain detail data collection), yet not participating in any way in the pedagogical

3.1 P RE - TEST

We used a questionnaire in order to obtain information about different cognitive variables and
aspects of the STs’ knowledge and abilities concerning electric conduction in metals involving:

• the knowledge of basic concepts related to electrical conduction;

• the knowledge of structural characteristics and components of microscopic models;
• the way STs use a specific model in order to predict and interpret phenomena.

The research described here is based mainly on qualitative research methods (Strauss & Corbin,
1998). Data have been analysed by using a phenomenographic approach (Marton, 1981; 1994).
The STs’ answers have been classified in categories (by two independent researchers) on the
basis of a close reading of their explanations within a framework provided by domain-specific
expertise. Through triangulation, we verified that model definitions came out of STs’ statements
and were not imposed on them.
The first part of the pre-test was aimed at obtaining information about macroscopic knowledge
with respect to conduction phenomena (relevant variables describing conduction, relationships

between them, Ohm’s and Joule’s laws). The results have shown that 52% of STs were able to
give description containing the basic concepts and laws of electric conduction in metals; 30%
showed poor capabilities in describing electric conduction or in building descriptions where
causal relationships between components of the systems were clearly described; 18% of the
sample showed a completely inadequate knowledge of the topic.
Four questions have been directed at characterizing STs’ microscopic models of electric con-
duction by using a double criterion of analysis and classification: a direct one, looking at com-
ponents and structure of individual models, and an indirect criterion, focused on the analysis of
the ways STs use the models in order to predict and explain phenomena and processes. This
second kind of analysis supplied us with a classification of STs’ interpretative abilities and
competencies, allowing us to investigate their conceptions about the functionality of models in
understanding the phenomenological world.
The categorization of STs’ answers to these questions has been performed by considering the
following model characteristics:

• Model components (electron, ions, atoms,. . . )

• Component properties
• Relationships between components (nature and ways of interaction)
• Evolution rules and general properties (laws defining the evolution of single components
and statistical properties)

We will separately analyse the two questions involving model structure and the ones involving
model functionality. Questions 1 and 2 are reported below.

1) The picture on the right shows a metallic

nail. Briefly describe its microscopic structure,
also by using some drawings.

2) Try to give a microscopic interpretation of

electric conduction in the nail if it is inserted in
an electric circuit, as in the figure. You can use
a drawing to evidence microscopic processes
taking place inside it.

Models of the structure of metals depicted by STs in question 1 can be classified in 3 categories:

• Models describing atoms in fixed positions due to their strong interaction forces

• Models where atoms oscillate with respect to fixed positions

• Ions fixed or oscillating and electrons free to move in the lattice space

However, many STs did not assign any role to the atomic lattice in the process of electric
conduction. On the basis of answers to questions 1 and 2, our sample has been divided in four
main categories:

• M1 : STs classified here describe conduction by using a model characterized only by

charge carriers. Typical answers classified as M1 are reported in Table 1. Some individ-
ual models can be described as “hydraulic”, since charge carriers (in a single case both
positive and negative) created by a battery flow freely from one pole to the other (exactly
as in a hydraulic flow). Some other STs describe conduction as due to a “domino effect”,
in which “free electrons collide between them and allow the energy flow (i.e. the electric

• M2 : STs classified here describe a model where conduction electrons are subject to at-
tractive interactions that limit their mobility. In some cases this model, that we call here
“at bond electrons”, seems to resemble the “Jumping Model” reported in the literature
(Wittman, 2002). In other cases, it is described as an “electrostatic” model, in which the
electric field makes positive and negative charges separate. Typical answers classified as
M2 are reported in Table 2.

• M3 : STs classified here evidence a model similar to the classic reference model of con-
duction in metals (the Drude - Lorentz one), with free electrons interacting by means of
collisions with lattice ions. Typical answers classified as M3 are reported in Table 3.

• M0 : STs classified here gave no answer to the questions.

The metallic nail contains free electrons. Free electrons in the nail flow from the
They push each other and this allows the en- negative pole to the positive one. This is
ergy flow (i.e. the electric current). the origin of the electric current.

Table 1: Examples of answers classified as M1 .

Electric conduction induces an electric cur- The metal structure is made of a fixed ion lat-
rent, i.e an electric charge flow. From a mi- tice. Electrons move in the lattice and go to a
croscopic point of view, we can think about lower energy level. The electron movement
atoms of the metal that have a charge distri- from an ion to another, by means of “jumps”,
bution deformed by the electric voltage and is the electric current.
then attract each other by means of opposite
sign parts.

Table 2: Examples of answers classified as M2 .

Conduction electrons, subject to the electric Electrons are free to move in a lattice of fixed
voltage, are accelerated by the force F due to ions. They are accelerated by the external
the electric field. A charge flow in the ion field and then collide with ions, dissipating
lattice can, then, be detected, for time and their energy.
surface unit, that we call “electric current”.

Table 3: Examples of answers classified as M3 .

Table 4 resumes the characteristics of M categories, with the classic model of conduction re-
ported as a reference.
M categories are characterized by a decreasing distance with respect to the reference scientific
model. M1 differs from the reference model mainly for the lack of an essential characteristic (the
ionic lattice structure). M2 takes into account both components of the classic model but differs
from it mainly in the interactions described for charge carriers. The M3 model is basically
equivalent to the reference classic model.
Figure 1 shows the distribution of models in our ST sample.
If we merge data related to models more diverging from the scientific model (i.e. M0 , no model,
and M1 , “simple” model), we note that individual models are almost uniformly distributed:
about 38% of STs are classified as M0 or M1 , 29% belong to the M2 category, and 33% of them
evidence the M3 model (the one closer to the scientific one).

Model Charge Characteristics Ions/lattice Rules Interactions
M0 Not present - Not present - -
M1 Present Free carriers Not present Classical None.
(hydraulic (electrons or physics Electron-
or “charges”) electron.

M2 Present Bond carriers Present Classical More or less

(bond physics intense attractive
electrons/ interactions be-
Jumping tween electrons
model) and ions
M3 Present Free carriers Present Classical Collisions be-
(free physics tween electrons
electrons - and lattice
Classic Electrons Free carriers Present Classical Collisions be-
model physics tween electrons
Drude- and lattice

Table 4: Summary of the M categories.

Pre-test questions 3 and 4 are the following:

3) What is, in your opinion, the microscopic interpretation of electric resistance in

4) Let’s consider a copper bar that is heated (by using a flame). Do you think that
the temperature increase can give measurable effects on the electric properties of
the system? If yes, what are the physical variables that change, and how does this
change happen? Explain your answer on the basis of the microscopic model you
have described at question 3.

These questions were aimed at obtaining data for a finer grain analysis of the functionality of in-
dividual models. The classification of answers has been performed by considering three differ-
ent categories of explanation. A descriptive one, called I1 , in which no element of microscopic
interpretation is present. An explanation category, I2 , that is based on a microscopic model yet
is mainly incompatible with scientific models (or with experimental evidence). A final one,
called I3 , at least qualitatively in accord with the reference scientific model (the Drude-Lorentz
one). Category I0 includes the STs not answering both questions.

Figure 1: The distribution of models.

Table 5 resumes these four categories and gives some examples of answers given by STs for
each of them.

Cat. Nature of explanation Example of STs’ answer

I0 Gives no answer -
I1 Gives a descriptive answer, not includ- Resistance is the force opposed by mate-
ing any microscopic analysis element rial to the current flow.
and not explaining any causal relation. When temperature increases electric con-
duction increases, too.
I2 Uses a microscopic model to pre- Electric resistance is due to strength of in-
dict/interpret the phenomenon, yet this ner bonds in matter.
explanation is not coherent with the When T increases electrons have a greater
scientific model (or with the actual ex- energy and so are more weakly influenced
perimental result related to question by the ions’ electric potential. So, the cur-
4). rent increases.
I3 Gives an explanation basically in ac- The temperature increase makes free
cordance with the classic scientific electrons gain more kinetic energy. This
model (Drude-Lorentz model). increases rate and number of electrons’
collisions with lattice and the macro-
scopic effect is an increase of the system’s
electric resistance.

Table 5: Some examples of answers in categories I0 to I3 .

Figure 2 shows the distribution of I-type classification in the ST sample.

Figure 2: Distribution of I-type classification.

About 30% of our sample evidences an interpretation level classified as I1 ; answers show a weak
capability in connecting the aspects of their microscopic knowledge to explain the proposed
About 28% of our sample STs are classified as I2 . They use their microscopic models, yet their
explanations are not coherent with scientific interpretations and are often in contrast with exper-
imental evidence. All STs but one classified as I2 in answers to pre-test questions 3 and 4 have
been classified as M2 (bond electron model) in answers to questions 1 and 2. In particular, they
elaborate their model and conclude that when temperature increases the electric resistance of
the conductor has to decrease (instead of increasing). This explanation is a direct consequence
of building model M2 .
24% of our STs give answers that are classified as I3 ; they are also classified as M3 , show-
ing that their knowledge of microscopic models, coherent with the reference scientific model,
makes them able to build correctly on their knowledge, by activating a correct set of causal
The remaining 18% of our sample STs did not supply any answer.



We analysed the pre-test results, with respect to both of the two different cognitive variables:
model structural characteristics (M classification) and model functionality or practical use (I
classification). In order to point out correlations between these variables, a contingency table
has been calculated (see Table 6) showing the number of STs classified in the four I categories,
on the basis of their M classification. The table gives detail about the significant correlations
between structural and functional characteristics of the microscopic model evidenced by our
Figure 3 graphically reports the results shown in Table 6, in percentage form.

M0 M1 M2 M3
I0 4 0 0 0
I1 0 4 1 1
I2 0 0 5 1
I3 0 0 0 5

Table 6: χ2 = 45.2, p < 0.001.

Figure 3: Histogram of results in table 3.

These results give evidence of significant correlations between the structure of microscopic
models evidenced by our STs and their use to explain a given phenomenological situation. In

1. STs evidencing M1 models show a low functionality level (I1 ). This result is not so
obvious, as model M1 (making use only of free charge carriers) could have been used to
give results not exclusively descriptive;

2. STs classified as M2 (bond electrons) make use of a microscopic model, yet very often
give explanations not in accordance with phenomenological experience. As an example,
some of these STs think that when temperature increases a metal’s resistivity decreases.
This is in accord with results in the literature, as in Wittmann’s “Jumping Model” (of
which the M2 category can be considered a generalization), where STs’ answers are de-
scribed as due to the bad results of previous instruction, not always to be linked to signif-
icant defects in basic knowledge of the subject;

3. the majority of M3 STs give explanations coherent with scientific ones. Yet, we report
two anomalous cases: two STs not giving correct explanations, with one of them also not
using the microscopic model to predict or interpret phenomena. Even if these STs possess
the required knowledge, they are not able to build on it in a coherent way.

This analysis seems to suggest a qualitative conclusion: an inappropriate representation of in-

teractions between a system’s microscopic components (i.e. the belonging to the M1 and M2
categories) can influence the way STs give interpretations to observed phenomena. Moreover
the lack of a microscopic model (M0 STs) seems to influence deeply the understanding of phe-
nomena, that then remain connected only to the mere recall of previous knowledge.


Taking into account STs representations of micro- models of electric conduction, the following
activities have been developed:

1. Recalling of basic concepts of electric conduction in metals (Ohm’s law, Joule effect, and
so on). This phase has been considered particularly necessary as a consequence of the
poor levels of knowledge shown by some STs with respect to the subject.

2. Analysis of a simple experiment evidencing the effect of temperature on conductor resis-

tance. This has been done in order to: a) show a phenomenon whose analysis can give
information on interaction mechanisms between microscopic components of a conductor
(electrons and ions); b) make evident that there is a conflict between the predictions of the
spontaneous model with bound electrons and the experimental results, in order to favour
the transition to an alternative model more in accord with phenomenology.

3. Building of microscopic models of electric conduction in metals and their implementation

through computer simulations. In this phase all elements obtained by macroscopic obser-
vation are analysed and organised in a coherent schema, in order to build a first model of
metal structure that can take into account the electric conduction phenomenon (the Drude-
Lorentz model). The model is implemented by means of the NetLogo environment and
simulation results are used, in order to correlate the microscopic and macroscopic vari-
ables. In particular, plots of resistivity, ρ, vs temperature, T , are compared with actual
experimental results.

4. Revision of the previous model, in the light of the non-accordance of the ρ vs T predic-
tions of the Drude model with experimental results. Introduction of the Wien hypothesis,
relative to the influence of lattice ion oscillations on the metal resistivity, and modification
of the NetLogo model, in order to deduce the new trends of ρ vs T and compare them
with the experimental data.

5. Evaluation of the model limits and the necessary corrections introduced by quantum
physics (Fermi-Dirac distribution).

The main objective of the various steps is to make evident that different microscopic models can
supply explanations or interpretations of experimental results with different degrees of accuracy.
The simulations implement the models, pointing out the main role of the electron mean free path
and how it depends on microscopic variables. Details of different models are reported elsewhere
(Bonura et al., 2008); here we briefly synthesize them.


In this model the electrons have a velocity distribution in accord with Maxwell-Boltzmann
statistics. Then, at a given temperature T , their mean thermal velocity, vm , is proportional to
the square root of temperature.
The lattice ions are considered at rest with a radius equal to the ion radius. The electron-
ion collisions are treated as perfectly elastic and no dissipation mechanisms are considered.
For not too high electric field strengths, the simulation performed by using NetLogo reaches
a steady state condition in which the dynamical quantities, such as drift velocity and mean
relaxation time, are constant. By changing the system temperature (i.e. the value of vrms ), a
plot of resistivity as a function of T can be obtained, that is well fitted by a function ρ = Axb ,
with b ≈ 0.5, a value not in accordance with the linear dependence of ρ vs T , shown by the
experimental results.


In this model we take into account the ion oscillations, by assuming that they depend on tem-
perature. From theoretical considerations it is possible to assume that the effective maximum
oscillation amplitude is proportional to T 1/2 . The effective ionic radius r is assumed to be equal
to the maximum oscillation amplitude. Electron-ion collision is assumed to be elastic, with
exchanges of energy and momentum. We also consider ions in equilibrium with a thermal bath
at constant temperature. The NetLogo simulation gives a plot of resistivity as a function of T
that is well fitted by a function ρ = Axb , with b ≈ 1.5, a value not in good accordance with the
linear dependence of ρ vs T , shown by experimental results.


The implementation of this model differs from the previous ones since here the quantum na-
ture of electrons is taken into account by considering the Fermi-Dirac statistic. Then, in the
temperature range of interest, the distribution of electrons velocity is almost independent of the
temperature. As in the previous models, the electron-ion interactions are considered perfectly
elastic. The NetLogo simulation gives a plot of resistivity as a function of T that is well fitted
by a linear function, ρ = Ax + B, in good accordance with the experimental results. For more
detail, see Bonura et al., 2008.


The evaluation of TLS activities has been performed by means of written questions administered
to STs during the TLS development. Oral questions were discussed individually and with the

whole group. In order to answer to our third research question, here we report how some STs
became able to construct a model able to describe and explain the variation of metal resistivity
as a function of temperature.
Table 7 reports some examples of answers given to question 4 of the pre-test at the end of the
TLS pedagogical activities.
ST1 and ST2 were classified as M1 and I1 on the basis of their pre-test results. Their attention,
at the beginning of the TLS, was mainly devoted to the effect of temperature on the electron en-
ergy, since they did not assign any role to lattice in the process of conduction. The visualization
supplied by simulation helped them to concentrate their attention on the frequency of collision
(electrons against lattice ions), although ST1 failed to point out the further effect of increasing
the electronic mean velocity (if electrons are considered as classical particles) and ST2 failed to
make note of the temperature effect on ionic oscillations.
ST3 was classified as I2 on the basis of the pre-test results. He gained a good understanding
(qualitative and quantitative) of the full-classic model, that gives an acceptable fitting of exper-
imental data in a short temperature range, near the environmental temperature.
ST4, classified as I2 in the pre-test, shows a good qualitative and quantitative understanding of
the model that has been presented as the more appropriate in order to interpret the variation of
resistivity as a function of temperature.
Almost all STs gained a qualitative understanding of the phenomenon, by justifying the causes
affecting the variations in microscopic behaviour of the different components of the models
used. At the end of the TLS activities, more than 60% were able to add to their interpretations
a quantitative analysis based on an accurate use of the mathematical aspects of the microscopic

The reported study is concerned with the guidelines and the results of the pilot testing of a
TLS on electrical conduction in metals. In particular, the initial phase of this study is aimed at
analysing student teachers’ spontaneous models of electrical conduction and at identifying sig-
nificant correlations between microscopic and macroscopic aspects of their knowledge. It then
analysed how the planned educational environment has been shown to be effective in making
these models evolve towards more scientific ones, useable to explain experimental evidence.
A pre-test, focused on macroscopic aspects of electric conduction, as well as on the building
of interpretative microscopic models, has been administered to 21 pre-service Physics teachers.
TLS results have been analysed by studying the ways STs reacted to pedagogical activities. We
reported the pre-tests results and discussed them with respect to some results about the STs’
improvement in microscopic model building ability. More detail of the whole TLS testing can
be found in Bonura, 2009.
Pre-test results give answers to our first two research question. In fact, STs supplied a clear
and detailed representation of the initial cognitive state of our sample, making evident the char-
acteristics (structure and functionality) of their micro-models of the electric conduction phe-
nomenon. These results gave also the possibility of demonstrating regularities and significant

Pre-test answers Answers at the end of TLS
ST1 When T increases electrons have a When T increases, electron find a greater resistance, due to an increased probability
greater energy and so the current in- to collide (during their motion in the lattice) with lattice ions that oscillate at a speed
creases. that depends on the temperature value. So, the current decreases as the temperature
ST2 When T increases electrons move Current decreases. Resistivity increases because the oscillations of ions with
with a higher velocity and conse- respect to their equilibrium position grow and the electrons are more and more
quently the current increases. obstructed in their motion. If x0 is the oscillation amplitude, the equipartition
principle gives

kx20 = 12 KB T → x20 ∝ T

As ρ = (m/ne2 )1/τ , where m, n and τ are the electron mass, the average

number of electrons per volume unit and the average time between two consecutive
collisions, respectively, we have 1/τ ∝ Avm , where A and vm are the electron-ion
cross-section and the electron thermal velocity, respectively.
√ 3
Since A = πx20 ∝ T and vm ∝ T , we have ρ ∝ T 2 .
ST3 The heating makes the charges When T increases the metal resistivity increases and from the experimental data we
move faster. But I am not sure if see that ρ(T ) ∝ T . Free electrons move with a velocity that does not depend on
this is related to the actual value of the temperature and collide with the ion structure, that vibrates with respect to their
the electric current . . . equilibrium position. This means that ρ = CAvm ∝ T , where A = πx20 ∝ T
(from the equipartition principle) and vm = vf for T not too high (from the discrete
spectrum hypothesis and from the Pauli exclusion principle).

Table 7: Comparison of answers to question 4 in the pre test and at the end of TLS.
aspects of the STs’ individual knowledge that supplied useful hints on the guidelines for the
subsequent TLS development.
More than half of our sample of STs showed an adequate knowledge of the phenomena of
electric conduction (Ohm’s and Joule’s laws), yet only 33% were able to supply an adequate
microscopic model able to describe and explain electric conduction phenomena. We verified
that some STs answered the fourth pre-test question on the basis of their recall of the phe-
nomenological law connecting resistivity and temperature, while other were able to think about
a model describing resistivity and use it to implement a line of reasoning operating on specific
physical variables. STs unable to supply appropriate microscopic models showed themselves to
be also unable to predict phenomenological results, as well as to give a qualitative interpretation
of them.
Our idea of preparing a TLS helping STs in constructing “mechanisms of functioning”, as
well as in connecting system characteristics at macro and micro levels, showed its pedagogical
validity during the various phases of our pilot test.
Our teaching and learning sequence, making use of a specific learning environment based on
experiments and modelling of experimental results, modified the initial conceptions of our ST
sample in a twofold direction:

a. to be aware of modelling as a powerful thinking process for the explanation and interpre-
tation of experimental observations of electric conduction in metals;

b. to appreciate qualitative reasoning as a starting point for more accurate comparisons be-
tween experimental data and model quantitative analysis, developed on the basis of clas-
sical or quantum physics.

Data reported here, together with the deep analysis of STs’ worksheets and final tests (reported
in Bonura, 2009) can allow us to give an answer to our third research question. In fact, about
85% of our STs showed themselves to possess, at the end of the TLS, a qualitative model similar
to the scientific one (i.e. a model evidencing the presence in metals of free charge carriers
interacting, by means of collisions, with a ion-lattice) and more than 60% were also able to
analyse quantitatively the various kinds of models, in order to correlate them with experimental

The funding of this research work was partially supported by the EU project “Docente Europeo:
move’in Science” - 134648-2007-IT-COMENIUS-CMP.

Bonura, A. (2009). Modelli della conduzione elettrica e connessioni tra livelli di descrizione: aspetti
metodologici e dinamiche di apprendimento. Ph.D. thesis, University of Palermo.

Bonura, A., Capizzo, M.C. & Fazio, C. (2007). Characteristic properties of semiconductors through
experiments and modelling. Selected papers of GIREP-EPEC International Conference “Frontiers of
Physics Education”, Opatija (Croatia), 26-31 August 2007, pag. 196-201
Bonura, A., Capizzo, M.C., Fazio, C. & Guastella, I. (2008). Electric conduction in solids: a pedagogical
approach supported by laboratory measurements and computer modelling environments. In Sidhart,
B.G., Honsell, F., Mansutti, O., Sreenivasan & De Angelis, A. (eds.). Frontiers of Fundamental and
Computational Physics, Proceedings of AIP 9th International Symposium, Udine and Trieste (Italy), 7-9
January 2008. pp. 227-230, ISBN 978-0-7354-0539-4. American Institute of Physics, Melville, New
York, 2008.
Borges, A.T. & Gilbert, J.K. (1999). Mental models of electricity. Int. J. Sci. Educ., Volume 21, Issue 1.
Duit, R., & Komorek, M. (1997). Understanding the basic ideas of chaos theory in a study of limited
predictability. Int. J. Sci. Educ., 19, 247-264.
Duit, R. & Komorek, M. (2004). The teaching experiment as a powerful method to develop and evaluate
teaching and learning sequences in the domain of non-linear systems. Int. J. Sci. Educ., 26(5), 319-633.
Eylon, B.S. & Ganiel, U. (1990). Macro-Micro Relationships: The Missing Link Between Electrostatics
and Electrodynamics in Students’ Reasoning, Int. J. Sci. Educ. 12, 79-94
Net-Logo, (2009).
Kattman, U., Duit, R., Gropengießer, H., & Komorek, M. (1995). A model of Educational Reconstruc-
tion. Paper presented at The NARST annual meeting, San Francisco, CA.
Kattman, U., Duit, R., Gropengießer, H., & Komorek, M. (1997). Das Modell der Didaktischen Rekon-
struktion - Ein theoretischer Rahmen für naturwissenschaftsdidaktische Forschung und Entwicklung
[The model of educational Reconstruction - A theoretical frame for science education research and de-
velopment]. Zeitschrift für Didaktik der Naturwissenschaften, 3 (3), 3-18.
Linn, M.C. & Songer, N.B. (1991). Teaching thermodynamics to middle school students: what are
appropriate cognitive demands? J. Res. Sci. Teach., 28, .885-918.
Marton, F. (1981). Phenomenography. Describing conceptions of the world around us. Instructional
Science, 10, 177-200
Marton, F. (1994). Phenomenography. In T. Husen & T. N. Postlethwaite (Eds.), The international
encyclopedia of education (2nd ed., Vol. 8, pp. 4424-4429). Oxford, U.K.: Pergamon.
Osborne, R. (1983). Towards modifying children’s idea about electricity. Res. Sci. Tec. Ed. 1(1), 73-82.
Shipstone, D.M. (1985). Electricity in simple DC circuits. In R. Driver, E. Guesne, A. Tiberghien (eds),
Children’s idea in science. Milton Keynes, England. Open University Press.
Stocklmayer, S.M. & Treagust, D.F. (1996). Images of electricity: how do novices and experts model
electric current? Int. J. Sci. Educ. 18 (2), 163-178
Strauss, A. & Corbin, J. (1998). Basics of Qualitative Research. Techniques and Procedures for devel-
oping Grounded Theory. II edition. Thousand Oaks, CA: Sage Publications, Inc.
Thacker, B.A. (2003). A study of the nature of students’ models of microscopic processes in the context
of modern physics experiments Am. J. Phys. 71 (6).

Thacker, B.A., Ganiel, U., & Boys, D. (1999) Macroscopic phenomena and microscopic processes:
Student understanding of transients in direct current electric circuits, Am. J. Phys. 67, Physics Education
Research Supplement, S25-S31.
Wittmann, M., Steinberg, R., & Redish, E. (2002). Investigating student understanding of quantum
physics: Spontaneous models of conductivity. Am. J. Phys. 70 (3).

Renaat Frans[1,2] & Jeroen Vanesser[1]
[1] Teacher Training College, Katholieke Hogeschool Limburg, Hasselt, Belgium
[2] Department of Physics, University of Antwerp, Antwerpen, Belgium

The connection between Music and Science is a very old one. Modern physicists
still refer to it as an important motivating factor in their work. Why isn’t it used
more in physics education? At least the need for introducing quantum physics in
secondary schools is for the authors an opportunity to do this by way of the musical
analogy. Indeed, like the discrete musical intervals on a string that can be heard, the
intervals between quantum fields can be observed as particles. This might open an
path for introducing quantum physics and reintroducing the importance of beauty
not only in physics itself but even more in physics education.


Music, Math’s and Physics are very related although we seem to have forgotten why. Music
and science tell about reality in a connected way. It was in ancient times that the hidden perfect
mathematical proportions behind the harmony of music, surprised the Pythagoreans. Plato
placed harmony and mathematical patterns in the center of all attempts to understand the world
(see Plato’s Timaeus, ca. 400 B. C.). Since then the search for beauty never left the adventure
of maths and science. Many great physicists expressed their fascination for beauty in their
scientific work.

“A theory with mathematical beauty is more likely to be correct than an ugly one
that fits some experimental data.” P.A.M. Dirac, cited in Scientific American, by
Hovis and Kragh (1963)

“For I have no higher aim than to work out the beauty of science. I put beauty before
science.” Erwin Schrödinger in a letter to Max Born, cited by Cropper, W.H. (2001)


Indeed the insights of modern quantum physics do remind us again of the deep bond between
music and science. An observed quantum particle is the result of an oscillating field, or at least
of changes between the discrete modes of oscillation of such fields. Like the discrete modes on

a string, the fields can oscillate only in discrete modes. Changing the oscillation mode means
a quantized energy transfer out of or into the field. These quantum exchanges are the observed
quanta of light and matter.
In music the series of tones, for example on a bound string, occur also as sudden quantized
changes in the oscillation mode. The discrete musical intervals, which are the differences be-
tween these oscillation modes, can be seen therefore as an audible analogy for the quantized
energy transfers between the quantum fields of light and matter. The heard intervals play the
role of the observed light or matter quanta. It was the purpose of our research to open, by this
musical analogy, the field of quantum physics to pupils of secondary schools.
We will lean mainly upon quantum theory as developed after 1920 by De Broglie, Schrödinger
and even Dirac. Strnad (1986), Jones (1991) and Klassen (2008) showed earlier how miscon-
ceptions of the photon as a particle can damage understanding. In the later Quantum Field
Theory however both radiation and matter are seen as quanta of respectively a radiation and a
matter field. So rather than introducing particles with associated wave functions, we will stress
the reality of fields right from the beginning (see Art Hobson , 2005 & 2007 )


The learning path consists of the following steps:

• The connection between musical harmony and perfect proportions

• The emergence of tone and the discrete series of tones

• Musical intervals as an audible analogy for the observed quanta transfers between the
fields of matter and light

In the rest of this article we will clarify what exactly we worked out in our learning path,
although the third part is still in development. Up till now, the learning path has been developed
in Dutch and we are testing it out in a few secondary schools. We are also incorporating it in
the physics course (and part of it also in the music course) in our pre-service teacher training



Pupils start by redoing the experiments of Pythagoras. On a monochord they strike two strings
of the same tension and length. First they tune them to the same pitch. The length of the two
strings are in the proportion 1:1 and they sound harmonic.
Then they start to shorten one string in comparison to the other. Let us say they shorten the
second string a bit in comparison to the first for example in a proportion 0.99:1. Now the
strings sound dissonant to each other. In search of a new harmony they try the proportion 2:1.
Now harmony is restored. They hear the musical interval called the octave. When they place

Figure 1: A monochord (source: R. Frans).

the wooden wedge slightly out of the middle on the shortest string, they hear dissonance again.
They look for the next harmony and will find them in the proportion 3:2. Now they hear the
beautiful musical interval of the fifth. Placing the wedge slightly out of the perfect proportion
again, the harmony of the sound will be destroyed. One can go on like this with proportions 4:3
(the musical interval of the fourth), 5:4 (the third), . . . etc.

Proportions of the length of the 2 strings Harmonic interval


Table 1: The harmonic proportions (source: R. Frans).



Where noise is an irregular thickening and thinning of air pressure in an elastic medium, sound
arises as a recognizable pattern that repeats itself in time. The number of repetitions in a unit of
time, is the frequency. The maximum displacement of a wave is called the amplitude.

By experiments like filling the inside of a small drum with smoke (don’t forget to make a hole
in the bottom) pupils can see the propagation of the sound waves as rapidly moving circles of
smoke (every time you hit the membrane). The basic relation between speed, wavelength and
frequency is explained (v = λf ). With shareware PC-measuring software like Visual Analyzer,
the wave pattern and the frequency can be measured.


As we have some idea now what sound waves are, we start to investigate the possible tones
you can generate on an instrument. We take for example a string and investigate the influence
of length, thickness and mass per unit length. All these do influence the pitch of the tone. By
altering these boundary conditions, you can produce virtually any pitch of tone you want. But
which tones can I play when I do not alter the tension, the length nor thickness of a string?
Which tones can I play if I do not open or close any holes in a flute?
By experimenting with sound tubes, flutes and monochords, one can discover:

• The existence of a discrete series of tones on an instrument (when you do not alter the
boundary conditions), 1st harmonic, 2nd harmonic, 3rd harmonic, etc.

• The frequencies of these harmonics are always integer multiples of the frequency of the
fundamental tone

f1 , 2f1 , 3f1 , 4f1 , 5f1 . . .

As you can see, we discovered a discrete row of tones: between every harmonic, a quantum of
frequency exists. It is not possible to generate a tone in between those tones (as long as you do
not alter the boundary conditions). This quantum of tone is fundamental to a given unaltered
instrument: its quantum equals the frequency of the fundamental.


Can we explain the existence of only this discrete series of natural tones? Why are the other
tones not present? We go intensively into this explanation since it will be very analogous to the
quantization of bounded quantum fields.
Let us take the example of the string. If one plucks a string in fact lots of waves with all kinds of
frequencies (and also wavelengths), start to propagate forwards and backwards from the point
of plucking. Because of Newton’s 3rd law, they reverse in direction but also in displacement as
they bounce back at the ends of the string. Since the superposition principle holds (as long as
the displacement is linear with the force), the net displacement on every point is the sum of all
these passing waves. Only waves from which half the wavelength fits into the total length of
the string an integer times, stay in existence. All other waves cancel each other out.

Figure 2: The discrete series of natural tones on an Alpine horn (source: Frans & Vanesser)




As Feynman stated already: the double-slit experiment is the heart of quantum mechanics and
it can hardly be replaced by anything else. Our try-outs confirm this view but we have learned
also that briefly showing a ‘real’ experiments as a start - from which pupils can ‘feel’ the need
for a new physics - can strengthen their attention. The characteristic colors in line spectra (of
elements) can be such a reality. We show it by experiments in flames, spectral lamps or even
astronomical spectra. How can matter radiate in such a discrete pattern of lines? At this point,
we are not going into details, but just announce that we will come back to atomic spectra later.
In order to understand what is going on in double-slit experiments, we start to investigate two
macroscopic double-slit experiments:

• one for particles: This can be done e.g. with spray paint. There is no interference pattern
and the intensity diminishes with greater angles from the slits. For particles, closing a slit
will always reduce the number of arrivals on a screen.
• one for waves: We choose to do a macroscopic double-slit experiment with sound. The
pupils can hear the interference pattern by walking from right to left in front of the slits.
(The slits can be replaced by sending the same tone through two separated speakers). For
certain small angles (where the first minima are), closing a slit will raise the intensity of
the sound! It proves that sound is waves.
The principle of superposition of waves, can provide a deeper explanation of the phe-
nomenon. As was the case for the standing waves on a string, the net displacement is the

Figure 3: The possible waves of the harmonic series of tones on a string (source: Wikipedia,
public domain)

sum of the separate displacements due to both the waves. Due to differences in path length
the waves arrive in different phase at different places. At some special points crests meet
crests (both waves in phase), at other places troughs meet crests (waves out of phase). At
most points the relative phase of the two waves is somewhere in between. There you hear
a moderate sound intensity.

Now we are ready to show the results of double-slit experiments for light and matter. We use
animations and film but also perform electron diffraction by a graphite crystal. We notice the
fringes but also the arrival of the ‘particles’ of light or matter one by one. We stress the fact that
only one ‘particle’ is in the apparatus and that no ‘intelligent’ collision between particles can
count for the pattern always observed.

Figure 4: Result of a double-slit experiment with electrons. During approximately half an hour
electrons arrive one by one, forming a clear interference pattern (source Tonomura, Hitachi

We can conclude that the double-slit experiment with light and electrons, both contradicts di-
rectly the particle model of matter but also the wave model for light (light arrives in lumps). We
should think of new physics to explain what we see.


Inspired by the parallel between the experiments of light and matter we formulate the following
new concepts:

1. Suppose fields are the fundamental concept in physics, not particles. Fields spread out
and can go through both slits. A field can interfere with itself and give rise to interference.
For light we think of the electromagnetic field. For electrons we suppose the existence of
a new matter field.

2. Fields (like magnetic, electric and gravitational ones) are real and can carry energy. Since
Einstein’s special relativity matter is energy: E = mc2 . It might be plausible that the
energy of matter is also carried by a field rather than by particles themselves.

3. Like oscillating strings in musical instruments, the fields have discrete modes of oscilla-
tions: each with a different but related frequency. Since the energy is proportional to the
frequency (E = hf ), the energy carried is quantized as the frequency is. The electromag-
netic field can only vary in the ‘harmonic series’ of discrete energies

1hf, 2hf, 3hf, 4hf, . . .

The field of matter can carry only discrete series of energies

1mc2 , 2mc2 , 3mc2 , 4mc2 , . . .

4. The fields can exchange discrete quanta of energy: That is what we observe as particles!
The intervals of the radiation field are observed as photons, those of the matter field as
matter particles like electrons.

5. The amplitude of the waves is a measure of the probability of quantization.


Although it seems a very strange theory, this quantum physics can be applied to the electrons in
the atom. Like a string bounded at the ends, the matter field is bound in the electrostatic field of
the nucleus. Because of these boundary conditions the matter field can only oscillate in discrete
modes. The energy decrements of this field corresponds exactly with the energy increments of
the electromagnetic field. The electromagnetic field can give off quanta on his turn, with the
same energy amount as absorbed earlier from the matter field: we see lines of precisely definite
colors in the line spectra of elements. Motivated pupils can calculate precisely the frequency of
the lines of hydrogen and explain the observed colors.

The harmonic series that exist on an unaltered tone generator like a string or a tube, is an
example how quantization arises in classical mechanics. Just as we recognize the transfers
from one oscillation mode to another as musical intervals used in melody, so do we notice the
intervals of the quantum fields as observed particles. Although it is too early to make definite
conclusions, the analogy between discrete musical intervals and quantized fields might help the
introduction of pupils to our most advanced but most contra-intuitive physical theory ever.

Chandrasekhar, S. (1987) Truth and Beauty. Aesthetics and Motivations in Science. University of
Chicago Press, Chicago.
Cropper, W.H. (2001) Great Physicists, Oxford University Press, Oxford, UK, pp. 281-82
De Broglie, L. (1924) Recherches sur la théorie des quanta, reprint in 1925 in the Annales de physique
3, Ed. Masson, Paris

Figure 5: The quantized field intervals in the hydrogen atom, observed as photons.

Dirac, P.A.M. (1983) The origin of Quantum Field Theory, in The Birth of Particle Physics edited by
Brown, L.M. and Hoddeson L. , Cambridge University Press
Einstein, A. (1905) Über einen die Erzeugung und Versandlung des Lichtes betreffenden heuristischen
Gesichtspunkt. Annalen der Physik, 17, 132-148
Hobson, A. (July 2005) Electrons as field quanta: a better way to teach quantum physics in introductory
general physics courses, Am. J. Phys. 73, pp. 630-634
Hobson, A. (February 2007) Teaching Quantum Physics without Paradoxes, The Physics Teacher, 45,
pp. 96-99
Hovis, R.C. & Kragh, H. (1963) The Evolution of the Physicist’s Picture of Nature, Scientific American,
May, 208, pp. 45-53
Johnston, I. (2002) Measured Tones, the interplay between physics and music, Taylor & Francis, New-
Jones, D.G.C. (1991) Teaching modern physics - misconceptions of the photon that can damage under-
standing, Phys. Educ. 28, p. 93-98
Klassen, S. (2008) The photoelectric effect: rehabilitating the story for the Physics Classroom, Proc.
Scnd. Int. Conf. on Story in Science Teaching, München
Strnad, J. (1986) Photons in introductory quantum physics. American Journal of Physics, 54(7), 650-65

Tonomura, A. (1998) The Quantum World unveiled by Electron Waves, World Scientific Publishing,

Petter Hubber
Deakin University, 221 Burwood Highway, 3125, Burwood, Victoria, Australia.

The research described in this paper argues that difficulties of learning scientific
concepts such as those associated with processes involving the Sun, Moon and
Earth, such as day and night, the seasons and phases of the moon, are fundamen-
tally representational in nature. There is a need for learners to use their own rep-
resentational, cultural and cognitive resources to engage with the subject-specific
representational practices of science. From this perspective students need to un-
derstand and conceptually integrate different representational modalities or forms
in learning science and reasoning in science. The researchers worked with two ex-
perienced teachers in planning a teaching sequence in astronomy using a teaching
approach that highlight representational issues and options in helping students ex-
plore and develop key conceptual understandings. Classroom sequences involving
the two teachers were videotaped using a combined focus on the teacher and groups
of students. Video analysis software was used to capture the variety of represen-
tations used, and sequences of representational negotiation. From a pedagogical
perspective the representational approach placed a significant agency in the hands
of students which resulted in structured discussions around conceptual problems.
Representations were used as tools for reasoning and communication to drive class-
room discussions and develop higher levels of understanding in the students. The
pre- and post-testing showed significant gains in students’ thinking from naïve to
more scientific understandings of astronomy.

The research literature that has focused on individual’s understanding of astronomical phenom-
ena is quite extensive (refer to bibliography by Duit, 2002) and has dealt with such astronomical
behaviour as the day and night cycle (for example, Dunlop, 2000; Kikas, 1998); the seasons (for
example, Danaia & McKinnon, 2008; Hsu, 2008); phases of the moon (for example, Danaia &
McKinnon, 2008; Trundle, Atwood & Christopher, 2002) and gravity (for example, Agan &
Schneider, 2004; Palmer, 2001). The main finding from this literature is that students at all
levels of schooling and as adults, including pre-service and in-service teachers, hold alternative
conceptions about astronomical behaviour (Kalkan & Kiroglu, 2007; Parker & Heywood 1998;
Trumper, 2001). Examples of such alternative conceptions include: day and night is caused
by the motion of the Sun around the Earth; the phases of the moon are caused by the shadow

cast on the Moon due to the Earth obstructing the light from the Sun; the seasons are caused by
variations in distance between the observer on Earth and the Sun; and, gravity does not operate
in the absence of air.
The prevalence of alternative conceptions across most age levels of individuals may suggest that
school science has limited impact in resolving them. From this perspective Danaia & McKinnon
(2008) suggest that teachers need to have some understanding of these misconceptions, includ-
ing ways of dealing with the alternative conceptions their students bring to the classroom. Bakas
& Mikropoulas (2003) point out that the sometimes limited success of conventional teaching
methods in overcoming students’ alternative conceptions may be due to a lack of appropri-
ate teaching aids in the form of representations that can intervene dynamically in the learning
process and modify it.
There is general agreement that even the most elementary astronomical phenomena are difficult
for students to understand since many of the ideas involve three dimensional spatial relation-
ships and orientations between celestial objects (Hegarty & Waller 2004; Padalker & Ramadas,
2008; Yu, 2004). Within the classroom students are expected to deal with visual-spatial repre-
sentations through the application of reasoning skills that involve the ability to imagine spatial
forms and movements, including translation and rotations, and perspective taking (Hegarty &
Waller 2004; Padalker & Ramadas, 2008). The difficulty of learning three dimensional per-
spectives of astronomical phenomena is compounded when one considers that most phenomena
are time-dependent and that much of the teaching materials in astronomy education are two-
dimensional in nature (Yu, 2008).
The difficulties encountered by individuals in learning astronomical phenomena point to the
need for a very strong emphasis of the role of representations in learning. There is a need for
learners to use their own representational, cultural and cognitive resources to engage with the
subject-specific representational practices of science. From these perspectives students need to
understand and conceptually integrate different representational modalities or forms in learning
science and reasoning in science (Ainsworth, 1999; Lemke, 2004). These researchers argue
that to learn science effectively students must understand different representations of scientific
concepts and processes, and be able to translate these into one another, as well as understand
their co-ordinated use in representing scientific knowledge and explanation-building. Classifi-
cation categories of representations are generally held to include textual, visual, mathematical,
figurative and gestural, or kinaesthetic understandings.
In encapsulating the key features of adopting a teaching sequence with a representational focus
and the roles played by representations in supporting reasoning and learning we are collaborat-
ing with other colleagues1 in developing a set of pedagogical principles. These principles draw
on literature that emphasise the active role of representational work in supporting learning in
science (Greeno & Hall, 1997; Ford & Forman, 2006) and formed the basis of advice to the
teachers in this study concerning the nature of a representational focused approach to teaching
astronomy. The principles are described as:

1. Teachers need to clearly identify big ideas, key concepts and their representations, at the
This study is part of a wider research project titled, The role of representation in learning science, conducted
at three university sites.

planning stage of a topic.

2. There needs to be an explicit teacher focus on representational function and form, with
timely clarification of parts and their purposes.

3. Representational generation and negotiation as the focus of teaching and learning:

a. Students need to be active and exploratory in generating, manipulating and refining

b. Activity sequences need to have a strong experiential context and allow constant
two-way mapping between objects and representations.
c. Students need to be supported to develop explanations that involve coordinating and
re-representing multiple modes.
d. There needs to be a sequence of representational challenges which elicit student
ideas, guide them to explore and explain representations, to extend to a range of
situations, and allow opportunities to generate representations and integrate these
e. Students need to understand that a single representation cannot cover all purposes,
but needs to have a selective focus.
f. There needs to be interplay between teacher-introduced and student-constructed rep-
resentations where students are challenged and supported to refine and extend and
coordinate their understandings.
g. There needs to be ongoing assessment (by teachers and students) of student repre-
sentations. The adequacy of a representation depends on the particular purpose or

4. Activity sequences need to focus on engaging students in learning that is personally mean-
ingful and challenging, through attending to students’ interests, values and aesthetic pref-
erences, and personal histories.

5. Formative and summative assessment needs to allow opportunities for students to gener-
ate and interpret representations.

In this project we worked closely and collaboratively with teachers to construct a teaching se-
quence in astronomy that validated the pedagogical principles. The topic’s content was outside
the teachers’ discipline expertise of biology so the challenge for them was a combination of
content knowledge and pedagogic content knowledge. This paper explores how the teachers
and students responded to the new approach, implied by the principles above, and the role of
representations in supporting learning and reasoning in astronomy.

This study relates to a classroom study of an astronomy teaching sequence taught by two teach-
ers, Lyn and Sally2 , to Year 8 students (13 years of age). The sequence content covered astro-
nomical phenomena as it pertains to the Earth, Sun and Moon systems surrounded by planets
and stars. These phenomena included: day and night cycle, the seasons, phases of the moon
and tides, constellations, and gravity.
The research question is: What are the effects of an explicit representational focus on the teach-
ing and learning of Year 8 astronomy?


In coming into this study Lyn and Sally were experienced practitioners who already had experi-
ence in a previous study (Hubber, Tytler & Haslam, in press) in the innovative use of strategies
based on the development of students’ representations. However, prior to the commencement
of the teaching sequence both teachers expressed some reluctance to teaching it. This stemmed
from a perceived lack of content knowledge about some aspects of the topic as well as a per-
ceived lack of sufficient pedagogical content knowledge (Shulman, 1986) to effectively respond
to the learning needs of the students. In the following transcript of a conversation had with Lyn
and Sally the reasons behind their reluctance to teach the topic were expressed.

Researcher: last year you said that you really didn’t want to teach this topic.
Sally: yes
R: So what was the main reason behind that? Was it a lack of content knowledge?
S: [yes] The content knowledge, the fact that I had never done it in school myself
and I have never learnt the topic myself, only read it through books and watching
movies and so on but it was the fact that it was the topic that was endless. I mean
the kids could ask you lots and lots of questions and I was aware that half the time
I may not have the answers straight away and would need to come back to them
later. I mean that was the fear factor.

Whilst Lyn had taught the topic of astronomy before she commented that “. . . it was the one I
was the least confident with and I have always avoided it because it has been difficult”.

Researcher: What sort of approach did you have in that topic in the past? Did the
topic have a lot of textbook orientated work?
Lyn: Yes, more delivery of facts rather than exploration of understanding.

In planning the sequence with the teachers the adoption of a representational focus puts stringent
demands on clarifying what knowledge was to be pursued. The teaching sequence therefore
needed to be informed by a clear conceptual focus. At the planning stage, in order to refine the
Pseudonyms have been given to teachers in this study. Where reference has been made to names of students
pseudonyms have also been used.

representational work, the research team collaborated with the teachers in identifying big ideas
or key concepts of the topic in addition to the students’ alternative conceptions reported in the
literature. The initial lessons in each teaching sequence focused on exploration of students’ prior
views, generation of students’ representations, and introduction of the scientific conventions that
underpin the topic of astronomy.
The sequence sought to develop a model of classroom practice that foregrounds students’ ne-
gotiation of conceptual representations. The teaching and learning activities and sequences that
were developed modelled pedagogies that take a representational focus to support student en-
gagement and learning, but more importantly were a vehicle to explore the representational
challenges and issues, the link between conceptual learning and representation, and the nature
of effective teaching and learning focusing on representations, and effective assessment from
this perspective.
The team held workshops with activities that had a representational focus and discussed the
implications of these for student learning bearing in mind possible alternative conceptions the
students might have. This experience gave the teachers more confidence in tackling the teaching
of astronomy. For example, when Sally was asked about her reported increased confidence
following the workshop she commented:

Sally: I think the role plays, how to actually go about teaching the topic and using
ourselves, using the models, and understanding of relative distances, and I think
the Power Points that showed the distances, and the size of the different stars and
so on. I think that really was a wow factor and I put myself in the shoes of the kids
and I thought, yes that would be something I could feel comfortable in teaching.

Rather than dealing with an “endless” topic, learning astronomy at a Year 8 level was now seen
as a set of key ideas about astronomical phenomena that arise from simple dynamic systems
such as the Earth, Moon and Sun connected by gravity. The representations of the key ideas
associated with these systems gave the teachers the pedagogical tools they felt comfortable
with in using in the classroom. The representations also gave the teachers the reasoning tools
to enhance their own content knowledge. For example, Lyn commented:

Lyn: . . . I found the workshops that we had were really powerful . . . I had a problem
with the moon didn’t I. The rotation of the moon and the revolution of the moon
around the earth and I couldn’t get it until I physically did it myself.


The teaching sequence lasted 14 lessons varying between 45 minutes and 90 minutes of class
time. The features of the sequence were:

1. Pre-test of key ideas associated with the astronomy topics to be taught. The test incor-
porated a slightly modified3 set of multiple choice questions used by Trumper (2001)
and Kalkan and Kiroglu (2007). The pre-test also included short answer questions where
students were to provide full explanations using text or drawings.
Modified in language to support a perspective of observers in Australia.

2. The students’ notebooks were different to their normal A4 lined notebooks but larger
sized project type notebooks which, when opened out, had a lined page on the left and an
unlined page on the right. This encouraged the construction of multiple representations
by the students.

3. Students were given representational challenges where they were to generate a represen-
tation with a particular purpose. For example,

a. Explain to a 10 year old how it can be night in Melbourne the same time that it is
daytime in LA.
b. Explain what would be observed by an observer on the Moon in respect of the Earth
and Sun over time.

4. Students were given representational challenges where they were to interpret a particular
representation by generating a single or multiple representations in other forms and/or
modes. For example,

a. Explain through diagram/drawing the mutual revolution of two physical objects.

b. Construct an observer’s view on Earth for summer and winter of the height of the
midday Sun given a diagram showing these two seasons from the perspective of an
observer in space.

5. Instances where representations where used in exploring astronomical behaviour. For


a. Animations showing the motion of celestial bodies to explain such phenomena as

the seasons, day and night, phases of the Moon and gravitational pull of objects near
b. Role play of the dynamic system that involves the Zodiacal Constellations, Sun,
Moon and Earth to explain the location of zodiacal constellations in the sky at certain
times of the year, time periods of day, month and year, and eclipses.
c. Scale models showing relative distances and sizes of the planets and the Sun.

In most instances representations did not stand alone but were integrated with others. For
example, gesture, everyday or scientific language and diagrams.

6. Classroom discussion generated by students’ questions. For example, “how come we can
see the Southern Cross every night of the year?”

7. Inquiry-based investigations whereby student groups were to collect some data, repre-
sent it in a form of their choosing, analyse and interpret this data with the purpose of
answering some questions. The students were to provide a written report as a summative
assessment task and present their findings to the rest of the class in whatever form they
chose. Examples of investigations involved:

a. Finding directions using the Sun: investigating the length and direction of shadows
over the period of a day.
b. Tide Watch: students investigate tide heights and phases of the moon over time by
accessing data from internet sources.

8. A Predict Observe Explain (POE, Gunstone & White, 1992) activity whereby students
constructed plasticine scaled models of the Earth and Moon relative to their size and
distance separation.

9. Post-test that included the same set of multiple choice questions given in the pre-test
and also included some of the short answer questions. In addition, further short answer
questions were given.


Data collected included: (1) video recordings of most classroom sessions and of student in-
terviews; (2) student workbooks; (4) pre- and post-tests; (5) transcripts of tape recordings of
teacher and student interviews and (6) researchers’ field notes.
The video sequences used two cameras, one tracking the teacher and the main classroom inter-
actions, and one focusing on a small group of students. The teacher and student group were ra-
dio miked. The student group’s microphone was transported with the group if it moved around.
Most lessons in the unit sequence were videotaped as many of the lessons had some part or
parts that had a representational focus. The videotaped lessons were coded using ‘studiocode’
software which has been designed for this type of analysis, to allow quick reference to rep-
resentational events and monitoring of classroom negotiation of representations. The analysis
reported here included triangulation between video data, transcripts of student and teacher in-
terviews, student work, pre- and post-tests and researcher field notes.

The sequence resulted in a number of insights into the efficacy of a representational focus
in planning and implementing a lesson sequence in astronomy. During the lesson sequence
representations were used by teachers and students in different ways. The findings described in
this section are categorised as:

• The partial nature of representations means that the process of achieving full understand-
ing is multi-representational.

• Representations can be used to challenge students thinking and facilitate reasoning.

• Formative and summative assessment are facilitated through representation.

• Student learning in a lesson sequence focused on representations.


The partial nature of representations was exemplified on many occasions during the teaching se-
quence. The students came to understand that a single representation cannot cover all purposes,
but needs to have a selective focus. On most occasions when a teacher generated or student gen-
erated representation was introduced to the class there was a discussion as to the extent of the
fit with the target phenomenon or process. For example, in the first lesson the teachers initiated
a discussion of the globe as a representation of Earth. They asked the students what features of
Earth are represented by the globe and what features are not represented by the globe. In Sally’s
class the students quickly generated the following list of features (see Table 1).

The Globe shows The Globe does not show

Earth is round Day & night
Earth has oceans Gravity
Earth rotates about axis Size
Earth is tilted Mountains
Weight of Earth
Table 1: Students’ responses to those features of the Earth that are shown or not shown by the

The response by the students that the mountains were not represented by the globe opened
up further discussion. Sally made explicit links between different modes of representation in
generating the view that because of the scaled size of the globe the mountains would not be able
to be shown on a globe. She did this by getting the students to explore the globe through sight
and touch. This raised an issue of conflicting findings: by sight it did not appear that mountains
were represented.
However, by touch the students could feel slight bumps on the globe in the region of the Hi-
malayas. This issue of whether the height of the bumps was accurately represented was then
explored by the class.
Sally introduced the mathematical idea of diameter gesturing its meaning on the globe and
illustrating this on the board (figure 1). She then explained the scaling process using hand
gestures linking the numerical values for the Earth and globe diameters and the height to the
highest mountain, Mount Everest, written on the board to the actual globe.
The scaled globe height for Mount Everest was given by Sally as 0.01 cm, which the students
converted to 0.1 mm. Finally, the students were asked to get out their rulers to then look at
them to see what distance 0.1 mm might look like. The question “Can the globe represent
mountains?”, was then emphatically answered by the students as no.
During class interactions the role of drawing out the specific purpose of a representation was
undertaken by the teachers as well as the students. For example, when discussing the planetary
status of Pluto the teachers presented the students with a diagrammatic representation of the
planetary paths (figure 2) as well as images of the planets indicating their relative sizes. The

Figure 1: Copy of Sally’s board work

Figure 2: Planetary paths around the Sun

students were expected to interrogate the representations in terms of their adequacy in providing
evidence to differentiate Pluto from the planets.
The students were then able to argue successfully the non-planetary status of Pluto when pre-
sented with the criteria for assessing the status of a planet on the basis of evidence from different


During the sequence both teachers regularly gave students representational challenges which
elicited student ideas, guided them to explore and explain, to extend to a range of situations,
and allowed opportunities to integrate their representations meaningfully. The students were
given opportunities to re-represent to extend and demonstrate learning. They were challenged
and supported to coordinate representations as a means to express coherent, defensible and
flexible understandings. For example, during the first lesson in the teaching sequence both
teachers gave the students a representational challenge to show, through the physical action of
their bodies, the motions of rotation and revolution. This task elicited students’ understanding
of these motions. From this initial challenge the teachers each set a further challenge for the
In Sally’s class the students were further challenged to show if it was possible to revolve around
each other. She then asked two students:

Sally: How would you show what you did on the board? I want you to think about
different ways of showing a representation of a concept or a phenomenon.

The students initially found this representational challenge difficult. In resolving this difficulty
they partially acted out the motion and came up with the drawing on the left in Figure 3 after
realising the need to have a central point of revolution. Sally then asked the boys to re-represent
the diagram showing just the paths of the feet; this is shown by the drawing on the right in
Figure 3. A realisation then came from the pair that the feet trace out intersecting circles.

Figure 3: Representations of two objects revolving about each other.

This activity led to a discussion about binary star systems and their prevalence in the universe.
In a post-topic interview Sally commented on one of the student’s effort with this challenge:

Sally: But I loved the fact that Henry4 was trying to press the pen hard on the board
and going, “got to get that to move this way how do I do it. Coz it is 2-dimensional.
It is not 3-dimensional.”

Throughout this activity the students created three different representations, reasoning with one
representation in constructing another in a process that was challenging for them.
In Lyn’s class the students were challenged to show if it was possible to pair up and to revolve
and rotate simultaneously. She found two pairs of students who represented the task in two
different ways. Each pair then demonstrated their motions to the rest of the class who evaluated
each for the purpose underlying the role play: to represent rotation and revolution simultane-
For one pair, one student kept facing his partner whilst making one full revolution of him.
Lyn asked the students, “I noticed two different styles . . . Is John doing it right [rotating and
revolving]?” Most students were unsure which led Lyn setting the task for all students, in pairs,
to role play this action whilst at the same time noticing which walls of the room were being
looked at. By undertaking this activity the students found that for each revolution each wall was
seen once and thus realised that they were rotating at the same rate as they were revolving.
Lyn then proposed some questions about what this representation might mean when mapped to
celestial objects. This is illustrated in the following verbal exchange:

Lyn: if Shane was the Earth [central object] and John another object, what would
be seen from Earth?
Student 1: you would only see one side.
Lyn: if Shane was the Sun and John the Earth, what would this mean?
Student 2: if the sun shines out of Shane this would mean only one country would
see the sun.

Lyn then linked this motion to that of the Moon’s motion about the Earth which resonated with
the students some of whom knew about the observation that one can only see one side of the
Moon from Earth. Evidence of learning through this role play is illustrated in the following
post-topic interview transcript:

Researcher: did any of the role plays help you?

Student: I found the orbiting and noticing which wall you were looking at. If the
moon was rotating, that helped, because up to that point I didn’t think the moon
was rotating. If you were looking at different walls then you knew it was rotating.

Rather than just being told that the moon rotates was not enough. The student recognised this
through the actual action of a role play. The kinaesthetic experience of the role play gave him
the reasoning tool to consolidate this idea.
Pseudonyms are given for students who participated in the study.

These particular classroom scenarios illustrate two key roles for representations. There was an
explicit teacher focus on representational function and form of the role play. Students then used
the representation of role-play as a reasoning tool to explore, construct and consolidate ideas.
The scenarios also highlight the manner in which the teaching approach can lead the class into
different directions. From this perspective the teachers saw that they needed to be flexible and
to plan accordingly. According to Sally “. . . you plan your lesson with a lot of possibilities. You
think about okay what if the students ask me this question, what kind of activities can I have.”
Another example of where the representation was used as a reasoning tool can be seen in the
following verbal exchange between the researcher and a student in a post-topic interview.

Researcher: this is a globe. [refer to figure 4] What would it be representing?

Student: the globe represents the Earth
R: we have a little man there, he is an observer . . . If you wanted to look at a
representation of day and night what other things do we need?
S: we would want the Sun represented by some sort of flash light.
R: we haven’t got a flashlight but we have got the tennis ball there . . . what time of
the day is represented here for the man?
S: well it would be somewhere between 5 and 6 pm, I would say.
R: how did you work that out?
S: wait [Student looks at the globe and starts to rotate it with his hand, 10 seconds
R: tell me what you are thinking?
S: I want to change my mind, maybe it’s a.m.
R: what made you change your mind there?
S: because I remembered in which direction the Earth spins. It generally likes to
spin this way [demonstrating this with the globe].
R: how did you know that?
S: because what we learnt in class. We get the sunlight before Perth. If we are here
[pointing to the man in Melbourne] Perth is that way so [point out Perth] so we
turn this way [turning the globe].
R: so having the model here the representation helped you with that?
S: yeah
R: so you have worked out the day. Could you work out the season?
S: yes I could . . . let’s see [Student looks at the globe and then the tennis ball and
gestures the orbit of the globe] it’s [5 seconds] . . . I’m going to say autumn.
R: where would I represent the globe in three months time?
S: um . . . I want to say over there. [globe is placed at that location] I know that’s

In this interview the student connected a known fact about Perth that he learned in class with
the physical manipulation of the globe in reasoning about an astronomical behaviour from the
perspective of an observer on Earth (time and season) and one in space (rotation and revolution).

Figure 4: Representation of day and night.



The teachers saw benefit in the knowledge gained from the pre-test in terms of targeting the
teaching in resolving misconceptions that arose and for the students to be made aware of their
own thinking as an important part of the teaching sequence.

Lyn: Because we have more understanding of the misconceptions we can teach

accordingly and we can single out misconceptions . . . we can tackle them straight
away . . . if you are aware of what the misconceptions could be, you are explicitly
telling the students that you know some people think this is so, it has a huge impact
because the kids will not then go along those lines . . . The pre-test was used as a
basis to begin discussions, it gave kids a good reference point.

There were many instances during the teaching sequence whereby the students were given the
opportunity to interpret and generate representations which gave the teachers a good sense of
student learning from a formative and summative perspective.

Sally: It’s good to give them a representation, but it’s more powerful when they
re-represent it . . . it helps in their reasoning.
Lyn: . . . what you’re seeing with representation is that you’re seeing what’s in their
brain, not what they’re regurgitating.

The teachers found that the representational focus placed more involvement by the students in
the classroom interactions. This view is illustrated by the following comment by Sally.

Sally: I found most of the lessons were more student-driven . . . it was built up be-
cause of their questions. I think the kids appreciate the fact that they could openly
discuss why they thought that was the answer and that discussion was really pow-
erful . . . they valued the fact that they wanted me to know what they were thinking.

For the astronomy topic there was a change in the type of notebook used by the students. In
changing from an A4 lined page book to a larger than A4 project type book Sally commented
that such a change was, “much better than what we used to do because the kids liked the fact that
there were these blank pages where they knew, ah okay, I can draw this here and write what is
on the other side.” It was in their project books where students often re-represented a particular
situation. For example, in the diagram below (figure 5) the students were to re-represent the
diagram on the left for a midday observer on Earth. This student’s re-representation, shown on
the right hand side of figure 5, made explicit links between the original representation and her
re-representation through numerical labels. The direction of sunlight is indicated through lines
and use of shadows.

Figure 5: Student’s representations of midday Sun in winter and summer.

In terms of summative assessment the provision of a space rather than lines for the students to
respond to short answer questions in the post-test gave the opportunity, permission and authority
to adopt a range of representational modes. This is illustrated by the sample of responses (see
figure 6) made by students to a post-test question which asked: “An astronomer investigating
the motion of Europa, which is a moon, or natural satellite, of the planet Jupiter, found that it
revolved as well as rotated. Use the space below to clearly explain what each of these motions
The representations in figure 6 are scientifically correct and yet show a significant variation
within and across different modes.
During the sequence there were many instances where students were expected to interpret a
particular representation and generate others in explanation. The following example is taken
from a question in the post-test (figure 7) with an example of one student’s responses.

To rotate is to spin. Rotation is done on the spot. To revolve is to
orbit or go around something. To revolve you need two objects: one
to be revolved around and the other to revolve around the first object.
So Europa must spin or rotate at the same time as it orbits or revolves
around Jupiter.†

Figure 6: Four students’ responses to a post-test question about rotation and revolution. † The
students’ written response was re-written to provide clarity.


The pre-test that incorporated multiple choice and short-answer type questions elicited several
alternative conceptions similar to those found by other researchers cited in the introduction.
In evaluating student learning over the period of the teaching sequence the pre- and post-tests
contained the same set of multiple choice questions which had previously been used in other

Q20: The image† opposite is a time-lapse
photograph showing the position of the
Sun on Antarctica’s horizon every hour
for 8 hours.

Use the space below to explain why

the Sun doesn’t set below the horizon.

Figure 7: Student response to a post-test question. Taken from http://www.classroom.

studies (Trumper, 2001; Kalkan and Kiroglu, 2007). Table 2 indicates the students’ results
to these questions and provides a comparison to results obtained by the Kalkan and Kioglu
(2007) study. These researchers pre- and post-tested 100 pre-service primary and secondary
education teachers who participated in a semester length course in astronomy. A measure of
comparison of pre- and post-test results is the normalized gain index, <g>, the ratio of the actual
average student gain to the maximum possible average gain: <g> = (post% - pre%) / (100 - pre
%), reported by Zeilik, Schau, & Mattern (1999). Gain index values can range from 0 (no gain
achieved) to 1 (all possible gain achieved). The mean gain reported by Kalkan & Kiroglu (2007,
p. 17) to be a “respectable 0.3”. In contrast the mean gain for this study was significantly higher
as 0.65.
Apart from the conceptual growth shown in the multiple choice questions there was also evi-
dence of growth shown in the students’ responses to the short answer questions. For example,
when asked, “If objects like apples fall to the ground then why do you think the moon doesn’t
also fall to the ground? Explain why.” One student responded:

Pre-test response: The moon is out of reach for the earth’s gravity to pull it to earth.
Post-test response: the moon is constantly falling but it is falling at a sertain (sic)

Year 8 Students Kalkan & Kiroglu
N=40 (2007) study
Item Pre-test Post-test Gain Pre-test Post-test Gain
% correct <g> % correct <g>
1 Day-night cycle 61 92 0.78 91 93 0.22
2 Moon phases 43 81 0.66 23 30 0.09
3 Sun Earth distance scale 9 49 0.44 18 22 0.05
4 Altitude of midday Sun 10 66 0.62 29 39 0.14
5 Earth dimensions 30 63 0.48 5 14 0.09
6 Seasons 13 63 0.57 54 82 0.61
7 Relative distances 70 85 0.51 46 71 0.46
8 Moon’s revolution 38 83 0.72 49 60 0.22
9 Sun’s revolution 86 97 0.79 61 77 0.41
10 Solar eclipse 31 86 0.79 26 42 0.22
11 Moon’s rotation 21 61 0.5 13 28 0.17
12 Centre of universe location 78 95 0.75 65 88 0.66
13 Seasons 73 97 0.89 67 88 0.64
mean <g> 0.65 mean <g> 0.31
Table 2: Correct answer ratio and gain index (<g>) according to pre- and post-test results for
two studies.

speed so it is orbiting the earth due to our gravity.

Post-test responses also showed evidence of students providing more representational modes
(refer to figures 6 and 7 for examples of responses) than was shown in the pre-test.
In the following post-topic student interview transcript student 1 indicated that that models, the
kinaesthetic action of role play and constructing and re-constructing the representations assisted
her learning of ideas. For student 2 she made reference to diagrams and class discussion. Her
last comment suggested that students were often asked to explain their thinking to the teacher
and other students.

Student 1: [before the lesson sequence] I never knew how the moon kind of orbited
the Earth.
Researcher: what sort of things gave you a good understanding of moon?
S1: like how my teacher made us actually stand up and act out the motions. It’s
better than sitting down and trying to learn it by words on the board.
S1: I learnt a lot when my teacher showed models and making us do the represen-
R: what sort of things really helped you learn?
Student 2: I like looking at diagrams I find they really help. I like it when we have
class discussions . . . When the teacher says something you may not get it but when
someone else asks a question and another kid answers it in a different way to how

the teacher said it that helps. Then you get two different answers to a certain ques-
R: do you think it’s helpful if you have to explain something back to the teacher or
other students?
S2: yeah, you often have to do that.

In this study the teachers confirmed the efficacy of a representational focus in teaching and
learning the key science concepts in astronomy. The conceptual focus in the planning of this
topic gave the teachers a framework of concepts, alternative conceptions and simple dynamic
systems with which to develop activities that had a representational focus; content was con-
ceived of as an interconnected set of ideas linked and understood by representations. This focus
reduced the teaching of the topic of astronomy from what Sally characterised as an “endless”
one to a more manageable one. It also increased the teachers’ pedagogical content knowledge,
in addition to their content knowledge, as they gained a greater understanding of what makes
the learning of specific topics in astronomy easy or difficult.
The approach they adopted reflected the pedagogical principles listed in the introduction. Rep-
resentations were generated by the teachers and students and played an active and central role
in relation to thinking and learning. They were used as tools for thinking rather than as sum-
marised pieces of knowledge to be learned. Such a view is consistent with that held by Greeno
and Hall (1997, p. 366) who suggest that representations are “essential tools for communicating
and reasoning about concepts” and not “ends in themselves”. By being challenged to coordinate
and negotiate representations students came to refine and sharpen their understandings.
This approach was acknowledged to place much greater agency in the hands of students, and
this brought a need to learn to run longer and more structured discussions around conceptual
problems. The greater involvement of students in classroom interactions and the interconnect-
edness of ideas through representations mean the need for a greater flexibility in terms of what
content is to be covered in teaching each lesson.
The teachers saw the knowledge gained from the pre-test results as beneficial in terms of target-
ing the teaching in resolving misconceptions and for the students to be made aware of their own
thinking as an important part of the teaching sequence. The resolution to resolving the students’
naïve conceptions was perceived as very much a representational issue in terms of the use of
representational challenges to drive classroom discussions. The students were given many op-
portunities to interpret and generate representations which gave the teachers a good sense of the
students’ learning from a formative and summative perspective. The provision of spaces in the
students’ workbooks and tests encouraged the generation of representations.
The study described in this paper supports the idea that learning is fundamentally a representa-
tional issue. The focus on representations opened up the pedagogy in classrooms where students
readily responded to the invitation to refine and make coherent their representations. The pre-
and post-testing of the students showed significant gains in understanding across the astronom-
ical areas covered in the teaching sequence. The set of pedagogical principles given in the
introduction provided a useful framework for the teachers to use and in doing so enhanced their

confidence in teaching the topic. This greater confidence was generated by the high levels of
student engagement in using representations to construct high levels of understanding the key
ideas of astronomy that formed the basis of the topic.
Given current concerns about the engagement of students in meaningful science learning, and
the relatively limited success of pedagogical approaches based on cognitive views of learning,
we would argue that the approach adopted in this study is worthy of adopting in other science

Agan, L., & Sneider, C. (2004) Learning about The Earth’s Shape and Gravity: A Guide for Teachers
and Curriculum Developers, Astronomy Education Review, 2(2), 90.
Ainsworth, S. (1999). The functions of multiple representations. Computers & Education, 33, 131-152.
Bakkas, C. & Mikropoulos, T. (2003). Design of virtual environments for the comprehension of planetary
phenomena based on students’ ideas, International Journal of Science Education, (25) 8, 949-967
Danaia, L. & McKinnon, D. (2008). Common alternative astronomical conceptions encountered in junior
secondary science classes: Why is this so? The Astronomy Education Review, 2(6), 32-53.
Dunlop, J. (2000). How children observe the universe. Publications of the Astronomical Society of
Australia, 17, 194-206.
Duit, R. (2002). Bibliography - Students’ and Teachers’ Conceptions and Science Education. http://www. accessed 24th June, 2009
Greeno, J. & Hall, R. (1997). Practicing Representation: Learning with and about representational forms.
Phi Delta Kappan, 78 (5), 361-368.
Hegarty, M., & Waller, D. (2004). A dissociation between mental rotation and perspective-taking spatial
abilities, Intelligence, 32, 175-191.
Hsu, Y-S. (2008). Learning about seasons in a technologically enhanced environment: The impact of
teacher-guided and student-centered instructional approaches on the process of students’ conceptual
change. Science Education, 92(2), 302-344.
Hubber, P., Tytler, R. & Haslam, F. (in press). Teaching and learning about force with a representational
focus: Pedagogy and teacher change. Research in Science Education.
Kalkan, H. & Kiroglu, K. (2007). Science and nonscience students’ ideas about basic Astronomy con-
cepts in preservice training for elementary school teachers. The Astronomy Education Review, 1(6),
Kikas, E. (1998). The impact of teaching on students’ definitions and explanations of astronomical
phenomena Learning and Instruction, 8(5), 439-454.
Lemke, J. (2004). The literacies of science. In E.W. Saul (Ed.) Crossing borders in literacy and science
instruction: Perspectives on theory and practice. Newark DE: International Reading Association and
National Science Teachers Association.
Padalker, S. & Ramadas, J. (2008). Modelling the round Earth through diagrams, The Astronomy Edu-
cation Review, 2(6), 54-74,

Palmer, D. (2001), Students’ Alternative Conceptions and Scientifically Acceptable Conceptions about
Gravity, The Australian Science Teachers Journal, 33(7), 691.
Parker, J., & Heywood, D. (1998). The Earth and beyond: Developing primary teachers’ understanding
of basic astronomical events, International Journal of Science Education, 20, 503-520.
Shulman, L.S. (1986). Those who understand: Knowledge growth in teaching. Educational Researcher,
15(2), 4-14.
Trumper, R. (2001). Assessing students’ basic astronomy conceptions from Junior High School through
University, Australian Science Teachers Journal, 47(1), 21-31.
Trundle, K., Atwood, R. & Christopher, J. (2002). Pre-service elementary teachers’ conceptions of moon
phases before and after instruction. Journal of Research in Science Teaching, 39(7), 633-657.
White, R.T., & Gunstone, R.F. (1992). Probing Understanding. Great Britain: Falmer Press.
Yu, K. (2005). Digital Full-Domes: The Future of Virtual Astronomy Education, The Planetarian, 34(3),

Zeilik, M., Schau, C., & Mattern, N. (1998). Misconceptions and their change in university-level As-
tronomy courses, The Physics Teacher, 36(1), 104-107.


Tomaz Kranjc[1] & Nada Razpet[2]

Dept. of Physics and Technology, Faculty of Education, University of Ljubljana, Kardeljeva ploscad 16,
1000 Ljubljana, Slovenia
E-mail: [1], [2]

In their conversations, especially before exams, students often discuss and review
various topics from the (physics) syllabus. Listening to their conversations, it is
interesting to see how they understand and explain some of the topics, after having
finished basic physics lessons in primary and secondary school. Their explanations
are often unusual, and not rarely wrong. There are issues that they have learnt
about at several levels of education, from elementary school to university. Yet it
seems that, despite several successive treatments of some of the topics, some basic
misunderstanding may “survive” without making students feel uncomfortable and
wanting to ask questions or seek further explanations.
In order to assess ideas and conceptions that students have about some of the
common physics topics, we administered, at the beginning of the school year,
an exhaustive questionnaire to freshmen who had combined majors (mathematics-
technology and mathematics-computing), and also students of primary school teach-
ing. A special emphasis was put on astronomy in order to mark and draw attention
to the International Year of Astronomy.

In the modern, technically highly developed society, it seems natural and necessary that every-
body has some understanding of the most common natural phenomena that we are continuously
witnessing, and also has an idea of the basic functioning of the many devices, apparatuses, en-
gines, gadgets, etc that are accompanying us, either visibly or in a hidden way, in our everyday
life. In view of the high complexity and sophistication, it is obviously not possible to understand
the vast area of technological processes in detail.
For people to function well in our modern society we believe that the following items are not
only necessary but also possible:

a) acquire the basic knowledge of “classical sciences”, i.e. understand the basic concepts
upon which they rely, and get familiar with the basic procedures that govern their use;
this includes the ability to test simple things, dispensing with precise instructions, as in
school experiments;

b) accept some basic facts, and their consequences, of the “modern physics” which, in the
first place, encompass the “structure of matter” (being an extension of the old “atomic
theory”), the theory of light, and their interaction as far as they can be explained using the
classical concepts;
c) accept some facts from the relativity theory as a theory underlying all natural processes,
d) accept some basic facts of the quantum theory such as, quoting Huang (Huang, 1987),
“all systems in nature obey quantum mechanics”.

The purpose of science education, and in particular of physics, from elementary school to uni-
versity, is to make every pupil and student acquainted with the most basic facts of the “fabric of
reality” as explained by science (Osborne et al., 1991). This does not only mean that students
have to accept definitions and facts about the laws of nature and their functioning as true be-
cause this is what they are taught at school and because they are supposed to believe in school
teaching. Rather, they should get insight into the way the science and the science-based proce-
dures work, and convince themselves that science is not magic but a “common sense” reasoning
based on experimentally and/or theoretically obtained principles.
Students should be aware that science has no pretentions to assume the status of the ultimate
truth of everything. Science is not able to explain everything that exists in the world; rather,
what science is able to explain, is most probably the most accurate and reliable explanation
available. In this sense, science helps us building a picture of the world that does not contradict
the (verified) findings discovered by scientific research. (We should like to stress that science
does not claim to cover all the reality of the world, albeit there are ambitions, among scientists,
to build the “theory of everything”, or the “final theory” (Weinberg, 1992). There are frontiers
of science, of both practical nature as of principle, beyond which science, at least for the time
being, offers no answers.)
For science to establish itself already in schools as a system of (experimentally and theoreti-
cally) confirmed and verified truths (with the disclaimer of “falsification of theories” (Popper,
1974)), it is necessary for it to prove as such while “in action”. Students must be offered the
opportunity to learn how simple basic laws are discovered and confirmed, and how they lead
via a sequence of steps to explanations and understanding of a variety of natural phenomena.
In science and in particular in physics school curriculum only a tiny part of the accumulated
knowledge can be presented. It is our opinion that the basic principles of classical physics
should be presented in such a way that a clear understanding and working knowledge are
achieved. As far as “modern physics” is concerned (including atomic theory, theory of rela-
tivity and quantum mechanics), only a coherent set of facts can be presented to students; they
are invited to accept these facts without a proof. The full justification of our belief in these
theories (based on insight and understanding) is far beyond the scope of school physics.
The teaching-learning process is a never ending endeavour to present successfully the picture of
the natural world as seen through the eyes of science, in the educational process. The teaching
process and conceptual emphases are changing because science is progressing and crossing
more and more new barriers; the perception of the world is changing accordingly.

This is not the only, and even not the main, reason why educational methods have to be searching
for new paths and paradigms, and to adapt to new (mental and practical) situations. The ability
and motivation of students to perceive and study science is changing and the school approaches
have to take into account to what degree students are interested (or non-interested) to devote
their efforts and time to study science.
It is obvious that in the last decades the natural sciences have largely lost their fascination and
appeal, as well as their social status. One consequence is that it is often difficult to motivate
students to study material that seems difficult, and at the same time unappealing and of little
Any teaching of science has two major components, inseparably intertwined with each other.
On one hand, it is necessary to learn and accept some facts about nature (following from obser-
vations and subsequent conclusions) - natural laws, which are a “concentrate” of our knowledge
about nature.
On the other hand, in order to discover laws governing natural phenomena, it is necessary
to make observations and draw conclusions (Cavendish et al, 1990; Razpet & Čepič, 2008).
Very often experiments conceived to make controlled observations of some phenomenon may
be difficult to perform. (The search for the Higgs Boson particle, subject of CERN’s ATLAS
experiment, is an example of what the extent of an experiment can be.) At school, however,
students are expected to perform simple experiments (or just observations) which do not need
any (expensive) requisites and apparatuses (Russell et al, 1990). Quite often students seek a
solution to a problem in (often unreliable) notes and/or books, rather then checking a conjecture
by performing a simple observation by themselves.
It is the aim of this presentation to describe some of the difficulties encountered by students
(and teachers) dealing with simple problems that only require simple reflection and observation
(Cavendish et al, 1990; Harlen, 1990; Russel & Harlen, 1990). To this end, a study was con-
ducted among freshmen having combination majors mathematics-technology and mathematics-
computing. It turned out that, despite several successive treatments of some of the topics, basic
misunderstanding often “survives” without students feeling uncomfortable and wanting to ask
questions or seek further explanations. The results of this study show that students, in the first
place, lack initiative, self-confidence and a bit of curiosity.
In the next section, the questionnaire is presented that was intended to test the ability of students
to figure out answers to some simple problems, and the results of the test are shown along with
comments. The last section contains conclusions.

The questions of the test were mainly intended to check the ability to “think out” the answer
by “figuring out” how things work, or to show the “feeling” for some quantities, rather than
using physics knowledge and problem-solving procedures. Some of the questions were old and
well-known, yet the answers still don’t seem to be obvious to students.
Figure 1 below shows the percentage of correct answers to the questionnaire. The result is rather
depressing. It was expected that the answers should mainly be correct.

Figure 1: The percentage of correct answers to the questionnaire.

Here we give some of the questions of the Questionnaire:

1. Can the Moon be seen at about 10 o’clock in the morning? If YES which lunar phase is
seen (first quarter, etc.); if NOT, why not?

2. Why are days (in the Northern Hemisphere) longer in summer than in winter? Explain
the answer.

3. Can a shadow be seen at noon?

4. A glass full of water (to the rim) has a piece of ice floating in it. What happens when the
ice melts? a) Some water flows from the glass, b) the water level does not change, c) the
water level drops.

5. Draw the rods in a vertical position such that they cast their shadows as drawn in the
picture (Figure 2).

6. Between stations A and B, trains are running all the time according to a fixed schedule:
at the start of each hour, a train sets out from each station towards the other station. The
journey of each train lasts exactly three hours. How many trains does a passenger see from
the window of the train during the journey from one station to the other (not counting the
one coming to the station at the departure, and not counting the one setting out from the
station at his arrival)? a) 1, b) 2, c) 3, d) 4, e) 5.

7. A balloon with a weight attached to it is floating just under the water level. The balloon
is then immersed deeper into the water and then released. Does the balloon a) rise back
towards the surface, b) not move, c) sink further?

8. One litre of water at 30o C and one litre at 50o C are mixed. What is the final temperature:
a) 20o C, b) 80o C, c) 40o C?
9. In a pot, there is 1 litre of liquid water at the pressure of 1 bar and temperature 100o C.
The water is heated to evaporate completely. What is the volume of the water vapour?
10. An ideal heat engine is operating between the temperatures of 30o C and 150o C. What is
its efficiency? a) 100 percent, b) 80 percent, c) 28 percent, d) 0 percent.
11. One litre of water is heated from 0o C to 100o C. To what height could the water be lifted,
if instead of transferring heat to it, the same amount of work was done to lift it? a) 100
m, b) 5 m, c) 100 km, d) 42 km.
12. Estimate the power dissipated in a person quietly lying on the bed.
13. What kind of waves is sound? a) longitudinal, b) transverse, c) electromagnetic.
14. Two light bulbs are connected in series between the terminals of a 3 V battery. The light
bulbs are made to operate at 3 V and have resistances of 2 Ω and 3 Ω, respectively. a)
Sketch a schematic diagram of the circuit. Sketch how to connect a voltmeter to measure
b) the voltage on the first bulb, c) the voltage on the first and the second bulb.
15. Sketch magnetic field lines of a straight current-carrying wire.
16. Step in front of a plane mirror attached to a vertical wall. Draw the rays to show how the
image in your eye is formed.
17. When a beam of white light passes through a glass prism, a rainbow is seen on a screen
behind. Why?

Figure 2: Figure to question 5. The students were asked to draw vertical sticks such that they
cast the shadows as indicated in the figure.

There is not enough space to discuss all questions and answers; we therefore limit ourselves to
commenting some of them.


Most students know that the Moon can be seen in the middle of the morning, but very few have
given an answer about its phase. Among the answers, only about a half were correct, however.
It is our experience that students have difficulties visualizing the illumination of the Earth and
the Moon during their motion around the Sun. Problems requiring sketching what they would
see, if viewing the paths of the Moon, the Earth and the Sun from a far away space vehicle,
are too difficult for most of the students. It turns out that even the figures, simulations and
demonstrations with models are of little value. Sketching the illuminated parts of the Moon for
various positions with respect to the Earth also seems to be difficult. While sketching correctly
the illuminated parts of the Earth, they get confused with the Moon. The most difficult part for
them is to see, from a configuration which they have sketched themselves, is how the Moon is
seen from the Earth at a given time of the day (Figure 3).

Figure 3: Phases of the Moon. Example of a sketch as expected to be drawn by students in

figuring out what phase of the Moon an observer on Earth can see. (The drawing is not to

An efficient tool to help students visualize the phases of the Moon has proven to be a simulation
described in Kranjc & Razpet (2009).


Most students link the length of the day (and night) to the distance of the Earth from the Sun.
Obviously, they connect the length of the day with temperature: in summer, days are longer
and temperatures are higher. The distance of the Earth to the Sun changes due to the elliptical
trajectory of the Earth, higher temperatures are due to the smaller distance, so is the length
of the day. - Students learn about the seasons and their cause not only in physics, but also in
geography. A lot of demonstrations are shown in classes in order to make clear that the change
in seasons originates in the inclination of the Earth’s spin axis with respect to its orbital plane,
and the length of the day is a part of it. The wide appeal for the wrong answer stems, probably,
from its simplicity. It requires a lesser mental effort to figure out the dependence on the distance
than on the tilt of the spinning axis of the Earth orbiting around the Sun. (It is a bit of irony that
actually (in the Northern hemisphere) the Earth is closer to Sun in winter than in summer.)


Children observe shadows in kindergartens and in elementary schools; later the topic seems to
be too simple to deserve any special consideration. However, this is one of the topics at the
yearly field lab for secondary school students. Before starting practical activities on shadows,
they have a discussion about some of the features of shadows and how shadows cast by objects
in the sunlight change during the day. Students know that shadows get shorter towards the noon,
at which point they are the shortest, and then they get longer again. They also know that the
direction of shadows changes. They even learn about sundials (gnomons) and how the shadow
cast by the Sun from a style (e.g. a thin rod) can be used to measure time.
When they are asked about the length of a shadow at noon, the most frequent answer is that
there is no shadow at all, or that it is VERY short (meaning the invisible shadow under the
object). At that point, they have to learn about the maximum altitude of the Sun (the angle
between the horizontal plane and the ray from the observer to the Sun); a frequent answer is
that, at noon, the altitude of the Sun is 90 degrees, and that the Sun is located directly over our
heads. After observations and further discussions they realize that at our latitudes the Sun never
reaches the zenith. However, this knowledge does not seem to become a part of their long-
term memory. As freshmen at the university they have already slipped back to their primordial
misconceptions. A partial explanation of why the notion that there is no shadow at noon (and
several other misconceptions) is deeply rooted and stubborn is that it is a general “strategy” of
students to choose “simple” solutions although they do not really fit in with the problem.


There were two problems arising in answering this question. The first is that students did not
understand (or did not draw attention to) what “vertical” means. The rods that should be put
vertically were drawn in several directions: perpendicular to the surface, parallel to the sun
rays, or in some other direction. This seems to be linked to the fact that students are used
to observe shadows on horizontal planes only; the inclined planes present a new problem to
them. The second problem concerns the geometric construction of shadows cast by a light
beam from an (opaque) object. In relation to this, it appears that students are not familiar

with the geometric notion of a shadow, and that they are not able to apply the notion of the
“perpendicular projection” that they know from geometry to a practical problem of shadows.
Two answers are shown in Figure 4. The first leads us to believe that the student did not really
read the question (in which the dispersed directions of the rods are supposed to all be vertical)
and also that he/she did not have any idea about the formation of the shadows by light beams.
The second answer shows a better idea about verticality and the shadow formation; however,
the student made the answer “easier” (and wrong) by relaxing the requirement of verticality and
adapting the formation of the shadow when his/her construction did not really fit to the known
shadow. It is our belief that with a little more perseverance and self-confidence the student could
arrive at the right answer.

Figure 4: Two answers to question 5.


In the physics curriculum, a great deal of time and effort including experiments and demon-
strations is devoted to the treatment of ideal gases. Every student knows the equation of state,
P V = (m/M)RT . The problem requires the calculation of V = (m/P M)RT for 1 kg of
water vapour at the pressure of 1 bar and temperature 100o C. There was not a single correct
answer. Apparently, the difficulty lies in the fact that the data (pressure and temperature) were
given for 1 litre of liquid water, and that the word “ideal gas” did not appear in the text of the
question. One of the major problems encountered by students (and teachers) is the “fragmenta-
tion” of knowledge: students may have good skills in different areas, but are not able to connect
these seemingly disjoint pieces of information. Starting with a litre of liquid water hides the
fact that the question is about gas.

2.6 Q UESTION 10

Physics relies largely upon idealizations. By avoiding unessential complications, they make
things simpler and help clarify basic notions. Idealizations must retain the basic features of a
physical situation; they can discard only minor effects accompanying the main phenomenon,
and extraneous to it. In this sense, idealizations are “realistic” descriptions of nature; they
show and foretell what physical properties and the behaviour of a system really are, despite
“idealization”, within a “margin of approximation”. The notion of the “ideal” engine is an
example of a “useful idealization”.
In this question, however, students consider “ideal” to mean “100 percent efficient”, or nearly
so, which does not represent the physical reality. This is what they imagine under the name of

an “ideal”, or Carnot, heat engine. Although they have to learn, and verify on various examples,
that “ideal” in this case means “within the limits allowed by natural laws, assuming idealized
working conditions (no friction, full reversibility etc.)”. When they learn that the efficiency of
a Carnot engine is ηc = 1 − Tl /Th , with all the necessary comment, they also learn that ηc may
assume arbitrarily low values if the temperatures Tl and Th are close to each other, ηc still being
the highest value allowed by nature for those Tl and Th .
Sometimes, taking into account that “nothing is perfect” in a real situation, they simply put the
efficiency to 90 percent. Having heard that the efficiency of heat engines is mostly low, some
students put the efficiency deliberately to some still lower value; an arbitrary numerical value
drawn from the memory and linked to the efficiency of heat engines is often their choice.

2.7 Q UESTION 11

The answer to this question can be obtained by a straightforward one-line calculation; as the
data are simple, no calculator is needed. It is difficult to see how students arrived at their
answers, but it seems that they did not use physical (or factual) reasoning. Rather, it seems that
they thought that 42 km was a big height, and that this probably could not be the right answer.
But then, what about 100 km which also appears among the possible answers? It might be the
cause of a Gaussian distribution of the answers.

2.8 Q UESTION 12

It is puzzling that not a single answer was even approximately correct. Many students explicitly
stated that no power is dissipated by a person at rest. They all learnt quite in detail, in biology,
chemistry and physics, about the many processes going on in a living being, just to keep it alive:
circulation of blood, respiration and the accompanying contraction and relaxation of muscles,
digestion and the accompanying chemical processes, nervous activities etc. Nevertheless, it did
not cross their mind that these processes use energy, like an idling car: although the car does not
move, the running engine consumes gas, and energy. The first law of thermodynamics, stating
that “nothing is free in natural processes”, is a “physical” law, so somehow the students get the
impression that it does not pertain to biological systems.
There is another point of some concern. Several students came up what seem to be arbitrary
numbers for the power dissipated for example, “20 per cent”.

It may come as a surprise that, after a thorough study of a topic (including discussions, ob-
servations, measurements), students give wrong answers after a relatively short period of time.
Therefore it seems that in lessons and hands-on experiments they only get a “short-term”, “shal-
low” knowledge: they understand the explanations given in the class, but after a while they are
no longer able to reconstruct the reasoning. In such a situation, they return, without much
hesitation, to the simpler, previously obtained old concepts and misconceptions.
Learning is basically getting used to the correct explanation of a topic. “Getting used” is a
process which includes understanding, memorization and reinforcement of knowledge, and re-
quires a lot of repetition and much time. Failing to do so, even in very elementary matters (like

the explanation of shadows or the lengths of days) students slip back to seemingly simpler, but
incorrect, explanations from their childhood.
In our opinion, there are various and disparate reasons for the bad results of the test.
The first, and possibly the main reason lies in the interests of the current students and the source
from which they are obtaining a great deal of their knowledge: their “mental” interests and
activities are largely linked to the Internet and Internet communications. The tools they use
are computers, cell phones, MP3s and similar. In consequence, these are the “instruments”
they are familiar with: this is what they keep practicing, exercising and repeating every day.
As the virtual reality is often more “real” for them than the “real reality”, they have no time
and no interest in observing “real” natural phenomena. Observing natural phenomena requires
more effort, patience and time than watching computer animations and/or simulations. But
failing to make observations and experiments also means that they do not get acquainted with
the necessary instruments and equipment, or get to prepare self-made tools, and they do not get
any skills in performing even the simplest experiments. Also, computer “reality” is adjustable
and errors and mistakes have no real consequences. Making a wrong choice or giving a wrong
answer does not seem of much importance.
We must not forget that, most often, children play computer games before going to school,
and they get used to the extraordinary state-of-the-art computer graphics. School computer
presentations are often far below the quality of professional video games.
The rhythm of life with ever present lack of time makes a systematic long-term goals-oriented
study less popular. Students prefer “fast knowledge” and therefore learning “from day to day”:
instead of studying books, they read lecture notes (often of their colleagues), or even short
excerpts; this takes little time but also leaves little room for a deeper understanding and good
memorization. If there are mistakes in lecture notes they learn them without questioning or
noticing the inconsistencies of the material, attributing them to the fact that physics is “difficult
to grasp”.
We believe that physics teachers make a lot of effort to motivate students. Despite good expla-
nations of the syllabus, there may perhaps not be enough time left for the discussions that are
necessary to clarify and solidify basic concepts. Through discussions teachers can see which
concepts students have difficulty understanding and why.
We may add that there is probably an additional reason for bad results: when taking a test just
for assessing knowledge without getting a grade, students do not make a big effort to really
show their knowledge. Knowledge-testing is not a big motivation for them.
Nevertheless, we believe that testing students regularly, and in particular performing pre-tests,
in order to reveal gaps and holes in their knowledge and remedy misconceptions rather than to
give scores, is an important and good means to improve the knowledge and give students further

Cavendish, S., Galton, M., Hargreaves, L. & Harlen, W. (1990) Observing activities, Paul Chapman
Publishing Ltd., London

Harlen, W. (1990) Practical Tasks, Paul Chapman Publishing Ltd., London
Huang, K. (1987) Statistical Mechanics, John Wiley, New York, p. 171
Kranjc, T. & Razpet, N. (2009) What is a Moon’s Trajectory? and Some Other Demonstrations of
Motions of Heavenly Bodies, to be published
Osborne, R., Freyberg, P., Bell, B., Tasker, R., Cosgrove, M. & Schollum B. (1991) Learning in Science,
Popper, K.R. (1974) Objektive Erkenntnis, Ein Evolutionärer Entwurf, Hoffmann und Campe
Razpet, N. & Čepič, M. (2008) Early introduction into scientific learning, Sodobne strategije učenja in
poučevanja, Pedagoška fakulteta Koper, Univerza na Primorskem
Russell, T. & Harlen, W. (1990) Practical task, Paul Chapman Publishing Ltd., London
Weinberg S. (1992) Dreams of a Final Theory, Pantheon Books, New York


Paul Logman[1], Wolter Kaper[2] & Ton Ellermeijer[3]

University of Amsterdam, AMSTEL Institute, Science Park 904, 1098 XH Amsterdam, Netherlands
[1], [2], [3]

How can we use Watts’ frameworks (Watts, 1988) for talking about energy to design ed-
ucation on the subject of energy? We showed 6 pictures also used by Watts and gave a
writing assignment to 17 fifteen year olds prior to any physics education on energy. We
refined the definitions of Watts’ frameworks in order to solve our problems with assigning
student-statements to frameworks and found the frameworks not to be mutually exclusive.
Results found include the frequency of the 7 frameworks in the studied group of students,
and the overlap between these frameworks. We also found that certain frameworks are in-
volved positively and/or negatively in reaching the desired goal of secondary school learn-
ing about energy. Finally we will present some ideas on how to use the results in the design
of educational materials which we hope to develop in the near future.

Watts, using interviews, established 7 different frameworks for talking about energy. He did so by
considering students’ verbal reactions to certain pictures combined with the question to decide if the
pictured situations illustrate their concept of energy and to give a reason why (Watts, 1983). He provided
typical examples for each of the frameworks, in the form of citations from students.
Various later studies have used Watts’ frameworks in different ways: in group discussion assignments
(video recordings) (Gilbert and Pope, 1986), in interviews (Kruger, Palacio et al., 1992, Summers, Kruger
et al., 1998), in writing assignments (Finegold and Trumper, 1989), analysis of TIMMS database (Liu,
2005), and in surveys through questionnaires (Kruger, Palacio et al., 1992, Swackhamer, unpublished).
As a preparation for a planned teaching experiment we chose to diagnose Dutch students’ prior ideas
about energy using Watts’ frameworks in combination with a writing assignment. We chose a writing
assignment because it’s do-able for reasonable numbers of students, while it does not force students to
say yes or no to sentences they would never construct themselves. This paper reports some methodolog-
ical difficulties that we experienced with this chosen setup, and solutions found for those difficulties.
Finally, we will also diagnose our target group of students and draw conclusions about possible teaching
Other studies based on Watts’ frameworks make adjustments to the 7 frameworks described by Watts
and emphasize some frameworks more and other frameworks less. Sometimes an extra ‘framework’ is
added in which students confuse the concept of energy with another physical quantity or phenomenon.
Usually Watts’ frameworks are interpreted as being mutually exclusive. Exclusivity is a desirable fea-
ture of a category system such as frameworks. It makes it easier to classify students’ statements. Watts

himself makes the impression that he thinks of his frameworks as being mutually exclusive by present-
ing some pairs of his seven frameworks as clearly demarcated and by interpreting each of his example
citations to fit into one of his frameworks and one only.
However Watts and none of the later studies have checked the seven frameworks for clear differentia-
tion. In their original formulation by Watts it is not clear that the frameworks are necessarily mutually
exclusive. Particularly for specific pairs it is very hard to regard them as exclusive.

“To lift a box up a hill one needs energy (body energy). This energy is acquired from food
and sleep. It is generated from nutrients by burning.[. . . ]” (Student 7 on picture A)

This citation shows the idea of energy being needed, as a cause to let things happen (framework 2) and
the idea of generation of energy out of stuff that is not energy, so non-conservation (framework 5)5 .

“Because something happens to a chemical reaction such as heating, cooling, mixing and
stirring energy can be generated. This can be heat or light, but also other things such as
electricity or battery acid” (Student 16 on picture B)

In this citation and even in the same sentence we see the idea of a trigger, namely, the heating, cooling,
mixing or stirring being needed to release energy that didn’t exist prior to being awakened (framework
3) and the idea of generation of energy (framework 5).
Therefore we ask the question whether we are justified in requiring the frameworks to be exclusive.
We decided to interpret Watts’ frameworks as non-exclusive categories for two reasons. It leaves room
for us to stay close to Watts’ original definitions of the 7 frameworks and we were curious about what
kind of results and difficulties would arise from doing so. The limitations of the writing assignment we
try to counter by using a three-valued logic: inclusion of ‘undecided’ as a third category.


The next table (Table 1) shows a selection of definitions by Watts side by side to our definitions leaving
out two frameworks which are not essential to our major conclusions (for the missing definitions we refer
to an internal report (Logman and Kaper, 2009) which will be available from our website).
In the following, we will comment on some of the changes we made in the definitions (Table 1). From
these comments it will become clear that refining the definitions has been an iterative process: in arguing
about our definitions we will sometimes have to refer to results that will be substantiated later in the
results section.
In defining framework 2 we focused on objects having energy and to let this energy be the cause of some
activity. Doing this we stayed close to Watts. In students’ products from the writing assignment we
found students stating either the first or the second condition for conformation while seeming to imply
the other condition, so we decided to categorize both as framework 2 as long as the statement didn’t
contradict the condition that was left out.

See Watts’ definitions as given in Table 1. This table also contains our own definitions, but we are not yet
referring to those.

Watt’s definition Logman et al. definition
Framework 2: A ‘depository’ model of energy
This is a model of energy that (Clemont 1978) A statement conforms to framework 2 if:
calls a ‘source of force’ model. From this point
of view, youngsters see some objects as having • energy is regarded as a necessary cause for
energy (and being rechargeable), some as ‘need- something to happen and (or)
ing’ energy (and simply expending what they • energy is regarded as present, stored in
get) and yet others as neutral (and whose activ- objects† , before the relevant activity
ities are somehow ‘normal’ or ‘natural’). En-
ergy, then, is a causal agent, a source of activity If one of these characteristics is clearly present,
based or stored within certain objects. This idea the other one often seems to be implied, so the
of energy has a long history (Elkana 1974). The “and (or)” is really meant to be an “and”, the
‘power’ within things that enables them to act has other aspect must not be contradicted! However,
often been called energy. If something happens, it’s not required that both are explicitely stated.
then it seems natural to look for some source that
causes it. A statement contradicts framework 2 if:

• something happens without energy be-

ing necessary for it.
Framework 3: Energy is an ‘ingredient’
In this framework energy is not necessarily a A statement conforms to framework 3 if:
causal agent, but a ‘reactive’ one. It is a dormant
ingredient within objects or situations that needs • a trigger (mixing substances, pushing a
some ‘trigger’ to release it. Solomon (1980) switch, lighting a fuel) is needed before the
notes this feature when pupils talk about food: energy becomes active.
‘pupils believe that energy is not stored in food,
it only “gives you energy when you eat it”’. In A statement contradicts framework 3 if:
a similar way some would argue that energy is
not stored in coal or oil (as in framework 2) but • things happen and the necessity for a
is ‘sparked off’ when either is burnt. Likewise, trigger is denied.
a book lying upon a table would not have en-
ergy unless something (or someone) came along
to push it off.
Framework 4: Energy is an ‘obvious’ activity
To many, it is outward overt displays of activ- A statement conforms to framework 4 if:
ity that are the sole means of identifying energy.
Moreover, the activities themselves are actually • an activity is equated with energy (this
called energy. Movement (of any kind) is widely usually takes a verb).
given as a reason for energy being involved: the
energy is the movement itself. This often indi- A statement contradicts framework 4 if:
cated, in the responses, by energy being equated
with a verb, a word-of-action. • a noun is used to identify energy, or
• energy is the cause of the activity‡ , or
• energy is a product of activity‡ .
Continued on next page. . .

Watt’s definition Logman et al. definition
Framework 5: Energy is a product
In contrast, this framework carries the sugges- A statement conforms to framework 5 if:
tion that energy is not an ingredient or a process
(as above) but a byproduct of the situation. In • energy (of a certain form) is the result
some senses it was rather like a waste product of a certain activity.
as with smoke, sweat or exhaust fumes - another
perspective identified by Stead (1980). Clearly A statement contradicts framework 5 if:
energy is non-conserved, as with the other frame-
works, and here it is treated as a relatively short- • an activity occurs while it’s denied that
lived product that is generated, is active and then (any form of) energy is involved as a product.
disappears or fades.
Framework 7: A flow-transfer model of energy
Aarons (1965) says ‘Energy is not a substance, A statement conforms to framework 7 if:
fluid, paint or fuel which is smeared on bodies,
rubbed off from one to another; we use this term • for an activity a source of energy is
to denote a construct. Numbers calculated in a mentioned, and
certain prescribed way, that are found by theory • for the same activity it is mentioned where
and experiment to preserve a remarkably sim- or in what form the energy will be transferred as
ple relationship in very diverse physical phenom- a result of the activity.
ena’. And yet Warren (1982) points out that ‘en-
ergy is a fluid’ is both an implicit and explicit A statement contradicts framework 7 if:
suggestion behind the way in which the concept
is commonly taught in schools. In this way en- • no energy source was needed for the
ergy is seen as being ‘put in’, ‘given’, ‘trans- activity, or
ported’, ‘conducted’ and so on. • energy disappears after the activity.
Table 1: Watt’s definitions as opposed to Logman et al. definitions. † “Objects” in the most general
sense: air, fire . . . To be an object something must occupy space and have some durability, because
energy must be “stored” (a state), so an object is not a process, but apart from that it can be
almost everything. ‡ Being either the cause of activity or the product of an activity, energy can’t
be equated with the activity itself.

As most students in our study only mention the production and do not write about what happens to
the energy produced after an activity ends, we decided to broaden framework 5 slightly. We added
the situation in which production of energy was clearly mentioned yet nothing was written about the
non-conservation of energy to this framework.
Framework 7 is designed by Watts to be a goal of education. Though not as abstractly as Aarons (1965)
and Feynman (1963) want it, the idea of energy conservation is clearly included in the ‘flow’ terminol-
ogy. When looking at how to differentiate framework 7 from the other frameworks, it occurs that both
frameworks 2 and 5 are missing an aspect: adherents to framework 2 concentrate on the energy “needed”
for an activity, while adherents to framework 5 concentrate on the energy as it “results” from the activity.
Anyone believing both and believing that both necessarily go together for a single activity would be

Figure 1: A general view of energy conversion within framework 7.

a believer in conservation of energy, according to us. Therefore, we took this to be the definition of
framework 7.
The following statement illustrates belief in the conservation of energy (framework 7), because both
the need for energy before the activity (framework 2), as well as the existence of new forms of energy
resulting from the activity (framework 5) are indicated:

“Energy comes from the battery and goes through the wire past the switch, if it is switched
on. After the switch the energy continues through the wire towards the light bulb where it
creates light (energy) and heat (energy).” (Student 17 on picture D)



Following this line of reasoning further, we saw that most of the frameworks could be interpreted as
partly fitting within framework 7: the flow-transfer model of energy (see Figure 1).
In Figure 1 we see some form of energy storage as a source of energy for the activity. In some cases
the activity can only start after a possible energy barrier is overcome. By the activity the energy is
transferred to other objects or transformed into other forms of energy storage. Statements in framework
2 (a ‘depository’ model of energy), or in framework 5 (energy is a product) occupy different places in
this picture (see Figure 2). Statements that mention both energy as a cause and as a product belong to
both frameworks 2 and 5, and such statements are possible candidates for framework 7. In a similar way,
also the other frameworks (e.g. framework 4: energy is an ‘obvious’ activity) might get a place in this
picture. However, not all of them can be thought of as consistent with framework 7.
Looking at the different frameworks as supplementing each other instead of contradicting one another
while fitting them into this general view of energy transfer might have some advantages. First of all it
suggests ways of teaching students to reach the level of this general idea of energy transfer which can
be used as an intermediate stage for students in learning the scientific concept of energy as suggested in
other studies (Kaper & Goedhart, 2002a, Kaper & Goedhart, 2002b). Problems students encounter while
their statements are restricted to only a part of this general view are possibly solved by assimilation of the
missing parts in their view as these missing parts aren’t viewed as necessarily contradicting the general
view and each other any more. Another positive thing about this view is that it may help us create a new
set of stages in which students move from one framework to another during the learning of the concept of

Figure 2: Framework 2 and 5 seen as part of the general view of energy transfer.

Figure 3: An overview of possible inclusion and exclusion areas of the 7 frameworks.

energy. However we have to be aware of statements that don’t fit this general view as they may indicate
bigger problems on the road to learning about energy and these may only be solved by accommodation.
Based on our re-definitions of Watts’ frameworks, many “overlap” areas between frameworks can be
expected. Figure 3 shows all the possibilities we came to expect. Detailed reasonings can be found in an
internal report (Logman & Kaper, 2009).
We now picture a framework more formally as the set of statements that conforms to the framework
definition. Statements by students can be represented as points in this diagram. To try and see in what
ways the other frameworks relate to framework 7 we divide every framework into two parts: contradicting
and not contradicting framework 7. This turns our Figure 3 into a Venn diagram. It’s obvious that
statements in every other framework can either contradict or not contradict framework 7.
For framework 7 (a flow-transfer model of energy) we drew two boundaries: the “inner” boundary con-
tains all statements that fully conform to framework 7 (dark grey area), the “outer” boundary contains
statements that neither fully conform nor contradict framework 7 (light grey). Outside of this outer
boundary are statements that actively contradict framework 7 (white).
We need a source of energy (by definition this will imply framework 2) and we will need production of

energy in a certain form or location (implying framework 5) to conform to framework 7 fully. Therefore
we will need conformation to framework 2 and 5 if we want to conform to framework 7. However, this
is not enough. Additionally our definition for framework 7 requires that both the “needed” energy and
the “produced” energy are connected by a single activity6 . Therefore framework 7 needs to be drawn as
a smaller area within the overlap of frameworks 2 and 5.

Our study used a writing assignment and not an interview as Watts did. We have shown the students the
six pictures also used by Watts (figure 4) and asked them to write as extensively as possible about the
pictures on the subject of energy. The exact question was “to write down, picture by picture, what each
has to do with energy and to do so as extensively as possible”.

Figure 4: The 6 pictures used by Watts and by us.

This feature of our definition is motivated by a wish that the work-energy theorem should somehow fit in
frameowrk 7 - framework 7 being the desired end product of secondary school learning about energy

Next we tried to categorize the writings of the students into the different frameworks. We used the
text written by a student concerning one picture as one statement. We assess the relation between each
statement and all 7 frameworks.
The relation between a statement and a framework can be one of three things:

• “conforms”: the statement contains all properties required in the framework-definition.

• “contradicts”: the statement actively denies, or implies a denial of one of the properties required
in the framework definition.

• “undecided”: if neither of the above is true, the statement neither conforms to nor contradicts
the framework. This includes the possibility of the subject not having any strong feelings for or
against the beliefs of the framework.

One statement can conform to multiple frameworks, can contradict multiple frameworks and can be
undecided about multiple frameworks. In categorizing all the statements by a student we want to show
a student’s relationship to the different frameworks at a certain stage in his learning. What does the
student believe, what does the student deny to believe and about which frameworks are we undecided.
We take into account that different situations may provoke different reactions by the students. A word
like ‘energy’ may have different meanings in different situations to the student depending on the situation
in which he learned about the concept of ‘energy’.
Within the text written about one situation we expect consistency, otherwise our choice of “statement”
would have been unwise. If it had not been consistent we then should have split the statement into smaller
parts. However that did not seem necessary.
After the both of us analyzed the same 4 students we compared our results to come to a sharper consensus
on how to differentiate between the possible outcomes of our analysis. In total we analyzed seventeen
15-year olds.

Having six pictures and seventeen students resulted in 102 statements to analyze. Out of these state-
ments 4 statements didn’t say anything about energy and a 5th statement could not be classified into any
framework and resulted therefore in no conformation or contradiction at all (see Figure 5: white).
In 33 statements conformation was found to only one framework (see Figure 5: white): 14 of these
statements concern framework 2 and the other 19 statements concern framework 5. Only framework 2
and 5 were encountered on their own.
In all 64 remaining statements conformation to two or more frameworks was found (see Figure 5: black).
This means that some of the overlap areas allowed by our definitions are indeed well filled and the
expectation that frameworks are not mutually exclusive is confirmed.
Looking at the results in the group of students studied we encountered some frameworks much more
frequently than others (see Figure 6).

Figure 5: Number of frameworks found per statement.

We see that in the group of students studied frameworks 2, 5, and 6 are very common whereas framework
1 and 4 are contradicted in many statements. Framework 7 is confirmed or contradicted in only a few
While looking for the expected overlap between frameworks 2, 5 and 7 we have found 12 statements
conforming to both framework 2 and framework 5. In only one of these statements was conformation to
framework 7 not found (see Figure 7). As expected, in this one statement the need for energy was for a
different activity than the one which produced a certain form of energy. The dark grey area is therefore
rightfully depicted smaller than the overlap of frameworks 2 and 5.
As framework 7 is the goal of our teaching, we separate all 102 statements now into the ones conforming
to framework 7, contradicting framework 7, and undecided about framework 7 (the large majority).
For each of these three groups of statements we show which other frameworks are involved in those
statements (Figure 8).
Conforming to framework 7. When designing education, these statements by students should be the ones
we are satisfied with, if indeed framework 7 is our goal. As required by our definitions, frameworks 2
and 5 are always conformed to when conforming to framework 7. Framework 4 is contradicted in all
these statements which one would expect considering its definition. More interestingly, framework 1 is
also contradicted by nearly all these 11 statements, while framework 6 is conformed to by 8 of these 11.
Undecided on framework 7. We have found 75 statements which were undecided on framework 7 (all
17 students showed this kind of statement). In this case either students did not state enough to place

Figure 6: Number of statements conforming to and contradicting each of the frameworks.

Figure 7: Number of statements conforming to the possible combinations of frameworks 2, 5 and 7.

Figure 8: Relations of the other six frameworks to framework 7.

their statements into framework 7 or they did not show their rejection of framework 7. We suppose that
students showing these statements may need to add knowledge to their understanding of energy (assim-
ilation), but they probably do not need to ‘unlearn’ much (accommodation). If we agree on framework
4 as being contradictory to framework 7 not much work needs to be done with this group of students
concerning that framework as most statements reject it already (see Figure 6). Framework 3 is optional
to framework 7; and as no contradiction to framework 3 has been found not much work is needed in that
area. On other overlap between frameworks and its consequences for education we refer to an internal
report (Logman & Kaper, 2009). If our goal is to have students’ statements agree to framework 7 we
mainly need to try and persuade the students to move towards both framework 2 and 5 in which it is clear
that storage of energy after an activity should get the most attention.
Contradicting framework 7. This always coincided with conformation to one of the other frameworks.
The big white outer area outside all the frameworks therefore seems to be empty which in its turn means
that the 7 frameworks may be collectively exhaustive. Statements contradicting framework 7 make us
suppose that some accommodation is needed. We notice that framework 5 is the biggest problem for the
group of students studied.

“By eating one receives new energy. The nutrients are transformed into energy.” (Student

Figure 9: An overview of inclusion and exclusion areas of the 7 frameworks.

15 on picture F)

The second sentence clearly implies that the nutrients did not contain energy, but energy is produced out
of them (framework 5), a clear case of non-conservation (contradicting framework 7).
In 4 of the statements framework 3 was also a problem and in only one statement was framework 4
considered to be the cause of contradiction. There was no instance where only one framework was a
problem. Whereas Watts implied that all frameworks were contradictory to framework 7 it appears that
for our target group of students mainly framework 5 is the culprit. The 11 statements are widespread as
they were made by 8 of the 17 students. Again we conclude that energy storage after an activity should
get the most attention in any education designed to counter this problem.

In looking at 102 statements made by 17 fifteen year olds on 6 different pictures previously used by Watts
we find that the 7 frameworks established by Watts are not mutually exclusive. In fact most frameworks
seem to be part of framework 7 (a flow-transfer model of energy) in a smaller or larger sense (see Figure
9: mainly frameworks 2, 3, 5, and 6). In this paper, focus has been on the overlaps between frameworks
2, 5, and 7. We refer to an internal report (Logman & Kaper, 2009) for the other relationships shown in
the figure.
Our view of the overlap of each of the frameworks with framework 7 has mostly been confirmed by
analyzing the students’ statements. Indeed to conform to framework 7, framework 2 (a ‘depository’
model of energy) together with framework 5 (energy is a product) seem to be a necessary part which is
an important observation for future educational design.
To reach the desired end product of secondary school learning about energy, framework 7, framework 5
is shown to be especially troublesome to our group of students, as it was often involved in contradicting
7. It is necessary to take this into account when designing education on the subject of energy.

The frameworks encountered may of course vary by student or group of students and may be used as a
starting point from which their education on the subject of energy can be designed. One way may be
to create small groups of students of a certain mixture to perform special tasks to do with energy. To
do so one would prefer a faster way of establishing a student’s set of frameworks however. Another
way in which it may help education design is that it is shown that students who encounter contradictions
to framework 7 may need some form of accommodation of their ideas to make a step forward, whereas
students who find their views within framework 7 may only need some form of assimilation to make their
next step. For instance the large group that conforms to framework 2, energy as a needed cause, needs to
learn to combine their ideas with framework 5, energy as a product: needed energy and produced energy
being connected through an activity.

The following studies are similar to ours in various degrees, which makes it useful to compare methods
and results.
Kruger (Kruger, Palacio et al., 1992), using interviews, found that about 20% of primary school teachers
believe in framework 1 and substantial numbers of them lacked belief in conservation of energy. They
didn’t expand on the 5 other frameworks.
Summers (Summers, Kruger et al., 1998), also using interviews, departed from Watts categories and
decided to divide up different views into life-world view, partially scientific view and scientific view.
These categories are too broad for our purpose, because they do not inform to the content of students’
ideas. Summers, Kruger et al. found that energy efficiency can be readily taught to a group of primary
school teachers who already grasped the concept of energy conservation.
Swackhamer (Swackhamer, unpublished) has created his own categories that are only partly inspired
by Watts’ frameworks. He is trying to establish an energy concept inventory to check on students’
development and to identify fundamental issues to address. He also made guidelines to develop a healthy
conception of energy. (e.g., avoiding forms of energy, explicit energy transfer, requiring conservation of
Finegold and Trumper (Finegold & Trumper, 1989), using writing assignments mainly found frameworks
2 and 7 and have found similar distributions in Watts’ frameworks as us, yet they treated them as mutually
exclusive per statement. They did treat the students as adherents to multiple frameworks. However on
15 different pictures shown they found only thirty-six percent of students using 2 or more alternative
frameworks. Unlike us, they found no framework 3 adherents in ninth grade but did find framework
3 adherents in grades ten to twelve. During their research they decided to subdivide framework 2 and
framework 7 (where the latter was divided into flow-transfer as opposed to the scientific view). The
example citations mentioned have been categorized into frameworks that we would have chosen as well,
but some we would have categorized in more than just the one chosen by Finegold and Trumper (1989).
They did not look for contradictions because they considered the statements to be categorized into one
framework only. Like us, they come to the conclusion that frameworks 2 and 5 are needed to form
framework 7. They concluded that the effect of education on energy wears off by seeing an increase in
framework 7 adherents immediately after education and a following decrease in framework 7 adherents
in the years after education.
Comparing the various results we can say that Finegold & Trumper (1989) agree with us that before
instruction framework 2 is very frequent. Also, before instruction they found significant numbers of
framework 5, though much less than we found. Education on the concept of energy has been proven to

have little lasting effect on students (Finegold & Trumper, 1989, Trumper, 1990, Kruger, Palacio et al.,
Using the assumption of overlapping frameworks we have been able to establish important relationships
between the different frameworks which may prove useful for future educational design. As our aim is
also to look at the manouverability of the energy concept in students, it is important to find an effective
way of establishing to which frameworks a student or a group of students adheres, so that a more lasting
and effective education can be designed and the development of students can be tracked.
Our method has some limitations. Because we used a writing assignment we can be more certain that
students used their own words and wrote their own sentences instead of reacting to ours. Therefore we
can be confident in identifying students’ conformation and contradiction to frameworks. On the other
hand the ‘undecided’ category will be larger in a writing assignment, because students show us a selection
of their thoughts with missing information as a result. Because sometimes we have to be undecided about
a specific framework we can not establish for sure the relationship between that framework and other
frameworks because we can not ask follow-up questions.
In the near future we will try to use our method with seventeen year olds and twenty-two year olds and
will keep tracking the group of students studied to check persistency. We will also use our findings
in a design for education on the subject of energy. In testing our results on education we will try out
other methods as well such as the energy concept inventory currently being developed by Swackhamer
(Swackhamer, unpublished).

Aarons, A. B. (1965). Development of Concepts of Physics. London, Addison Wesley.
Clement, J. J. (1978). Mapping a student’s causal conceptions from a problem solving protocol. Mas-
sachussetts, Department of Physics and Astronomy, University of Massachussetts, USA.
Elkana, Y. (1974). The discovery of the conservation of energy. S.l., Harvard University Press.
Feynman, R. P., Leighton, R.B., & Sands, M.S. (1963). The Feynman lectures on physics, New York:
Finegold, M. and Trumper, R. (1989). Categorizing pupils’ explanatory frameworks in energy as a means
to the development of a teaching approach. Research in Science Education 19(1): 97-110.
Gilbert, J. K. & Pope, M. L. (1986). Small Group Discussions About Conceptions in Science: a case
study. Research in Science & Technological Education 4(1): 61 - 76.
Kaper, W. H. & Goedhart, M. J. (2002a). ‘Forms of Energy’, an intermediary language on the road to
thermodynamics? Part I. International Journal of Science Education 24(1): 81 - 95.
Kaper, W. H. & Goedhart, M. J. (2002b). ‘Forms of energy’, an intermediary language on the road to
thermodynamics? Part II. International Journal of Science Education 24(2): 119 - 137.
Kruger, C., D. Palacio, et al. (1992). Surveys of english primary teachers’ conceptions of force, energy,
and materials. Science Education 76(4): 339-351.
Liu, X. (2005). Developmental growth in students; concept of energy Analysis of selected items from
the TIMSS database. Journal of Research in Science Teaching 42(5): 493-517.

Logman, P. S. W. M. & Kaper, W. H. (2009). Elaborations on the overlap of frameworks on energy. Ams-
terdam, AMSTEL Institute, University of Amsterdam. Available from:∼logman
Solomon, J. (1980). The growth of the child’s concept of energy in the secondary school. Science in a
Social Context. Harlech, North Wales.
Stead, B. F. (1980). Energy. Learning in Science project. University of Waikato, Hamilton, New Zealand.
Summers, M., C. Kruger, et al. (1998). Developing primary teachers’ understanding of energy efficiency.
Educational Research 40(3): 311 - 328.
Swackhamer, G., & Hestenes, D. (unpublished). An energy concept inventory. Arizona State University.
Trumper, R. (1990). Being Constructive - An Alternative Approach to the Teaching of the Energy Con-
cept. 1. International Journal of Science Education 12(4): 343-354.
Warren, J. W. (1982). The nature of energy. International Journal of Science Education 4(3): 295 - 297.

Watts, D. M. (1983). Some alternative views of energy. Physics Education 18(5): 213-217.

A.J. Rissanen
National Defence University, Department of Military Tech., P.O. Box 7, Fin-00861 Helsinki

At the National Defence University, a two semester physics course was lectured to
a non-physics majors group. The course survey and lecturer’s notes on the students’
learning methods showed a need for course reinforcement. This study shows how
design research methodology is applied to solve the intervention needs and, later,
how the artefact was constructed.
For non-majors in physics, there was an essential need to lower obstacles to learn-
ing, because the subsequent sensor course, oriented towards applications, and later,
the ballistics course, are relatively theoretical, and dependent on utilization of con-
cepts from the basic physics course. One way of mastering physics, is the ability
to discover relationships between facts and patterns. To achieve such a level of
mastery, the following knowledge is needed: basic definitions, constructing and
understanding equations, and the use of mathematics in physics.
One important feature of a physics textbook is to provide a systematic description of
the course contents and the context for each topic. Students’ ability to read the text
easily is equally important. One of the contributions of the intervention is to provide
the lecturer with the opportunity to concentrate on provoking students’ thinking
and reasoning instead of merely presenting theory and numerical examples for the
students to copy. The project produced a course book set, using design research
methods, consisting of five steps: theoretical work, design procedure, empirical
phase, outcome of the design, and evaluation.

At the National Defence University in Finland, a two semester introductory physics course was
lectured to cadets. Analysis, based on a course survey and lecturer’s progress notes, showed
needs for course reinforcement and a demand for more specific and focused course material.
Learning material for physics courses was improved in this research and in the connected learn-
ing material enhancement project. The project produced a very course specific and customized
physics textbook set by using design research methodology, as a developmental frame, in ma-
terial design and in production.


The design oriented research tradition underlines bonds between research work and instruc-
tional practice (Duit, 2006). The design-based research (DBR) approach has been considered

a solution in filling the gap between science education research and science teaching. Design-
based research connects empirical educational research and the theory-driven design of learning
environments. This means that a designer (researcher) and a practitioner (teacher) both have an
important role in the development process for the aimed artifact (Design-Based Research Col-
lective, 2003).
A design process often begins with a perceived problem or arises through an opportunity and
an idea describing how to respond to it. The problem analysis characterizes the goals, need,
or opportunity that the design is intended to address and evolves over the course of a design
process incorporating information from a variety of sources (Edelson, 2002). Typically design-
based research contains cyclic and iterative stages. The design-based research is formed by the
following phases: the phase of theoretical problem analysis, the phase of empirical findings, the
phase of design procedure, and the phase of design solution (the artefact). Iterative cycles lead
to a better understanding of the intervention process. With its iterative cycles the design and
the re-design phases are more transferable and probably more useful results may be achieved
(Amiel & Reeves, 2008). Formative evaluation is critical in design research because it can
identify inadequacies in the problem analysis, the solution, and the design procedure that cannot
be exposed through analytical processes alone (Edelson, 2002).
In design based research, the designer has to break knowledge down into smaller components
for the purpose of design (Ruthven et al., 2009). Among members of the DBR community, the
DBR method is characterized differently but the following opinions seem to be shared (Walker,

• an emphasis in authentic, natural educational context

• a pluralist approach in respect to theories, research designs, methods, and procedures

• the desire for research to have a practical impact on the development in education

• emphasis on theory guided educational interventions

• focus on intervention evaluation.


Cadets at the NDU learn essential concepts of mechanics and electricity for military applications
and service safety. Students will develop permanent interest towards science and the means it
offers to develop understanding and skills to manage both daily life and battle field phenomena.
Students will get to know physics as a scientific discipline. Generally, the basic professional
education in technology subjects consists of the basic features of ammunition, sensors, naviga-
tion, communication systems, and general properties of materials (figure 1). These are utilized
in five major concepts of warfare.
Nowadays university level physics education has many challenges. In the cadets’ case physics
education is part of professional knowledge of how modern technology is utilized in concepts of
warfare which are further applied in successful operations and tactics. Therefore, it is especially

Figure 1: Connecting discipline and warfare concepts from the viewpoint of technical studies.

necessary to promote these links to those students who do not have much experience in or
interest on physics (non-majors in physics). The extent of the course in the standard physics
was valued to represent 2.5 credit units (equal to 2.5 weeks as full time work) and was divided
into 96 learning hours. Table 1 shows a condensed curriculum for the general physics course
(non-majors in physics).

lecturing exercises self study

Mechanics 20 6 2
Thermodynamics 14 4 2
Light and Wave Propagation 14 6 2
Electricity and Magnetism 16 6 4

Table 1: Condensed curriculum for the general physics course.

The curriculum included two examinations. The first one was held after the thermodynamics
lectures at the end of the spring term. The second test was at the end of the course, at the end
of the second autumn term. For planning reasons, self-study was marked in the plan, but not
implemented in the weekly schedules. Participation in lectures was compulsory. The average
estimated time for homework, preparation for lectures, and finally studying for examinations,
were not explicitly measured. Generally, common practices and procedures in Defence Forces
are guided by archive coded permanent documents. Order and daily routines are controlled by

commands. On the other hand, in learning and for example, in the contents of academic learning
materials, formal instructions and regulations, as well as principles of scientific formulation of
matters, are applied.


Figure 2 shows the development phases in this research and reveals the methodology in the
connected development project. The development process contained areas of problem identi-
fication, theoretical problem analysis, design procedure, empirical problem analysis, and the
design solution phase (an artefact), which were typical phases for the DBR method. These
phases were then completed with the evaluative phase and the maturation phases. Iterative cy-
cles were emphasized at the design and development, as well as during the maturation phase
when the product is in use. All feedback signals are notified and informal feedback with formal
evaluation results are forming a basis for improvement and maturation procedures.


The work started with a preliminary description of needs. The lecturer’s own observations
showed the need for an educational material development project. The development project
started with an analysis of environmental constraints which covered evaluation of structural
facets, discipline-based and environmental factors.
The theoretical frames investigated in the project were pedagogical frames, process-connected
frames, and frames concerning substance and development policy. University pedagogy un-
derlines an approach, which is crossing borders of earlier knowledge (Bransford et al., 2000).
In-depth knowledge of the most common theoretical pedagogically emphasized frames is not
a necessity for the learning material producer, but may widen the point of view and would
help to categorize these issues. This does not mean necessarily that the production policy is
only leaning on traditions. With customization procedures, a proper theoretical basis is likely
to open eyes for seeing and handling these issues in relation to current needs. Also effective
maturation processes may exist, when know-how and research are forming the working plat-
form. Sometimes teams are concentrating only on the frames in forming the substance, such
as the learned subject and educational aspects. It would be wise to consider the management
based frames (process-connected frames) and idioms in the design phase, so that they can be
applied smoothly to the project. This means that frames controlling the project work and design
process are worth consideration in resource intensive projects. Table 2 presents environmental
constraints and the theoretical frames behind the development paradigm.
In this project, a lean featured approach was selected as a frame for project work. The lean
ideology comes from industrial management, but has some examples applying also in the field
of education (Emiliani, M. 2005; Emiliani, M. 2004; Smith, 2008). Lean ideology underlies
“waste gap definition” under project work which helps to identify unnecessary development

Figure 2: The six stage flow model utilizing the Design Based Methods.

In the design phase, the prescriptions for the design and for the design solution were developed.
In this phase also, plans for the investigations to be made in the empirical phase were structured.
The results from the empirical phase guided the area of design protocols further. Therefore,
internal feedback process formed an important portion of the iterative development cycle. The
design phase contained plans for product design and also guidelines for project managerial
A customized and case-streamlined theory combination was utilized in the production team’s
toolkit. Individual innovational tasks were added in this phase for the design process. These
items were finally refined to a practical solution which was the intended text book set. Brans-
ford et al. (2000) present model pedagogical laboratories for testing pedagogical principles. It

Environmental E • Environmental factors and structure
constraints • Discipline based factors; curriculum, and orders
• Resources

Pedagogical T • Pedagogical approaches, methods and tactics


Substance and de- T • Substance connected frames; demands for a physics text book
velopment policy today
• Quality theoretical frames
• Creativity and innovation policy
• Design research methodology with its applying procedures

Process-connected T • Management aspects; project work, intensity dilemma in pro-

frames duction
• Re-engineering policy in development and networking idea
• Evaluate frames
Table 2: Environmental constraints and theoretical frames.

is not always possible to test a model in a strict laboratory context, therefore a lighter testing
and validating procedure may prove useful in method selection and refinement. Customization
is a procedure which helps the utilization of theoretical framing. We recommend the term “cus-
tomization”, which, we think, has a little different emphasis than the term “theory refinement”
usually used in DBR. We think that customization allows more space for personal innovations
and points to protocols which are typical and useful in case-based design. Customization is
considered as a set of procedures which streamline the frames in a way that they can be applied
with elasticity in the intended case or development procedure. Customization is widely needed
because for example, pedagogical frames mostly offer guidelines only for teaching activities but
give less support for other purposes for example for learning material design. Customization
also allows individual and case specific solutions (table 3).
According to the project experience, iterative procedures and quite free tailoring policy may
reduce the gap between theory and practice. Theoretical framing may inspire development
work plans, but individual innovations and specifications are needed to gain usable and situation
bounded products. This also guarantees that theories and approaches are not hindering new
innovational steps in educational design (figure 3).


Empirical results were gathered from three main sources:

• Student feedback study

• Observations, made at lectures and during lecturing protocols

• Educational experiments.

Frames/models Customization work Why customization?
• pedagogical frames • cutting, featuring, tailoring, mixing • some models or approaches are designed to
support educational activities, but do not cover
educational material production
• process-related frames • emphasizing certain features • some features may be altered
• quality and creativity policy • omitting non-relevant features • not all features are relevant for the needs of

the focus group
• other developmental frames • adding individual innovations • every development project is unique; indi-
vidual tailoring or streamlining
• validating and streamlining • space for own innovations

Table 3: Reasons for customization in this study.

Figure 3: Reasons for customization work.

The empirical basis was planned in a way that it would give a comprehensive view of the current
situation. Moreover, this choice reduced the validity problems in each part of the project’s em-
pirical phase. The findings concerning this phase, formed the key development areas and special
attention was given to critical tasks which arose: for example some of the students in the group
claimed that the education was too theoretical, a fact which reduced interest among certain stu-
dents (quantitative part of the study; see also figure 4). In the qualitative part of the study, the
following tasks were noticed: some claims concerning the course material, insufficient number
of demonstrations, and some improvement needs in course arrangements.

Figure 4: Expanding results from each question.

As a summary, in the empirical phase, the following areas were identified as areas requiring

• course material: need of a more course specific text book

• lecturing methods: presentation too theoretical and teaching protocols requiring more
exercises and demonstrations

• course arrangements: need of more level/ranking specific groups


After prototyping with its iterative phases a three part customized physics text book set for
cadets was the final outcome of this project. The set consisted of following parts: Mechanics and
Thermodynamics (Rissanen, Ojala, 2001), Electricity and Magnetostatistics (Rissanen, 2001b),
and Equations and Tables (Rissanen, 2001a) as a supplement product (figure 5).

Figure 5: The textbook set.

The main idea was to produce a set of text books, which followed the course structure and
content in an aligned way. The 2.5 credit extent and the further course utilization need meant
that the textbook set was made very course specific. The desire for a compact product meant
that only the most essential, yet necessary items and topics were included in the textbooks.
This meant that compression was needed in areas like graphic presentation and text generation.
However, the improved early stage customer satisfaction and slightly better examination results
showed that the resulting product was more appropriate than the preceding product, a more
general and demanding physics textbook which was not read by the cadets.

A lean type approach gave inspirational hints to the product framing and to the product fea-
turing. Lean underlines “waste gap definition”. The most obvious waste gaps in project work
turned out to be the following two tasks: a timing schedule too open in production and a missing
life cycle estimation of the product. Both of them are factors which could increase production
costs unnecessarily. Reasonable life cycle estimation of the product gives a good schedule frame
to the production policy and it is wise to take this fact into account in changing and demanding
educational environment. An efficient, but not too heavy evaluation mechanism keeps the costs
in use and maintenance during the life cycle of the product to an optimal degree.
University pedagogy (Bransford et al., 2000) underlines the importance of formative assessment
also in educational development. Hence, the evaluation system and its features as a part of an
educational quality system, were in focus all the time within the production. In this study
an open product means an evaluation system that collects feedback data and ensures that the
product is open to develop with changing demands. The light appearance of the product and
easy resample policy and lean integration made this feature possible.
The main advantages of the new product were the following:

• the text book set was course specific which enhanced the students’ learning processes
• for the teacher: more choices in how to construct lectures
• the product was compact enough and convenient for lecture use
• the compactness of the textbook decreased the impression of a difficult course

At the theoretical phase, the demands for a modern physics textbook were investigated. Accord-
ing to the team’s self-evaluative analysis these desires were partially achieved in this project.
The following table 4 shows how these areas were achieved. Demands concerning compact-
ness, clarity, and customization of the product were captured successfully. Resource savings
were achieved because course specific text books reduced the amount of additional lecture con-
nected routine work. The innovativeness of the product and the product’s feature in helping the
understanding of the nature of physics were also areas of progress.

Cobb et al. (2003) has underlined selected features in design experiments in educational re-
search. Some of those features have been emphasized as basic features in this study (table 5). A
highly practice oriented project and iterative work was applied from the design phase onwards.
Maintenance was conducted according to evaluations. The actual product is based on analytical
theory framing, but because of customization procedures, the space for individual innovational
solutions was guaranteed. We think that the term “customization” is proper to represent these
useful procedures. The design research methodology gives a useful frame which connects the
production and theoretical research phases to a formula. For the designer group this formula
clarifies how to apply production protocols in a systematic way. Design research methodology
is an approach, which provokes the designers to approach systematically different questions in
development which connects the basis of application and theory. Hence, it can give remarkable
support to the whole production process.

Desire Emphasis in the textbook project (from *** to *)
Compactness Successful, compact outlook***
Innovativeness Lean adaptation, light appearance of the product**
Clarity Contents coherence & clearness and systematic***
Moderate costs Resource savings because of reduced routine course work***
Differentiation Taken into account in exercises routines*
Nature in physics Self-evaluative questions in the exercise set, internal logic of the
textbook series**
Customization Highly course customized product***

Table 4: Desired features gained in the textbook project. *** high, ** average and * at certain

Theoretical problem analysis selections for theory framing

Design procedure theory customization, iterations, design based procedures,
risk calculation
Empirical problem analysis interventions and data gathering for support of production
Design solution practice oriented solution

Table 5: Features of design research methodology in this study.

Amiel, T., & Reeves, T. (2008) Design-Based Research and Educational Technology: Rethinking Tech-
nology and the Research Agenda, Educational Technology & Society 11(4), 29-40.
Bransford, J., Brown A., & Cocking, R. (2000) How people learn: Brain, Mind, Experience and School.
Washington (DC), National Academy Press.
Cobb, P., Confrey, J., diSessa, A., Lehrer, R., & Schauble L. (2003) Design experiments in educational
research. Educational Researcher, 32(1), 9-13.
Design-Based Research Collective (2003) Baumgartner, E., Bell, P., Brophy, S., Hoadley, C., His, S.,
Joseph, D., Orrill, C., Puntambekar, S., Sandoval, W., & Tabak, I. Design-based research: An emerging
paradigm for educational inquiry. Educational Researcher, 32, 5-8.
Duit, R. (2006) Science education research - An indispensable prerequisite for improving instructional
practice. Paper presented in ESERA Summer School, Braga. Retrieved 18.11.08 from:
Edelson, D.C. (2002) Design research: What we learn when we engage in Design, The Journal of The
Learning Sciences, 11(1), 105-121.
Emiliani, B. (2005) The time is right for higher education administrators, faculty, and staff to applying
Lean management to their business. The consequences of not doing so could be fatal. The Center for

Lean Business Management, LCC, Kensington (CO). Retrieved 18.08.09 from:
Emiliani, M.L. (2004) Improving business school courses by applying lean principles and practices,
Quality Assurance in Education, 12(4),175-187.
Rissanen, A. (2001a) Kaavoja ja Taulukoita (Equations and Tables), Edita.
Rissanen, A. (2001b) Sähkö- ja Magnetostatiikka (Electricity and Magnetostatistics), Edita.
Rissanen, A. & Ojala, H. (2001) Mekaniikka ja Lämpö-oppi (Mechanics and Thermodynamics), Edita.
Rissanen, A. (2008). Utilizing elements of Lean philosophy in course material preparation, In E. van
den Berg, A.L. Ellermeijer, O .Slooten (Eds.), Modelling in Physics and Physics Education, 928-931.
Amsterdam, AMSTEL Institute, University of Amsterdam.
Ruthven, K., Laborde, C., Leach, J. & Tiberghien, A. (2009) Design tools in didactical Research: In-
strumenting the epistemological and cognitive aspects of the design of teaching sequences. Educational
Researcher, 38(5), 329-342.

Smith, T. (2008) Lean Manufacturing Methods Applied to the Development of an Online Educational
Program. Online Educa Berlin. 14th International Conference on Technology Supported Learning &


Pavlo Antonov
V.N. Karazin Kharkiv National University

This work is an example of the development of the use of projects in problem-
based education. In the study of physics in universities and colleges it is necessary
to stimulate the creative activity of students. This can be addressed by means of
problem-formulated tasks.
The problem: Natural cataclysms and accidents (e.g. typhoons, hurricanes, El-Nino
and so on) pose a constant threat for many countries. The problem put to students
is to model the atmospheric processes with soap bubbles.
The purposes of the educational task are to:

1. Study physical laws (optics and thermodynamics) using the soap bubbles;
2. Use interdisciplinary communications (geography, mathematics, meteorol-
ogy, chemistry);
3. Activate the interest of students;
4. Train in modelling;
5. Work in small groups;
6. Use video- and photo- equipment, computer technology etc.

It is known that the behavior of turbulent flows on the surface of a soap bubble is
equivalent to turbulent flows in Earth’s atmosphere1.
Some physical methods which I have used in this project provide an opportunity to
observe the progress of turbulent flows on the bubble surface for comparison with
photographs from weather satellites.
The result of the project is the performance of laboratory work in which students
get acquainted with interference and reflexion phenomena. Also they will come
nearer to understanding the occurrence of hurricanes.
1 servlet/GetabsServlet?prog=normal&id=PRLTAO000100000014144501000

In teaching physics at universities and colleges it is essential to stimulate the creative activity of
students. This can be addressed by means of projects.
This work is an example of the development of the use of projects in the sphere of problem-
based education.
The use of projects is a way to achieve of the didactic purpose through detailed working out
of problems (technologies) which should include with real, tangible practical results. The main
objective of the use of projects consists in granting by the student the possibility of independent
acquisition of knowledge in the course of solving practical problems or problems demanding
integration of knowledge from various subject domains. If we speak about the use of projects
in terms of pedagogical technology, this technology assumes the totality of research, search,
problem methods and is inherently creative. To a teacher within the project is assigned the role
of developer, coordinator, expert and adviser.
The essence of the use of projects is the stimulation of the interest of students in certain prob-
lems, assuming possession of a certain sum of knowledge, and through the design activity
providing the solution to one or a variety of problems, to show practical application of their
The basis of the use of projects is the development of the research skills of students, the ability
independently to acquire knowledge, the ability to be guided in a research field, and the devel-
opment of critical thinking. The use of projects is always focused on the independent activity
of students, individual, pair or group, which pupils carry out during a certain interval of time.
Individual and collective works are combined in such a manner that allows everyone to make
a contribution to that area in which they are strong, and on the contrary, to use the support and
help of others as required. Participation in the project may give students a real reason to use
a foreign language. The use of projects always assumes that the solution of a problem will
provide, on the one hand, the use of various methods such as tutorials, and on the other the
integration of knowledge from various areas of a science, including techniques, technology and
The important part of the predesign is to gather the materials for design work and literature
analysis. Sometimes, it depends on the supervisor whether a pupil will manage to grasp research
methods: to explore encyclopaedic dictionaries, to use library catalogues and indexes, abstract
journals and to look through the periodical literature. The use of the Internet can give much
useful information too. All these sources allow collection of data on the level of scrutiny of
the point in question and its urgency, and acquaintance with the literature concerning the task,
study of the solution of similar problems and to estimation of the level and efficiency of their
solution. As a result in the report there can be historical inquiry on the problem, a statement of
the theory and much other material anticipating empirical research [1, 2].
Projects can be used for learning some basical laws of physics. As an example I suggest a
project which consists in research and modelling of atmospheric anomalies with soap bubbles.

The purpose of the project was to:

1. Study physical laws (optics and thermodynamics) using the soap bubbles;

2. Use interdisciplinary communications (geography, mathematics, meteorology, chemistry);

3. Activate the interest of students;

4. Train in modelling;

5. Work in small groups;

6. Use video- and photo- equipment, computer technology etc.

The urgency of forecasting the origin and character of atmospherical anomalies is driven by the
fact that hurricanes lead to many human victims and enormous economic losses.
The natural abnormal phenomena, storms, typhoons, hurricanes, El Nino and La Nina are inter-
connected, but prediction of them is almost impossible.
The main reason why it is so difficult to forecast the weather for a long time is atmospheric
turbulence. Size of the greatest cyclones is a thousand kilometers. The atmospheric depth,
in comparison with cyclones, is very small, only 40 kilometers. For this reason progress of a
cyclone is a two dimensional hydrodynamic problem.
The derivation of turbulence from the hydrodynamics equations has become one of the Seven
Millennium Prize Problems.
Modern conditions in this area of science are the following:

1. Some results on the detection of the causes of hurricanes’ causes have been drawn;

2. Experiment does not provide a mathematical method by which one could describe the
origin of a hurricane.

3. The equations of hydrodynamics describing a climate are so difficult that they have not
been solved.

Tropical cyclones appear essentially in the intratropical convergent zone above an overheated
ocean surface. For example, figures 1 and 2 show photographs of typhoons made from satellites.
The conditions of the origin of a hurricane are well known:

1. The first factor is a high temperature (more than 26 o C). It causes an evaporation of the
ocean surface and as a result a satiation vortex.

Figure 1: Typhoons Saomai and Bopha approaching China, 2006.

Figure 2: The “eye” of the typhoon Elena, 1985.

2. The second condition is a small velocity differential over height of the hurricane. This
does not allow the vortex to breakup.

3. Some attendant factors are known: a sharp temperature contrast on the surface of the
ocean, congestion of cumulus clouds etc.

There are also correlations of hurricanes with other weather phenomena such as wind circulation
in the stratosphere, rains in Western Africa and El Nino.
A hurricane (from the name of Indian God of wind Huracan) is a free-running thermodynamic
system, in which there are two thermal levels: high (temperature of the ocean) and low (tem-
perature of the highest tropospheric layer). The heat carrier in such system is a water stream.
Energy is scooped from the thermal energy of the ocean and potential energy of the high-rise
instability of the atmosphere, passing to kinetic energy of a whirlwind. While a hurricane is
over the ocean the power continues to increase. But on the land the link with the source of
energy vanishes and hurricane subsides, wrecking everything around.
There are hundreds of theoretical methods of hurricane suppression and management of its
trajectory. A lot of them have been checked in practice.
The interconnection with hurricanes and the El Nino phenomenon was realized only in 1970.
El Nino is an intersification of the monthly or seasonal fluctuations in the air pressure difference
between Tahiti and Darwin (Australia), caused by the warming of the surface waters of the
tropical Eastern Pacific Ocean that occurs each three to eight years.
The mechanisms which sustain the El Nino cycle remain a matter of research. But it is asso-
ciated with disruption of the Pacific trade winds and a stronger than usual so-called Madden-
Julian oscillation, which is a frequent and regularly occurring eastward progression of tropical
rainfalls over the Pacific.
Methods for prediction of El Nino include the analysis of such parameters as:

1. The number of storm days in the northwestern part of the Pacific ocean in the typhoon
seasons. Their number before El Nino are double these in a calm season.

2. The quantity of strong typhoons: their coordinates and time are observed. Figure 3 depicts
the hurricanes for the last fifteen years.

3. The areas and magnitudes of the positive thermal anomaly of a surface of ocean at 40 -
45 degrees of northern latitude;

4. The Indices of southern fluctuations (differences of atmospheric pressure between Darwin

and Tahiti).

The first El Nino prediction the year before its start was very successful. Of course, for final
estimates numerous statistics for such predictions are required, which will take scores of years
Nowadays a number of physical laboratories in the USA, India, France and China studies turbu-
lent flows in soap films. In this way they model the processes going in atmosphere of the Earth,
for example, the formation of cyclones.
In the laboratory it is difficult to model the beginning of hurricanes, because of their huge
sizes. Therefore the French scientists have proposed original experiments with soap bubbles in

Figure 3: Points correspond to cyclone positions in 6 hour intervals.

their work “Thermal Convection and Emergence of Isolated Vortices in Soap Bubbles”, almost
equivalent to replacing these atmospheric phenomena. A soap bubble is a perfect model for
turbulence in the gas layers of planets since such physical parameter as the ratio of the thickness
of a soap film to the diameter of a bubble is equivalent to the ratio of the thickness of the
atmosphere to the diameter of the planet [4].
The method of the experiment was very simple. Half of the soap bubble, which was at room
temperature of about 20 o C, and with radius in different variations of experiment from 8 to
10 cm, was heated up by special ring covering the equator of the bubble. This half of wire
was connected to a alternating voltage source (about 5V). Heat propagated by convection from
equator to poles, producing a difference of temperatures ∆T . Students irradiated the bubble
by white light. They observed an interference picture from which it seemed that at the largest
difference of temperatures between equator and pole, a whirlwind similar to an atmospheric
cyclone was originating.
As was mentioned above, students were divided in several workgroups. There was a group
which collected information about this problem. Others found and prepared the ingredients
necessary for strong soap solution, tried to take photographs of turbulent flows on the bubble
surface and so on.
The scheme of the experiment is presented below.
Students should notice the fact that changing of poles of the power supply causes a change of
direction of turbulent whirlwind motion. Also they must observe that both factors, a difference
of temperatures and current magnetic field, give rise to these flows.
One of these groups can research how modern meteorological systems watch for weather changes
all over the world.
At the end of the project, students present their results in presentations or posters.

The use of projects is powerful tool to attract students to science. As a result of applying
this method to modelling of atmospheric phenomena using the soap bubbles a high interest

Figure 4: Principle scheme.

of students in science was stimulated. The method assumes possession of a certain sum of
knowledge, providing the solution of a variety of problems through a design activity. Practical
application of the acquired knowledge was demonstrated.




[2] "   #

"  %  % &. / #


[3] &  +

-  ./'  0/' 1.  %  23  +
/ 6  2
[4] Thermal Convection and Emergence of Isolated Vortices in Soap Bubbles [Electronic Resource] / F.
Seychelles, Y. Amarouchene, M. Bessafi, H. Kellay // Physical review letters. - 2008 - #4 - Access mode:

Onne van Buuren, Peter Uylings & Ton Ellermeijer
AMSTEL Institute, University of Amsterdam, The Netherlands

Although the use of computer modelling and of measurement with video and sen-
sors in teaching is promising, the results are not always as good as expected. More
insight is needed into the way students can develop an understanding of mathe-
matical and physical concepts and methods by means of modelling. Our aim is to
develop a learning path on which students not only can learn how to model, but also
develop such an understanding by a combination of modelling and experimentation.
Three pilot studies were carried out with upper level students with a focus on the
problems they encountered in a variety of modelling tasks. In a fourth pilot, lower
level students (13-14 years) studied different representations of movement by com-
paring video with model-driven animation. This combination appeared to stimulate
learning, as it acted as a mechanism for self-correction.

Modelling is a key activity in scientific enquiry and engineering. Therefore it might be ex-
pected that in education explicit attention should be paid to it. According to Hestenes, it should
be the central theme of physics instruction (Hestenes, 1987, 1992). But until recently, explicit
attention to models and modelling in educational practice has been rare, at least in the Nether-
lands (Lijnse, 2006). During the last decades of the twentieth century however, educators in
science and mathematics started to realise that modelling on computers could be useful in ed-
ucation. Educational researchers began to make suggestions in favour of computer modelling
as a learning activity. Within these suggestions, three categories can be distinguished (Löhner,
2005). The first category concerns learning about modelling. Students should be introduced to
modelling because it is a professional activity of scientists and engineers. They have to learn
about the possibilities and limitations of computer models and should learn how to make and
use models themselves. The second category concerns an expected improvement of scientific
reasoning skills. Modelling might encourage higher level skills, like predicting, developing hy-
potheses, testing hypotheses, analytical reasoning and explaining. The third category concerns
learning domain content by modelling. Because the computer can deal with a great part of the
mathematical difficulties, students should no longer be confined by their mathematical capabil-
ities. This should make it possible to study more realistic subjects and to solve problems that
would be too complex otherwise.
Almost without saying, our focus has shifted from modelling in general towards the type of
computer modelling mostly used in physics courses. This type of modelling is often referred

to as ‘dynamical modelling’. From a mathematical point of view, this mainly concerns the nu-
merical solving of differential equations, especially initial value problems and boundary value
problems. In this study, the focus will be on this type of modelling.
The importance of computer modelling has now been recognized in many countries, but the
integration of modelling in curricula is still rather poor. Until recently, Dutch secondary stu-
dents were not involved in modelling activities until they reached the highest level of secondary
education (age 17-18, pre-university level). Even at this level, computer modelling only occurs
at a modest scale in the curriculum. Now the situation is changing. New curricula are under
development for the upper levels of Dutch secondary education for all the sciences. Modelling
is meant to be part of all of these new curricula (Savelsbergh et al., 2008).
In spite of all promising suggestions, the educational results of modelling activities do not
always meet expectations. Students do not automatically connect models, experiments and
underlying theories and concepts with each other (Doerr, 1996; Hodson, 1993; Schecker, 2005).
Claims on positive learning effects are proven only occasionally (Hwang, 2006; Löhner, 2005).
But as the importance of modelling in education is clear, the question is not whether, but how we
should use modelling in teaching. We need to develop a learning path into modelling. Therefore
we need more insight into the way students can develop modelling skills and an understanding
of mathematical and physical concepts and methods by means of modelling (Hwang, 2006;
Lijnse, 2006).
A related question is at what age such a learning path may start. So far most research has been
focussed on the highest levels of secondary education and on modelling lessons on a project
base, not on modelling as an integrated part of a curriculum. As far as we know, little research
has been done on the learning of students with modelling activities extending over a longer
interval of time. Nevertheless, it might be advantageous to start with modelling at an earlier
age, based on the following assumptions:

1. It gives students more time to get acquainted with modelling.

2. More realistic problems can be tackled at an early stage of education.

3. Modelling makes it possible to study traditional subjects in a different, possibly more

effective way.

4. More students will be acquainted with modelling.

5. Modelling might turn out to be a more natural activity for students if they are introduced
to it in the initial stages of their science education, when they are not yet used to the
standard approaches of doing physics.

Possible disadvantages of starting with modelling at an early age may be, that

1. Modelling and modelling tools may be to abstract for young students. In that case not
only will the success rate be low, but an early negative experience may also hamper the
learning process at a higher age.

2. Students do not yet have the required skills and knowledge.

3. Modelling may have a negative influence on the development of traditional skills such
as calculating with formulae. Not only does modelling take time, which cannot be used
for traditional exercises but students even might become ‘spoiled’, because they might no
longer see the relevance of doing traditional calculations.

4. The situations that are to be modelled may be too complex for young students when too
many concepts are involved.

5. With computer models, an additional representation of physics will be added to the cur-
riculum. This may take extra time and can be an extra complication.

Some research has indicated that it is possible to involve students in quantitative modelling at
the age of 14-15 years (Mulder, Slooten, Uylings, & Wieberdink, 2008). Schwarz & White
(2005) did some research with seventh grade children (age 12-13 years), but their software was
domain specific and limited, as children had to express their ideas qualitatively by answering
multiple choice questions. Their answers were translated into a quantitative, hidden model,
used for simulation. Lawrence (2004) describes a strategy for modelling using VnR-software,
based on his experiences with 11-14 years-old children, but his approach is semi-quantitative.
We want to examine if starting a learning path into quantitative modelling in the Netherlands is
possible in the year in which students get their first lessons in physics. At most Dutch secondary
schools, this is the second year (age 13-14 years).


The question for us is which skills and knowledge are required for modelling. To answer this
question we will examine the modelling process more closely. The requirements proposed by
the commissions for renewal of the Dutch science curricula form a general description of the
modelling skills a science student must possess after finishing secondary school (see for exam-
ple: Commissie Vernieuwing Natuurkundeonderwijs havo/vwo, 2006). The proposal offers a
framework that can be useful for describing the modelling process:

“The student must be able to analyse a situation in a realistic context and reduce it
to a manageable problem, translate this into a model, generate outcomes, interpret
these outcomes, and test and evaluate the model.”

This has been represented schematically in figure 1. In this figure the modelling process is
represented as a sequence of six ‘states’ and six ‘transitions’ between states.
Similar descriptions can be found elsewhere, especially in research on mathematical modelling
(Blum & Leiss, 2005; Galbraith & Stillman, 2006; Maass, 2006). The process is often called
the modelling cycle, as for the evaluation of the process it is necessary to revisit the realistic
context. If the outcome does not prove to be appropriate, particular steps or even the entire
modelling process need to be revised.

Figure 1: Modelling process.

In practice, the process is far from linear and not unidirectional (Galbraith & Stillman, 2006;
Haines & Crouch, 2007). This is not surprising. As we will see below, during most transitions
knowledge of subsequent transitions and stages is required, and vice versa.
The advantages of the description are that it can serve as a map on which to put the required
skills and knowledge, and that it gives direction to the process, not only for the researchers, but
also for students (Maass, 2006).
Sometimes, the term ‘modelling’ is used in a narrow sense, limited to the translation of the
problem into a model and the generation of outcomes. We consider this approach to modelling
as ‘incomplete’, as it lacks the necessary links to reality. Our aim is ‘complete’ modelling, in
which the whole modelling process is involved.
We shall examine the stages and transitions of the process more closely below.


Without sufficient knowledge of the realistic context, neither analysis and reduction nor inter-
pretation, testing and evaluation will be possible. Students’ understanding of this situation can
be a major hurdle (Galbraith & Stillman, 2006; Hwang, 2006; Maass, 2006).
Therefore, this situation must first be studied. Doing experiments can be an appropriate way
to get a better understanding of the situation and to collect data on which analysis, reduction,
interpretation and evaluation can be based.


Some basic knowledge of the domain content is required or must be learned at the beginning
of a modelling task, because the situation must be analysed in terms of its possible applicable
theories and because variables, and relations between them, must be determined. Part of the
analysis can be an estimation of magnitudes of variables on which a proper reduction can be
based. Estimations and predictions can provide a base for evaluation during the whole process.
In order to reduce the situation to a “manageable” problem, the modeller must recognize the
problem as manageable. This requires some experience and sufficient skills relating to the
subsequent steps of the modelling process. This is congruent with the findings of Galbraith
& Stillman (2006) and Gellert, Jablonka, & Keitel (2001). They found that the mathematical
knowledge of the students and the knowledge of the capabilities of the technological tools at
their disposal have a direct impact on how these students tackle a problem.

For more advanced students, part of the reduction can be the splitting up of the problem into
partial problems, which can be solved subsequently (Lijnse, 2006).


For the translation of the problem into a computer model, students must be able to manipulate
the relations between variables, in order to adapt them for the model. They must understand
some principles of numerical modelling, e.g. iteration processes. They must know how to use
their tools and how to build models. It is recommended that they learn to recognize standard
patterns in order to use these patterns as building blocks (Lawrence, 2006; Lijnse, 2006). In
the case of graphical modelling tools based on the stock and flow approach developed by J.W.
Forrester in the early 1960s, such as Stella or PowerSim, some understanding of this approach
is required. The software we have used, Coach 6, offers a mode based on this approach (Heck,
Kedzierska, & Ellermeijer, 2009).


Without interpretation, the model cannot be tested or evaluated. The output of a model usually
consists of tables and graphs. As graphs provide the best overview, we consider the ability to
read and interpret them as one of the most important skills.
In figure 1, evaluation is positioned at the end of the process. However, in practice competent
modellers constantly evaluate intermediate result during all stages of the process. The inability
to do so can form a blockage for students (Galbraith & Stillman, 2006).
When evaluating a model, a modeller must compare model outcomes with their counterparts not
only in the realistic world, but also in the ‘manageable problem’. The values of key variables
often will be somewhat different in these different stages, as a result of simplifications and the
numerical method. There can be even more different values corresponding to each variable
when the model is developed in multiple cycles with an increasing level of complexity. For
an appropriate evaluation, students must be able to keep all approximations and corresponding
values of variables apart.


More can be said about the skills and knowledge required for modelling. For example, for most
skills, levels of increasing abstraction and difficulty can be distinguished. Much is yet to be
uncovered. The description given above provides us with a first overview.
Summarizing, we conclude that basic skills and knowledge for modelling are: knowledge of the
realistic situation, some content knowledge and the skills to use this knowledge, some knowl-
edge of the modelling tool, and the ability to interpret graphs.

Our final goal is the development of a learning path into modelling. As mentioned above,
students do not automatically connect reality, experiments, models and underlying theories and
concepts with each other. Therefore, our first research question is:

How can a combination of measuring and modelling help to bridge the gap between
realistic contexts and the relevant mathematical and physical concepts and methods
in school practice?

The development of basic skills and knowledge is part of a learning path. Related research
questions are:

• What is a good sequence of steps on a learning path to modelling?

• What conditions must be fulfilled for each step?

• What problems and what new opportunities for learning are encountered by students and
teachers on a learning path to modelling?


We carried out four pilot studies. Three of them involved upper grade students. These pilots will
be summarized below. The fourth involved lower grade students. This pilot will be described
in more detail in section 5. As an educational tool we mainly used Coach 6. This integrated
computer working environment is used at the majority of Dutch secondary schools. It integrates
the facilities required for doing experiments with sensors, video analysis, modelling and model
driven animation, thereby making it possible to combine these activities (Heck et al., 2009).


In this pilot we observed two upper grade students (age 16-17 years) with little experience of
modelling, doing a multiple-cycle modelling research project (60 hours) on celestial mechanics.
We wanted to investigate what blockages occurred. Blockages reported by the students were
analysed and discussed.
We encountered 4 kinds of potential blockages:

• Blockages caused by a lack of knowledge of the realistic situation. Students were not
able to evaluate some of their results because they could not perceive if their results were

• Blockages caused by misconceptions about the realistic situation. In some cases students
did not perceive that their outcomes were correct.

• Blockages caused by a lack of experience with debugging. Especially in cases where two
or more bugs occurred simultaneously, debugging turned out to be very difficult.

• Blockages that had to do with the differences in perspective between traditional teaching
and modelling. Models are mainly built on fundamental equations, such as the definitions
of velocity and acceleration, and Newton’s second law. In traditional teaching, much
emphasis is put on solutions of fundamental equations for special cases, such as h = 12 gt2 ,

which is only valid for free fall with zero initial velocity in a constant gravitational field.
Another example is the supposed circularity of planetary orbits. The students tended
to use these solutions as laws on which the model was to be built, instead of using the
fundamental equations.

The first two of these kinds of blockages stress the importance of a careful introduction of the
realistic situation. For the third, explicit attention must be paid to a learning path. The last
blockage might be overcome by an early introduction of modelling, prior to or simultaneously
with exact solutions for special cases.


In this pilot we involved a class of 16 upper grade students (16-18 years) in parts of the mod-
elling process. Most of them had not been involved in modelling activities before. The lessons
were part of an experimental course on weather and climate. The pilot concerned the first chap-
ter, in which the pressure in the atmosphere was modelled in a few stages with an increasing
level of complexity. In the last stage, the students had to modify an existing computer model,
after an introduction of the iteration process, in which a few iterations had to be calculated by
hand. Partial research questions were:

1. Can these students understand the iteration process?

2. Can these students separate the effects on key parameters of the different approximations
and assumptions connected to successive steps in the modelling process?

From classroom observation, analysis of their exercise-books, and a questionnaire among the
students, it appeared that most students were able to understand the iteration process. However,
from an analysis of their exercise-books, and a test, it appeared that most of them had great
difficulty in separating the effects of the different approximations and assumptions. This may
be considered as a cause for blockages during evaluation.


The third pilot was with the same group of students as the second, and the lessons were part
of the same series of lessons. A given graphical model, for the temperature of the surface of
a square meter of the Earth during a day, was used as a tool for simulation. The graph for
the surface temperature was to be matched to experimental data, provided in written material.
Therefore some of the model parameters had to be adjusted. The model output consisted both
of the graph of the surface temperature and graphs of two underlying quantities.
Partial research questions were

1. What do students learn about the underlying model from a simulation in which parameters
must be adjusted?

2. How do students use the model and the underlying quantities in their reasoning?
3. How do students use the realistic context in their reasoning?

From a post-test it appeared that most students were able to reproduce the main structure of
the model. They also succeeded in describing qualitatively the influence of a change of the key
parameter on the target, the graph of the surface temperature. However, most of them failed in
describing the influence of this key parameter on underlying quantities, and in describing the
effects of other parameters on the target. Apparently, they had not learned much more than the
direct influence of the key parameter on the graphs they had to match, in spite of the fact that
the entire model had been introduced to them in the preceding lessons.
From classroom observations, screen recordings and sound recordings, it appeared that during
the task students did not use the underlying model for reasoning and did not pay attention to the
graphs of underlying quantities.
It also appeared that few students used the realistic context in their reasoning. Even more,
some students were seen adapting their view on the realistic context instead of questioning their
model input after making some mistakes. Apparently, an orientation on the realistic situation
by reading about it is too weak for these students.




Research questions for this pilot study are

• Is it possible to involve students of 13-14 years into complete modelling?

• Can a combination of video analysis and model-driven animation stimulate students to
interpret and evaluate their models?
• Can this combination stimulate students to use or acquire new domain skills?
• What differences in behaviour are there between well skilled and poorly skilled students?


5.2.1 Choice of a starting point for a learning path. In accordance with section 2, in order
to make a start on a learning path on modelling, students must at least be able to interpret
and evaluate graphs and they must know how to use formulae. At the school at which we
planned this pilot project, these skills are usually taught in the middle of the year. During a first
course on kinematics, graphs of distance versus time (x,t-graphs), graphs of velocity versus time
(v,t-graphs) and the formula relating mean velocity to displacement and time are introduced.
Therefore, this seems to be an adequate starting point.

5.2.2 Research method. As we wanted to integrate modelling into the curriculum, we devel-
oped a series of lessons on kinematics for this purpose. These lessons were to be tested and
are to be improved in subsequent cycles. This approach can be indicated as educational design
research (van den Akker, Gravemeijer, McKenney, & Nieveen, 2006; Plomp & Nieveen, 2009).
We tested the materials in two different groups of 30 students each. After testing in the first
group, it was possible to adapt part of the materials for the second group. Therefore we went
through two research cycles.

5.2.3 Design of the lessons The 5 lessons of 80 minutes each consisted of the following

1. Reading & interpreting x,t-graphs

2. Formula for mean velocity

3. Video analysis: start of a runner

4. Reading & interpreting v,t-graphs

5. Combining video analysis & model-driven animation

In the first two lessons, basic skills were introduced. In the third lesson, the realistic context
was introduced by means of a video which was to be analysed by the students. Also, in this
lesson Coach was introduced. The fourth lesson was meant for basic skills (v,t-graphs). In the
fifth lesson modelling was introduced. The design of the third and fifth lesson will be described
in more detail below.

5.2.4 Design of the third lesson: video analysis As a context we chose the start of a runner.
We used an already existing video, not only for logistical reasons, but also because this made it
possible to adapt the modelling activity and the video measurements to each other, as we shall
see later.
In this lesson Coach had to be introduced. Whereas higher grade students usually do not have
problems working with Coach for the first time, younger students can have some problems if too
many new features have to be learned. Therefore, we introduced only the features we considered
most important. These were: starting the measurement; measuring by clicking on the target in
the video; scaling the video; scaling diagrams; scanning data in a graph; and saving results.
We consider scaling important, because scaling confronts students with real magnitudes in the
real situation. In addition, a wrong scale can make effects in graphs invisible. Coach offers
the option to scale graphs automatically, but if data are not correct, automatic scaling can put
students on a wrong track.
All instructions were on screen, and instructional pictures were added (see figure 2). Because
the results of this approach were poor, as we shall see later, we replaced these instructions by a
short (2 minutes) instructional video.

Figure 2: Video analysis in Coach.

The measurements of the students resulted in an x,t-graph. The students had to study qualita-
tively the relation between the velocity of the runner and the slope of the graph. They also had
to determine quantitatively the mean velocity of the runner over different intervals of time.
5.2.5 Design of the fifth lesson, combining video analysis & model-driven animation
Set-up of the lesson
In the last lesson, modelling was introduced. The start of the video runner was to be modelled
in a simple way. The model should have the same starting place and the same mean velocity as
the video runner. The students had to determine the right values for these parameters from the
x,t-graph of the video runner, just as they had done in the lesson on video-analysis.
The targets of this lesson were: introduction of modelling, improvement of understanding of
graphs, and clarification of the difference between the concepts of mean velocity and velocity
as a function of time. Students often have difficulty with these concepts, as could be seen in
tests in the years before this pilot. Even higher grade students often make mistakes with these
concepts. One cause for this is the confusion of distance with position. Therefore, we did not
use x = 0 as the starting place of the runners.
As we wanted to put emphasis on interpretation and evaluation, the model drove an animation
which we wanted to run simultaneously with the video. Because of problems with timing, this

Figure 3: Screenshot during the run.

was not possible. Instead, we used the data of the video as input for a second model, which
drove a second animation. Now, we had two models which ran simultaneously. We told the
students about this trick, but the second model itself was kept hidden. While the models were
running, x,t-graphs and v,t-graphs were drawn. In this way, the runs of both the video and
the model were represented in three ways. A screenshot, made during the run, can be seen in
figure 3.
By means of questions and tasks the students were invited to compare the video and model in
each representation, and to compare different representations with each other. A fourth rep-
resentation was offered by the formula, which the students had to use to determine the mean
velocity of the video runner. This velocity had to be entered into the model. The run of the
model provided them with feedback on the results of their calculations.
Choice of the modelling mode
Coach offers three modes for constructing and viewing models: the graphical mode, the equa-
tions mode and the text mode. Only the graphical mode permitted us to drive an animation
while keeping the second model hidden. A disadvantage of this mode at this stage might be
that it would take considerable time and effort to explain the stocks and flows approach to our
students, especially because this approach is quite abstract in the case of kinematics. Because
no numerical knowledge is needed to understand this simple model, forcing students to learn

the numerical method would not stimulate them. Therefore, we decided not to explain the nu-
merical method yet. Instead, we offered a ready-made model, in which the initial values had to
be adapted. This model can be seen in the upper left corner of figure 3. Although the model was
used as a simulation, in this case the difference between simulating and complete modelling is
rather small, because the students still had to use the basic relation for the model in order to
determine the mean velocity. An advantage of our approach is that it provides the students with
a first orientation on graphical modelling. We used an instructional video of a few minutes in
order to introduce our students to the required features.
Extra activity
Finally, an extra activity was added for the fastest students. In this activity, students were asked
to change the model. We let them use an if-then condition for this purpose, as their mathematical
skills were not suited to other ways of changing the model. By means of an if-then condition the
movement of the runner can be split into two parts, each with a different velocity. The students
were asked to keep the distance between the two runners as small as possible.


We did classroom observations during all lessons and analysed a number of exercise books
and some of the questions in the post-test. During the fifth lesson, six pairs of students were
observed more closely. In addition, audio recordings were made.


5.4.1. Results of the third lesson, on video analysis For organisational reasons the lesson
on video analysis in the first of the two groups was started earlier, halfway through the second
lesson. This first group appeared to have great difficulties using features of Coach. Much
scaffolding was required and students did not finish this activity in time. The main reason
appeared to be poor reading. After this lesson, we decided to make an instructional video in
which the use of the software was demonstrated. Attending this video took only two minutes.
We showed it at the start of the next lesson. This video solved most of our problems. Students
who still had difficulties after seeing the video could easily watch it again. Students became far
more enthusiastic about the activity.
Most students were able to answer the qualitative questions about the relation between the
slope of the x,t-graph and the velocity of the runner. The determination of mean velocities over
different intervals of time proved to be more difficult.

5.4.2. Results of the fifth lesson, combination of video-analysis and model-driven anima-
tion Most students were able to finish the greatest part of the activity in time. After watching
the instructional videos, students had few problems using the modelling environment. They
clearly favoured video instructions above written ones. However, even with video instruction,
we observed that students who do not like to read encountered more blockages and used more
trial and error. On the other hand, some of these students discovered and used more features of

In the first group, not being able to use the formula appeared to be a major reason for block-
ages. Therefore, in the second group, the students were briefly reminded of the formula at the
beginning of the lesson. We observed fewer blockages in this group.
The activity stimulated students to interpret and evaluate their model. The combination of
animation and video acted as a debugging tool. Students still made mistakes in determining the
mean velocity, but they clearly detected these and other mistakes by watching the animations
and the x,t-graphs, and they were able to correct these mistakes. Therefore we conclude that the
combination of video and model-driven animation can act as a mechanism for self-correction.
However, the ways students corrected their errors were different. Some of them discovered and
understood the cause of their errors themselves, other students invoked the help of the teacher,
but others corrected their errors by trial and error.
The activity stimulated students to use or acquire new domain skills. We observed students

• comparing representations in order to get a better understanding of the more abstract ones;

• discussing different representations of the same movement;

• making and testing hypotheses about different representations;

• getting a better understanding of the relation between velocity and slope of the x,t-graph
by experimenting with different values of the velocity.

Differences were seen in behaviour between well skilled and poorly skilled students. More
poorly skilled students more often made use of the animations and their own experience with
running. They seemed to avoid the more abstract and new representations. They also made more
use of trial and error. After making a mistake, better skilled students still used all representations
in order to correct their error, whereas poorer skilled students showed some regression to the
less abstract representations.
Many students worked quickly enough to accomplish the extra task. These students were able
to use the if-then condition. However, as far as we could observe, they mainly used trial and
error to fulfil the task. This is not surprising, as this is actually an efficient method for this
Test results on qualitative reasoning with both x,t-graphs and v,t-graphs were good. 76% of the
students scored adequately on questions. 40% of the students could determine a mean velocity
from an x,t-graph. As this was a major hurdle in preceding years, this can be considered an
improvement. However, this number dropped to 19% when the students were offered a choice
between an x,t-graph and a v,t-graph from which to determine the mean velocity. 17% chose
the x,t-graph but did not correct for the starting position. 40% of the students tried to use the
v,t-graph. Among this last group, three different types of answers were given. 12% used the
velocity at the end of the v,t-graph, 10% divided the end velocity by the time, and 19% took an
average of velocities from the v,t-graph.


Our main result is that it is possible to involve students in this modelling activity. The combina-
tion of video and model can be used to acquire new domain skills and does stimulate students
to interpret and evaluate their model more than video analysis alone. The combination acts as
a mechanism for self-correction. However, attention must be paid to the way students correct
Students still had difficulty in determining a mean velocity when offered a choice between
an x,t-graph and a v,t-graph. This might be because we did not pay explicit attention to the
relation between mean velocity and v,t-graphs in our modelling lesson, and the paper lesson
on v,t-graphs was near the end of the course, so probably not all students had enough time to
understand it sufficiently. In the next version of the course, more advantage can be taken of the
mechanism for self-correction to improve students’ understanding of v,t-graphs.
A question is: is the different behaviour of more poorly skilled students is caused by a lack
of knowledge? We observed students who, having the formula written down and filled-in, still
hesitated to use their results in the model. In many cases, encouraging students to use new
representations was enough. Therefore, the different behaviour may also be caused by a lack of
confidence or a lack of experience. Another question is: who learned the most from the activity,
the better-skilled students of the more poorly skilled ones?

To our main research question, how can a combination of measuring and modelling help to
bridge the gap between realistic contexts and the relevant mathematical and physical concepts
and methods in school practice, we can say that, in the case of video-analysis and modelling,
this combination stimulates students to compare the results of the measurements in a realistic
context with the outcomes of the model, thereby acting as a mechanism for self-correction.
Regarding the sequence of steps and the conditions to be fulfilled for each step on a modelling
path, we can say the following:

• The realistic situation that is to be modelled has to be introduced carefully before, or at

the beginning of the activity, because a lack of knowledge or misconceptions about it
can cause blockages. An introduction by means of written material only is possibly not
• A lack of basic domain knowledge and skills can be a cause for blockages. At the begin-
ning of a learning path this knowledge and these skills must be developed to a sufficient
level before the modelling activity commences.
• The integration of modelling into the curriculum may have consequences for the way
physics is taught, because modelling needs more emphasis on fundamental equations
instead of emphasis on exact solutions for special cases.

To the learning problems encountered by students and teachers, much has already been said
above. We add to this:

• A lack of debugging skills can cause blockages;

• Students may have difficulty in separating the effects of different approximations and
assumptions on key variables.

van den Akker, J., Gravemeijer, K., McKenney, S., & Nieveen, N. (Eds.). (2006). Educational Design
Research. London and New York: Routledge.
Blum, W., & Leiss, D. (2005). How do students and teachers deal with modelling problems? In C.
Haines, P. Galbraith, W. Blum, & S. Khan (Eds.), Mathematical Modelling (ICTMA 12): Education,
Engineering and Economics. Presented at the ICTMA 12, London: Horwood Publishing Limited.
Commissie Vernieuwing Natuurkundeonderwijs havo/vwo. (2006). Concept Examenprogramma natu-
urkunde havo. Retrieved from
Doerr, H. M. (1996). Stella ten years later: A review of the literature. International Journal of Computers
for Mathematical Learning, 1(2), 201-224.
Galbraith, P., & Stillman, G. (2006). A framework for identifying student blockages during transitions
in the modelling process. Zentralblatt für Didaktik der Mathematik, 38(2), 143-162.
Gellert, U., Jablonka, E., & Keitel, C. (2001). Mathematical literacy and common sense. In H. Atweh, B.
Nebres, & H. Forgasz (Eds.), Sociocultural research on mathematics education (pp. 57-77). Routledge.
Haines, C., & Crouch, R. (2007). Remarks on a modelling cycle and interpretation of behaviours. Pre-
sented at the ICTMA-13, Bloomington, Indiana.
Heck, A., Kedzierska, E., & Ellermeijer, T. (2009). Design and implementation of an integrated computer
working environment for doing mathematics and science. Journal of Computers in Mathematics and
Science Teaching, 28(2), 147-161.
Hestenes, D. (1987). Toward a modeling theory of physics instruction. American Journal of Physics,
55(5), 440-454.
Hestenes, D. (1992). Modeling games in the Newtonian World. American Journal of Physics, 60(8),
Hodson, D. (1993). Re-thinking Old Ways: Towards A More Critical Approach To Practical Work In
School Science. Studies in Science Education, 22(1), 85.
Hwang, F. (2006). An Instruction Model for Modeling with simulations: How to help student build
their own model with simulations. In E. van den Berg, T. Ellermeijer, & O. Slooten (Eds.), Modelling
in Physics and Physics Education (pp. 91-100). Presented at the GIREP conference 2006, Amsterdam:
University of Amsterdam.
Lawrence, I. (2004). Modelling simply, without algebra: beyond the spreadsheet. Physics Education,
39, 281-288.
Lawrence, I. (2006). Challenges and Opportunities in computational modelling. In E. van den Berg, T.
Ellermeijer, & O. Slooten (Eds.), Modelling in Physics and Physics Education (pp. 137-148). Presented
at the GIREP conference 2006, Amsterdam: University of Amsterdam.

Lijnse, P. (2006). Models of / for Teaching Modeling. In E. van den Berg, T. Ellermeijer, & O. Slooten
(Eds.), Modelling in Physics and Physics Education (pp. 20-33). Presented at the GIREP conference
2006, Amsterdam: University of Amsterdam.
Löhner, S. (2005). Computer Based Modeling Tasks: the Role Of External Representation. University
of Amsterdam.
Maass, K. (2006). What are modelling competencies? Zentralblatt für Didaktik der Mathematik, 38(2).
Mulder, C., Slooten, O., Uylings, P., & Wieberdink, T. (2008). Starten met grafisch modelleren in klas 3.
Werkgroep Natuurkunde Didactiek. Retrieved from
Plomp, T., & Nieveen, N. (Eds.). (2009). An introduction to Educational Design Research. Enschede:
Savelsbergh, E., Drijvers, P., van de Giessen, C., Heck, A., Hooyman, K., Kruger, J., et al. (2008). Mod-
elleren en computermodellen in de ß-vakken: Advies aan de gezamenlijke ß-vernieuwingscommissies.
Schecker, H. (2005). Learning Mechanics with Modelling Systems. Presented at the Proceedings of the
Fifth International ESERA Conference on Contributions of Research to Enhancing Students’ Interest in
Learning Science, Barcelona. Retrieved from

Schwarz, C. V., & White, B. Y. (2005). Metamodeling Knowledge: Developing Students’ Understanding
of Scientific Modeling. Cognition and Instruction, 23(2), 165.


Sergej Faletič[1] & Gorazd Planinšič

University of Ljubljana, Faculty of mathematics and physics, Jadranska 19, 1000 Ljubljana, Slovenia

Liquid crystal displays (LCDs) emit linearly polarised light. In addition to light being
polarised, LCDs can emit coloured light the hue of which can be easily adjusted. We can
use this property of LCDs in several experiments with the help of the computer. In this
article we describe a set of experiments that can be done when birefringent material (in
our case adhesive tape) is placed between a monitor and a polarizer. We explore in detail
how changes in polarisation result in changes of colour and what we can measure in these


One of the properties of almost all liquid crystal displays (LCDs) is that they emit polarised light. This
is because the layer of liquid crystal in these displays is put between two polarizers due to the very
characteristics of liquid crystals and how they react to electric field (Kortemeyer, 2008). There are
special LCDs that emit light which is not linearly polarized, but those are rare.
When hearing LCD, one usually first thinks of computer or TV screens. LCDs are used in a large
variety of devices ranging from computer and TV screens to mobile phones, handheld computers, pocket
calculators, wrist watches, displays of several toys and many more (Kortemeyer, 2008). All of these
LCDs emit linearly polarised light.
When we talk about LCDs as computer screens, we have additional features available that we can use to
our advantage. Firstly, we can choose different colours. Many applications allow us to make the entire
monitor or part of it glow in a chosen colour. Secondly, with an LCD as computer monitor, we can
use custom applications to predict colours, perform calculations, draw, etc. all at the same time while
observing the phenomena.
Before we proceed, let us briefly consider how the computer screen produces colours. Coloured LCD
monitors mostly use an RGB (Red, Green, Blue) colour scheme. This means that all colours are produced
by additive mixing of different proportions of these three, so only these three are actual colours with well
defined wavelengths. For example, the yellow colour does not have a wavelength spectrum usual for a
yellow colour but is rather a mixture of wavelengths from the red and green spectra.


We can use any LCD to perform the basic experiments on light and polarisation (Fakhruddin, 2008).

Figure 1: Basic polarisation experiment that is usually done with two crossed polarizers. As shown in
the figure the same experiment can be performed with an LCD and one polarizer. a) The transmissive
direction of the polarizer is parallel to the monitor’s polarisation. b) The transmissive direction of the
polarizer is perpendicular to the monitor’s polarisation.

By using a lens (or dropping a droplet of water onto the screen), we can observe the colours that make
up each pixel and see that they are indeed only red green and blue.
We can perform the classic polarisation experiments with crossed polarizers (Figure 1). Since we already
have a polarised light source we only need one polarizer (the analyser) to see the effect.
If the analyser is positioned parallel to the monitor polarisation, almost all light is transmitted. If it is in
a perpendicular (crossed) position, no light is transmitted. With this experiment we can also determine
the polarisation direction of the LCD, if we know the polarisation direction of the analyser.
Next we can perform the well known experiment with three polarizers (Figure 2). In this experiment we
put a second polarizer between the monitor and the analyser.
If we rotate the middle polarizer (say for 45o ), some of the light is transmitted even if the analyser is in
the crossed position.
This is due to the fact that the polarizer in the middle transmits the projection of the original polarisation
onto its transmissive direction. Since this is not perpendicular to the analyser’s, some of the light is



With the LCD, we can perform some advanced experiments, some of which we will describe here and
are the focus of this article.
We repeat the previous experiment, but this time using a piece of plastic foil that we put between the
monitor and the polarizer (Figure 3) (Fakhruddin, 2008).

Figure 2: Pictures of a basic polarisation experiment with a polarizer between the source of the polarised
light and the analyser. Using an LCD we can perform it with only two polarizers instead of three. a) The
polarizer in the middle lets all light pass, but the analyser blocks it. b) The polarizer in rotated by 90o .
Now it blocks all the light itself. c) The polarizer in the middle is at a 45o angle with both the monitor’s
polarisation and the analyser. It lets some light pass and since this light is now polarised at a 45o angle,
some of it can also pass through the analyser.

Figure 3: Pictures of the advanced polarisation experiment with a piece of plastic foil between the
source of polarised light and the analyser. We can see that the foil behaves neither as glass, nor as a
polarizer. a) The foil is placed so that the analyser blocks all the light. b) The foil is rotated for 90o . The
analyser again blocks all the light. c) The foil is rotated by 45o . We see that now some of the light is
transmitted through the analyser. d) The analyser is rotated by 90o . Now it lets all the light directly from
the monitor pass, but the light that first passes through the foil is partially blocked.

We observe a similar effect. However, there are important differences. The plastic itself, in contrast to
the polarizer, does not alter the monitor light in any apparent way. No matter how we turn it, virtually all
light will be transmitted, much as if we used glass. Yet, if we use the plastic and the analyser together,
at some orientation of the plastic there is no effect; at others, however, the plastic allows some light to
be transmitted through a crossed polarizer or attenuated when passing through a parallel polarizer. This
does not happen with glass.
From this we could interpret the foil as having two perpendicular transmissive directions; therefore,
incident light polarised in any direction splits into two perpendicular components, something our eyes
would not notice. So we do not observe any effect with the foil alone. Now, what if the two components
were not in phase, but one would travel a little slower than the other? Then the resulting polarisation
would not be linear, but rather elliptical. We can see this if we simply model (figure 4) the result with
this assumption. In elliptically polarised light, the electric field vector describes an ellipse instead of a
line: again something our eyes would not notice.

Figure 4: Pictures showing the polarisation at 8 different depths inside a birefringent material. The
original polarisation is as shown at z = 0. z0 is the period thickness for the material and wavelength. At
every z = nz0 , where n = 1, 2, 3, . . . the pattern is repeated.

However, by actually modelling this behaviour, we can see that the ratio between the two axes of the
ellipse varies with distance travelled in the medium (figure 4). This means that the intensities of the
perpendicular components vary with the thickness of the plastic. So when light reaches the crossed
analyser its polarisation may have a component in the transmissive direction of the later: thus some light
may be transmitted.
From this we can see, that our model of plastic, that it has two polarisation axes and that light, polarised
along one of the axes travels at different velocity with respect to light polarised in the other direction,
can explain qualitatively the phenomenon we just observed.


In reality the results of the experiment as described so far are not exactly what we observe. We do not get
a grey shade in the intersecting region, but usually we get magenta-like or a cyan-like colour, depending
on the position of the analyser (Figure 5).
If we rotate the analyser by 90o , we get the complementary colour of whichever colour we got in the
original position, just as black and white are complementary when we use the analyser alone. The usual
colours we get are as mentioned above. But it is possible, as we will see later on, to achieve different
pairs of complementary colours.
To explain this we have to look into the real polarisation properties of plastic (Babič, Čepič, 2009). Plastic
foil, adhesive tape and many similar materials are birefringent. This means they have different optical
densities (indices of refraction) for two perpendicular polarisations, as suggested before. This gives rise
to another effect. The phase difference between the two perpendicular components of polarisation is

Figure 5: Pictures showing that when using plastic foil, we do not get shades of grey depending on
the position of the foil, but rather a cyan-like or magenta-like hue. The two colours are complementary,
meaning that together they form the original (white) colour. With different plastics we might get different
pairs of colours, though. a) The analyser is perpendicular to the monitor’s polarisation. If we put the
plastic foil between them, a magenta-like colour is transmitted through the analyser. b) If we rotate the
analyser by 90o , we get the complementary colour.

different for light with different wavelengths, which results in different colours having different elliptical
polarisations (different ratios between the two axes of the ellipse) when emerging from the material. We
can see this, too, by simply modelling the behaviour of the polarised light inside a birefringent material
for different wavelengths. The analyser lets pass only the projection of the ellipse onto its transmissive
direction, and as the ellipses are different for different wavelengths, their projections are also different.
Due to this it can easily happen that some colours are better transmitted through the analyser than others,
which results in observed changes of colours or their intensities. If we rotate the analyser for 90o , we
allow transmission of colours that were blocked before and block those that before were transmitted.
Therefore, we get the complementary colours.


By using adhesive tape instead of a plastic foil we can take this experiments one step further.
To explore the dependence of colours on the thickness of the material, one can use layers of adhesive
tape, (see Figure 6) (Babič, Čepič, 2009).
The practical reason is that we can stick the tape in different layers onto a glass surface. We can assume
pieces of tape to be of the same thickness. This allows us to explore what is the relation between the

Figure 6: The picture above shows the colours we observe if we put a strip of adhesive tape with varying
number of layers (one on the left through six on the right) between a white LCD and an analyser.

thickness of the material and the resulting colour, and further allows us to test our model on experimental
data. Again, if we rotate the analyser by 90o we get complementary colours.
To perform this experiment and analyze it does not require knowledge about birefringence. We can take
it as a purely phenomenological experiment. We can also try to produce our own model in order to
predict or test the results.


The advantage we have using a colour LCD is that we can choose colours, i.e. wavelengths, and explore
the phenomenon for each of the basic colours separately.
An interesting and not too complex task is to determine a characteristic parameter of the adhesive tape
that we chose to call period thickness. It turns out that when we observe layers of tape in each colour
separately we notice that at a certain number of layers the colour appears very close to the original and
the effect repeats periodically (see Figure 7). This implies that at certain thickness (or around it) the
situation is again the same as at the beginning (before the beam enters the material). We may call this
distance a period. And accordingly, we will call the respective thickness, period thickness.
We can also see that at certain numbers of layers we get a black colour, meaning the transmission is close
to zero. Occasionally these may be easier to spot than the layers with almost perfect transmission. We
may assume that these black hues occur at thicknesses exactly in the middle between two thicknesses of
perfect transmissivity. This may also be used to determine the period thickness.
We can estimate the period thicknesses for the three basic colours just by using logic and common sense.
When we use the red colour from LCD we see a picture as in Figure 7.

Figure 7: a) A screenshot of our JavaScript program that, assuming birefringence, from the estimated
period thicknesses calculates and displays the resulting colours. b) The picture shows the colours we see,
if we put layers of adhesive tape (1 layer at the top, increasing by one, to six at the bottom) in front of an
LCD glowing in red.

We see that the nearly perfect transmission occurs at approximately two layers and the pattern is also
repeated for approximately every two layers, giving the period thickness of around two layers. This is
just an estimate so far.
From the fact that the transmission at four and six layers is not as perfect as for two we can deduce either
that there are non-negligible losses in the material, or that the period thickness is not exactly two layers.
It turns out that losses are negligible, so the second explanation is the right one, but there will be a chance
to make small corrections later.
Looking at the green colour (Figure 8) we see that the almost perfect transmission does not occur until
five layers.
We might mistakenly think that the transmission at three layers is close to perfect, and therefore mistak-
enly think that three layers might be the period thickness, but the pattern before it is not repeated after
it. For a particular thickness to be the period thickness, the entire period (the pattern of colours between
two almost perfect transmissions) must also repeat. And this does not happen around three layers. On
the other hand, if we estimate the period thickness to be five layers, we see that the hue at six layers is
indeed very similar to the one at one layer, just as we would expect if the pattern would start to repeat
itself. Therefore, we can make a first estimate for the period thickness to be five layers. But, from the
results that we will get observing the blue colour (namely the fact that the period thickness for blue is
around three layers and green should be between red and blue, because green wavelengths are between
red and blue) we will see that this is probably two period thicknesses rather than one, giving a period
thickness of 2.5 layers which is also consistent with our first estimate, as 2.5 layers is quite close to the
deceiving three layers.
It may be also noticed that the yellowish colour at two layers is green with a little addition of red, while
the cyan-like colour at three layers is green with a little addition of blue, which is on the opposite side of
the spectrum than red with respect to green. This may further support our estimate, that at 2.5 layers we
would indeed get green alone.
We can also see that the green colour at five layers is indeed very close to the original green. This points
to the possibility that losses are negligibly small in the material and that the darker hue that we observed

Figure 8: a) A screenshot of our JavaScript program that, assuming birefringence, from the estimated
period thicknesses calculates and displays the resulting colours. b) The picture shows the colours we see,
if we put layers of adhesive tape (1 layer at the top, increasing by one, to six at the bottom) in front of an
LCD glowing in green.

for red at six layers is indeed due to the fact that the period thickness is probably not exactly two layers
rather than due to losses.
In our particular setting it was better to analyse the green colour after the blue, because of the clues we
mentioned above, but in general, all these conclusions are derived from observing all three colours and
comparing the results.
Looking at the blue colour (Figure 9), we see that almost perfect transmission occurs at three layers.
The pattern that should be repeated after this is not exactly the same, but the colour sequence appears to
be right, albeit a little darker. So it is safe to estimate that the period thickness for blue would be around
three layers.
From this we get clues about the period thickness for green that we mentioned before.
We have already noticed that the second series is darker than the first. Also, transmission at six layers,
which would be around two period thicknesses, is much poorer than it was for both red and green. We
already excluded the possibility of losses in the material, so we must assume that the approximation of
three layers carries an error that is even bigger than the one for red. But again, there will be a chance to
explore these corrections later.


As mentioned at the beginning, one advantage of using an LCD computer monitor for doing experiments
is that one can also simultaneously use all the other capabilities of the computer. In our case, we wrote a
short program (script) that takes as input the period thicknesses and intensities for each colour, calculates
the colours for each layer and displays them in a way that is easy to compare with what we observe in
the experiment (Figure 10). The script can help us check our model, check our estimates, and determine
the period thickness more accurately.
Taking as input both intensity and period thickness allows us to model the outcome both for the white
colour (figure 10) and for each of the basic colours separately (figures 7, 8 and 9). Furthermore, when

Figure 9: a) A screenshot of our JavaScript program that, assuming birefringence, from the estimated
period thicknesses calculates and displays the resulting colours. b) The picture shows the colours we see,
if we put layers of adhesive tape (1 layer at the top, increasing by one, to six at the bottom) in front of an
LCD glowing in blue.

Figure 10: a) The results of our short JavaScript application that displays the colours by calculating
them using as input the period thickness for each colour. The application is opened in a web browser and
can be used simultaneously with observing the colours through the polarizer. b) The colours as we see
them through the polarizer.

observing one colour, it also allows us to add a little of the other two colours to simulate its spectral
spread. The program is written in JavaScript, and embedded in an HTML web page, so that it can
be viewed with any web browser that supports JavaScript. This means it can be used on virtually any
platform that supports browsing the internet. Additionally we let the program display the calculated
values of red, green and blue for each colour.
We can use the program to compare the calculated colour shades with what we see and vary the period
thicknesses (program parameter) to achieve a result as close as possible to what we actually observe in
the experiment. These are the corrections we mentioned when observing the colours of the adhesive tape
for each of the three basic colours. In figure 10 it can be seen that instead of values 2 for red, 2.5 for green
and 3 for blue, we ended up with values 2.05 for red, 2.5 for green and 3.3 for blue. Just as expected,
the estimate for green was the best, since for green the colour at period thickness (actually, two period
thicknesses) was the closest to the original green. And again, just as expected, blue needed the biggest
correction (10%).

At this point, let us make a note on losses. One thing we have not considered so far is that losses
could depend on wavelength, but that would have shown with white colour, as the adhesive tape would
appear coloured, if it indeed absorbed different wavelengths differently. This does not happen, so we
may assume the losses to be the same for all wavelengths in the visual spectrum and, from what we
determined before, negligible.
We have observed earlier that we do not get clear red, green and blue hues for a red, green and blue
glowing screen respectively, but rather mixtures of all three of them. (The phenomenon is most prominent
with green.) The reason is that the spectra of the three basic colours vary greatly between different LCD
producers. Some are very sharp, while some are very spread (Artamonov, 2007). In the later case, the
spectrum of each colour might have tails in the regions of the other two, thus producing hues of all three

In this article we presented how the LCD can offer a wide range of new experiments. Its main advantages
are that it emits polarised coloured light.
If we choose a computer LCD monitor, we have the additional advantage of being able to use different
software at the same time and in the same view as the experiments. Moreover, we can set up e-learning
tasks. Either as part of distant learning projects or as an additional activity in traditional school settings.
There are various questions and problems we can pose: for example, how are colours dependent on the
tilt of the plastic (the angle between the planes of the plastic and the analyzer)? What happens, if we
bend the plastic? What happens if we tilt the polarizer or the plastic in different directions, and how can
we explain all this?
All of the above offers opportunities to explore phenomena, form hypotheses, test them and use previ-
ously acquired knowledge to form new ones.
We started this project from a simple wish to explain colours observed when a plastic foil is placed
between two polarizers. All the tasks described here derive from the questions we asked ourselves and
wanted answered to explain the phenomenon. In doing so we developed a strong so called ownership of

knowledge. We started from the knowledge that there are birefringent materials, in which the velocity
of light is different for two perpendicular polarisations. We derived all further implications from this
and checked if this phenomenon can explain our observations. This enabled us to go through the entire
procedure ourselves just as students might.
One interesting conclusion that we reached is that, for the experiments with colours from layers of
adhesive tape alone, assuming optical activity instead of birefringence yields the same results in the
simulation. This is due to the fact that both yield a cosine square dependence of transmitted intensity
on thickness. This can be used to show, how a wrong hypothesis can explain phenomena, until further
observations are available. In fact, the experiments with foil between the screen and the analyzer (figure
3) can not be explained with optical activity, but only with birefringence.

Artamonov, O. (2007), Contemporary LCD Monitor Parameters: Objective and Subjective Analysis,, Pages 15-16
Babič, V. & Čepič, M. (2009), Complementary colours for a physicist, European Journal of Physics, Vol.
30, Pages 793-806
Fakhruddin, H. (2008), Some Activities with Polarized Light from a Laptop LCD Screen, The Physics
Teacher, Vol. 46, Pages 229-231

Kortemeyer, G. (2008), A Polarizer Demo Using LCDs, The Physics Teacher, Vol. 46, Page 58


Cristina Mariani[1] , Federico Corni[2] , Tiziana Altiero[3] , Carlo Bortolotti[4], Enrico Giliberti[5] ,
Laura Landi[6] , Mauro Marchetti[7] & Alberto Martini[8]
[1] Istituto Comprensivo di Tione, Via Circonvallazione 44, I-38079 Tione (TN), Italy
[2] Department of Physics, University of Modena and Reggio Emilia, Via G.Campi 213/A, I-41125 Mod-
ena, Italy
[3] Department of Animal Biology, University of Modena and Reggio Emilia, Via Campi, 213/D, I-41125

Modena, Italy
[4] Department of Chemistry , University of Modena and Reggio Emilia, Via G. Campi, 183, I- 41125

Modena, Italy
[5] Social, Cognitive and Quantitative Sciences, University of Modena and Reggio Emilia, Viale A. Alle-

gri 9, I-42121 Reggio Emilia, Italy

[6] Primary School “Ca’ Bianca”, Via Cattalupa 1, I-42100 Reggio Emilia, Italy
[7] Department of Earth Sciences, University of Modena and Reggio Emilia, Largo Sant’Eufemia 19, I-

41100 Modena, Italy

[8] Laboratory Project “Laboratorio Galileiana”, Liceo Malpighi, Via Isaia 77, I-40122 Bologna, Italy


The project “Little scientists in the lab: Experiments & Models for science learn-
ing in primary school”, funded by the Ministry of Education and currently under
development and designing in Italy, is addressed to primary school teachers and to
pupils. It proposes a “Model-Centered Learning Environment” of pilot activities
for teaching and learning Physics and Sciences, based on experiments and mod-
elling. A website supports teachers’ school activities, facilitating and promoting
communication and exchanges of materials between teachers and researchers of
the University of Modena and Reggio Emilia, as well as between teachers them-
selves. In this paper, we illustrate the general features of the project and focus on
preliminary results of a training intervention for in-service teachers on fluids and


The project aims at designing and testing innovative ways of teaching Physics and Science
in primary schools. It promotes the exploration of curricular subjects employing elementary
models. Its pillars are teachers’ training and follow up activities, targeted on supporting lesson

Teachers’ training paths are designed to provide a model-oriented teaching and learning sys-
tem. They are based on laboratory activities, modeling and content-specific knowledge using a
cooperative learning method. Such an approach is inspired by the Karlsruhe Physikkurs (KPK)
(Hermann, 1995) and by the Continuum Physics Paradigm (Fuchs 1997a; 1997b; 1998). From
this point of view, we offer to primary school teachers an opportunity for training in Physics
focusing on elementary concepts and models (Hestenes 1997, Gilbert & Boulter 2000) and
constructing a shared language. These concepts are recurring in curricula and cross all scien-
tific disciplines. They are key to exploration and understanding. Teachers should then plan
and carry out their classroom teaching actions autonomously. A website and specific training
support their everyday activities to prepare didactical pathways. Experimental and multimedia
materials, which have been tested during the training courses, are available for teachers to be
used as teaching tools in class.
For in-class activities, the project supplies 3 integrated tools: i) animated stories to boost moti-
vation, with a character named Leo; ii) Leo’s case, containing a set of didactical tools to carry
out experiments under the teacher’s guidance; iii) modelling tools (software, cards, drawings,
role games) to promote thought and further discussion. These tools are integrated as follows:
the teacher shows in class one of Leo’s animated stories, then guides students to design and
perform experiments with the help of the materials included in the case, and finally leads the
students to discuss the experimental activities performed.
In the following chapters, we will describe, as an example, a training intervention for in-service
teachers on fluid and electric circuits. The key is presenting analogy to teachers as a way to
move from a familiar context, problem or domain, to a related unfamiliar one (Gentner D. &
Gentner D.R.,1983; Reeves L.M., Weisberg R.T. ;1993; 1994). Thus, we suggest, as other au-
thors have investigated, (Black & Solomon, 1987; Van den Berg & Grosheide, 1993; Cosgrove,
1995; Heywood & Parker, 1997; Paatz et al. 2004; Chiu & Lin, 2005) the use of analogy to
pass from the fluids context to electricity.
We will analyze teachers’ results in terms of the elementary models employed and difficul-
ties encountered. Finally, we will focus on some preliminary topics emerging from such an


Scientific knowledge must be actively built up by the learner (Driver et al. 1994). In this process,
the teacher’s role has important elements. One of these “is to introduce new ideas or cultural
tools where necessary and to provide the support and guidance for students to make sense of
these for themselves” (Driver, 1994). Research has investigated teachers’ elementary ideas on
Physics topics (Shipstone et al. 1988; Kruger, 1990; Kruger et al., 1992; Webb, 1992; Green-
wood 1996; Stocklmayer & Treagust, 1996; Atwood et al. 2001; Heywood & Parker,1997;
Testa & Michelini, 2006), showing that they often share the same alternative conceptions as
their students.
For these reasons, our training strategy is centered on putting teachers in the same situations (i.e.
laboratory and modelling activities) they could offer to their students (Pontecorvo et al., 1987;

Leinhard, 1988; Chaiklin & Lave, 1993). Teachers directly experience questions, difficulties
and urges of their students and can experience the effective advantage of training activities.
According to the literature, students’ alternative frameworks and reasoning schemes about elec-
tric circuits can be briefly catalogued as follows: current is consumed when passing through
a resistance (Tiberghien, 1984; Shipstone et al., 1988; McDermott & Schaffer, 1992); current
provided by a battery is independent of the circuit topology (Cohen et al. 1983; Shipstone et
al. 1988; McDermott. Et al., 1992); potential difference is confused with current or energy
defined as “strength” of a battery or “force of the current” (Psillos et al. 1988; Duit & von
Rhoneck, 1998); potential differences within a circuit depend on its topology (McDermott &
Shaffer, 1992); no current implies no potential difference (Cohen R. et al. 1983); parallel re-
sistances decreases circuit resistance (McDermott & Shaffer,1992); resistance considered only
as “obstacle” to current (Cohen et al. 1983); difficulties on Ohm relation (Liegeois & Mul-
let, 2002); local reasoning in analyzing a circuit (Shipstone et al. 1998; Duit & von Rhoneck,
1998); and sequential reasoning in analyzing a circuit (Duit & von Rhoneck, 1998). Moreover,
students often fail to understand that potential differences in a circuit depend on its topology
(McDermott L. C. & Shaffer P.S 1992); finally, one common idea among students is that across
an open switch there cannot be a potential difference because the current is zero (Cohenet et al.
1983). Finally, data in the literature support the fact that students (Duit & von Rhoneck, 1998)
as well as teachers (Testa & Michelini, 2006) focus their attention on one point in the circuit
and ignore what is happening elsewhere.
Keeping in mind these results, suggested by Physics education research, the project we are car-
rying out for teachers’ training would suggest and test some improved primary education paths,
in different specific content areas, to help teachers deal correctly with disciplinary contents, to
motivate them in teaching Physics and Science and to probe their difficulties.
Teachers’ training is structured along these steps:

• energy as regulator of natural phenomena, as a physical quantity that is conserved, i.e.

can be calculated by using a comprehensive set of rules (Feynman et al. 1963 pp4-1 to
4-80) and that can be stored and transferred from a system to another (KPK);

• extensive and intensive quantities: the difference of potential (dop) as the “driving force”
for the flux of an extensive quantity (treated as a “substance”) in natural processes;

• elementary concepts to study various processes: amount of moving substance, current of

substance, difference of potential, resistance and capacitance;

• interpretation, based on the analogy to fluids, of electric processes in terms of the elemen-
tary concepts.


A theoretical lesson provides the teachers with a shared language to describe the model of
energy carriers and exchangers. Such a shared language is applied to simple examples taken
from everyday life. Teachers, working in groups (Figure 1) are subsequently invited to explore

simple artifacts (hand torches, steam boats, solar vehicles, wind and hot mills, rockets fueled
by pumps or chemical reactions, solar moving animals, etc) and interpret them in terms of
energy, drawing the relative diagrams of energy flow. To support reasoning and communication,
teachers use cards or graphic software based on icons with combination rules coherent with the
approach of the course.

Figure 1: Teachers working in group, constructing energy models with cards or software.

As individual homework, teachers analyze complex processes in different scientific contexts

(photosynthesis, alimentary chains, water cycle, respiration, digestion, blood circulation etc.)
and build the corresponding energy flow diagrams.


The elementary models are introduced to teachers in three steps: execution of experiments with
a specific aim, discussion of similar experiments and collective workshops.
Teachers are assigned to 3 groups. Each group, organized into two subgroups, performs an
experimental subroutine concerning the same elementary concept in two different physical con-
texts: water and heat (Table 1).
The available equipments allow the experiments to be easily performed (Figure 2).
The experimental activities are guided by handouts to help teachers identifying the “substance”,
the independent variable, the dependent one, the parameters, how they can measure the flowing
substance, the relation between independent and dependent variables and what happens if the
parameters change.
After the experiments, each pair of subgroups meets to compare and discuss the subroutines.
This effort is aimed at identifying the elementary concepts of current and potential difference,
current and resistance, capacitance and potential difference, and their relationships, keeping the
other variables constant as parameters.
Then a plenary section enables the three groups to share their results and considerations, and
to identify and discuss the relationships between elementary concepts involving the three vari-
ables: Ohm’s law relating current, potential difference and resistance (Figure 3(a)), and the law
of capacitance relating potential difference, amount of substance and capacitance (Figure 3(b)).
A triangle helps teachers remember and combine the various quantities.

Groups Group 1 Group 2 Group 3
Elementary Experiments to intro- Experiments to intro- Experiments to intro-
concepts duce the potential dif- duce resistance and its duce capacitance and
ference and its relation- relationship with cur- its relationship with the
ship with current rent potential difference
Experiments 1(a) the amount of wa- 2(a) the amount of wa- 3(a) the level reached
in water ter flowing through a ter flowing as a function by the same amount of
context hole and a pipe to the of section and length water poured in cylin-
bottom of a cylinder as of a pipe connected to drical container of dif-
a function of the wa- a hole to the bottom ferent sections;
ter level in the cylinder of a cylinder (maintain-
during a certain time ing constant the water
period; level) during a certain
time period;
Experiments 1(b) the heat transfer to 2(b) the heat trans- 3(b) the temperature
in heat con- water placed on a ther- fer to water placed on reached by water in a
text mostat as a function of a thermostat during a container heated for a
the heater temperature time period as a func- certain time period as a
during a certain time tion of the contact area function of the quantity
period. and of the thickness of of water.
the layers of refractory
stone placed between
the container and the
Table 1: Experiments to introduce elementary models.



The hydraulic circuits consist of a source (reservoir filled by a pump) with three taps at different
heights, fans (flow-meters), vertical pipes open at the top that can be inserted along the circuit
(pressure probes), various tube connectors and a basin collector. Working air bubbles out of the
circuit is recommended.
The handout, that teachers follow, requires identification of the basic concepts: energy exchang-
ers, current, potential difference, resistance and capacitance in the hydraulic system. Then it
proposes 13 experimental situations with questions, including schematic drawings. For each
experiment the requirements are:

• to identify the parts of the schematic model and mark them with tags: T: energy exchang-
ers; I: Current; V high and V low potential difference, R: Resistance; C: Capacity;
• to write a prediction about the outcome of the experimental situation, supported by graph-
• to briefly motivate the prediction using the basic concepts;

(a) 1a (b) 1b (c) 2a

(d) 2b (e) 3a (f) 3b

Figure 2: Experimental setups for step 2 activities.

Figure 3: (Left) Triangle and relationships of the Ohm’s law. (Right) Triangle and relationships
of the capacitance law.

• to carry out the experiment;

• to mark the predictions that do not fit with the experimental observations using a pen
of a different color, noting what consideration it would have taken to make the correct

The first group of inquiries (1-4) is designed to help teachers recognizing and applying the
basic concepts to various elements of the circuit. The second group consists of experiments
on circuits with one resistor (5-7), two resistors in series (8-10) and two resistors in parallel
(11-13), in configurations of open and closed circuits (Table 2, first and second columns).



The electric circuits are made by a battery and bulbs linkable in various configurations. A
voltmeter is supplied to measure the potential difference. The handout for teachers supplies the
scheme of the electric circuit to be considered and, by analogy, the corresponding hydraulic
circuit. To promote the use of analogy as a means of explicit reasoning, the questions remain
the same as in the hydraulic case. The first question for the electric circuit corresponds to the
fifth step of the hydraulic circuit.
Each row of Table 2 shows the drawings of the hydraulic circuit and the analogous electrical
one, together with the question referring to the hydraulic circuit only.

HYDRAULIC CIRCUIT Description of the experiment ELECTRIC MODEL

Schematic drawing and question Schematic drawing
Flow of water from holes at dif-
ferent heights - Fill completely the
reservoir making sure that all taps
are closed. Open the three taps:
how are the jets?
Flow of water from holes and fans
at different heights - Connect three
fans to the three taps. Open the
three taps. How do the fans move?

Flow of water from holes and fans

connected through a tube to the
basin collector - Open the taps: how
do the fans move? Why?
As in the previous case, but now a
segment of one of the three tubes is
raised over the level of water in the
reservoir. How does the fan move?
Table 2 Continued on Next Page . . .

Table 2 – Continued
HYDRAULIC CIRCUIT Description of the experiment ELECTRIC MODEL
Schematic drawing and inquiry Schematic drawing
Connect a tube to a tap, then a verti-
cal pipe, a fan, another vertical pipe
and a tube connected to the basin
collector through a tap. Close the
tap of the reservoir and open the tap
of the basin collector. What will
happen? Why?

As in the previous configuration:

open the reservoir tap and close the
basin collector tap. What will hap-
pen? Why?

As in the previous configuration:

open both taps. What will happen?
What is the function of the vertical
pipe? Why?
Insert another fan and another verti-
cal pipe in series with the previous
circuit. Close the tap of the reser-
voir and open the tap of the basin
collector. What will happen? Why?

As in the previous configuration:

open the tap of the reservoir and
close the tap of the basin collector.
What will happen? Why?

As in the previous configuration:

open both taps. What will happen?
Connect a tube to a tap of the reser-
voir, then a vertical pipe, two fans
connected in parallel, another verti-
cal pipe and a tube connected to the
basin collector through a tap. Close
the tap of the reservoir and open the
tap of the basin collector. What will
happen? Why?
Table 2 Continued on Next Page . . .

Table 2 – Continued
HYDRAULIC CIRCUIT Description of the experiment ELECTRIC MODEL
Schematic drawing and inquiry Schematic drawing

As in the previous configuration:

open the tap of the reservoir and
close the tap of the basin collector.
What will happen? Why?

As in the previous configuration:

open both taps. What will happen?
Table 2: Questions referring to the hydraulic circuit only, and drawings of the hydraulic
and the analogous electrical circuit.

Teachers’ predictions, made before the experiments and the reasoning they claim to support
them, will be analyzed in the next section to investigate the role and the kind of models used to
interpret the circuit behaviors.



The results presented here are from about 90 in-service teachers (having different ages, years of
service, curricula) belonging to 15 different schools in three different cities: Modena, Reggio
nell’Emilia and Parma.
The first 4 questions have allowed teachers to begin focusing on the elementary concepts of
drop, current and resistance in the hydraulic context. For the first question, designed to intro-
duce dop and current, teachers presented 81% correct predictions (9 groups out of 11). Only
two groups with wrong estimates, make a wrong prediction for the second question too (2/11 =
Data analysis of the first question shows the following wrong prediction for the drop that how-
ever allow teachers to make correct predictions on the water jets: drop as due to

• the height of the reservoir (2/9 = 22%);

• the difference in level between taps (2/9 = 22%);

• the difference in level between the water level in the reservoir and in the basin collector
(3/9 = 33%);

• the difference in level between each tap and the water in the basin collector (2/9 = 22%).

In only two cases did participants make wrong predictions on the water jets. Teachers say
respectively that they will observe “the same jets because the pressure is equal at all points” and
that “increased flow (is) what comes out of the higher tap because the water pressure is greater”.
The second experiment, which differs from the first by an additional fan, aims at introducing
the concept of resistance. In this case, teachers suppose that the flow is slowed down (9/11 =
The two groups (2/11 =18%) that make a wrong prediction are the same that had not recognized
the existence of pressure drop. The group (1/11 = 9%) who stated that “the pressure is equal
in all points” draw the jets still in a similar manner to the test 1 (the jet leaving the higher tap
goes further) even though their justification is correct: “the pressure is greater at the bottom of
the reservoir”. The other group now correctly draws the jets, but predicts that the fastest fan is
placed on the highest tap. In this inquiry, the teachers reflect on the role of the fan, recognized
as resistance and at the same time as an indicator of current: the higher the current, the greater
the angular velocity of the blades of the fan (10/11 = 90%).
The third and fourth questions are proposed to reinforce the concepts previously introduced
through situations inducing cognitive conflict. In the third experimental situation 7 groups out
of 11 (64%) say that the tube, representing a resistance, slows the flow, 2 groups of 11 (18%)
say that it is irrelevant because the tubes are of the same section as the hole on the reservoir
(1/11 = 9%) or because “the tube is inserted after the fan, so that it does not change the fan
speed in relation to previous experiments”, (1/11 = 9%); 2 groups (2/11 = 18%) say that the fan
with the tube runs faster, but does not give any explanation.
In the fourth situation, the group predictions indicate that 6 groups (6/11 = 54%) focus only on
one part of the tube: the upward section (4/11 = 36%) that leads them to conclude (4/11= 36%
) that “lifting the tube, the fan slows down, due to the greater slope” or the downward section
(2/1 = 18%) that leads them to hypothesize that the fan runs faster, compared to the previous
situation, because more drop is now created.
The analysis of 58 forms from 5 to 13 on the hydraulic model provided by 11 groups, shows that
in almost all cases the predictions are phenomenological descriptions because teachers consider
the water level in the tubes (54/58 = 93%), the movement of the fan (21/58 = 36%) and water
circulation or lack thereof in the tubes (15/58 = 26%).
In the 5th and 6th questions, which refer to open circuits with a single fan and a closed tap, the
model of communicating vessels (32%) has prompted the teacher to estimate the water levels
achieved in the vertical pipes before and after the fan. Continuing with enquiries, teachers no
longer mention this model and try thinking in terms of constant potential or zero drop.
The concept of potential and potential difference has been recorded in 67% (39/58) of cases of
open and close circuits and in 89% of them (35/39) it is cited in the explanations. The current is
quoted in 22/58 = 38% of cases overall, considering both predictions and explanations and by
a small minority (2/22 = 9%) it is considered to be the only motivation. A deeper analysis of
the teachers’ responses mentioning the drop as cause of flux, shows that the drop is cited as the

only justification in 41% of cases (16/39), while in 15% of cases (6/39) it is cited as a cause of
current too, and in 28% of cases (11/39) it is cited along with the current. Anyway the current
is never considered the cause of pressure drops.
In closed circuits, the fan resistance is identified together with the drop as the variable that
determines the current (10/33 = 30%), while in very few cases (2/33 = 6%) an estimate of the
speed of rotation of the fan is made to indicate the intensity of the current.
In the hydraulic circuit with two fans in series and in parallel, the teachers discuss neither the fan
speed, nor the total resistance of the circuit. In almost all of these situations they do not write
predictions but prefer to present them only verbally. In any case, they constrain themselves to
consider only the water level in the vertical pipes or to assume that the two fans are running at
the same speed, without adding any comparison between circuits with a single resistor, in series
or in parallel.
In the following we report the few cases of wrong predictions. They are related to experiments
7 and 13. In experiment 7, related to a closed circuit, a group makes the prediction that the
water rises to the same height in the vertical pipes, although it is recognized that the fan turns
to the passage of current.
The two failed predictions, related to the parallel circuit (question 13), consider twice the overall
effect of the resistance compared to circuit with a single fan.
The analysis of teachers’ works on the electric model is made based on 81 answers to questions
5 to 13, provided by 9 groups.
Teachers’ predictions are focused on bulbs’ light turned on or off (59/81 = 73%), on potential
difference along the circuit (50/81 = 63%) and on current (21/81 = 26%).
A peculiar result of the electrical circuit is the presence of the drop only as prediction (50/77 =
65%), while current as motivation (37/58 = 64%). 50% of this 64% include explicitly current
as cause of drop. However, this situation needs to be further investigated because in many cases
teachers do not show a correct use of implication symbols or of logical connectives.
In closed circuits, the electrical resistance of the bulbs and their brightness are cited in 16/27 =
59% and 14/27 = 52% of cases respectively. In 66% of predictions for bulbs in series, teachers
recognize that the overall resistance of the circuit increases. Only 33% of the groups consider
resistance properly in the case of bulbs in parallel. Both in the circuit in series and in parallel
there is an incorrect prediction of 2/9 = 22%.
Finally, we report some teachers’ mistakes. The group that misses the 5th question predicts the
voltage of the battery at all points of the circuit. In the series closed circuit, 2 groups (2/10
= 20%) state that the bulb closer to the positive pole is brighter than the other. In the closed
parallel circuit 2 groups (2/10 = 20%) say that the circuit resistance increases.


All the considerations to interpret the results are related to sequencing the hydraulic circuit,
followed by the electrical circuit. As a consequence, at the moment, we are unable to definitively
distinguish what is determined by the specific context from what is due to the sequence.

In the following we discuss the results of the two investigated contexts and draw some general
In almost all hydraulic experimental situations, teachers’ estimates are focused on phenomeno-
logical descriptions. The first experimental situation identifies 4 naive ideas of pressure drop,
but since all of them imply a suggestion of gradient, they have allowed teachers to make accu-
rate representations of water jets. Our results suggest that teachers’ inclination is to consider
each circuit element in a reductionist way (Testa & Michelini 2006). For example, in question
3 the tube is considered only as a resistance, forgetting that having one entrance and one exit
can vary the drop depending on the difference in height it creates; in the 4th experiment the
upward part of the tube is thought separated from the downward one; the fan is only considered
as a barrier to current and not also as an opportunity to move, especially in the test with parallel
However the hydraulic model has created the opportunity for trainers to learn properly the
concept of pressure drop as the cause of currents both for open circuits (communicating vessels)
and for closed circuits. Moreover the fans are correctly recognized as resistors which cause a
fall in potential and a decrease of current.
In regard to the analogous electric circuit, we can draw the following conclusions: the teachers
make use of potential drop, but differently from the hydraulic case. Indeed the mention of
potential drop, not only in reasoning but also in prediction, suggests that teachers consider
potential drop as a phenomenological aspect of electrical circuits just as bulbs lightening and
their brightness. This result could be due to the fact that in electrical circuits potential drop is
the result of a measurement.
Anyway teachers analyze properly the open circuits and they are well oriented to provide the
correct behavior of closed circuits, even with numerical indications of potential drop along the
circuit, before and after the bulbs, with the exception of the parallel circuits.
For an understanding of parallel circuits it seems to be necessary to overcome the reductive
model of resistance as a barrier to current and to consider it rather as an opportunity for current
In conclusion, experiments on the basic elementary models (1-3) and the first 4 questions on
hydraulic circuits have allowed teachers to focus on the concepts of potential drop, current and
resistance, even if the analysis of results suggests that the last two are adequately understood
only if confined to the hydraulic model, but not so sufficiently consolidated to be always cor-
rectly transferred to the electrical situation. This suggests the inclusion of further training path
specific situations to create conceptual conflict for resistance and current in hydraulic context in
a way similar to what we did with questions 3 and 4 concerning the concept of potential drop.
When teachers already have an interpretative model previously established, they are able to ori-
entate themselves, whereas when the model is absent or only mathematically introduced, they
are not able to transfer it to others contexts. Elementary consolidated patterns of interpretation
(communicating vessels, potential drop as the cause of current) enable teachers to make predic-
tions and to interpret correctly the behavior of the majority of hydraulic and electric circuits,
despite their differences and peculiarities.

The electrical circuit is certainly far away from the sensitivity and the propensity of primary
school teachers, but the opportunity to analyze it in analogy to what is done in the hydraulic
circuit has enabled them to address it. Teachers have in fact benefited from models borrowed
from hydraulic circuits and, by analogy, have made in most cases successful predictions on
electrical circuits.
Finally, the analysis, of which and how many variables are considered, underlines the difficulties
of teachers to consider three variables at the same time. We believe that further situations of
conceptual conflict where teachers are obliged to consider three variables simultaneously will
help them to build a mental model for Ohm’s law and the concept of resistance.

Atwood, R.K., Christopher, J.E., & Trundle, K.C. (2001). Are middle school science teachers prepared
to teach standards-based heat and temperature concepts? Paper presented at the Annual Meeting of the
National Association for Research in Science Teaching, St Louis, MO, 25-28 March.
Black, D. & Solomon, J. (1987). Can Pupils Use Taught Analogies for Electric Current? School Science
Review, 69, 247, 249-254.
Driver, R., Asoko, H., Leach, J., Mortimer, E. & Scott, P. (1994). Constructing Scientific Knowledge in
the Classroom Educational Researcher, Vol. 23, No. 7., pp. 5-12.
Duit, R. & von Rhoneck, C. (1998). Connecting Research in Physics Education with Teacher Education,
I.C.P.E. Book ©International Commission on Physics Education.
Chaiklin, S. & Lave, J.(1993). Understanding practice. Perspectives on activity and context, Cambridge
University Press.
Cohen, R., Eylon, B. & Ganiel, M. (1983). Potential difference and current in simple electric circuits: A
study of students’ concepts Am. J. Phys., 51, 5, 407.
Cosgrove, M. (1995). A study of science-in-the-making as students generate an analogy for electricity.
International Journal of Science Education, 17, 3, 295-310.
Fuchs, H.U. (1997a). The Continuum Physics Paradigm in physics instruction I. Images and models of
continuous change. Zurich University of Applied Sciences at Winterthur.∼fuh/
Fuchs, H.U. (1997b). The Continuum Physics Paradigm in physics instruction II. System dynamics mod-
eling of physical processes. Zurich University of Applied Sciences at Winterthur.∼fuh/
Fuchs, H.U. (1998). The Continuum Physics Paradigm in physics instruction III. Using the Second Law.
Zurich University of Applied Sciences at Winterthur.∼fuh/
Gentner D. & Gentner D.R. (1983). Flowing Water or Teaching Crowds: Mental Models of Electricity.
In Gentner D., Stevens A. (Eds.). Mental Models. Hillsdale: Erlbaum.
Gilbert J.K. & Boulter, C. (2000). Developing Models, In Science Education, Dordrecht, Kluwer Aca-
demic Publisher.

Greenwood, A. (1996). When it comes to teaching about floating and sinking, preservice elementary
teachers do not have to feel as though they are drowning! Journal of Elementary Science Education, 8,
1, 1-16.
Herrmann F. (1995). Der Karlsruher Physikkurs, Aulis, Köln.
Hestenes D. (1997). Modeling Methodology for Physics Teachers. In E. Redish & J. Rigden (Eds.) The
changing role of the physics department in modern universities, American Institute of Physics Part II. p.
Heywood, D. & Parker J. (1997). Confronting the analogy: Primary teachers exploring theusefulness of
analogies in the teaching and learning of electricity. International Journal of Science Education 19, 8,
Kruger, C.J. (1990). Some primary teachers’ ideas about energy. Physics Education, 25, 86-91.
Kruger, C.J., Palacio D. & Summers, M. (1992). Surveys of English primary teachers’ conceptions of
force, energy, and materials. Science Education, 76, 339-351.
Leinhard, G. (1988). In Situated Knowledge and expertise in teaching in J Calderhead, Teachers’ pro-
fessional learning, Falmer Press, London 1988.
McDermott, L.C. & Shaffer, P.S. (1992). Research as a guide for curriculum development: An exam-
ple from introductory electricity. Part I: Investigation of student understanding. American Journal Of
Physics, 60, 11, 994-1003.
Pontecorvo, C., Ajello, A.M. & Zucchermaglio, C. (1987). I contesti sociali dell’apprendimento, LED,
Paatz R., Ryder J., Schwedes H. & Scott P. (2004). A case study analysing the process of analogy-based
learning in a teaching unit about simple electric circuits. International Journal of Science Education, 26,
9, 1065-1081.
Psillos, D., Koumaras P. & Tiberghien A. (1988). Int. J. Sci. Educ, 10, 1, 29.
Reeves L.M. & Weisberg R.T. (1993). On the concrete nature of human thinking: content and context in
analogical transfer. Educational psychology. 13, 245-258.
Reeves L.M. & Weisberg R.T. (1994). The role of content and abstract information in analogical transfer.
Psychological Bulletin. 115, 381-400.
Schwedes, H. & Dudeck, W.G. (1996) Teaching electricity by help of a water analogy (how to cope with
the needs of conceptual change). In G. Welford, J. Osborne, P. Scott (eds),Research in Science Education
in Europe: Current issues and themes, 50-63.
Shipstone, D.M., Rhöneck, C., Jung, W., Karrqvist, C., Dupin J.J., Johsua, S. & Licht, P. (1988). A
study of student understanding of electricity in five European countries. International Journal of Science
Education, 10, 3, 303-316.
Stocklmayer, S.M., & Treagust, D.F. (1996). Images of electricity: How do novices and experts model
electric current? International Journal of Science Education, 18, 2, 163-178
Tiberghien A. (1984). Critical review of research concerning the meaning of electric circuits for students
aged 8 to 20 years. In Research on Physics Education: Proceedings of the First International Workshop.
Editions du Centre National de la Recherche Scientifique, La Londe les Maures, France, 1-18.

Testa, I. & Michelini, M. (2006). Prospective primary teachers’ functional models of electric and logic
circuits: results and implications for research in teacher education Proceedings of International GIREP
Conference (Amsterdam) Full-Text available on-line
Van den Berg, E. & Grosheide, W. (1993). Electricity at Home: Remediating alternative conceptions
through redefining goals and concept sequences and using auxiliary concepts and analogies in 9th grade
electricity education. In Proceedings of the Third International Seminar on Misconceptions and Educa-
tional Strategies in Science and Mathematics Misconceptions. Trust, Ithaca, New York.
Webb, P. (1992). Primary science teachers’ understandings of electric current. International Journal of
Science Education, 14, 4, 423-429.


Gorazd Planinšič[1] & Leoš Dvořák[2]

[1] Faculty for Mathematics and Physics, University of Ljubljana, Jadranska 19, 1000 Ljubljana, Slovenia.
[2] Faculty of Mathematics and Physics, Charles University in Prague, V Holešovičkách 2, 180 00 Prague

8, Czech Republic. E-mail:

The paper reports on work in progress. The aim of our work is to design sequences
of structured problems based on simple experiments that will form complete ‘sto-
ries’. The important feature is that the depth and cognitive level of each story
gradually increases from primary to university level. Connections with real world
situations and technological applications are also an important ingredient of the
material. The material will be designed so that teachers can take and use just parts
of such stories.

The main rationales that led us to the idea outlined above are the following.
Competent teachers need subject knowledge that extends beyond the frame of the curriculum.
This is especially true for applying active learning methods where teachers are suppose to stim-
ulate discussion and inquiry based learning. In the available literature simple experiments are
normally explained (if at all) at the level depending on the target users or readers. Pedagogical
aspects and suggestions for effective implementations of simple experiments in teaching and
learning are also limited to one level (typically primary school level), if present at all. Very sel-
dom can one find explanations and analysis of the physics background of simple experiments
that would address also the limits of simple theoretical models and suggest improved models.
It is equally rare to find descriptions of simple experiments that include real life examples that
are based on the same principles that are applied in the simple experiments.
To prepare activities for students with different abilities teachers need a broad repertoire of
problems, tasks and themes for each topic. If teachers know several variations of simple ex-
periment and are familiar with the underlying theory it will be easy for them to give additional
tasks at the appropriate levels to gifted students who solved the regular task well before their
class mates. Alternatively teachers may use simple experiments as a basis for mini-projects for
students to work on at home and present their results in school.
Distance learning is becoming increasingly important in modern society. Distance learning of
science subjects relies very much on simple experiments. In particularly there is a need for
simple experiments that can be performed with everyday objects but also offer possibilities for

simple measurements and quantitative analysis. With the help of technology (in particular ICT)
the area of measurements that can be called ’simple’ is constantly growing. For instance, a
sound card, which is an integral part of every computer, together with appropriate software
(excellent programs are available as freeware) offers possibilities for measurements of sound or
other dynamic phenomena.
Simple experiments can open channels for better communication between university teachers
and students (but also between academics themselves). University teachers are often not aware
of the fact (or neglect it) that even the advanced concepts and achievements of modern technol-
ogy can be initially presented by means of very simple experiments. Research has shown that
only 25 - 30% of the students entering University have developed abstract thinking. All the rest
are concrete thinkers and for them simple experiments used at an appropriate time can represent
an important link between concrete and abstract thinking.
The proposed idea has been illustrated in a sequence of experiments that include different pipe-
like objects that emit sound. The collection includes plastic and aluminium tubes of different
lengths, a plastic bottle and glass bottles with long necks (figure 1).

Figure 1: Pipe-like objects used in the presented simple experiments.

The following simple experiments and corresponding pedagogical applications have been pre-
sented (see table 1). Note that in some countries the names of the school levels may not cor-
respond to the same age groups as stated here. The list presented only illustrates the approach
and is shorter than it will be in the final version.

Level Goals Simple experiments Suggestions for teaching/learning task (pedagog-
ical implementations)
Primary level To recognize the relationship be- Hitting plastic pipes with palm. Throwing Questions: How to make them sing? Which is the best
(age 6-15) tween the size of the object and Al pipes on the floor. Blowing on the bot- way to make a particular object sing? Are size of the
pitch of the emitted sound (for these tles + varying amount of water in the bottle object and pitch of the tone somehow related?
particular objects!)
Solidify knowledge, motivate, con- Colour labelled palm pipes with specific
nect science and art. lengths + colour coded notes: children
playing song on palm pipes (see Table 2).
To apply knowledge (longer pipe- Blowing on the plastic bottle half filled Predict how the pitch will change in time. Real world
lower pitch) in different situation. with water which is leaking from a hole on example: trombone.
the bottom.
Secondary level Introduction of new concepts, for- Experiments with microphone attached to Vertical pipe is initially closed at the bottom and half
(age 15-19) mal thinking, critical thinking (e.g. computer sound card† : 1) clapping pro- filled with water. You start blowing on the pipe to pro-
concepts of spectrum and standing duces “white” noise; 2) if microphone is duce the sound and at t1 you open the bottom end so that
waves etc.); using graphs . . . inserted into pipe, pipe will respond to water starts flowing out. At t2 the pipe is empty. Sketch
clapping with resonant frequencies. a graph that will show qualitatively time dependence of
fundamental frequency emitted by the pipe.
Undergraduate Combine topics, go deeper into the Two equal bottles are partly filled with Predict - observe - explain sequence (White, 1992)

university level theory (what does the speed of equal amount of water. How will the
(age 19-21) sound depend on?); critical thinking pitches of the sound emitted by the bottles
+ ability to judge which phenomena compare if I put an effervescent tablet into
are dominant in particular case; one bottle and wait until the bubbles stop
Improved or new models are intro- Blow on the bottle and measure the sound Improved theoretical model is based on the Helmholtz
duced that give more accurate de- frequency. Large discrepancy between resonator (Greenslade, 1996; Gluck, 2006). Real world
scription of reality. measured and calculated values (if simple example: “bass reflex”. Correct explanation: sound is
pipe model is used). Frequencies emitted produced by transversal vibrations of aluminium tubes;
by aluminium pipes (when thrown on the there are two nodes in fundamental mode; analysis re-
floor) do not match at all with those calcu- quires more advanced theory (Lapp, 2003). Real world
lated using the pipe model. example: Wind chimes, tubular bells.

Table 1: Example of mulitlayered simple experiments and corresponding suggestions for teaching/learning tasks. † We recommend Soundcard Scope which is free
of charge for private and non-commercial use in educational institutions and can be downloaded from ‡ The level of the
liquid in the bottle with tablet is a little higher than the level in the other bottle, so one may expect that the pitch of the bottle with tablet will be higher. But since the
effervescent tablet produces carbon dioxide and since the speed of sound in CO2 is about 78% of that in air at the same temperature, the pitch of the bottle with the
tablet is notably lower than the pitch of the other bottle.
C white 328
D red 290
E orange 258
F yellow 243
G green 216
A blue 193
H(B) violet 170
C black 159

Table 2: An example of colour codes and lengths of pipes that cover one octave as typically
used in schools.

Authors wish to thank Irena Dvořáková for useful discussions in particularly on the topics for
primary level.
This work was partly financed by bilateral project Slovenia - Czech Republic BI-CZ/09-010-
007 (MEB090907).

Greenslade, T.B. (1996), Experiments with Helmholtz resonators, Phys. Teach, 34, 228-230.
Gluck, P., Ben-Sultan, S., & Dinur, T. (2006), Resonance in flask and pipes, Phys. Teach, 44, 10-15.
Lapp, D.R. (2003), Building a copper pipe ‘xylophone’, Phys Educ. 38, 316-319.
White, R. and Gunstone R (1992), Probing Understanding, The Falmer Press, London.


Saša Ziherl[1,∗] , Jurij Bajc[1] & Mojca Čepič[1,2]

[1] Faculty of Education, University of Ljubljana, Slovenia.
[2] “Jožef Stefan” Institut, Ljubljana, Slovenia.

Wood is transparent to microwaves and due to its anisotropic structure has anisotropic di-
electric properties. In the microwave range a piece of wood acts as a uniaxial crystal. A suit-
ably cut block of wood can be used for interesting microwave experiments demonstrating
optical anisotropy. The selection of microwaves and wood in comparison to, for example,
liquid crystals and visible light, is useful, because it is easier to work with microwaves due
to their wavelengths and because wood has a pronounced and easily observable structure,
which leads to anisotropic properties.
We show a simple experiment, in which the consequences of anisotropy are demonstrated.
The experiment shows anisotropic properties generally found in crystals in the optical re-
gion. This laboratory experiment with a microwave transmitter and receiver can easily be
done in schools for the qualitative demonstration and/or quantitative measurements of the
birefringence of wood. A wood plate with variable thickness is presented.

Anisotropic materials are increasingly more and more significant in science and technology (for example
liquid crystals in LCDs) and it is important that one introduces anisotropy to students. Using wood and
microwaves is a good choice from several points of view. It is evident to students that the anisotropy of
wood results from its anisotropic structure. Furthermore, it is very easy to identify the symmetry axis of
the wood structure by simply looking at it with no special equipment. The evidence of anisotropic struc-
ture serves the visualization of physical phenomena, which is very important for students’ understanding.
As wood is transparent to microwaves, many optical properties can be illustrated experimentally.
Usually, microwaves are discussed during undergraduate courses. The interference and diffraction (Perk-
alskis & Gluck, 2007), Bragg reflection (Amato & Williams, 2004), evanescent absorption (Planinsic &
Vollmer, 2008), and other phenomena (Lozano-Rogado et al., 2001) are discussed during lectures or
studied in the laboratory. Microwaves are rarely used in schools for showing polarization and anisotropy.
In fact, we did not find many school experiments with anisotropy. And as we said, wood is very good
material for demonstrating anisotropy. The most common polarization experiment with microwaves is to
determine the polarization of the microwaves after crossing a wire-grid polarizer. During our literature
search we found only one study of anisotropic properties of wood for educational purposes - the construc-
tion of a microwave retarder (λ/4 plate) using wood (Perkalskis and Freeman, 1995), and one comment
on the same paper, which points out the importance of linear dichroism making the construction of the
retarder more elaborate (Chu and Noble, 1997). The aim of this paper is to present the construction of the
microwave retarder using wood and to present an experiment that allows for measurement of the birefrin-
gence of wood. The experiment is based on the anisotropy of wood and it may serve as a demonstration
for explaining the anisotropic properties of crystals in the visible region.

x, P

a b

z y, ^

Figure 1: Left: Sample of wood appropriate for the measurements. Right: Experimental geometry. The
dark lines indicate wood fibers. The x axis is chosen parallel to the fibers and y axis is perpendicular to
the fibers. Arrows denoted by P and A show the transmission directions of polarizer and analyzer.

In a typical piece of wood phloem vessels are elongated and annual rings form a layered structure. The
layers are nearly parallel and one can easily identify three orthogonal directions: one is directed along
the wood fibers, one is perpendicular to the layered annual sheets and the third is perpendicular to these
two directions. For dielectric properties, the most important is the orientation of the fibers and uniaxial
symmetry is expected. Therefore the refractive indices for polarization parallel and perpendicular to the
wood fibers differ, as is pointed out in the paper by Perkalskis & Freeman (1995). Our measurements
were made with geometrically simple samples like the one in Fig. 1, without knots and with the wood
fibers along the annual ring lines.
The amplitude of the electric field for the incident polarized wave may be written in the coordinate system
(Fig. 1) that corresponds to the wood sample as:

 in = E0 (cos β, sin β) = E
E  + E
 ⊥ = (E , E⊥ ) , (1)

where β is the angle between the transmission axis of the polarizer P and the x-axis parallel to the wood
fibers (Fig. 1). A polarized microwave is decomposed into two components with mutually perpendicular
polarizations that are oriented along the eigenvectors of the piece of wood: one polarization is parallel to
wood fibers and the other is perpendicular to the fibers. As the refractive indices of wood are different
for each polarization, the beam with one polarization is retarded with respect to the beam with the
polarization perpendicular to it.
After transmition through the wood the incident linearly polarized wave

 in (t, z) = (E0 , E0⊥ ) cos(ωt − kz)

E (2)

becomes elliptically polarized

 out (t, z) = (E cos(ωt − kz − ϕ ), E⊥ cos(ωt − kz − ϕ⊥ )) ,

E (3)

where ϕ and ϕ⊥ are the phases gained by the microwaves of both polarizations while passing through
the sample of wood. The phase difference between the microwaves polarized parallel and microwaves
polarized perpendicular to the wood fibers is

2π(n − n⊥ )
δ = ϕ − ϕ⊥ = d, (4)

where n and n⊥ are refractive indices for microwaves polarized parallel and perpendicular to the wood
fibers, respectively, λ0 is the wavelength of the microwaves in air, and d is the thickness of the wood.
Resetting the origin of time

t = t − n⊥ (5)
allows us to express the outgoing microwave at the position z = d as

 out (t , d) = (E cos(ωt ), E⊥ cos(ωt − δ)),

E (6)

which proves that the outgoing wave is elliptically polarized.

Anisotropic materials are usually anisotropic in more than one property. For the transmition through
wood, the anisotropy in refractive index as well as in absorption, have to be included in the analysis. We
assume that wood is uniaxial in microwave region, so the intensity of the transmitted electromagnetic
wave is

I = 2
ε0 c Eout,A
= ε0 c E02 [e−2λ d cos2 α cos2 β + e−2λ⊥ d sin2 α sin2 β (7)
− e−(λ +λ⊥ )d sin 2α sin 2β cos δ] ,

where α and β are the angles defined in Fig. 1, λ and λ⊥ are the absorption coefficients for the par-
allel and perpendicular polarization respectively, and Eout,A is the amplitude of the electric field of the
outgoing microwave.
As our receiver detects only the electric field along one polarization direction, it acts as an analyzer.
So the analyzer in Fig. 1 is in fact the receiver used for the analysis of the transmitted wave. The
detected electric field of the transmitted wave is just the component of the field linearly polarized along
the detectable direction eA of the receiver.
As there are in general two types of electromagnetic waves detectors one has to be sure which one is used.
The output signal of one type of detector is proportional to the electric field amplitude of the incoming
microwaves and the output of the other type is proportional to the intensity of the incoming microwaves.
The signal of the detector that we used is proportional to the electric field amplitude.

Microwave transmitters emit polarized microwaves and the receivers detect only the component of the
microwaves polarized in one specific direction. Therefore, the polarizer and the analyzer in Fig. 1 are
considered symbolically as indicators for the direction of the incident microwave polarization and for the
detectable polarization of the receiver.
The experimental setup is seen in Fig. 2. One is allowed to rotate separately the transmitter and the
receiver. The transmitter used was IPhO Microwave transmitter (A) (IPhO, 2006). The signal was
measured by IPhO Microwave receiver (B) (IPhO, 2006). Each is equipped with a goniometer with 5◦
marks. The transmitter used emits microwaves with wavelength 2.8 cm. The intensity of the signal was
measured with digital multimeter. Measured current is proportional to the amplitude of the electric field.
The sample of wood was a spruce board with variable thickness. Details about this board are described
in the continuation.

Figure 2: Experimental setup. Instrument “A” is the microwave transmitter and instrument “B” is the
microwave receiver. The geometry of the setup is shown in Fig. 1


A board of variable thickness is made from two triangular-shaped pieces of wood with the same fiber
axis as suggested in the paper by Perkalskis & Freeman (1995). The easiest way one can get these two
pieces of wood with an identical fiber axis is to cut diagonally one block of wood. The thickness of the
sample is then controlled by clamping together the two pieces appropriately (see Fig. 3). The thickness
of the board depends linearly on the length (l) defined in Fig. 3

t= l, (8)

where a is width of the wedge, b is length of the wedge, and l is the horizontal length between the two
tips of the triangular-shaped pieces of wood (Fig. 3).

a t



Figure 3: Side view of three boards of different thicknesses made of two wedges that are used to adjust
the thickness: a thick board (bottom), a thin board (middle), and a medium thickness board (top). Clamps
are drawn in gray. Letter a denotes the width of the wedge, b denotes the length of the wedge, l is the
horizontal length between the two tips of the triangular-shaped pieces of wood, and t is the thickness of
the board, used in measurements.

Our spruce wedges have length 31.0 cm and width 3.5 cm. If one wants to calculate minimum and max-
imum useful thicknesses of the board from these equations, one has to include in calculation also the
minimum “working” length of the board (Fig 3). In our experiment the minimum length was approxi-
mately 15 cm (the width of the horn of the transmitter plus clamps). So the minimum thickness of the
board, which was appropriate for measurements, was 1.7 cm and maximum thickness was 5.3 cm.


To measure the birefringence of wood in the microwave region, the incident microwaves are polarized
in the direction 45◦ with respect to the wood fibers (β = 45◦ ). The analyzer is rotated and α takes the
values between 0 and π in steps of 5◦ . Equation (7) becomes

I = 2
ε0 c E0A
= ε0 c E02 [e−2λ d cos2 α + e−2λ⊥ d sin2 α − e−(λ +λ⊥ )d sin 2α cos δ] . (9)

The measured dependence of the received signal on the angle between the polarization direction of the
receiver and the wood fibers (α) for 8 different thicknesses of spruce wood is plotted in Fig. 4. The un-
determined parameters are the absorption coefficients and the birefringence. They are determined by the
minimization of the difference between the observed and theoretical intensities for all wood thicknesses.
To get the value of the phase difference between the two polarized microwave beams one has to vary
cos δ to obtain the best agreement between the measurements and the theoretical values (Fig. 4). We
used trial and error to find the best single ∆n for all thicknesses of spruce samples, where ∆n is related
to δ by equations

δ = 2π ∆n , ∆n = n − n⊥ . (10)

According to the literature, the average refractive index for dry wood is around 1.5 (Bucur, 2003). Di-
electric properties of wood depend on water content and therefore the quantitative results of different
measurements (in the classroom) usually differ. In spite of this, general anisotropic properties (e.g. bire-
fringence and anisotropy in absorption) are well observable and can be quantitatively determined for
each set of measurements. The measured birefringence for our wood sample is 0.14, which indicates a
relatively large anisotropy. For example liquid crystals, which are known to have very large birefringence
in optical region, can have ∆n up to 0.3 (Chandrasekhar, 1992).
As mentioned before, the incident wave becomes elliptically polarized after transmition through the
wood. With known value of the phase difference δ one can plot an ellipse in parametric form (from
Eq. 3)

Ex = E0 cos(ωt) ,
Ey = E0 cos(ωt − δ) , (11)

where the absorption is not included. Ellipses for all thicknesses of the wood are plotted in Fig. 5.
The absorption coefficient for polarization parallel to wood fibers is consistently larger than the ab-
sorption coefficient for the polarization perpendicular to the fibers. The ratio between the absorption
coefficients is significantly larger than one, as already suggested in the comment by Chu & Noble (1997)
and it is around 3 for the wood studied in our case. Including absorption in the calculation leads to

Ex = E0 e−λ d cos(ωt) ,
Ey = E0 e−λ⊥ d cos(ωt − δ) . (12)

Two ellipses, one taking the absorption into account and one not taking the absorption into account, for
the selected thickness of wood are plotted in Fig. 6.

detector signal [a.u.]

1.8 21mm
1.4 fit_30
1.0 51mm



0 20 40 60 80 100 120 140 160 180
alpha [°]

Figure 4: Measured intensities in dependence on the angle (α) between the analyzer and wood fibers
for 8 thicknesses of spruce wood. The polarizer is rotated by 45◦ (β = 45◦ ) with respect to the fibers.
The fit of this measurements according to Eq. (9) is given by the corresponding solid curves.

In contrast to the usual optical experiments with visible light, where anisotropy is not obvious to students
because it is impossible to show the structural reasons for the anisotropy, wood and microwaves offer an
ideal system where anisotropy is easily understood. We have presented a simple experiment that allows
for measurements of the birefringence of wood. This experiment can easily be used for demonstration
purposes or as laboratory work in undergraduate courses. We have already used this demonstration with
our students. No formal measurements of an improvement in understanding were made yet, but the
students’ response was very positive.

Amato, J.C. & Williams, R.E., (2004) Rotating crystal microwave Bragg diffraction apparatus, Am. J.
Phys. 77(10), 942-945.
Bucur, V. (2003) Nondestructive Characterization and Imaging of Wood. Springer-Verlag Berlin Heidel-
Chandrasekhar, S. (1992) Liquid crystals. Cambridge University Press, 2nd ed.
Chu, K.C. & Noble, J.D. (1997) Comment on ‘Demonstrating crystal optics using microwaves on wood

targets,’ by Perkalskis, B.S. & Freeman, J.R. Am. J. Phys. 63(8), 762 - 764 (1995)]," Am. J. Phys. 65(8),
797 - 797 (1997).
Equipment used in the experimental problem in the 37th International Physics Olympiad in Singapore in
2006 (see:
Lozano-Rogado, J., Miranda-Pantoja, J.M. & Sebastian, J.L. (2001) Measurement of velocities in noisy
environments with a microwave Doppler-effect radar. Eur. J. Phys. 22(3), 249-255.
Perkalskis, B.S. & Freeman, J.R. (1995) Demonstrating crystal optics using microwaves on wood targets.
Am. J. Phys. 63(8), 762-764.
Perkalskis, B.S. & Gluck, P. (2007) Demonstrating edge diffraction with microwaves. Eur. J. Phys.
28(6), 1091-1095.

Planinsic, G. & Vollmer, M. (2008) The surface-to-volume ratio in thermal physics: from cheese cube
physics to animal metabolism. Eur. J. Phys. 29(2), 369-384.

21mm 0.8
30mm 0.6



-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8





Figure 5: Plotted elliptically polarized waves for different wood thicknesses.




-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8





Figure 6: Plotted ellipses for 21 mm thick wood, one taking the absorption into account and one not
taking the absorption into account, for the selected thickness of wood.


Marialuisa Aliotta[1] & Simon P. Bates[2]

School of Physics and Astronomy, The University of Edinburgh, JCMB - King’s Buildings, Mayfield
Road, Edinburgh, UK, EH9 3JZ
[1] [2]

Electronic Voting Systems have become widely spread in Higher Education as a way to
promote greater students’ engagement in the cognitive process. Indeed, the introduction
of student-generated content in (either formative or summative) assessment processes may
provide yet another means to further fostering a deeper and long-lasting learning. This
paper reports on a case study where student-generated multiple choice questions were used
for an end-of-course class competition using clickers. The results from the competition,
together with students’ feedback and lecturer’s personal reflection on the effectiveness of
this approach are discussed.

It has long been recognised that the traditional lecturing style involving a one-way transfer of informa-
tion from instructor to students is not ideal in supporting students’ engagement and learning experiences.
As a consequence, both lecturer and students are left with a sense of mild dissatisfaction as to the ef-
fectiveness of some of the most widespread teaching modes. Indeed, efforts to address this problem
have resulted in a variety of approaches which all aim at a more student-centred learning by fostering
students’ responsibility as independent learners. Some of these approaches include enhancing student-
instructor communication, for example using electronic voting systems (Bruff, 2009); peer instruction
(Mazur, 1997); and collaborative learning (e.g. through team/group activities) both in and out of classes.
The advent of wireless technology has enabled the creative introduction of Electronic Voting Systems
(EVS) in a variety of teaching contexts (lectures, workshops, revision classes) as a way to enhance stu-
dents’ participation and interactivity, particularly in scientific disciplines. Albeit with mixed results –
see for example (Preszler, 2007) and references therein – electronic voting systems have been used in
different disciplines such as biology (Knight and Wood, 2005); microbiology (Suchman, 2006); phys-
iology (Paschal, 2002); and physics (Bates, 2006), to cite just a few. A comprehensive list of subject
deployments in the UK is maintained at the University of Glasgow (Draper, 2009).
While the use of such technologies is often met with enthusiasm by students, who consistently report a
level of greater interest and engagement in the learning process, they are by no means a ‘magic bullet’ for
effective learning. A key challenge is deploying them as part of a coherent pedagogy, and a major part of
that is designing effective questions (Beatty, 2006). Clearly, whether learning benefits are to be gained
depends not on the latest technological kit, but rather on what specific teaching method is associated
with it. The adoption of EVS in conjunction with more student-centred strategies seems to have led to
increased conceptual gains on standard diagnostic tests (Hake, 1998).

Perhaps, the most widely used approach involving EVS is in conjunction with multiple choice questions
(MCQs) and as such probably “associated with the lowest kind of learning of disconnected facts” (Draper,
2009a). Yet, it may be argued that there is an intrinsic value in this approach as it fosters students’
attention and reflection on their own progress. In addition, MCQs develop students’ self-questioning
(as well as self-monitoring) about certainty and in some cases provide practice of some examination
methods. It thus appears that attention should be paid to the pedagogical approach and learning design
in order to foster deeper cognitive processes.
A way to try and attain greater involvement and larger learning gains seems to involve students playing
a direct role in the conceptual development of the assessment strategies. Student-generated content
has previously been shown to have a beneficial effect on students’ understanding of material and to
enhance later performance on end-of-course examinations. An excellent recent example of this approach
is provided by the PeerWise system, an online tool that involves students in the process of creating,
sharing, answering and discussing multiple choice questions (Denny, 2008).
This paper presents a case study where EVS has been used for an end-of-course class competition, where
the content of over half of the MCQs used for the competition was generated by the students themselves.
The context within which the competition took place, the results obtained, and useful considerations on
the success of this initiative are provided in the following sections.

Many different types of Electronic Voting Systems, colloquially referred to as “clickers”, are currently
available on the market and a quick Google search would give plenty of options to choose from. Typi-
cally, the complete set comprises a base station (receiver), wireless keypads (one for each participant),
and the vendor software. The most widely spread systems employ wireless hardware and data are trans-
mitted using either infrared or radio frequency technologies (a useful overview of both can be found, for
example, in:, accessed on Nov 10th 2009). The receiver
resembles a USB memory stick and can be plugged directly into a desktop or laptop on which the vendor
software will have been installed. The software can be used either as a stand-alone presentation platform
or as a plug-in to PowerPoint presentations and it enables the lecturer to collect students’ data, display
graphical polling results, and export the data to be used in reporting and analysis.
The most common application of EVS involves the lecturer using a computer presentation to display
questions with several possible answers, commonly referred to Multiple Choice Questions (MCQs).
The students select the answer they believe to be correct by pressing the corresponding key on their
individual wireless keypad. After a set time, or when all participants have cast their preference, the
system closes the polling for that particular question. Data are collected by the USB receiver and results
are immediately displayed in a graphical form for all to see, indicating the percentage of responses for
each of the possible answers. Entry-level clickers also have additional keys allowing the students to
select a true/false statement. More sophisticated handsets can take text-based or numerical input from
the students or allow students to rate their degree of confidence for their given answer. Depending on
the purpose of the polling, data can be either collected anonymously (e.g. as in the case of a vote, or in
a survey) or traced to the individual participants (e.g. in classroom tests, homework, or questions that
count towards a student’s course grade). Incoming data can finally be stored in a database on the host
computer for later retrieval and analysis.
There are several advantages of using EVS over non-electronic approaches such as, for example, a show
of hands or the use of colour cards to select one of the possible options. The most common benefits

include: improved attentiveness by increasing students’ engagement in the classroom; increased knowl-
edge retention (Mazur, 1997; Hestenes, 1992); anonymity of the voters; possibility to track responses;
and immediate display of polling results. As a consequence, the lecturer can effectively and immediately
judge the audience’s understanding of the key concepts being tested, thus adjusting their teaching to the
immediate needs of the students, either by confronting and resolving misconceptions or by encouraging
peer discussion before re-opening the poll for the same question. In general, the use of EVS helps to
create an interactive and fun atmosphere where the students are genuinely interested in seeing the results
to a given questions and in self-assessing their own performance in comparison to that of the rest of the
Although EVS has been mostly used in large-class settings chiefly to test students’ understanding, the
system is very versatile and can be used equally successfully for surveys, assessment, and feedback, both
in large- and in small-class settings, with its use being limited only by the lecturer’s creativity. This paper
presents a novel use of EVS as a class competition where an enhanced engagement was sought through
asking the students to generate about half of the MCQs to be used in the competition, as described in the
following section. The competition took place as a last revision class ahead of the final written exam for a
Senior Honours optional course in the School of Physics and Astronomy at the University of Edinburgh.
Feedback was then elicited from the students to try and assess the extent to which this use of EVS can
effectively promote learning.

Nuclear Physics (NP) is an optional Honours-level course (‘Level 10’ in Scottish Qualification Frame-
work) offered to 4th (final year Honours) and 5th (Integrated Masters) year students within the School
of Physics and Astronomy at the University of Edinburgh (one of the authors, MA, being the course or-
ganiser and lecturer). The course, typically attended by 45-55 students, consists of 22 one-hour lectures
over 11 weeks. It runs in the second semester (from January to March) and is complemented by weekly
two-hour workshops where students work in groups of 4-5 students. About two thirds of the workshop
time is devoted to traditional problem solving, with students working at pre-assigned problem sheets.
The last 30-40 minutes are used by the lecturer to present and discuss solutions to the problems with
particular emphasis on clarifying concepts and addressing any misunderstanding. The atmosphere of
the workshops is typically very relaxed and students are encouraged to have discussions both with their
peers, with the lecturer and with teaching assistants (typically one or two PhD students who have taken
the NP course in previous years). Indeed, the format of the workshops has been particularly useful in
addressing students’ engagement in the learning process and has resulted in a far greater participation
than achieved in previous traditional tutorial classes. At the request of the students, additional workshop
classes are offered as a revision opportunity well after the course has finished and shortly before the final
In 2006, a novel approach to the revision process was included as an end-of-course class competition
using clickers. Following the students’ positive feedback, a similar competition was run again in the
academic year 2008-09. This time, however, a new element was introduced as a way to further promote
students’ engagement and learning: namely, students were asked to produce about half of the MCQs that
would be used during the competition. These were collated by the lecturer during the last two weeks
of delivery of the course. For each question, and based on their level of difficulty, both the time for the
students to vote and the appropriate marks for a correct answer were set by the lecturer and stored in the
PowerPoint presentation, containing embedded EVS questions.
The competition took place about one week before the final written exam with 23 students attending and
taking part out of the 58 enrolled. Students were asked to arrange themselves in 5 groups of 4-5 and

to stick to their group for the duration of the competition. The session consisted of two distinct parts,
or “rounds”. In the first, students were given a series of 15 MCQs of varying difficulty (with correct
answers worth 1, 2, or 3 marks, totalling 22). The questions were answered individually by students,
with no discussion within their group, using their own clickers. The questions for the first round had been
devised individually by students and selected (and in some cases slightly edited) by the course organiser.
Questions centred mainly on factual recollection of definitions, understanding of key concepts, true/false
statements, and, occasionally, very brief calculations. The second round of the quiz consisted of a series
of questions that were answered collaboratively by students within their own group. The questions for
the second round had been devised by the course organiser and covered a range of topics and concepts
that had been part of the course. This round comprised 11 questions, again of varying difficulty (with
correct answers worth 1 to 5 marks), with a possible total of 25 marks. Student groups discussed the
questions and responded using a single, group clicker. Examples of students’ and lecturer’s questions are
shown in Figures 1 and 2, respectively.
A group’s total score was determined by summing the scores from the individual rounds. From the first
round, the group was given the average of the individual members’ scores, while from the second round
the members of each group collaborated to produce a single score per group. Prizes (£ 25 Amazon
vouchers) were awarded to each member of the highest scoring group, and to the student with the highest
individual score (14 out of 22) from the first round.

The results from the two rounds of the competition are shown in Figure 3 (round 1) and Figure 4 (round
2), respectively. Figure 3 shows a distribution of marks with an average at 7.7 (out of 22) and a stan-
dard deviation of 3.7, corresponding to 35% and 17%, respectively. (Here the total number of entries
corresponds to 22 students as one of them reported a faulty clicker.)
The results in Figure 4 show a distribution of marks averaged at 15.4 (out of 25) with a standard deviation
of 4.5, corresponding to 62% and 18%, respectively. Note that two groups scored the same result (13/25)
and they appear under the same histogram bar.
Clearly, the available low-number statistics do not allow for any quantitative conclusion on the results
obtained and only some qualitative comments can be made.
Given that the questions in the second rounds (i.e. those set by the lecturer) were slightly more chal-
lenging than those devised by the students, the higher average mark associated with the second round
cannot be associated with an easier round. Rather, it is likely to be ascribed to the fact that students
were allowed to discuss questions with peers within their group before selecting their answer. This is
clearly not surprising as it can be argued that the knowledge of many is often greater than that of single
individuals and there is clearly a benefit to the group as a whole. However, while a learner’s response can
often be regarded as a very early step in the cognitive process leading eventually to a deeper learning, the
average performance of individual students remains somewhat disappointing given the stage at which the
competition took place.
Yet, the ultimate aim of running the competition was to try and assess whether such an activity can
be used as a type of assessment that promotes learning and whether it provided an enhanced learning
experience for the students. To this end, students were asked to provide open written statements at the
end of the competition. Overall, student feedback was fairly positive as the competition was seen as
a good test of knowledge, despite not leading to any new learning. Students found it useful to see the
whole class response as it gave an idea of individual knowledge with respect to the rest of the class,

Figure 1: Examples of student-generated MCQs used in the competition.

Figure 2: Examples of lecturer’s MCQs used in the competition.

Figure 3: Individual student’s score distribution. Average mark is 7.7 (out of a total of 22) with
a standard deviation of 3.7.

and allowed them to assess the progress made with individual revision. A comment was made by a
student that longer questions should be reserved for group discussion only, while shorter ones are more
appropriate for questions to be answered individually. A student stated that the competition was a good
idea, but that it should not be run so close to the exam. Other statements included “it gets you thinking
and it is more fun than [traditional] revision”; “it [the system] is easy to operate (not the questions!)”;
and, interestingly, “multiple choice [questions] can be more misleading than just writing answer”.
Once again, the number of comments received as well as the fact that these were open-ended (and thus
not necessarily addressing a specific aspect) does not allow for any statistically meaningful conclusion.
However, some general considerations can be made based on anecdotal evidence as well as on the per-
sonal reflection of the lecturer.
During the two to three weeks preceding the end of the course the lecturer was approached by a few
students asking for some feedback on the questions they were preparing. Also, on a number of occasions
the students were seen to discuss MCQs within their own group during the workshop sessions. Both these
factors seem to indicate a good level of engagement and the desire to contribute good quality MCQs to the
final quiz. One student, in particular, approached the lecturer with a specific question regarding a topic
which – according to the student’s expectation – may have come up in the oral exam he was about to take
in another course that had been running in parallel to NP. After the viva, the student prepared an MCQ
related to this specific topic (which had indeed come up during the viva). In fact, it turned out that his
question was perhaps the most challenging amongst student-devised ones. Indeed, after the competition,
this student was identified as being the one with the highest individual mark and was thus rewarded with

Figure 4: Groups score distribution. Average mark is 15.4 (out of a total of 25) with a standard
deviation of 4.5. Note that one of the two groups scored the same result (13/25) and they appear
under the same histogram bar.

one of the Amazon vouchers. Again, this episode can be taken as an indication of an increased level of
engagement and it appears likely that similar activities can be further exploited in creative ways to try to
promote commitment.

Despite the lecturer’s perceived impression that setting questions was a useful activity to engage students
in their learning experience, another aspect of the whole exercise seemed puzzling at first: namely, that
student-devised questions turned out to be relatively “easy”, contrary to the lecturer’s initial expectation.
There is a number of possible explanations as to why this was the case: 1) the students were asked to
prepare MCQs at a time where the course was still taking place and thus it can be argued that some
concepts were still not fully understood and/or no revision had taken place yet; 2) students may have felt
self-conscious about venturing into more complicated questions for fear of producing wrong answers; 3)
students did not receive any training into what constitutes a good MCQ; and 4) the time in the semester
was getting close to the submission deadlines for Honours Project reports, that is major pieces of assessed
work which would have affected almost the whole cohort. As a consequence, students may have decided
to prioritise their studying thus deciding to spend less time on an activity not strictly required and also
not formally assessed.
From these considerations there are perhaps a few lessons that can be learned for improved, future ex-
ploitation of this type of activity as suggested below:

1. The process of creating MCQs can be integrated earlier on into the lecture course, for example at

the end of each self-contained topic. This would give students more time to devise suitable ques-
tions that would also be more focused and thus (hopefully) more complex than simple true/false
statements or factual recollection.

2. Students may be asked to produce follow-up questions asking for reasons supporting any of the
answers given in the preceding question.

3. More input and guidance need to be provided by the instructor on what constitutes a good MCQ.

4. Credit can be given according to the quality of the questions set.

5. Students may be encouraged to work in groups to produce more difficult MCQs.

6. The entire format of the competition may be changed into a challenge of two groups (each com-
prising half of the class) competing against one another, much like the BBC University Challenge
TV show. The expectation here is that each group would try to set more difficult questions to make
it harder for the counterpart to win the competition.

Clearly, the greatest limitation of this study consisted in the relatively small class size and thus no sta-
tistically sound conclusion can be drawn on the effectiveness of the use of competitions with student-
generated content as a viable pedagogic tool. Only the involvement of far larger cohorts will allow for a
quantitative analysis of knowledge and conceptual gains attained. Improved ways of collating feedback
from the students may involve asking students to provide their views on specific issues, such as the ef-
fectiveness of the approach, to improve their engagement and lead to a deeper learning. This could be
done, for example, directly after the competition and using the very EVS technology.

EVS are widely used in Higher Education as a tool to promote students’ engagement and interactivity.
In this paper we presented a case study for an end-of-course class competition using, in part, student-
generated content. Although the number of students taking part in the competition was too low to draw
any meaningful statistical conclusion on the effectiveness of such an activity in fostering students’ en-
gagement, the positive feedback received from the students as well as personal positive evidence gathered
by the lecturer on the course, seem to indicate that similar approaches can be used successfully with both
small- and large-class settings. Suggestions on how to improve on students’ engagement were given to-
gether with ways on how the measurement of the effectiveness of a student-led competition can be better

Bates, S.P., Howie, K and Murphy, A.StJ. (2006) The use of electronic voting systems in large group
lectures:challenges and opportunities. New Directions: the Journal of the Higher Education Academy
Physical Sciences Centre (ISSN 1740-9888) - Issue 2 Dec pp1-10.
Beatty, I.D., Gerace, W.J., Leonard, W.J. and Dufresne, R.J. (2006) Designing effective questions for
classroom response system teaching. Am. J. Phys. 74(1): 31-39.
Bruff, D. (2009) Teaching with Classroom Response Systems: Creating Active Learning Environments.
Wiley, NY.

Denny, P., Luxton-Reilly, A. and Hamer, J. (2008) The PeerWise system of student contributed assess-
ment questions. in Simon and M. Hamilton, editors, Tenth Australasian Computing Education Confer-
ence (ACE 2008), volume 78 of CRPIT, pages 69-74,Wollongong, NSW, Australia, ACS.
Draper, S.W. (2009) Electronic voting systems and interactive lectures [online]
∼steve/ilig/, accessed 12th November 2009.
Draper, S.W. (2009a) Catalytic assessment: understanding how MCQs and EVS can foster deep learning,
British Journal of Educational Technology, vol.40 no.2 pp.285-293.
Hake, R.R. (1998) Interactive-engagement vs traditional methods: A six thousand- student survey of
mechanics test data for introductory physics courses. Am. J. Phys. 66(1): 64-74.
Hestenes, D., Wells, M. and Swackhamer, G. (1992) Force Concept Inventory. Phys. Teach. 30, 141-158.
Knight, J.K., and Wood, W.G. (2005) Teaching more by lecturing less. Cell. Biol. Educ., vol. 4, 298-310.
Mazur, E. (1997) Peer Instruction. Prentice Hall, NJ.
Pascal, C.B. (2002) Formative assessment in physiology teaching using a wireless classroom communi-
cation system. Adv. Physiol. Educ., vol. 26, 299-308.
Preszler, R.W., Dawe, A., Shuster, C.B., and Schuster, M. (2007), Assessment of the Effects on Student
Response Systems on Student Learning and Attitudes over a Broad Range of Biology Courses. CBE-Life
Sciences Education, vol. 6, 29-41, Springer.

Suchman, E., Uchiyama, K., Smith, R., and Bender, K. (2006) Evaluation the impact of a classroom
response system in a microbiology course, Microbiol. Educ., Vol. 7, 3-11.

S.P. Bates[1,2] & K.A. Slaughter[1,3]
[1] School of Physics and Astronomy, University of Edinburgh, Mayfield Road, Edinburgh, EH9 3JZ.

The Colorado Learning Attitudes about Science Survey (CLASS) has been widely
deployed as a valid and robust measure of student ‘expertness’ in thinking about
their chosen subject of study. It has been extensively used in North American
colleges and universities, but far less-widely adopted in Europe. The survey was
administered pre and post instruction of the first year physics courses at the Uni-
versity of Edinburgh. At first glance, our findings confirm the general picture that
is frequently found: after a year of instruction, the students were seen to decline in
expert-like thinking. However, further analysis reveals some interesting differences
compared to previously published data. Firstly, our students’ degree of overall ex-
pertness prior to commencing their university studies was relatively high. Secondly,
we found that the physics majors at Edinburgh did not become significantly less
expert after instruction, the non-majors became dramatically, and significantly, less
expert. We consider the implications for future delivery of these first year courses.

Over and above their content knowledge, there is evidence that the way students think about a
subject has a profound effect on performance in that subject. It has been suggested that from
as early as primary school children develop misconceptions which can shape their attitudes
and beliefs of how they view science in general (Linn, 1991). It has long been established
that external social factors and stereotypes create a series of beliefs in students that can affect
performance (Eccles, 1986). In fact it has been suggested in previous research that attitudes
and beliefs may be a better indicator of student ability than examination results (Adams, 2006).
This has lead to increased research into students’ attitudes and beliefs in a variety of subject
In order to better understand student attitudes, the University of Colorado has developed an
instrument which enables student attitude shifts after a period of instruction to be examined.
The Colorado Learning Attitudes about Science Survey (CLASS; Adams, 2006) measures the
attitudes and beliefs of university students through a series of attitudinal questions and cal-
culates their level of “expert-like” thinking by comparing the answers to those of faculty and
other professionals. Slightly different versions for students studying Physics and Chemistry are
available: we have used the Physics version.

The survey has been used at many institutions across North America but has been far less
widely adopted in the UK and Europe. The survey is usually administered both before and after
a specific course is taught or period of instruction, thus allowing the effect of the course on
students’ attitudes to be assessed through the shift in the students’ answers.
Whilst most institutions who use the survey with their students report a decrease in expert-like
thinking of between five and ten percent (Redish, 2008) it is increasingly being used as a tool
to measure the effectiveness of course changes where courses have been redesigned to reflect
students needs (Otero, 2008; Wells, 2008).
Due to the different learning needs of students studying different degree choices the CLASS
survey and its precursor MPEX have been adapted and revised for use by biology majors taking
calculus based courses (Redish 2008). Here too, when a course was redesigned to teach to the
needs of biology students and adapted to better suit the skills set of biologists gains were seen
in the attitude surveys.
An interest in epistemological development of students has led to the use of CLASS to look
at longitudinal development of student expertise (Gire 2009). It was found in a study at the
University of California - San Diego that student levels of expertise remain approximately con-
stant through the first 3 years of study and increase in the final year of a degree and graduate
school. However these conclusions are tentative due to the small numbers of students surveyed,
particularly in the latter years.


By UK standards, the first year Physics class at the University of Edinburgh is large, comprising
approximately 300 students. First degrees at Scottish Universities usually take 4 years (com-
pared to 3 in the rest of the UK) and the nature of the first year of study is intentionally broad.
This results in approximately half of the first year Physics class intending to study degrees
other than the various ‘flavours’ of Physics. These students are taking the course as an elec-
tive option, but like all students on the course, need to have attained the pre-requisite grades in
school-leaving examinations in both Physics and maths (typically, but not always, at the highest
grade of ‘A’).
The course has had a long history of innovative teaching developments, including the extensive
use of electronic voting handsets (or ‘clickers’) as a tool for interactive engagement in large
lectures. It is a well-received course by students, and pre- and post-instruction testing with
the Force Concept Inventory (Hestenes, 1992) yields average normalized gains (Hake, 1998)
of 0.3-0.5 over several years. Instead of a traditional small group tutorial, the course uses a
‘studio’ based approach to additional classes (we call them ‘workshops’; Bates, 2005). In these
weekly workshop sessions, students work in small groups of five or six. They are engaged in a
variety of different tasks designed to foster not just the ‘hard’ skills associated with the subject
matter and content of the course (e.g. via ‘tutorial’-style problems) but also a wider-ranging of
skills such as thinking questioningly, arguing systematically etc.

The Colorado Learning Attitudes about Science Survey was developed in order to look at the
effect instruction can have on student attitudes. Building on previous work such as the MPEX
survey of Redish and co-workers at the University of Maryland (Redish, 1998) it gauges ex-
pertness of attitude towards and beliefs about the subject. The survey consists of 42 statements,
each of which students indicate their response to on a 5-point Likert Scale, from strongly dis-
agree to strongly agree. The questions are not a measure of content knowledge but as to how an
individual thinks about physics and what they believe ‘doing’ Physics entails.
Each statement can be examined individually but in addition the majority are categorized into
one or more of 8 categories; real world connections, sense-making/effort, personal interest,
conceptual connections, applied conceptual understanding, problem solving, problem solving
sophistication, and problem solving confidence.
In order to ensure the data is of good quality, a student’s response must pass a series of checks.
A “failsafe” question is used; question 32 of the survey asks students to select option four
‘agree’ if they are reading the question. Six percent of the surveys collected at Edinburgh failed
the failsafe question and so were discounted from the data. This is comparable with the 7-12%
(inversely related to the level of the course) seen in previous studies (Adams 2006). In addition
there is also a requirement that at least a minimum number of questions have to be answered,
together with a minimum number answered in each of the eight categories.
The survey has gone through extensive tests in order to ensure its validity and reliability includ-
ing tests of the clarity of statements as perceived by students and experts. The University of
Colorado have also carried out interviews with students to establish whether they are answering
the question with what they believe or what they think the interviewer wants to hear (Adams


The survey was administered during the first week of student workshops, before any instruction
had taken place, and again at the end of second semester teaching, before students’ end of year
examinations. In order for each individual’s shift in attitude to be calculated there must be
matched pre- and post- instruction survey data for each student. Due to difficulties in collecting
data in some cases and to students leaving the course, whilst the physics class comprised 300
students at the beginning of the year we were only able to collect matched data for 149 students.
One of the outputs of the analysis of the data collected is the percentage of student responses
that agree with those of experts in the discipline. (Note that in doing this analysis, the 5-point
scale is compressed to a 3-point scale of Agree-Neutral-Disagree). Over all students, and all
questions, this value of agreement with expert views was found to decrease from 72% pre-
instruction to 68% post-instruction. This is a similar trend as seen in results of administering
the survey elsewhere (Adams 2005).
Results for the same quantity of percentage agreement with expert views for each of the cate-
gories can be found in Table 1. Whilst the patterns of decrease were the same as those found in
previous studies it has been noted that the students at Edinburgh started with a very high level of

expert-like thinking and also did not drop as much as has been seen at some institutions (Adams

Category PRE (%) POST (%) Standard Error on mean

N = 149
Overall 72 68 1.1†
Personal Interest 76 69 1.9
Real World Connection 74 69 2.3
Problem Solving General 81 78 1.6
Problem Solving Confidence 78 77 2.2
Problem Solving Sophistication 75 70 2.0
Sense-making-effort 82 76 1.8
Conceptual Understanding 76 71 2.0
Applied Conceptual Understanding 63 59 2.0

Table 1: Breakdown of favourable CLASS scores Pre- and Post- Instruction. † Calculated using
standard deviation of the mean score divided by the square root of the number of students who
answered. The variations in the standard error are due to the differences in numbers of scored
questions for each category.

An alternate way to visualize these results is to plot them on an agree-disagree plot, as shown
in Figure 1. The x-axis represents the proportion of unfavourable answers (those that disagree
with the expert answer) and along the y-axis is the percentage of student responses that agree
with expert views. The percentages do not add up to one hundred, as the neutral responses are
not included. A shift towards the top left-hand corner would be towards becoming more expert.
As can be seen from Figure 1, in each of the eight categories, as well as overall, the students
become less expert-like in their thinking about Physics.
For each of the 42 statements on the survey statistical tests were carried out to determine if
the shift in expert-like thinking was significant. Statistically significant shifts for 13 statements
were found, of which 11 resulted in a move towards novice thinking and 2 where the students
became more expert. (For a full list of the statements which resulted in a statistically significant
change, see Appendix A.) The two statements where students became more expert-like were “To
understand physics I discuss it with friends and other students” and “It is possible to explain
Physics without mathematical formulae”. Both these statements represent intentional learning
aims of the course and are points of particular emphasis throughout the instruction.
We have also performed further analysis of the data, splitting it so as to allow gender effects to
be discerned and also so that the differences between the students intending to study physics
degrees (“majors”) and those taking physics as an elective course (“non-majors”) could be ex-
In terms of degree intent we were surprised to discover that not only did those students intending
to study physics as a degree start at a higher level of expert-like thinking, they also did not
show a statistically significant drop in expert-like thinking after the period of instruction (from
72.2% to 71.0%). In contrast those students taking physics as an outside course not only started

Figure 1: Agree-Disagree Plot for CLASS data from Physics class 2008-09. The x-axis repre-
sents percentage of unfavourable thinking, compared to expert response; the y-axis percentage
of favourable answers.

at a lower level of expert-like thinking but also showed a drop of nearly 10% over the same
period of instruction, a decrease which is statistically significant (p < 0.0042). These results
are illustrated graphically in Figure 2, plotting percentage of students agreeing with expert
responses for all questions. Error bars are calculated from the standard deviation of the results,
divided by the square root of the number of students in the sample (the standard error). The
differences over the period of instruction between these two groups is somewhat surprising as
the students are all in their first year of university, all have broadly similar grades at Physics and
Mathematics school-leaving exams, and all attend the same lectures and workshops.
A similar pattern was seen when looking at the male-female split, as shown in Figure 3. Male
students started with a higher level of expert-like thinking and did not show a statistically-
significant attitude drop whilst female students started lower and markedly decreased, though
in this case the decrease was not found to be statistically significant (p < 0.08).
The differences between male and female students are not unexpected. It has been seen in stud-
ies by the University of Colorado that female students are generally less expert in the categories
of ‘problem solving’ (both confidence and sophistication), ‘personal interest’ and ‘real world
connections’ they were slightly more expert in ‘sense making/effort’ type questions (Adams
It has also been seen in research by Colorado (Adams 2005) that that physics courses which
comprise predominantly non-scientists do not show a significant decline in attitudes and be-

Figure 2: Breakdown of CLASS data for Physics 1A class 2008-09 by Degree Choice. Pre-
and Post- instruction are shown for both physics majors (N = 111) and non majors (N = 38),
plotted against percentage of expert-like thinking.

Figure 3: Breakdown of CLASS data for Physics 1A class 2008-09 by Gender. Pre- and Post-
instruction are shown for both male (N = 114) and female (N = 35) students, plotted against
percentage of expert-like thinking.

liefs. However in the courses where the non-scientists were surveyed the issue of attitudes and
beliefs was directly addressed in the teaching. Where our results differ from those of most pre-
vious studies is that the physics majors at Edinburgh do not significantly decline after a year of

In a direct comparison between students at UCSD starting an engineering degree and those
starting a physics degree (Gire 2009) it was also seen that physics majors are more expert than
non-majors. However this was a different scenario to that represented here as the engineers were
not taking physics courses, but did have the same entry requirements as the physics students.


We have described the delivery of the CLASS survey of attitudes and beliefs to the 2008-09
first year Physics cohort at the University of Edinburgh. We have found, in line with what
has been recorded at many universities in North America, that a year of instruction has an
overall negative effect on student attitudes and beliefs, even when the nature of the instruction
is judged to be successful in terms of student performance (evidenced by learning gains on
standard diagnostic test instruments) and feedback. The only questions in the survey where
students were seen to become more expert were particular strengths of the course design at the
University of Edinburgh. This raises interesting possibilities for future course design.
Despite the overall similarity to previously-published results, there are some interesting dif-
ferences in our collected data, when analysed according to degree intention and gender. In
contrast to what was seen in other studies (Adams, 2006) we have found that Physics majors
do not become statistically significantly less expert over a period of instruction, while those
taking physics as an outside course do so markedly. We have also found a similar, though less
pronounced effect as a function of gender, with male students decreasing in expertness far less
than females.
We are currently collecting data on this year’s cohort of student to assess the repeatability of
our findings. If the results we have presented here are confirmed, this could have significant
implications for the design and delivery of the first year curriculum in the future, including
possibly having to address the difficult issue of how to effectively teach majors and non-majors
most successfully within the same class, or whether they are more effectively taught separately.
In addition to repeating the survey with current first year students we are also undertaking a
more longitudinal study to assess how expertise develops immediately prior to, over the course
of and beyond an undergraduate degree. We have delivered the survey to students in their final
year of school, those engaged in undergraduate study, postgraduate students, post docs and
right through to members of physics faculty. As these are all different groups of people it is not
possible to calculate the shift in expert-like thinking but we will be able to gain a ‘snapshot’ of
where each group is at that point in time. This will be reported elsewhere.

We gratefully acknowledge financial support from the University of Edinburgh’s Principal’s
Teaching Award Scheme.

Summary of those statements from the Colorado Learning Attitudes about Science Survey
which show a statistically significant shift post instruction:

Statement Change
1. A significant problem in learning physics is being able to memorize all the Novice
information I need to know.
5. After I study a topic in physics and feel that I understand it, I have difficulty Novice
solving problems on the same topic.
11. I am not satisfied until I understand why something works the way it does. Novice

14. I study physics to learn knowledge that will be useful in my life outside Novice
of school.
15. If I get stuck on a physics problem on my first try, I usually try to figure Novice
out a different way that works.
18. There could be two different correct values for the answer to a physics Novice
problem if I use two different approaches.
19. To understand physics I discuss it with friends and other students. Expert
23. In doing a physics problem, if my calculation gives a result very different Novice
from what I’d expect, I’d trust the calculation rather than going back through
the problem.
25. I enjoy solving physics problems. Novice
30. Reasoning skills used to understand physics can be helpful to me in my Novice
everyday life.
35. The subject of physics has little relation to what I experience in the real Novice
36. There are times I solve a physics problem more than one way to help my Novice
38. It is possible to explain physics ideas without mathematical formulas. Expert

Table 2: Questions from CLASS survey which show a statistically significant shift post instruc-
tion for the 2008-2009 Physics class.

Adams, W.K., Perkins, K.K. & Dubson, M (2005) The design and validation of the Colorado Learning
Attitudes about Science Survey, Physics Education Research Conference, 790, 45-48.
Adams, W.K., Perkins K.K. & Podolefsky N.S (2006) A new instrument for measuring student beliefs
about physics and learning physics the Colorado Learning Attitudes about Science Survey, Physical
Review Special Topics-Physics Education Research 2, 010101.

Bates, S.P. (2005) Reshaping Large Undergraduate Science Courses; the Weekly Workshop.
Available at
Retrieved 12/11/2009
Eccles, J.S. & Jacobs, J.E (1986) Social Forces Shape Math Attitudes and Performance, Journal of
Women in Culture and Society, 11, 367-380.
Gire, E. & Jone, B (2009) Characterizing the epistemological development of physics majors, Physical
Review Special Topics-Physics Education Research, 5, 010103.
Hake, R.R. (1998) Interactive-engagement versus traditional methods: A six-thousand-student survey of
mechanics test data for introductory physics courses, American Journal of Physics, 66, 64-74.
Hestenes, D., Wells, M. & Swackhamer, G. (1992) Force Concept Inventory, Phys. Teach, 30, 141-158.
Linn M. C. & Songer N. B (1991) Cognitive and conceptual change in adolescence, American Journal
of Education, 99, 379-417.
Otero, V.K., & Gray, K.E. (2008) Attitudinal gains across multiple universities using the Physics and
Everyday Thinking curriculum, Physical Review Special Topics-Physics Education Research, 4, 020104.
Redish, E.F., Saul, J.M. & Steinberg, R.N. (1998) Student expectations in introductory physics, American
Journal of Physics, 66, 212-224.
Redish, E.F. & Hammer, D(2009) Reinventing college physics for biologists: Explicating an epistemo-
logical curriculum, American Journal of Physics, 77, 629-642.
Wells, L., Valenzuela, R., & Brewe, E. (2008) Impact of the FIU PhysTEC Physics Reform of Introduc-
tory Labs, Physics Education Research Conference, 1064, 227-230.

Marion Birch[1] & Niels R. Walet[2]
School of Physics and Astronomy, The University of Manchester, Manchester M13 9PL, UK.
[1] [2]

We consider the reliability of different methods of identifying weaker students upon entry
to a university Physics course. To that end three possible indicators of ‘at risk’ students
are analysed: (i) Mathematics ‘A’ level module scores, (ii) Force concept inventory results
and (iii) Mathematics diagnostic test marks. The results of all three assessments have been
compared with the end of semester examination results for mathematics and for dynamics.
Neither ‘A’ level module grades, nor the force concept inventory appear to be good indi-
cators of potential weak performance. The mathematics diagnostic test, taken upon entry
to university, is slightly better but the nature of the data precludes it from being used as a
reliable predictor of weak final examination performance.

The School of Physics and Astronomy at the University of Manchester has an annual intake of approxi-
mately 230 students. The majority of these students enter with grades of at least AAB in three subjects at
‘A’ level (the final English high school examination), including both mathematics and physics. (Grades
at A level range from A to E, A being the highest, with an unclassified grade U.) During their first week
at university, but before the start of teaching, all physics students take a mathematics diagnostic test. The
aim is to identify those students who may be weak or rusty at mathematics - they may have had a gap
year for instance - in order to offer them extra mathematics support during their first year.
This type of mathematics diagnostic testing upon entry to university is currently relatively common
practice in the UK for physics, mathematics and engineering students (LTSN, 2003); we have been doing
it for thirteen years now at Manchester. However, there are still some students each year who fail the
mathematics exam at the end of the first semester who were not identified as ‘at risk’ by the diagnostic
test. These students are generally those who also struggle with the dynamics (Newtonian mechanics)
course in the first semester because of the mathematical nature of this subject. Therefore in 2008-09 we
started an investigation whether there might be a better way of identifying ‘at risk’ students when they
arrive at university.

Three possible ‘at risk on entry’ indicators have been considered:

(i) Mathematics ‘A’ level scores.

The UK Mathematics ‘A’ level comprises six modules; four compulsory core modules and two
option modules. The academic year 2008-09 was the first year that universities in England were
provided with full data on students’ individual module results for their Mathematics ‘A’ level.

(ii) Mathematics diagnostic test marks

(iii) Force Concept Inventory (FCI) results

The Force Concept Inventory is a test devised by David Hestenes (1992), which analyses con-
ceptual understanding of Newtonian mechanics. All the 2008-09 students undertook the FCI test
during their first week at university, and again during their mid-semester study week.

In addition to considering the students’ entrance qualifications and their performance on tests undertaken
during their first week at university, we also looked at the information we could obtain a little later. We
considered the results of mid-term tests which the students sat in semester one after just five weeks of
study, at which point we also re-administered the FCI test.
Finally, we investigated whether there was any correlation between students’ end of module examination
results with their participation in tutorials. All first year students are timetabled for one hour’s mathemat-
ics and dynamics tutorials each week. These tutorials are generally delivered by postgraduate students.
The first year undergraduates have weekly problem sheets in mathematics and dynamics and they submit
their work for marking before the tutorial so that any difficulties are identified by the tutors and can be
discussed in the tutorials. Students’ participation in this process is recorded in terms of their submission
of work for marking and their attendance at tutorials.

Complete data were not available for all students in the 2008-09 cohort. The most limited data set was
for the A-level modules, which contained data for only 167 out of 230 students. No A level data were
available either for those students who had taken a ‘gap’ year between school and university. Some
of the A level data were suspect and were discarded. Of course, not all students enter the university
with A levels: some students had completed the University’s Foundation, or “Year 0”, course, and most
international students enter with alternative qualifications. In addition there were inevitable absences
from some of the in-university assessments.
No data on a particular student were discarded if they were not complete. As a consequence some graphs
contain more data points than others, because in each case all the available data have been included. We
believe this is justified given the large number of students and hence adequate statistics.


3.1.1 Mathematics A level Grades Initial investigation focussed on the grades students had ob-
tained for the four core mathematics modules, known as C1, C2, C3 and C4 because all students take
these modules. In addition to the core modules, students take two optional modules. All six modules
are studied during the last two years at high school. Students are allowed to retake the module exams
multiple times until they are combined in a single A level upon leaving high school.
Fig. 1 shows that there is a significant distribution in the grades for the C4 module only (a grade A is
the highest grade and a grade ‘U’ represents the lowest “unclassified” mark.). Since most students take
this module at the end of their A level course, it is the one which they are least likely to have had the
opportunity to resit.
We therefore looked to see whether there was any correlation between the C4 grade and the students’
performance in the final end of module mathematics examination taken at the end of semester one. Fig. 2

C1 C2
150 150

100 100

50 50

0 0
C3 C4
120 100
100 80
80 60
20 20
0 0

Figure 1: Mathematics A level Core Module Results. (The number of students is plotted on the
y-axis against the module grade on the x-axis, for each of the core modules)

shows that there is weak correlation (linear correlation coefficient, r = −0.42). (Data points which
represent more than one student are larger and of deeper colour. This representation applies to all the
scatter plots in this paper.) It is clear that the C4 grade could not be used to identify reliably those students
who are likely to fail the mathematics examination. The pass mark for the examination is 40%, and there
are six students who obtained a grade A in their C4 module yet failed the end of module examination. At
the other end of the scale, there are eight students who passed the mathematics examination despite only
achieving a grade D or lower for C4.

3.1.2 Mathematics Diagnostic Test Fig.3 shows the mathematics diagnostic test results versus
mathematics examination performance. The blue line is the line of equal marks. The brown line is the
normal least-squares regression line for y = ax. The green line is an orthogonal regression line. [We
prefer to use orthogonal regression, since our data has scatter in both abscissa and ordinate.] These data
give a stronger correlation (r = 0.52) than those of Fig. 2. However, it is much easier to identify those
students who will perform well than those who are likely to perform badly.
The 20 students with the lowest diagnostic test marks, i.e. those with 46% or less, were invited to attend
additional weekly mathematics support sessions. Not all students took up the offer, but those who did
passed the Mathematics examination.
Again it can be seen that there are students who did very well on the diagnostic test, and hence were not
invited to attend the maths support sessions, but they failed the mathematics examination.

3.1.3 Mathematics mid-term test Next we investigated whether it would have been possible to
have identified ‘at risk’ students from their mid-term mathematics test results. Fig. 4 shows the math-

Math exam 
Figure 2: Mathematics examination mark versus C4 module grade.

Maths Exam 

0 20 40 60 80 100
Figure 3: Mathematics examination marks versus mathematics diagnostic test marks.


Math Exam 

0 20 40 60 80 100
Maths Mid 
Figure 4: Mathematics mid-term test marks versus mathematics examination results.

ematics mid-term test marks against the final mathematics examination results. The various the lines
drawn on the graph are colour coded in exactly the same manner as those in Fig. 3. The linear correla-
tion coefficient is r = 0.53 in this instance. It is clear from this figure that it is not possible to identify
the ‘at risk’ students even half way through the semester. There are still students who did well on the
mid-semester test, who ended up failing the mathematics examination. In fact there would appear to be
more students in the lower right hand region of this plot than in Fig. 3.
The fact that the orthogonal regression line (green) is below the 1-1 correspondence line (blue) might
suggest that the mid-term test was a little too easy. Maybe if it were tougher, identification of the weaker
students might be more likely. Some students could have been lulled into a false sense of security by
their good mid-semester marks. On the other hand, some of the students who performed badly on the
mid-term test but did well on the final examination may have been spurred on to increase their effort.
Fig. 5 shows that there are very few students who performed badly on both the diagnostic and the
mid-term tests (linear correlation coefficient, r = 0.5). There are many students who did well on the
diagnostic test but not on the mid-term test.


3.2.1 Mathematics A level Grades There are several optional mechanics modules in the UK Math-
ematics A levels, and a student may study anywhere between zero and five modules of mechanics, if they
also take Further Mathematics A level. The majority of first year physics undergraduates at the Univer-
sity of Manchester have generally studied either one or two mechanics modules, but there are a number
who have not done any. In the 2008-09 cohort there were three students who had done four mechanics
modules. Fig. 6 shows the distribution of grades obtained within the four mechanics modules.
The following scoring system was used to produce a numerical scale to represent the level and extent
of students’ previous experience and demonstrated ability in A level mechanics upon entry to university.

Math Mid 
0 20 40 60 80 100
Figure 5: Mathematics mid-term test marks versus diagnostic test marks.

M1 M2
120 50
100 40
80 30
60 20
20 10
0 0
M3 M4
12 1.0
10 0.8
8 0.6
4 0.4
2 0.2
0 0.0

Figure 6: Distribution of mechanics module grades within A level mathematics. (The number
of students is plotted on the y-axis against the module grade on the x-axis, for each of the
mechanics modules).


Dynamics exam 

0 5 10 15 20 25 30
Mechanics A
Figure 7: A level mechanics module(s) score versus dynamics examination results.

For each A level Mathematics mechanics module that a student had studied, they were given a score of:
6 for a grade A, 5 for a grade B, 4 for a grade C, 3 for a grade D, 2 for a grade E and 1 for a grade U,
i.e. unclassified. The scores for each mechanics module studied were then added to give an overall score
for experience and ability in A level mechanics. Those students who had not studied any mechanics
modules within their A level mathematics scored zero. Fig. 7 shows the resulting score for each student
versus their end of module examination mark for dynamics. As usual, the blue line represents the 1-1
correspondence and the green line is the orthogonal regression line. There is little correlation but the A
level mechanics module(s) score does appear to act as a reasonable predictor of the lower achievement
level in the dynamics exam.

3.2.2 Force Concept Inventory Scores Fig. 8 shows the results of the Force Concept Inventory
test, which the students sat during their first week at university, against their mark for the end of mod-
ule dynamics examination. The latter was also held at the end of semester one, like the mathematics
examination. Again there is very little correlation (r = 0.4) between the FCI score and the dynamics
examination mark.
The variation of Force Concept Inventory scores with achievement in the optional mechanics modules
studied as part of the mathematics A level has also been investigated. This comparison is shown in Fig.
9. There is very little correlation again upon entry. This might indicate that the students are able to score
quite well on the Maths A level mechanics modules without a complete understanding of the physical
concepts of Newtonian mechanics.
It might be possible to argue that the A level mechanics score gives a reasonable prediction of the min-
imum achievement level attained on the Force Concept test, as it does for the end of module dynamics
examination (Fig. 7) but otherwise it is of little use as a predictor of ‘at risk’ students.


Dynamics Exam 

0 20 40 60 80 100
FCI 1 
Figure 8: Force Concept Inventory scores upon entry to university versus dynamics examination

FCI test 1 

0 5 10 15 20 25 30
Mechanics A
Figure 9: Mechanics A level module scores versus Force Concept Inventory scores.

FCI test 2  60
0 5 10 15 20 25 30
Mechanics A
Figure 10: FCI score after five weeks instruction in Newtonian mechanics versus mechanics A
level module scores.

When the FCI test was repeated after just five weeks of the dynamics course, the mean score improved
significantly from 71% upon entry to 81%, with those students who had not studied any mechanics
modules previously at A level showing the most significant improvement. This can be seen in Fig. 10.
This suggests that we can remedy a lack of conceptual understanding quite quickly.


Figs. 11 and 12 show the performance in the mathematics and dynamics end of module examinations
respectively, versus tutorial participation. Neither graph shows much correlation, but that for dynamics
(r = 0.34) is very much weaker than that for mathematics (r = 0.52). Also, it is noticeable that all those
students who had 100% tutorial participation passed the mathematics examination but the same is not
true for dynamics.

The findings of the present analysis are in reasonable agreement with those of Lee et al (2008) who at-
tempted to predict the overall performance of 107 first year engineering students at Loughborough Uni-
versity, as well as their performance in first year mechanics. They considered fourteen possible predictive
variables including mathematics and mechanics diagnostic tests and concluded that the mathematics di-
agnostic test was a much more significant predictor than either ‘A’ level points score or mathematics A
level overall grade. Interestingly, it was also more significant than an in-house mechanics diagnostic test
in predicting performance in first year mechanics. They also found that whether or not students availed
themselves of extra mathematics support was also a significant predictor of successful performance.


Maths Exam 

0 20 40 60 80 100
maths tut participation 
Figure 11: Mathematics examination marks versus tutorial participation.

Dynamics Exam 

0 20 40 60 80 100
Maths tut participation 
Figure 12: Dynamics examination marks versus tutorial participation.

Clearly they will be many other factors which influence the performance of students at university, for
example personal, financial, none of which we have considered here.

It is very difficult to identify those students who are at risk of failing the end of module mathematics or
dynamics examinations, either when they begin university, or even after the first five weeks of the first
A level grades for the core modules in mathematics do not appear to be a good indicator for future
performance in mathematics but it must be acknowledged that this conclusion is based upon a relatively
small sample of students and from just one academic year.
The mathematics diagnostic test taken upon entry to university is a slightly better indicator, but whilst it
does identify some of the ‘at risk’ students, it does not identify them all and provides no guarantee that
all those students who would benefit from extra mathematics support, receive it.
Those students who are identified as at risk by the diagnostics test and do attend the extra mathematics
support sessions generally perform satisfactorily in the final examination. It therefore behoves us to
encourage even more strongly, those ‘at risk’ students who are identified by the diagnostic test but do not
avail themselves of the extra support, to do so.
There is little correlation between the number of mechanics modules studied in A level mathematics and
the grades obtained in them, and performance in the dynamics examination at the end of the first semester.
Neither is there much correlation of the mechanics A level module scores with the understanding of the
concepts of Newtonian mechanics as measured by Hestenes’ Force Concept Inventory. However, there
is also very little correlation between the Force Concept Inventory scores upon entry to university and
the final examination performance in dynamics.
There is some correlation of performance in the mathematics examination with tutorial participation, but
little correlation of performance in the dynamics examination with tutorial participation.

LTSN (2003) Learning and Teaching Support Network, Diagnostic testing for mathematics, MathsTEAM
Project Booklet, ISBN0704423731, also at,
last accessed 16/11/09.
Hestenes, D., Wells, M and Swackhamer, G. (1992). Force Concept Inventory. The Physics Teacher, 30,
141-151. Revised 1995 Halloun, I., Hake, R., Mosca, E.

Lee, S., Harrison, M.C., Pell, G., and Robinson, C.L., (2008) Predicting performance of first year engi-
neering students and the importance of assessment tools therein, Engineering Education: Journal of the
Higher Education Academy Engineering Subject Centre, vol 3, no.1.


Morag M. Casey[1] & Stephen McVitie

Dept. of Physics & Astronomy, University of Glasgow, GLASGOW G12 8QQ, UK

At the beginning of academic year 2007-08, staff in the Department of Physics & Astronomy at the
University of Glasgow began a systematic set of improvements to the level 1 physics undergraduate
class in order to address concerns about student academic performance and progression. The effects of
low attendance at compulsory continuous assessment components had already been observed to have
a significant effect on the academic performance of students sitting in the middle of the grade curve.
However, analysis of data from the 2007-08 class showed that even some nominally high-achieving
students achieved lowered grades due to the effects of lowered attendance. By the start of the academic
year 2008-09, increases of approximately 8% and approximately 10% in the pass rate and progression
rate respectively had been observed and it was believed that the academic and other support measures
put in place in 2007-08 had played a part.
This paper contains an overview of work published by the authors in greater detail in Eur. J. Phys. 30
(2009) pp 1153-1161.

It is generally accepted that the retention and associated completion rates for first year classes are an area
of concern for UK universities (Crosling, Thomas & Heagney 2008); level 1 physics at the University of
Glasgow is no exception. Classes are often large and, as a result, student integration on academic and
social levels can be difficult to achieve; from a staff perspective it can also be difficult to track students
who disengage from the process of studying. It had already been observed that students attaining lower
grades within the level 1 physics class frequently failed to present themselves for all available assessment
tasks. Until this point, however, the effect of incomplete presentation for assessment on the academic
performance of all students in the class had not been quantified
In order to address these issues, staff in the Department of Physics & Astronomy at the University of
Glasgow began a systematic set of improvements of the level 1 physics undergraduate class. One of the
authors (Casey) was charged with the responsibility of directing learning support and worked closely
with the class head (McVitie).


In order to understand how best to address the issue of supporting physics undergraduates, it is informa-
tive to place the physics degree structure in the wider contexts of both undergraduate science degrees at
the University of Glasgow as well as the Scottish educational system.


The structure of the Scottish schools’ educational system (The Scottish Government 2008) is such that
students are normally able to enter university after a minimum of twelve years of primary and secondary
education. Therefore, undergraduate honours science degrees at Scottish universities normally take a
minimum of four years to complete. Academically able students may select to complete a further year
of advanced study at undergraduate level.


The University of Glasgow, in common with some other Scottish universities, grants students entry
to a Faculty rather than an individual course (University of Glasgow 2009). Students study a broad-
based level 1 curriculum and degree specialisation occurs from level 2 onwards. As a result, typically
approximately 50% of level 1 physics students intend to study the subject beyond level 1. Thus, after
taking account of the level 1 pass rate, a direct progression to level 2 physics of approximately 37% is
typical. The question was not whether this should be improved but whether that aim was possible, given
the constraints of Faculty Entry.


In 2007-08 level 1 physics consisted of two modules (P1X and P1Y), each assessed in separate semesters.
Students were required to achieve at least D-grade in both modules in order to progress to level 2 physics.
Students were also required to achieve satisfactory grades in the level 1 maths classes, taught separately
in the Department of Mathematics.
The relevant statistic for level 1 physics is that fraction of the class attaining D-grade or above by the
end of the academic year. On that basis, the average notional ‘pass rates’ over a 5-year period from
2002-2007 were 77.2% for P1X (σP 1X = 1.9) and 72.4% for P1Y (σP 1Y = 2.2). The aims, therefore,
were to raise both pass rates and narrow the gap in performance between P1X and P1Y.


It had already been observed that the effect of incomplete presentation for continuous assessment tasks
was quite large for the weaker students (those who normally achieved D/E grades). However, another
observation had been made during the marking of final exam papers: some students did not answer all
the questions on the paper, thus throwing away 7.5% of their final mark for each question not attempted.
Similarly, every time students missed a laboratory class or other continuous assessment task they were
throwing away anything between 1% and 4% of their final grade.
As a means of quantifying this effect, a scatter-graph of the percentage of assessment that students pre-
sented (x-axis) for against their final percentage mark (y-axis) was made. The plot is shown in Figure 1
using the Physics 1Y 2007-08 data. This was the first year for which a comprehensive attendance data
set existed and, therefore, post-dates the introduction of attendance monitoring described in Section 4.1;
the explanation is given here for clarity. Similar trends trends also appeared in the 2007-08 Physics 1X
data set
In Figure 1, triangles A, B, C and D indicate four regions of interest in the data. The maximum mark
a student can attain must be equal to or less than the amount of assessment they attempt and, therefore,




Final Mark (%)




30.0 C


0.0 20.0 40.0 60.0 80.0 100.0

Assessment Attempted (%)

Figure 1: Academic performance versus the percentage of assessment attempted for students in
Physics 1Y, 2007-08.

region A must remain unpopulated. The D/E-grade boundary for Physics 1Y is set at 40% and, there-
fore, regions B and C indicate respectively those students who had ‘passed’ and ‘failed’ Physics 1Y. The
significance of region C is that students whose academic performance lies within this region could been
placed in region B (a ‘pass’ grade) had they attempted more assessment tasks. Students whose academic
performance placed them in region D could not have exceeded a 40% final mark even if they had max-
imised their attendance. This effect is present at all grade boundaries and, in fact, a small number of
students slip through two major grade boundaries due to non-attendance. This has a more subtle effect
than might first be imagined. Although the grades a student achieves at level 1 do not contribute to
their final degree classification, they do determine entrance (or otherwise) to the advanced honours study



The primary aim in introducing a comprehensive set of improvements was to enhance the Department’s
effective learning environments by identifying students at risk of disengaging from the process of learn-
ing. Secondly, it was hoped that the provision of extra targeted academic support would assist all students
in achieving their best possible grade (Casey 2007, Casey & McVitie 2009). The hypothesis was that
poor attendance at compulsory attendance components was one cause of lowered grades in the student
population and so it was decided that attendance should be taken at every class component.
All students at the University of Glasgow are issued with credit-card sized identification cards which
have a barcode printed on them. Portable handheld barcode scanners were obtained from a commercial
outlet (Opticon Sensors Europe 2003) and used to monitor attendance at every lecture, laboratory and
continuous assessment task. Using this attendance data, students were contacted on every occasion that
they missed a compulsory assessment task. The lecture attendance data was used to further inform
student attendance profiles.


The specific aims in introducing extra academic support to level 1 physics were to address the historic
drop in pass rate between P1X and P1Y as well as improve the performance of the weaker students (those
with C/D grades).
Traditional methods for extra academic support typically consist entirely of a series of extra revision
classes or large-class tutorials. However, revision classes are usually optional and tend to attract the only
academically committed who do not necessarily need the extra help (Biggs 1999). It was decided that
all students would be able to make use of optional drop-in tutorials run by a few members of staff who
would provide one-to-one tuition. In addition, those students who appeared to be underachieving were
sent personal invitations to attend the tutorials.
Attendance at the drop-in tutorials was typically ∼ 10% of the level 1 physics class, and consisted
primarily of B-D grade students. Those who attended the tutorials showed an average increase of ∼ 3%
between P1X and P1Y grades and some managed an increase of ∼ 15%. With this encouraging outcome,
provision of drop-in tutorials was expanded in 2008-09 and similar results were observed.

Figure 2: Percentage of the class attaining A-D passes after the resits in Physics 1X and Physics 1Y by
academic year.



In terms of progression to level 2 physics, students must attain at least a grade-D (40% or above) in both
Physics 1X and Physics 1Y although this can include results from the resit exams. Data from years 2002-
03 to 2007-08, shown in Figure 2, show that both significant improvements in pass rates for Physics 1X
and Physics 1Y and a reduction in the historical gap in attainment between the two modules had been
achieved. Compared to the historical averages, the P1X passrate had increased by 5.7% (and 3 σP 1X ) to
82.9% while P1Y had increased by 8.5% (and 4 σP 1Y ) to 80.9%.


At the start of 2007-08, 50% of level 1 students had self-identified an ‘intention to progress’ to level 2
physics. Based on the pass rate, a direct progression rate of ∼ 40%, in keeping with previous years,
was predicted. In fact, data from the level 2 class for the following year demonstrated that the direct
progression rate was 47%, a significant improvement on previous years.


The main aims in the restructuring of the level 1 physics course appear not only to have been achieved
but, on some levels, to have been surpassed. The 2007-08 pass rate in both Physics 1X and Physics 1Y
increased by approximately 8%, a move exceeding 3 standard deviations from the historical averages for
the classes. More significantly, the direct progression rate from level 1 physics to level 2 physics was
increased by approximately 10%.
In 2008-09, the University of Glasgow introduced significant changes to the timetabling of all its under-
graduate classes. One consequence was that a decision to merge the Physics 1X and Physics 1Y modules

into a single year-long course was taken. As a result, it has been not yet been possible to make direct
correlations between the 2007-08 data set and subsequent academic years as some patterns of student
attendance and performance appeared to relate primarily to the existence of single-semester modules and
have not yet been fully quantified. A systematic quantification of data pertaining to students taking only
one or other of Physics 1X or Physics 1Y must be carried out before comparison with the 2008-09 data
set is made. This is currently in progress.
On a broader note, it is possible to report that the 2008-09 pass rate for the full-year course was approxi-
mately 10% above the combined pass rate for Physics 1X and Physics 1Y. This is relevant because passes
in both modules were required for progression to level 2 physics. Secondly, the direct progression rate
from the 2008-09 Physics 1 class to the level 2 physics class was approximately 13% above the 2006-07
Future work will centre around methods to streamline the data capture techniques so that these may be
rolled out to other classes within the Department, notably level 2 physics. In particular, an investigation
into understanding academic performance in level 2 physics and the related level 2 to level 3 progression
statistics is now being undertaken. A number of longitudinal studies will be made, starting with the
2007-08 level 1 cohort.

Biggs, J. (1999), Teaching for Quality Learning at University, SRHE / Open University Press, Bucking-
Casey, M.M. (2007), Supporting level 1 physics & astronomy students at the University of Glasgow, in
‘Proceedings of Frontiers of Physics Education, GIREP-EPEC Conference 2007’, Sveuc̆ilis̆na Knjiz̆nica
Rijeka, pp. 283-287. ISBN: 978-953-55066-1-4.
Casey, M.M. & McVitie, S. (2009), ‘Academic performance and student engagement in level 1 physics
undergraduates’, European Journal of Physics, 30, 1153-1162.
Crosling, G., Thomas, L. & Heagney, M., eds (2008), Improving Student Retention in Higher Education:
The Role of Teaching and Learning, Routledge, Abingdon. ISBN: 978-0-415-39921-0.
Opticon Sensors Europe (2003), SQA Attainment and School Leaver Qualifications in Scotland: 2006/07,
Statistics Publication Notice: ISBN: 1479-7569. URL:

University of Glasgow (2009), Undergraduate Prospectus, University of Glasgow, Glasgow. URL:

Sally Jordan[1] & Philip Butcher
OpenCETL, The Open University, Milton Keynes, United Kingdom, MK7 6AA.

This paper describes the use of online interactive computer marked assignments (iCMAs) in a range of
physics and general science modules at the UK Open University (OU), and an evaluation, jointly funded
by the Physics Innovations Centre for Excellence in Teaching and Learning (piCETL) and the Centre
for Open Learning of Mathematics, Science, Computing and Technology (COLMSCT) into student en-
gagement with these iCMAs. Although the work has been done in the context of one particular large
distance-learning university, the software used is open source and the evaluation has wider implications
for those seeking to find the most appropriate way to use e-assessment to support their students’ learning.
The Joint Information Systems Committee (JISC) define e-assessment as ‘the end-to-end electronic as-
sessment processes where ICT is used for the presentation of assessment and the recording of responses’
(JISC, 2009). There are many exciting ways in which e-assessment can be used to support learning,
including the use of e-portfolios and the assessment of students’ contributions to online discussions
(Buzzetto-More and Alade, 2006). Here we describe the use of online quizzes, where a student enters his
or her responses and receives instantaneous feedback. Note however that the range of question types in
use goes far beyond the selected response multiple-choice questions that are commonly associated with
assessment of this type.

Throughout the Open University’s 40-year history there has been some blurring of the summative (‘for
measuring’) and formative (‘for learning’) roles of assessment. OU undergraduate students are typically
(though no longer exclusively) adults, studying part-time alongside other commitments and they have a
wide range of entry qualifications from previous higher education qualifications to, literally, none. Many
have not studied for many years and so they may be particularly lacking in confidence. The students study
at a distance, but the OU’s model of supported distance education means that they are usually supported
by a tutor. This tutor will provide occasional tutorials (face-to-face or using a range of synchronous
and asynchronous electronic communication technologies) and be available to support student learning
by telephone or email; however a substantial part of the tutor’s contracted time will be spent in grading
and providing feedback on ‘tutor-marked assignments’ (TMAs). The fact that this task is described as
‘correspondence tuition’ reflects the importance that is placed on the feedback provided by tutors to
their students in this way; this is particularly important in a distance-learning organisation, where many
students never meet their tutor and opportunities for informal feedback are extremely limited. However
TMA scores have usually contributed substantially to students’ overall course grades.
The use of e-assessment also has a long history at the Open University. TMAs have long been supple-
mented by computer-marked assignments (CMAs), initially comprising solely multiple choice questions,
with students’ responses entered on machine-readable forms and submitted by post. Now, in many Sci-
ence Faculty courses, students complete online interactive computer-marked assignments (iCMAs) from
their own computers at home.

It is widely recognised that rapidly received feedback on assessment tasks has an important part to play
in underpinning student learning, encouraging engagement and promoting retention (see, for example,
Rust, Donovan and Price, 2005). Online assessment, with its instantaneous feedback, can been seen as
providing ‘a tutor at the student’s elbow’ (Ross, Jordan and Butcher, 2006, p125), of particular impor-
tance in a distance-learning environment. For high-population modules and programmes, e-assessment
can also deliver savings of cost and effort. Finally, e-assessment is the natural partner to the growth
industry of e-learning.
However, opinions of e-assessment are mixed and evidence for its effectiveness is inconclusive; indeed e-
assessment is sometimes perceived as having a negative effect on learning (Gibbs, 2006). There are more
widely voiced concerns that e-assessment tasks (predominantly but not exclusively multiple-choice) can
encourage memorisation and factual recall and lead to surface-learning, far removed from the tasks that
will be required of the learners in the real world (Scouller and Prosser, 1994).
Thus when e-assessment is used, careful evaluation is required to ensure that it is being used to optimum
effect and having a positive not a detrimental effect on student learning. There are a number of related
questions, for example:

• What sorts of e-assessment tasks have the best potential to support student learning?

• What mode of use is most effective: summative, formative, thresholded etc?

• What sort of feedback on e-assessment tasks is most useful?

• Does it matter that the feedback is generated by a computer rather than by tutors, peers or the
students themselves?


The iCMAs included in the current evaluation all use the ‘OpenMark’ web-based assessment system
(Butcher, 2006) within the Moodle virtual learning environment (Butcher, 2008). Question types include
those requiring free text entry of numbers, letters, and words in addition to more conventional drag and
drop, multiple choice, multiple response and hotspot questions. In most cases, students are allowed mul-
tiple attempts (usually three) at each question, with increasingly detailed and tailored prompts allowing
them to act on the feedback whilst it is still fresh in their minds and so to learn from it (Sadler, 1989;
Gibbs and Simpson, 2004), as illustrated in the simple question shown in Figure 1. The hints frequently
refer students back to relevant course material (which might be a book, a video sequence or an inter-
active activity). Feedback can also be provided on the student’s demonstration of learning outcomes
developed in the preceding period of study. OpenMark has good accessibility features and wherever
possible, questions exist in several variants. In summative use this enables different students to receive
different assignments, whilst in formative-only use, the different questions provide extra opportunities to
The range of question types has been extended still further to include those in which students have to
construct a response in the form of a phrase or sentence of up to 20 words. The answer matching for
these questions was initially written using a commercial linguistically-based authoring tool (Jordan and
Mitchell, 2009); answer matching of equivalent accuracy can now be obtained using OpenMark’s own

Figure 1: A simple numerical OpenMark question, with increasing feedback given at each attempt.

‘PMatch’ algorithmically-based system. Whichever software is used, the answer matching is developed
using previous responses from students, and the questions sit within the OpenMark framework, so mul-
tiple attempts are allowed, with increasing feedback (Figure 2).
Within the Open University Science Faculty, iCMAs are embedded in courses’ assessment strategies in
a wide variety of ways, for example:

Case 1: The level 1 10 CATs point course S151: Maths for Science has a single summative
OpenMark end of course assignment (available to students for 5 weeks) with instantaneous
feedback given to students on individual questions. The students are not told their mark. A
similar practice assessment (Case 1 PA) is available for the duration of the course.

Case 2: The introductory 10 CATs point course S154: Science Starts Here has two very
short tutor marked assignments and two very short summative but low stakes iCMAs (Case
2 SA) plus a purely formative iCMA (Case 2 PA) available for the duration of the course.

Case 3: The Science Faculty’s major 60 point foundation course S104: Exploring Science
(the introduction to physics, chemistry, biology and Earth science) has nine summative but
low stakes iCMAs, eight tutor marked assignments and an end of course assignment.

Figure 2: A short answer free-text question, with increasing feedback given at each attempt.

Case 4: The level 3 physics course SM358: The quantum world has regular formative
iCMAs, clearly embedded within the course alongside tutor-marked assignments. Students
are told that there will be similar questions to those in the iCMAs in the final examination.
From 2010, all TMAs and iCMAs on this course and other level 3 physics and astronomy
courses will be formative only, but thresholded i.e. students will have to complete a certain
number of the assignments, where satisfactory completion of an iCMA will be deemed to
be a score of 30% or more.

Case 5: iCMAs are used for diagnostic purposes in a series of ‘Are you ready for?’ quizzes,
designed to help students to decide whether or not they are sufficiently prepared to study a
particular course.

The Moodle Gradebook enables students to monitor their own progress, encouraging sustainable self-
assessment practices (Boud, 2000), and the tutor’s view of the Moodle Gradebook encourages discussion
between students and tutors.

Figure 3: The number of students attempting each question in a summative but low-stakes iCMA
(Case 3).

Evaluation methodologies have included surveys of student opinion, observation of students in a usability
laboratory, a ‘success case method’ approach (Hunter, 2008) and a comparison of accuracy of computer
and human marking (Jordan and Mitchell, 2009). The systems have been found to be robust and accurate
in marking and most students report enjoying the iCMAs and finding them useful. However there are
some anomalies. For example, whilst more than 90% of students report finding the feedback provided
useful and, when observed in a usability laboratory, some students were seen to read the feedback and
then to adjust their answer in a sensible way, others do not make use of the feedback in the way that
iCMA authors would hope.

The current project is seeking to ‘observe’ student behaviour remotely, by means of a quantitative and
anonymised analysis of the data captured when students attempt iCMAs. Tools have been produced to
extract summary information from the databases. It should be noted that the quantitative data extracted
in this way are reliable since the student populations of the courses in question are large, for example,
the course identified as ‘Case 3’ above has two presentations each year with 1500-2000 students per


Not surprisingly, when iCMAs are summative (even if low stakes), students are highly motivated to
attempt all the questions, as shown in Figure 3 below.
However, when an iCMA is formative-only, usage drops off in a characteristic way, as shown in Figure 4.
The top bar-chart in Figure 4 shows the number of students who attempted each question; the lower
bar-chart shows the number of separate uses of each question (so each user attempted each question an
average of three times). Note that this particular iCMA includes forty-two questions; usage drops off

Figure 4: The number of students attempting each question in a formative-only iCMA (Case 1 PA).

less for iCMAs with fewer questions, however if there are then several separate iCMAs spread over the
duration of the course, there is then a drop in use from iCMA to iCMA, resulting in a similar drop in use
from the first question in the first iCMA to the final question in the final iCMA. Typically, the number
of users has dropped to around half by half-way through the iCMA or course and to around a quarter by
the end. In addition, some students view the questions in the iCMA but never attempt any; for the iCMA
illustrated in Figure 4 and over the same time-scale, the iCMA was viewed by around 4500 students.
There appear to be particular aspects of iCMA design that can contribute to a marked decline in use
(which is not recovered in subsequent questions); this is often linked to questions that are deemed to
be complicated (perhaps with multiple boxes to complete) or time-consuming (though not necessarily
difficult) or which require the student to access a piece of multi-media or even perhaps just to use their
calculator (as illustrated in the example shown in Figure 5 below, which is question 19 in the iCMA under
consideration in Figure 4). However use can be encouraged by making it clear which questions relate to
which section of the course (as shown in the navigation panel to the left hand side of Figure 5); Figure 4
shows that students who had not attempted previous questions were nevertheless sufficiently motivated
to attempt the questions linked to Chapter 7 (starting with question 27) and Chapter 10 (starting with
question 39). Identifying which questions relate to which sections of the course (and reminding students
to attempt them at appropriate times, by notes in the course texts or website) is now considered to be
good iCMA design, although it is not practical in all situations, for instance when iCMA questions have
a deliberately synoptic role.

Figure 5: The iCMA under consideration in Figure 4, showing Question 19 and the navigation panel.


Summative iCMAs are usually made available to students for a period of several weeks and within that
time scale students are allowed to spend as long as they would like to on the questions; if a student closes
their browser and returns to an iCMA at a later stage, the system will remember where they were up
to. However most summative iCMAs have a ‘hard’ cut-off date; this is a deliberate policy, designed to
encourage OU students (who frequently have many competing demands on their time) to keep up to date
in their studies. In Case 3, the cut-off date for each iCMA is a few days after students are scheduled to
have completed their study of the relevant course material. Figure 6 shows that the cut-off date is clearly
effective in encouraging students to attempt the iCMA but most complete the questions in the few days
immediately before the cut-off date. The three graphs in Figure 6 are for an early question (Question 1),
a late question (Question 9) and the combined usage of all 10 questions in the iCMA; thus the behaviour
is similar for all questions.
The situation for Case 2 is rather different. The purely formative practice iCMA has 88 questions and
is available to students throughout the course’s 10-week duration. Figure 7 shows the number of actions
per day for an early question (Question 2), a question around half-way through the iCMA (Question
40), a late question (Question 88) and all the questions combined. The relatively uniform overall usage
appears to be attributable to the fact that students are attempting different questions at different times.
This iCMA, like the one shown in Figure 5, has questions linked to different chapters of the course, and
students are reminded after each chapter to try the relevant questions.

Figure 6: Number of actions on iCMA questions per day, for a summative iCMA with a hard cut-off
date (Case 3).

Figure 8 shows the number of actions per day for the summative iCMA for the same course (Case 2
SA), for an early question (Question 1) a late question (Question 9) and all the questions combined.
The course teams who produced the courses in Case 2 and Case 3 had designed the summative iCMAs
of the two courses (which are linked; Case 2 is a precursor to Case 3) to be similar; the iCMAs have
similar weightings, both have 10 questions, they both have hard cut-off dates and they are available to
students for a similar length of time. So it is surprising that Figure 6 is rather different from Figure
8; in the latter case students again appear to be attempting different questions at different times. One
possible explanation of this is purely that this is what students are advised to do. The questions in the
first summative iCMA assess three chapters of the course. On completing Chapter 2, students are advised
to attempt the relevant formative and summative questions, and similarly for Chapters 3 and 4.
Most of the courses included in the study have a final assessed component that is completed in the
student’s own home and with access to course material. Case 4 (the level 3 physics course) ends with
an examination, taken at an exam centre. It is clear from Figure 9 that many students are making use
of the iCMA questions on this course for revision, so it appears that use of formative iCMAs can be
encouraged simply by saying that the practice and feedback will be useful for the examination. Figure 9
shows the use of an iCMA primarily intended for use earlier in the year. Selections of earlier questions
had been combined together into three ‘revision’ iCMAs and these were also heavily used in the run up
to the examination, though the fact that some students chose to use the earlier iCMAs perhaps indicates
that they were using iCMAs in their revision of specific topics. The reasons for this behaviour will be
explored further by interview.


In addition to looking at all actions on a particular iCMA question, it is possible to inspect the way in
which individual students progress through the iCMA, and three typical student behaviours are shown
in Figure 10 (all for Case 2 SA). ‘Student 1’ and ‘Student 2’ are typical for all summative uses: many
students attempt all 10 questions on the very last day an iCMA is open (like Student 1) whilst some
attempt a question, then attempt another, then return to the first question etc. in an apparently chaotic
fashion, sometimes with a period of several days between consecutive uses of the same question (like
Student 2). However graphs such that shown for Student 3 were observed frequently for Case 2 SA

Figure 7: Number of actions on iCMA questions per day, for a formative-only iCMA (Case 2 PA).

Figure 8: Number of actions on iCMA questions per day, for a summative iCMA with a hard cut-off
date (Case 2 SA).

Figure 9: Number of actions on iCMA questions per day, for a formative-only iCMA on a course with
a final exam (Case 4).

Figure 10: Days on which three students made attempts at the questions on a summative iCMA (Case 2

but never for Case 3, and this is another illustration of the behaviour shown in Figure 8. This student
has attempted the 4 questions that assess Chapter 2, then presumably worked through Chapter 3 and
attempted that chapter’s questions, then similarly for Chapter 4.


Inspection of the actual responses entered by students, in particular to free-text responses in summative
use, has been used to learn about common student misconceptions (Jordan, 2007). Student responses can
also provide valuable insights into more general factors concerning the use of iCMAs by students.



Student responses to short-answer free text questions in summative use have generally been found to be:

• more likely to be correct

• more likely to be expressed as sentences (as requested in the course guide and the introductory
screen in Case 3)

• longer

than the responses to the questions in formative-only use. Figures 11 and 12 compare the length of
responses obtained to the question shown in Figure 2 (without the wording shown in italics).
In formative-only use (Figure 11) the peak at one word corresponds to the answer ‘balanced’ and the
peak at 3 words corresponds to the answers such as ‘they are balanced’ and ‘equal and opposite’. In
summative use (Figure 12) the peak at 4 words corresponds to answers such as ‘the forces are balanced’
and the peak at 8 words corresponds to answers such as ‘the forces acting on the snowflake are balanced’
and ‘there are no unbalanced forces acting on it’. It is quite common for the distribution of lengths to be
bimodal; in other questions there is sometimes a peak for answers that simply answer the question (e.g.
‘the force is reduced by a factor of four’) and another for answers that add an explanation (e.g. ‘the force
is reduced by a factor of four since it depends on dividing by the square of the separation’).
Unfortunately some excessively long responses were received (up to several hundred words) to early
summative versions of short-answer free text questions, and these frequently contained a correct answer

Figure 11: Distribution of length of 888 responses to the question shown in Figure 2 (without the
wording shown in italics) in formative-only use.

Figure 12: Distribution of length of 2057 responses to the question shown in Figure 2 (without the
wording shown in italics) in summative use (Case 3).

Figure 13: Distribution of length of 1991 responses to a question in summative use (Case 3) with filter
and additional wording on the question.

within an incorrect one. Responses of this type are recognised as being the most difficult for comput-
erised marking systems of any type to deal with and for this reason, from February 2009, a filter has been
introduced to limit the length of responses to 20 words. Students who give a longer answer are told that
their answer is too long and are given an extra attempt. The filter was initially accompanied by the text
shown in Figure 2 (You should give your answer as a short phrase or sentence. Answers of more than 20
words will not be accepted).
The introduction of the filter and explanatory text reduced the number of students who added text to pre-
vious answers without thought to the sense of the response so produced. It also dealt with the excessively
long responses that were difficult to mark, and increased the number of students giving their responses as
sentences. However, for all questions, the addition of the filter and explanatory text resulted in an overall
increase in median response length (see the distribution shown in Figure 13).
A possible explanation of this effect is that more students were heeding the advice to give their answer as
a sentence, now that this advice was given in the question. A less desirable explanation is that students
were interpreting the advice to use no more than 20 words as indicating that they should be writing
exactly or almost twenty words. From July 2009, the advice accompanying the filter has been changed
to ‘You should give your answer as a short phrase or sentence.’ The question shown in Figure 2 has
not been used since then, but length distributions for other questions illustrate that the latest change of
wording appears to have had the desired effect, reducing the median length and the number of responses
that are exactly or just under 20 words in length.



One of the anomalies of previous evaluations of e-assessment at the Open University and elsewhere is
that the vast majority of students report finding the feedback provided on iCMA questions useful, but yet
some are observed to make no use of it. In a first attempt to interpret evidence of actual behaviour, graphs
have been plotted to show the number of incorrect responses that were unchanged at second and third
attempt. Figure 14 illustrates the number of repeated responses for the same question in summative and

Figure 14: Blank and repeated responses for the same question in diagnostic, formative-only and sum-
mative use.

formative-only use (for Case 2) and in diagnostic use (Case 5). In the figure, grey shading indicates blank
responses, green shading indicates correct responses; red, orange or yellow shading indicates incorrect
responses; an identical colour from first to second attempt or from second to third attempt indicates an
unchanged response.
Thus it can be seen that a proportion of responses are left blank and in some circumstances almost half
are left unchanged for a second and third attempt even when users have been told that their previous
answer was incorrect. Not surprisingly, the proportion of blank and repeated responses varies markedly
with mode of use, with very little of this behaviour in summative use, more in purely formative use
and more still in diagnostic use. The proportion of responses that are blank and/or repeated also varies
considerably with question type: students are more likely to repeat responses when they cannot guess
the answer (in a multiple choice or drag and drop question). It is possible that students are deliberately
leaving responses blank or repeating them in order to receive the feedback provided at a later stage.
Reasons for this behaviour will be explored further by interview.

In general terms, students appear to engage with iCMAs in a deeper way when the questions carry
some summative weight. However, in summative use, students become obsessed with the minutiae of the
grading, as witnessed by many emails from students who believe - usually wrongly - that ‘the computer
has marked them incorrectly’. The use of thresholding or a ‘carrot’ (e.g. having similar questions in
an unseen exam) may provide an alterative mechanism for encouraging students to engage with iCMA
questions and so to learn from the feedback provided.
Things are not always as simple as you think they will be. At face value, use of the iCMAs in Case 3
and Case 2 SA should be similar, but this has been shown not to be the case. The differences appear
to be entirely attributable to students’ interpretation of what they have been told to do (in Case 2 SA
they have been told to do the questions after studying the relevant part of the course; in Case 3 emphasis
has been put on the importance of checking access to the iCMA in plenty of time, even if the student
doesn’t attempt any of the questions). Similarly, the increase in average length of responses to free text
questions in response to the instruction that responses should be no more than 20 words in length, points

towards a student interpretation that ‘no more than 20 words’ means ‘nearly 20 words’. The fact that
responses were more likely to be in complete sentences suggests that students may be more likely to
read instructions when they are provided within the question, rather than hidden away in an introductory
screen or in the course guide.


The results reported here are just some of the outcomes from the quantitative analysis of the data captured
when students attempt iCMAs. Other factors that are being investigated include the length of time
students spend in answering iCMA questions of various types and in different modes of use, and the
order in which they answer the questions. The effect of variables such as start date, finish date, elapsed
time and active time on performance will also be analysed, in order to investigate whether students who
engage with iCMAs in different ways perform differently. For example, are early and late completers
of iCMAs equally successful? Linked to this will be further investigation into whether students behave
differently in different situations. For example, do students exhibit similar behaviour when they are
revising to that when attempting an iCMA for the first time?
A comparison of iCMA scores, TMA scores and overall course performance will investigate whether
iCMAs are a good predictor of success in other areas.
In autumn 2009, online questionnaires were sent to students on SM358 The quantum world (Case 4)
and a range of level 1 courses, with the aim of discovering the reasons for some of the behaviour that
has been observed, for example students leaving responses blank and repeating unchanged responses at
second and third attempt. A subset of the students who completed the questionnaires will be interviewed.
From February 2010, four level 3 physics and astronomy courses (including SM358 The quantum world)
will have a radically different assessment strategy, with all assignments being formative-only but thresh-
olded i.e. students will have to complete a certain number of TMAs and iCMAs, where satisfactory
completion of an iCMA will be deemed to be a score of 30% or more. A similar quantitative analysis of
the responses to these courses’ iCMA questions, and a comparison with findings from the current work
(especially for SM358 in 2009 and for courses in which the iCMAs are in summative use) will be one of
the mechanisms used to evaluate the impact of the change in assessment strategy.
For further information on this project, and updates, see

The authors gratefully acknowledge the financial support of the UK Higher Education Funding Council
via the Centre for Open Learning of Mathematics, Computing, Science and Technology (COLMSCT)
and the Physics Innovations Centre for Excellence in Teaching and Learning (piCETL), the assistance of
many people associated with COLMSCT and piCETL, especially Spencer Harben and Richard Jordan,
and the co-operation of the course teams involved in the investigation.

Boud, D. (2000) Sustainable assessment: rethinking assessment for the learning society, Studies in Con-
tinuing Education, 22(2), 151-167.
Butcher, P.G. (2006) OpenMark examples. Available online at
index.shtml [accessed 14th Nov 2009].

Butcher, P.G. (2008) Online assessment at the Open University using open source software: Moodle,
OpenMark and more. 12th International CAA Conference, Loughborough, UK, July 2008.
Buzzetto-More, N.A. & Alade, A. J. (2006) Best practices in e-assessment. Journal of information
technology education, 5, 251-269.
Gibbs, G. (2006) Why assessment is changing, in C. Bryan and K.V.Clegg (eds) Innovative Assessment
in Higher Education. London: Routledge, 11-22.
Gibbs, G & Simpson, C. (2004) Conditions under which assessment supports students’ learning. Learn-
ing and Teaching in Higher Education, 1, 3-31.
Hunter, A.G. (2008) Using an online formative assessment framework to enhance student engagement:
a learning outcomes approach. Paper presented at the Improving Student Learning Symposium, Dublin,
3-5 Sept 2007.
JISC (2009) Accessed from, 14th
Nov 2009.
Jordan, S.E. & Mitchell, T. (2009) E-assessment for learning? The potential of short free-text questions
with tailored feedback. British Journal of Educational Technology, 40, 2, 371-385.
Jordan, S.E. (2007) The mathematical misconceptions of adult distance-learning science students. CETL-
MSOR Conference, Loughborough, 11-12 September 2006. Maths, Stats and OR Network, 87-92.
Ross, S.M., Jordan, S.E & Butcher, P.G. (2006) Online instantaneous and targeted feedback for remote
learners. In C. Bryan and K.V.Clegg (eds) Innovative Assessment in Higher Education. London: Rout-
ledge, 123-131.
Rust, C., O’Donovan, B & Price, M (2005) A social constructivist assessment process model: how the
research literature shows us this could be best practice. Assessment & Evaluation in Higher Education,
30(3), 231-240.
Sadler, D.R. (1989) Formative assessment and the design of instructional systems. Instructional Science,
18, 119-144.

Scouller, K.M. & Prosser, M (1994) Students’ experiences of studying for multiple choice question
examinations. Studies in Higher Education, 19(3), 267-279.


Robert Lambourne
piCETL, The Open University, Milton Keynes, MK7 6AA, UK

Students in general and more particularly those studying through distance education
can benefit from clear and concise explanations of scientific terminology of the
kind provided by an authoritative physics glossary. The evolution of glossaries
developed by the Open University to meet this need is described along with some
educational applications of glossaries and expectations for the future development
of this particular aspect of educational technology.

This paper reports work conducted over a number of years centred on the construction course-
based physics (and applied maths) glossaries to support student learning at the UK Open Uni-
The Open University (OU) is a distance teaching university with about 180 000 students and a
reputation for the production of high quality teaching resources in a range of media. It has a
national HQ at Milton Keynes, in the English midlands, and 13 regional offices that organize
local tutorial support for its geographically dispersed students.
A typical OU student is aged about 30, has a family and a full-time job and wants to study for
a degree part-time and at a distance. The OU provides a range of purpose written study materi-
als, assessments, tutorials, practical courses and exams leading to accredited degrees. However,
the typical OU physics student is unlikely to have regular contact with other OU physics stu-
dents with whom they can discuss physics. So, despite the benefits of modern communications
technology, each OU student still works substantially on his or her own.
As a consequence of this relative isolation, Open University students are in particular need of
support regarding the use of physics terminology. Physics texts (including those written for
OU students) are rich in words that must be precisely understood if physics concepts are to be
clearly communicated and physics itself is to be properly comprehended.
Some of these words, such as adiabatic, eigenfunction and entropy, are likely to be unfamiliar
to a new student. Others will be familiar but will be used in unfamiliar or technically precise
ways: wave, second, and energy are examples that come immediately to mind. All such terms
are certain to be properly defined and discussed when they are introduced in the relevant course
materials and their first appearance is usually highlighted in bold type so that students are made
aware that an important new term is being defined or a familiar term is being made more precise.

Nonetheless, students find it remarkably easy to forget or overlook such definitions and are
certainly in need of concise and effective reminders of the key definitions.
The Open University response to this need is to provide the students with extensive glossaries
tailored to the specific needs of each course. At first these glossaries differed little from typical
commercially produced dictionaries of physics terminology but over the 40 years of Open Uni-
versity physics teaching the course glossaries have been repeatedly defined and have developed
a number of features that make them valuable and worthy of attention in their own right.


The following short extract from an introductory OU physics text on particle mechanics illus-
trates the terminological challenge that faces the student.

Fx = −Cx (13)

This is Hooke’s law. The constant C is called the force constant, and x is the
displacement of the particle from the equilibrium position at x = 0. The minus
sign ensures that Fx is a restoring force, one that pulls the particle back towards
its equilibrium position.

As the emboldening indicates, this section of text introduces three new concepts; Hooke’s law,
force constant and restoring force. In addition, the text also includes several other concepts that
will have been introduced earlier, either in the course concerned or in a prerequisite course: con-
stant, displacement, force, particle and equilibrium position to say nothing of the mathematical
concepts implicitly required to interpret the equations.
The problem of language in physics teaching, and in science teaching more generally, is well
described in the book Words, Science and Learning by Clive Sutton (Sutton, 1992) where the
author says that “Teachers have tended to use words as labels - e.g. acid, vitamin, force -
rather than as a means of interpreting ideas.” The Open University response to the plethora
of technical terms that the student must negotiate is (or was in the case of older discontinued
courses) glossary definitions such as the following.
From a Level 1 (introductory) astronomy course:

electromagnetic spectrum The full family of closely related kinds of electromag-

netic radiation, usually arranged in order of increasing or decreasing wavelength.
Traditional divisions of the electromagnetic spectrum include (in order of decreas-
ing wavelength); radio waves, microwaves, infrared radiation, visible light, ultravi-
olet radiation, X-rays and gamma rays.

From a level 2 (intermediate) astrophysics and course:

Silk damping The dominant source of angular power in the angular power spec-
trum of the cosmic background radiation for multipole numbers in excess of l =
1000 (or angular scales less than 0.1 degrees or so). The effect arises from the sup-
pression of acoustic waves of very short wavelength due to the free movement of
photons between encounters with charged particles.

From a level 2 (intermediate) physics course, where italicisation is used to indicate other terms
that may be found elsewhere in the glossary:

amplitude In the context of an oscillation; the amplitude is the magnitude of

the maximum displacement of the oscillator from its equilibrium position. In the
particular case of a simple harmonic oscillator described by the equation x(t) =
A sin(ωt + φ), the amplitude is represented by the parameter A.

All the above examples use relatively simple ‘headword and body text’ format (also known as
term-explanation format), though the last is significantly more sophisticated in terms of its style
and structure. However, more complex forms of definition have been used as the next section
makes clear.

The development of physics glossaries at the Open University has been strongly influenced by
the work of the educational technologist R. S. Zimmer, particularly through the development of
relational glossaries. The defining properties of a relational glossary are:

1. Connectivity
2. Closure
3. Consistency

Detailed explanations of these properties can be found in (Zimmer, 1981) but put simply:
Connectivity implies that all significant logical relationships between terms are represented.
Closure implies that every defined term should be explained in common language or using terms
that are themselves defined in the glossary.
Consistency implies that there should be no conflicts between definitions even when they in-
volve several different aspects.
An example of a relational glossary is the one used for a 3rd level (advanced) course in space,
time and cosmology. The course involved the development of Newtonian mechanics, special
relativity and general relativity, so many ‘elementary’ concepts, such as mass and distance, were
progressively developed in the course making consistency a particularly challenging aspect of
the glossary. In this case the glossary was printed in a landscape format with different aspects
of definitions being explicitly identified. The following is just part of one of those relational
definitions which contained four aspects in its full form.

The ENERGY is a function E(x1 (t), x2 (t), v1 (t), v2 (t), t) 3 4.3
FUNCTION of an which in principle depends on time and the
isolated physical positions and velocities of the particles.
system of two par- As time passes by, the particles move and
ticles viewed from x1 (t), x2 (t), v1 (t), v2 (t) and t all change, but
an inertial frame of the energy function is chosen so that its value
reference remains constant. (This is the conservation
of energy.)
has a form which in Newtonian mechanics 3 5.3
completely characterizes the dynamical be-
haviour of the system. The energy function
is therefore such a fundamental quantity that
it is required to be form invariant under all
transformations listed under the principle
of relativity.
has a value which is assumed to be invariant 3 5.3
under the transformations of translation in
time, translation in space and rotations in
space. However, the value of E is not in-
variant under uniform linear boosts because
the value of the kinetic energy function of
a particle depends on the motion of the ob-
is, in Newtonian mechanics, the sum of two 3 5.3
terms; a kinetic energy function which de-
pends only on the speeds of the particles and
a potential energy function which depends
only on the distance between them. For ex-
ample, if two particles interact according to
Newton’s law of universal gravitation then
[See kinetic energy function, potential en-
ergy function and conservation of energy.]

The implementation of the relational aspect of a glossary can be checked using software or
by hand. The connectivity can even be measured numerically by the number of times a term
cites another term as a cross-reference. According to Zimmer, citation of a term in a subject
phrase is counted for each aspect of meaning in which the subject phrase participates and every
appearance of a term counts.
The numerical measurement of connectivity is simple only if the glossary is produced in the
subject-predicate format with aspects of a definition clearly separated. Despite this advantage
several glossaries have been produced that have tried to combine the advantages (and discipline)
of a relational glossary with the familiarity and simplicity of a conventional term-explanation


For the purposes of combining glossary formats it is useful to recognize the range of aspects of
meaning that are typically found in a glossary or even in a conventional physics dictionary. For
example, a force might be defined qualitatively (i.e. conceptually) as a push or a pull arising
in some specific situation, or it might be defined quantitatively in terms of a certain number of
newtons and a particular direction of action. Both are aspects of the definition of force.
A detailed examination of several existing glossaries showed that the following aspects of defi-
nitions had been included in various combinations.

1. Concepts

2. Units

3. Quantities

4. Properties

5. Laws, principles, hypotheses, postulates

6. Equations, inequalities, approximations.

7. Phenomena and effects

8. Processes

9. Devices and components

10. Conditions and special states

11. Constants and special values

12. Proposals, theories, conjectures, philosophical stances

13. Classification system

14. Fields of physics

15. Notations symbols and abbreviations

16. Problems

17. Cross references

Though this list is probably incomplete as far as the whole of physics is concerned, for the
range of (undergraduate) courses surveyed it represents the ‘parts of speech’ that were found
in the relevant course glossaries. Each aspect of definition that was examined could be seen
as positioning or helping to position the term being defined in relation to one or more of these

Work on the development of glossaries at the Open University is ongoing. On the one hand there
is a desire to combine all the course glossaries that are currently used into a single glossary so
that gaps, conflicts and inconsistencies between courses or across an entire programme of study
can be identified and rectified.
In pursuing this there is a desire to gain a more complete understanding of the best way to
structure a definition and to ensure that all of its aspects are properly represented with the
maximum of efficiency and clarity. This means that further analysis of existing dictionaries
and glossaries is required but also a wider survey of the work of psychologists and linguistic
philosophers as well as further reflection on the practice of physicists and the meaning of the
concepts they use.
Another line of development in the effective use of glossaries is in through their role in assess-
ment. It is well known that assessment often drives student learning. Convinced of the value
of glossaries in distance education some Open University exam setters have included questions
specifically devoted to the glossary work. In the introductory Open University physics course
S207 The Physical World, as much as a sixth of the total examination time is devoted to short
answer questions that relate directly to definitions. Not just simple “write down the definition
of . . . ” type questions that might be found in any exam paper but also more probing exercises
that require students to criticise and possibly improve a given definition in a way that shows real
understanding of the concept concerned.
Yet another direction of recent and future work concerns the learning and teaching implications
of electronic glossaries. Almost all Open University glossaries are already available in some
kind of electronic format. In some cases this is just a pdf version of an existing printed glossary
but in other cases there is a least hyper linking to other parts of the glossary or linking to and
from a glossary and the related course material. An even more ambitious example is shown in
Figure 1 which is a screen shot from the Physica programme developed by John Bolton. Physica
builds on the Mathematica program develped by Wolfram Research. It uses the Mathematica
help system to house a physics glossary. Figure 1 shows one of the definitions with a number
of cross-references indicated.

Figure 1: A definition from an electronic glossary developed as part of the Physica project.

More remains to be done here with a clear need to ensure that the student’s ability to easily
follow links should not allow them to become lost or even to find themselves following red
herrings or being led to dead ends.
Much remains to be done in this area and the author is pleased to acknowledge useful conver-
sations with Josip Slisko with whom he expects to undertake some of these tasks in the near

Sutton, C.R. (1992) Words, Science and Learning, Open University Press

Zimmer, R.S. (1981) The relational glossary: A tool for effective distance learning. In F. Percival, & H.
Wellington (Eds.), Aspects of Educational Technology XV, Distance learning and evaluation (pp. 73-79),
Kogan Page.

Robert J Lucas
PiCETL, The Open University

We are currently piloting a range of computer simulated science experiments as
3-D virtual environments. These are rendered on a PC in 3-D and use photographs
of specific parts of the actual apparatus as textures to add realism to the simula-
tion. In particular, photographs are used to represent the consequential views of
an experiment. These particular views may also be animated depending on the
state of the experiment. The work combines the photographic approach of the In-
teractive Screen Experiments (ISEs) with the advantages of a fully simulated 3-D
environment where the user can interact with the apparatus in a more natural and
intuitive way. The potential advantages are that users can quickly adapt to the envi-
ronment and in particular the controls. They gain realistic views of the physicality
of the experiment as they are not just seeing it from a particular viewpoint, but from
wherever they see fit to place themselves within the experiment’s scene. They are
immersed in the experiment in a way that mitigates some of the objections to online
as opposed to real laboratory experimentation. It is also the case that the results of
an initial calibration or setup carry over into the main part of the experiment. This
is perceived as an extremely important teaching element of Physics practicals as the
user learns that care in setting up an experiment is an essential part of being able
to get good results. Furthermore there is no need to represent scales, read-outs or
controls as separate parts of the interface; these can all be rendered at their correct
physical positions within the experiment. The first of these experiments based on
the use of a diffraction grating has been fully implemented and has been evaluated
with a Physics A level class. The application and its evaluation will be presented. A
more complicated experiment using a spectrometer has also been modelled which
raises issues of complexity. These issues will also be discussed.

2-D simulations are used extensively in Physics. We can find simulations of a simple pendulum
to an atomic reaction in this form. However, there are situations when 2-D is not enough and to
gain a true insight into how something behaves we need 3-D. This is the case for the programs
described below. The Celestial E-Sphere (2008) was developed to help students understand
celestial coordinates and various aspects of the motion of celestial bodies, and it is impossible
to see how this could have ever worked as anything but a 3-D program.
The Meade Simulator (Lucas & Kolb, 2009) is used to help students studying the Open Uni-
versity course, Observing the Universe (SXR208, accessed 2008), to gain familiarity with con-
trolling a telescope. It simulates the night sky, a computer controlled telescope and its hand

controller. It is the 3-D that gives this program the look and feel of the actual telescope and
enables the students to come to terms with how to control it, enabling them to advance much
more quickly when faced with the real thing. This has been in use for several years and has
proved itself to be of enormous value in getting the students to a stage where they can master
the controls of the telescope without wasting precious dark time at the observatory.
The 3-D Immersive Screen Experiments are a natural progression from these. Having imple-
mented these 3-D programs and then facing the 2-D Interactive Screen Experiments it was at
once obvious that the many difficulties that were inherent in the 2-D approach could be entirely
avoided by rendering these experiments in their own 3-D virtual world.


The idea of virtual laboratory resources that can be used on a computer is described in Hatherly,
Jordan & Cayless (2009) and in particular the Interactive Screen Experiments (ISE) and several
implementations are described, several of which are also discussed here.
ISEs (Bronner et al. 2009; Hatherley, Jordan & Cayless 2009) are perhaps the most common
form of experiment that students can interact with on a personal computer. These use pho-
tographs to give 2-D views of the actual apparatus in use during an actual experiment. They are
basically interactive movies and as such give a comforting feeling of reality. However, due to
the combinatorial explosion of the necessary photographs there often needs to be compromises
made in the degrees of freedom that the user can access. Despite this there have been many suc-
cessful implementations and the OU and other institutions are actively engaged in producing
these. This lack of ability to cope with multiple degrees of freedom leads to poor flexibility and
scope as pointed out in Altherr et al. (2004) where an ISE showing a Michelson-Interferometer
is compared to a video which scores the same low value for flexibility but actually does better
than the ISE for scope. The terms ‘scope’ and ‘flexibility’ are defined within this paper.
It is interesting to note that the following observation about simulations made in Gorghui et al.
(2009): this approach (simulations) can lead to the oversimplification of real experiments. As
a simple consequence of this fact, students may run into difficulties facing the complexity of the
real world. In Bronner et al. (2009) we find a similar but stronger statement that Simulations
are optimized to an idealized view on how nature would behave according to the accepted
theory. Although a simulation may do this to the detriment of the learning experience and this
leads to the oversimplification noted above, a simulation does not have to do this; it is free
to use photographs of actual experiments in the same way that an ISE does. The simplification
inherent in the paradigm of the ISE is of a different nature. The simplification is due the inability
to represent multiple degrees of freedom and it is this that has a direct bearing on the usefulness
of the ISE as a teaching tool.
Although ISEs claim not to be simulations - “Note that an ISE is emphatically NOT a simula-
tion” (Hatherley & Macdonald, 2009), in fact many of them are. It is emphatically not the case
that every frame that a user sees is a photograph of a state of an experiment, but is most usually
a photomontage where the selections, positions and orientations of the photographs are con-
trolled by a program that is simulating the behaviour of the original experiment. This is true of
the ISEs discussed here. Whether it is admitted that the ISEs are simulations or not, there is an

inherent oversimplification in the ISE paradigm due the inability to represent multiple degrees
of freedom and this has a direct bearing on the usefulness of the ISE as a teaching tool. This
point will be taken up later in the paper.
3-D Immersive Screen Experiments also use photographs of a real experiment but these are
reserved for consequential views such as a spectrum, rather than the mundane views, such as
the back of a voltmeter. The apparatus of the experiment is reproduced as 3-D models. These
models are programmatically animated to simulate their real-life behaviour. The user’s eye is
implemented as a software camera that can be positioned anywhere within the scene in exactly
the same way that most computer games are programmed. Indeed a games language is used
as it provides many useful facilities such as collision detection. Clearly this approach is much
more complicated to program than the 2-D ISE. In fact the modelling of the apparatus needs
engineering accuracy so that all parts work properly together. This is where there is a large
departure from the games approach, which mostly attempts to fool the user into believing that
the operation of a device is realistic by often exaggerating its behaviour, although there are an
increasing number of games, particularly of the racing type, that strive to emulate the actual
underlying physics.

To demonstrate some of the challenges of reproducing an experiment on-screen, a concrete
example of a simple experiment is used. Figure 1 shows the set-up for a diffraction grating
experiment as used on the OU level one course, Exploring Science (S104, accessed 2009). This
experiment can be constructed by the student on a dining table and used to obtain the angles of
various coloured light diffracted by the grating seen here mounted into a 35mm slide. A bedside
or office light illuminates the inside of a shoebox. A slit in the shoebox allows light to reach the
diffraction grating that then diffracts the different colours of the light into various directions.
A paper protractor is used to measure angles. A pin and length of cotton is a sighting device
that the user looks along when determining the direction of a particular colour. The angle of
diffraction can then be read from the protractor by examining the pin’s position. The diffraction
pattern consists of a bright white central region with rainbow coloured regions on either side
that repeat themselves several times. The white pattern in the centre is called the zero order and
each rainbow of colours is called an order. They are counted outwards with the left and right
ones next to the centre being the first order, the next pair moving outwards is the second order
and so on.
If this were to be faithfully reproduced by a set of photographs we would need to deal with all
the possible configurations. For the sake of performing an experiment we will consider only
those that bear directly on the experimental results that are to be obtained. For example, in
reality it is always possible when performing an experiment to stand on our heads, but if this
has no bearing on the useful visualisation of the experiment, the degree of freedom that allows
us to perform such a manoeuvre will be excluded. One might imagine an experiment in which
standing on one’s head was useful. It is just a question of choosing those degrees of freedom
that are of use for the experiment being considered. In reality and ignoring pointless degrees of
motion there are three degrees of freedom for the user’s eye, up/down, left/right, in/out. There
is one degree of freedom in the shoebox that can move to the left or right to centre the zero order

Figure 1: The diffraction grating experiment.

diffraction pattern so that its position reads zero on the paper protractor. Finally, the pin can
be moved along the edge of the protractor adding one more degree of freedom. If one hundred
photographs are used for each degree of freedom, 10 billion photographs would be needed to
represent the entire experiment in all its possible configurations. Note that the required number
is the product of the number required for each degree of freedom. This perhaps is why it
is sometimes said that ‘the whole is more than the sum of its parts’ as in a very real sense the
whole is the product of the parts. This can be simplified by not allowing the eye to move forward
or backwards or up and down. A further simplification would be to have the experiment already
calibrated. So what remains are only the eye and pin moving in a plane. This would need 10,000
photographs. In practice this proves too many and the ISE that was actually produced combines
the pin movement with the eye movement (Hatherly, Jordan & Cayless 2009). See figure 2,
where the image of the slide is superimposed onto an aerial photograph of the experiment’s
current configuration and the eye is assumed to be on a line from the centre of the slide to the
This is perhaps one compromise too far, as we now have the diffraction pattern changing as
the pin is moved which is emphatically not the case. The diffraction pattern seen depends
only on where the eye or camera is. This is an example of how the combinatorial explosion at
the heart of the ISE forces such compromises often with unfortunate effect. Some might say,
and exactly this has been heard, that it is obvious that the viewpoint is along a line through
the pin to the slide. This was said by a lecturer of Physics at university level, so doubtless
it was obvious to him. However, such assumptions should never be made of a student who
potentially comes to this experiment with no expectations of its behaviour at all. Additionally,
there is nothing intuitive about how the image on the diffraction grating appears to move as the
viewpoint changes, if anything it is counter intuitive upon first viewing.
In practice the purely photographic approach of the ISE restricts us to reproducing just one
degree of freedom at a time. Bronner et al. (2009) states ‘Obviously, it is not feasible to include
all degrees of freedom in a single ISE’, this would seem to be an understatement and rarely more
that one degree of freedom is seen in an ISE. Reproducing the experiment as a 3-D simulated
environment allows all useful degrees of freedom to be retained but at the expense of a much
more complicated simulation.

Figure 2: The diffraction grating ISE.

Figure 3 shows the same diffraction experiment rendered as a 3-D graphics simulation. Here
the user can move his eye (the software camera) in all three directions. It is possible to move
behind the shoebox and see what kind of light bulb is being used (the experiment is commonly
performed with an ordinary tungsten filament and an energy saving bulb which produce quite
different spectra). Note that there is no way of knowing what bulb is being used in the ISE other
than reading the notes which are not part of the simulation. The pin can be moved independently
and the shoebox can be moved left and right to perform the calibration. The final point is
significant because the position of the zero order is one of the points plotted on the graph used
to calculate the frequency of the various colours of light. And it is the accuracy of this point
that needs to be taken account of when plotting a best-fit line. That there is an issue of accuracy
concerning this point is very hard to appreciate if it is assumed that the apparatus has been set up
with no possibility of moving this position. It is also necessary to perform this calibration with
some care. This is a point worth emphasising because it is a major learning outcome that taking
care in any set-up or calibration procedures affect the quality of the results that can subsequently
be obtained.
This ‘1-degree of freedom at a time’ often leads to a ‘flattening out’ of the experiment into
a linear set of objectives where one degree of freedom is used to achieve the first goal and
once achieved the experiment moves on the achieving a second goal using another degree of
freedom now completely divorced from the first. Physics experiments are not like this, they are
overwhelmingly built from many interacting facets.
Figure 4 shows a spectrometer. The tube with the square appendage is the collimator used to

Figure 3: The diffraction grating 3D Immersive Screen Experiment.

produce a narrow beam of light. The other tube is a telescope with a reticule eyepiece used
to align accurately on whatever image is produced by what is placed on the central platform.
Commonly a prism or diffraction grating is placed on the table and the telescope is moved on the
central vertical axis to view the spectrum. Essentially the spectrometer can be used to perform
a very accurate version of the shoebox-and-protractor diffraction experiment described above.
Performing the experiment with a spectrometer, however, means that the user must know how
to use this piece of apparatus and be able to control it.
Before measurements can be made with the spectrometer it is important to focus the telescope
on a distant object. In reality we point the telescope out of a window in the laboratory and focus
on a distant tree or building.
The ISE of the spectrometer experiment requires that the user focus the telescope before being
allowed to continue with the measurements. Once focused, the telescope is always focused.
It is no longer possible to un-focus it. In reality it is the quality of the focus obtained by the
student that has a direct bearing on the results he obtains. Part of the knowledge acquired by
the student is how to deal with this parameter, the quality of the focus. He learns that a deal of
care in performing the focus pays off in the result he gets. He may not learn this the first time
he uses a spectrometer, but it is all part of the experience that he gains in a real laboratory. The
ISE will always give him the results of a perfectly focused telescope, hence the user will have

Figure 4: A commercially available spectrometer.

no experience of trying to garner readings from a less than perfectly focused instrument and is
unlikely to learn either the significance of the focusing or the consequences of it. The user is
simply being made to jump through a hoop to achieve the necessary focus to proceed.
Figure 5 shows the spectrometer experiment rendered as a 3-D graphics program. Here the
telescope is being focused on a distant pylon visible through the window of the laboratory.
The user is able to move his eye/camera in all directions. The various parts of the spectrometer
can be moved by either dragging with the mouse or using keys. The various knobs, such as the
focus knob, can be turned by placing the mouse over them and then using the mouse wheel.
This gives a very intuitive feeling for the focusing in particular; note that whatever focus is
achieved is carried over into the experiment itself. If the telescope is poorly focused the user
will find it difficult to obtain accurate readings as in the real experiment.
Another compromise that we find with the purely photographic ISE approach is that the view

Figure 5: The Spectrometer 3D Immersive Screen Experiment.

presented of the experiment will not necessarily yield a usable view of the scale that is being
used for data collection. In the case of the spectrometer (Hatherly, Jordan & Cayless 2009) it
is necessary to represent the vernier that is used for measuring the angle of the telescope by a
different window. This is an artifice that does not shed light on the workings of the spectrometer
but obscures them. The student will almost certainly know how a vernier works but seeing it in
its rightful place and being able to see its readings change when the telescope is moved gives the
student a much better opportunity to understand it within the context of the experiment. Figure
6 shows how the vernier of the 3-D Immersive Screen Experiment is read by simply moving the
eye or camera up to it.

The models used in the shoebox diffraction grating experiment were extremely easy to imple-
ment. However the spectroscope demanded some complex programming and modelling.
The spectroscope must be modelled as individual components that are assembled and animated
by the program. Every moving part must be constructed as an individual 3-D model. Each
component must accurately fit with all associated components so that they work together. For
example, the main scale seen above, must be capable of moving accurately against the vernier
scale which must give accurate results at all positions. This requires a high polygon count to
achieve smoothness and a high degree of precision so that components fit together well. The
spectroscope was created using a 3-D modelling program called Milkshape (Milkshape 2009).

Figure 6: Reading the vernier scale.

Figure 7 shows a view from within this program of the base and main scale. Construction of
such a model is a painstaking business. This model took a total of about 30 hours to build
and another 30 to animate correctly. The fixed base and scale account for over two thousand
polygons. Although this may seem like a large amount, modern computers equipped with
graphics cards can easily cope with an order of magnitude more than this. The knobs and the
focus tubes are modelled separately so that turning the focus knob causes the focus tube to move
in or out. This enables the student to understand how the focus is achieved and reinforce his
understanding of the lens equation.
The telescope is modelled as a single lens magnifying glass. There is a second software camera
(the first is the camera used for the user’s eye) inside this telescope pointing towards the centre
of the spectroscope. This camera copies its view to a buffer in memory. The lens equation is
used to determine how the pixels of the telescope’s memory buffer are to be dispersed over the
eyepiece lens and it is this simulation that gives the focusing its realistic behaviour.
Figure 8 shows an exploded view of the spectroscope.
The Blitz3D games language (Blitz3D 2009) is used to bring all the components together and
simulate their behaviour.

We evaluated the 3-D Diffraction simulation with two lower sixth form physics classes consist-
ing of a total of 22 students. All students successfully used the program to obtain the necessary
readings that enabled them to plot the graphs and calculate the wavelengths of blue, red, and
green light. This in itself is very encouraging. The students were immediately able to explore
the room containing the experiment to check what kind of bulb was being used and position

Figure 7: The Milkshape 3D modelling application.

themselves (i.e. the software camera) and the pin in order to take the necessary readings in the
space of a single lesson.
They were given two sets of questions. The answers to all questions were given as a number 1
to 5, where 5 indicates ‘strongly agree’ and 1 indicates ‘strongly disagree’. The first set tested
their understanding and the second set, asked for their views of the simulation. The questions are
given in appendix I and the results in appendix II. On ease of use the responses were extremely
positive with all students bar one recording a score of 4 or 5. On being as realistic as the real
experiment only 2 out of the 22 rated it poorly as 2 or 1, but there were 7 undecided here.
On the question of having the same look-and-feel of the actual experiment there were only 5
negatives in total but 8 undecided and 9 positive. Thus on look-and-feel the positives outweigh
the negatives by nearly two to one. This is perceived as encouraging for a first attempt at an
immersive 3-D experiment where the user interface was restricted to simple key presses. There
is clearly scope for improving upon this.
On the idea of replacing real experiments with 3-D simulations they were rather negative which
was expected. This was the same for whether they thought that they could learn as much from
the simulation as a real experiment.
On all the other questions they were very evenly split, but if divided into two groups, one that
did well on the test and those that did not do so well, there is an interesting result. Those that
did well are very positive about the use of the simulations for revision and preparation but do
not rate the simulation as realistic as those that did not do so well, and this group is less positive
about the possible uses for revision and preparation. For example, of the weaker group, 7/10
were positive about the realism but only 2/10 were positive about the usefulness for revision. In
the stronger group, only 5/12 were positive about the realism whereas 9/12 were positive about

Figure 8: Exploded view of the spectrometer.

the usefulness for revision. This is perceived as quite logical; those who are very competent can
perceive the weaknesses of the simulation (the degree of realism in this case) but also understand
its strengths too (that despite the lack of total realism it still has useful potential as a revision
aid). In contrast the weaker students are more easily convinced of the realism but perhaps lack
the necessary degree of abstraction to perceive its use as a revision aid.

The crucial part of many physics experiments is how all the varying parameters interact, in
particular how the set-up or calibration of an instrument has an effect on subsequent measure-
ments. We are clearly able to reproduce this with the Immersive 3-D Screen Experiments that
proves very difficult if not impossible with the ISE approach due to the combinatorial explosion
of the necessary photographs and this is a problem that impacts on all ISEs. Currently there
are no ISEs that have set-up or calibration procedures of any complexity that carry forward
the accuracy achieved in this initial phase into the main body of the experiment. The nearest
behaviour to this is in the Diffraction from a steel ruler ISE (SteelRule2, accessed 2010) where
leaving the laboratory light on causes a reduction in the quality of the results (as clearly the
spectra are easier to see in the dark). This causes a doubling of the required photographs which
remains feasible. A binary set-up procedure such as this is clearly of limited value as set-up

or calibration procedures are rarely this simple. It is situations such as achieving a focus (i.e.
there is a continuous range of values) that are extremely hard if not impossible for the ISE ap-
proach because of the huge number of images needed to implement a second or third degree of
A common criticism of simulations is that they do not reflect reality, but here we are able to
exploit photographs from real experiments as textures within the 3-D environment and thereby
preserve the realism from the actual experiment just as the 2-D ISEs do.
Furthermore, we do not need to create programming artefacts to represents scales or controls
of any kind. These can be represented at the positions that they are found in reality. Classroom
trials clearly show that students are receptive to this approach although healthily sceptical that
such programs can replace the real thing.
An immersive 3-D simulation using photographs allows us to implement a highly realistic ex-
perience of an experiment where we can utilise many degrees of freedom allowing us to retain
the physicality of the original experiment as well as the essential feature that the quality of
the experimental results are dependent on the care taken with the initial set-up and calibration

Questionnaire for evaluation of 3D-ISE Diffraction


These questions should be answered with a Y for yes or a N for no.

(1) The zero order spectrum is visible at zero degrees from a line drawn from the diffraction
grating at a right-angle.
(2) All the diffraction patterns occur in pairs.
(3) When sin(θ) is plotted against the order, the gradient gives the wavelength of the light
(4) Moving the position of the eye (camera) alters the diffraction pattern that can be seen.
(5) Moving the position of the sighting pin alters the diffraction pattern that can be seen.
(6) The higher order diffraction patterns are brighter than the lower order ones.
(7) Each of a pair of diffraction patterns constituting an order occur at the same angle of view
but on opposite sides of the zero line.
The following questions require more than just yes or no.
(8) At the zero order how are the various colours diffracted?
(9) List the sources of the uncertainties in your measurements.
(10) Why are the uncertainties not necessarily the same for all the measurements?


These questions should be answered by circling one of the values given where 5 is ‘strongly
agree’, 1 is ‘strongly disagree’.
(1) I found the controls easy to use. ( 5 4 3 2 1 )
(2) The simulation was a realistic simulation of the real experiment. ( 5 4 3 2 1 )
(3) Using the simulation improved my understanding of the experiment. ( 5 4 3 2 1 )
(4) The simulation had the look and feel of the actual experiment. ( 5 4 3 2 1 )
(5) I would find such simulations useful as a way of revising certain experiments. ( 5 4 3 2 1 )
(6) Simulations like this could usefully replace actual experiments. ( 5 4 3 2 1 )
(7) I would learn as much from this simulation as from the real experiment. ( 5 4 3 2 1 )
(8) It would be useful to do this simulation before doing the actual experiment. ( 5 4 3 2 1 )


Please feel free to put any comments you like here. In particular we are keen to hear of any
suggestions which will help us to improve the simulation.


Student A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 B1 B2 B3 B4 B5 B6 B7 B8
1 1 1 1 1 1 1 1 0 1 1 4 4 5 4 4 3 3 3
2 1 1 1 1 1 1 1 1 1 1 5 2 2 2 4 1 2 3
3 1 1 1 1 1 1 1 1 0 0 5 3 4 4 5 3 3 4
5 1 1 1 1 1 1 1 1 1 0 5 5 5 5 5 2 2 4
6 1 1 1 1 1 1 1 1 1 0 4 5 4 4 5 1 3 5
8 1 1 1 1 1 1 1 1 0 0 3 3 4 3 4 2 2 3

10 1 1 1 1 1 1 1 1 1 0 4 3 3 3 4 1 1 2
11 1 1 1 1 0 0 1 1 1 1 4 3 4 3 4 4 3 4
14 1 1 1 1 1 1 1 1 1 1 5 4 2 3 3 1 4 5
16 1 1 1 1 1 1 1 1 1 0 5 5 5 1 1 1 3 5
20 1 1 1 1 1 1 1 1 1 0 4 3 3 3 3 1 2 3
22 1 1 1 1 1 1 1 1 0 1 5 3 2 2 4 3 2 4

Table 1: Students who did well.

Student A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 B1 B2 B3 B4 B5 B6 B7 B8
4 1 1 1 0 0 1 1 1 1 0 4 3 2 3 3 1 1 2
7 1 1 1 1 1 1 0 0 1 0 5 4 1 2 3 1 1 1
9 1 1 1 1 1 1 1 0 0 0 4 5 2 4 5 2 4 3
12 0 1 0 1 0 1 1 0 0 1 5 5 3 5 3 5 2 2
13 0 1 0 1 1 0 1 1 1 0 4 4 3 3 2 1 1 2

15 0 1 0 1 1 0 1 1 1 0 4 4 4 3 3 3 4 3
17 0 1 0 1 0 1 0 0 1 1 4 5 3 4 5 5 3 4
18 0 1 0 0 1 1 1 0 0 0 5 1 5 1 1 1 1 5
19 0 1 0 1 1 0 1 0 0 1 4 3 4 5 3 4 4 5
21 0 1 0 1 1 1 1 1 0 1 5 4 3 4 3 1 2 4

Table 2: Students who did not do so well.

Altherr, S., Wagner, A., Eckert, B. & Jodl, H.J. (2004) Multimedia material for teaching physics (search,
evaluation and examples). Eur. J. Phys. 25, 7-14.
Blitz3d (2009) (accessed April 2009)
Bronner, P., Strunz, A., Silberhorn, C. & Meyn, J. (2009) Interactive screen experiments with single
photons. European Journal of Physics. 30 (2009) 345-353.
Exploring Science (2009) Open University course S104,
Gorghui, G., Gorghiu, L.M., Suduc, A.M., Bizoi, M., Dumitrescu, C. & Olteanu, R.L. Related Aspects
to the Pedagogical Use of Virtual Experiments. (2009) Research, Reflections and Innovations in Inte-
grating ICT in Education VOL 2. ISBN 978-84-692-1790-0 Ed. Mendez-Vilas 1, Salano Martin A, Mesa
Gonzalez J.A. and Mesa Gonzalez J. Formatex, Badajoz, Spain.
Hatherly, P., Jordan, S. & Cayless, A. (2009) Interactive screen experiments - innovative virtual labora-
tories for distance learners. Eur. J. Phys. 30 (2009) 751-762.
Hatherley, P. & Macdonald, J. Interactive Screen Experiments.
activities/theme.php?themeId=460261fae1b28 accessed (2009)
Lucas, R. & Kolb, U. (2009). Use of 3-D virtual environments in teaching astronomy and physics.
Frontiers in Science Education Research. Ayhan Bilsel & Mehmet U. Garip (Ed.). Eastern Mediterranean
University Press 2009.
Milkshape (2009)
Observing the Universe Open University course SXR208,
(accessed December 2008)
SteelRule2, Measuring laser wavelength with a steel rule. accessed (February 2010)
The Celestial E-Sphere, BBC Sky at Night cover disk, July 2008. Also
/esphere (accessed 2009)

D. J. Raine[1] & C. P. Hurkett[2]
Centre for Interdisciplinary Science, University of Leicester, Leicester, LE1 7 RH, UK
E-mail: [1], [2]

In problem-based learning students are presented with a problem as the motivation
for group research. The presentation of the problem attempts to address the issue
of ownership of content by both providing a context for knowledge acquisition and
requiring a transformation of material to address the problem. The transforma-
tional aspect in principle provides the opportunity for creativity. We present two
case studies, with different outcomes, and the results of a workshop, in which vi-
sual imagery was used as the hook for a PBL problem with a view to encouraging
creativity in the responses.

Science education begins with an emphasis on discovery. Very soon however we realize that no
normal pupil is going to discover Newtonian mechanics without some help. And then we dis-
cover that there isn’t time for discovery, so we just provide the answers. But scientific research
isn’t really about answers except in as much as these raise further questions. So there is a long
period of learning during which students are required to suspend judgment on the usefulness of
what they are learning until, finally, they become research students and at last they are doing
science again. This is a very long period of suspension and not to everyone’s taste.
Problem-based learning (PBL) is an attempt to address his situation by providing a starting point
in which a question generates the need-to-know in the form of a research issue. In tackling the
problem students are guided to the acquisition of new knowledge in a meaningful context (e.g.
Raine & Symons, 2005 and references therein). The essential motivation is that PBL requires a
transformation of material that captures the need for ownership in a way that the student essay
does not.
However, even within PBL in a science programme there is a tension between the need to cover
curriculum content and the freedom to foster creativity. Very often the desire on the part of
students to demonstrate creativity is focused on the medium and not the science content. Thus,
we get lovingly produced powerpoint presentations, videos, reports, but with superficial subject
matter. The conference workshop provided an opportunity to explore PBL with a particular
focus on the creative transformation of the problem material. We provide two contrasting case
studies from undergraduate courses and discuss the results of the workshop exercise.
The two case studies and the workshop focus on using visual material as the hook for a PBL
problem. In these cases the problem is usually one of decoding of the visual image and has a

correct answer. In our cases we wanted to provide a visual stimulus to creativity while at the
same time maintaining a set of learning objectives.

We presented students with a set of images representing the passage of a light beam from the Sun
to Mars and thence to the eye at the eyepiece of a telescope. The images of Mars were altered
to represent various defects of vision, the intention being that these should prompt students to
consider the way images are processed in the eye. It was made clear, or so we thought, that
the solution of the problem posed by the images was an account of the fundamental science of
optics and the eye. The creative aspect of the problem was to figure out a consistent story. This
should have required a detailed knowledge of the nature of light and visual processing in the
retina. However, it is easier to “google” the medical aspects of defects of vision, which were
not part of the learning objectives. The background information provided to facilitators proved
inadequate to deflect students from addressing mainly the medical aspects. Consequently the
student effort focused on treating each individual picture as a medical case, thinking they had
solved “the problem” once they had identified the nature of the defect and pasted a description
from the internet. There was no significant transformation of the material.

We presented students with the following problem statement:

A for Andromeda was a made-for-TV Science Fiction serial broadcast in 1961. The
basic plot is the discovery of a radio message received from a distant civilisation
with coded instructions on how to make a living being. Unfortunately the tapes
of all but one of the episodes have been wiped so the coded message has been
lost. In order to advertise their remake of the series TV Remakes Unlimited have
decided to publicise the supposed discovery of a “real” coded message received
by Jodrell Bank in 1987 which it claimed had been hushed up by the military and
government. The code in the original broadcast production was alpha-numeric but
it was decided to give the supposedly newly discovered one a glyphic form to make
the deciphering more interesting. You must decide what each glyph symbolises and
therefore in which order the “message” should be assembled in order to create the
life form.

Figure 1: Some of the glyphs for case study 2.

Some of the glyphs are shown in figure 1. The aspect of creativity here is to understand the
science well enough to argue for a logical ordering of the pictures. This allows many answers
while requiring students to explain the science behind each picture. Some examples of learning
objectives were: to describe the origin of molecular bonding, to discuss the structure and role
of amino acids and proteins and of the shapes of molecules.
The feedback from the problem was positive. Students attempted to master the learning objec-
tives at an appropriate level, while the proposed ordering of the glyphs varied between groups.
Some of the weaker groups gave an order without an argument, as if this were self evident from
their descriptions of the pictures.


The PBL questions posed in the workshop were somewhat ambitions. We presented each group
of participants with a choice of three specially painted pictures (figure 2) as their PBL hook.

Figure 2: The images used in the conference workshop.

Of course, we had some idea of what we would like, having specified a physics context: from
left to right we thought of the problems as fire and ice, electrical breakdown and Archimedean
spirals or, more physically, light transmission through a glass window. The groups chose the
first and last of the pictures. The first produced an interesting discussion of how an ice crystal
could survive in the vicinity of a flame, for long enough to be visible. (Not all flames are hot,
although that is perhaps chemistry rather than physics.) On the other hand the third picture
produced a discussion on material heating up as it accretes into a black hole. Of course, this
is a creative connection, but not one on the intended outcomes. It is also inconsistent with the

Scientific discovery depends on making creative connections that transform content. In the con-
text of PBL such transformations develop ownership of the material and hence promote deep
learning. Visual prompts can be used to promote the type of creative imagination without the
unrealistic expectation that students can make scientific discoveries. Such problems need care-
ful preparation and close facilitation. However, there is an ever-present danger that the results
are superficial repetition of existing knowledge and do not promote new learning. Hooks of
this type may be more appropriate in contexts where there are no specific scientific learning

objectives, but are simply triggers for a piece of in-depth research chosen by the students. Al-
ternatively, in a defined context, such as the second case study, one can think of these hooks
as prompting a mind-mapping of content knowledge. In this context they have the advantage
over traditional mind-maps of providing both the constraints and the freedom necessary for true

Raine, D.J. & Symons, S. (eds.) (2005) PossiBiLities: A Practice Guide to Problem-based Learning In
Physics and Astronomy. Hull, UK: The Higher Education Academy Physical Sciences Centre.


Edward F. Redish & Ayush Gupta

Department of Physics, University of Maryland, College Park, MD 20742-4111 USA.

Physics makes powerful use of mathematics, yet the way this use is made is often
poorly understood. Professionals closely integrate their mathematical symbology
with physical meaning, resulting in a powerful and productive structure. But be-
cause of the way the cognitive system builds expertise through binding, experts
may have difficulty in unpacking their well-established knowledge in order to un-
derstand the difficulties novice students have in learning their subject. This is par-
ticularly evident in subjects in which the students are learning to use mathematics,
to which they have previously been exposed in maths classes, in complex new ways.
In this paper, we propose that some of this unpacking can be facilitated by adopting
ideas and methods developed in the field of cognitive semantics, a sub-branch of
linguistics devoted to understanding how meaning is associated with language.

Many sciences, particularly physics, make extensive use of mathematics. Even introductory
physics texts include thousands of equations and a graduate student may find that equations
cover nearly half the pages in an advanced textbook. Though the procedures and equation
manipulations might appear the same, what physics does with maths is dramatically different
from what one learns in a maths class. Physics requires us to integrate our understanding of
the physical world with the symbolic relations of mathematics, adding meaning and structure to
both. Physics puts meaning to the maths, adding additional levels of structure, interpretations,
and even tools.
Instructors in upper division physics classes often complain that even students with satisfactory
grades in their maths courses “don’t know enough math.” Sometimes these instructors call for
the students “to take more math classes,” or suggest, “we should teach the math ourselves.”
Indeed many in the latter group feel that the maths is “reasonably simple” and “could be taught
in a week or two.”1 It appears that the issues of how maths is used in physics is not simple and
needs some careful analysis.
In order to understand better the use of maths in physics so that we can teach it more effectively,
we need to analyse the cognitive components of expertise and build an understanding of how it
develops. Since the critical issue appears to be “making meaning” with maths, we turn for
help to the subject of cognitive semantics - a subfield of linguistics that is concerned with
Personal communications over many years with many colleagues.

how ordinary language is imbued with meaning2. Researchers in that field are attempting to
answer the question, “What do we mean by meaning?” They appear to have made considerable
progress on this issue in the past twenty years and they have developed a number of concepts
and a terminology that may help us deconstruct the way meaning is put to maths in science.
In this paper, we argue that the barrier to effective use of mathematics in physics is not an
issue of deficiency of skills or knowledge that can be solved by ‘more and better instruction in
maths’. Using ideas from cognitive linguistics, we argue that the network of knowledge that
expert physicists use to put “meaning” to mathematics is different from the ways that novices
interpret mathematics in physics.
We begin by presenting some examples that illustrate how physics is added to mathematics to
create new structures, ones that construct meaning beyond the mathematical. We then outline
some of the most relevant basic ideas from cognitive semantics, in particular, encyclopedic
meaning and blending. We then present the case of two students, both of whom are good
at solving a physics problem, but who use different approaches to thinking about the maths.
We present a comparative analysis using the principles of cognitive linguistics to unpack the
differences in the ways that they look at the same equation. Finally, we consider the implications
of our analysis for research and instruction. Note that this is essentially a theoretical paper.
Although we cite examples, they are anecdotes, not data, chosen to reify and help the reader
make sense of the ideas. The value of this approach will have to be evaluated by its productivity
in leading to new ideas for observations and the analysis of data.



When we use maths in physics - or in any science - we are modelling some aspect of the physi-
cal world by specifying the way physical measurements are related and how they behave under
changes of perspective (Poincare, 1905). These translations from physical-causal relations be-
tween entities in the real world to mathematical relations between symbols is what we call
mathematical modelling.
In order to explicate the various components of the modelling process for the purpose of think-
ing about teaching mathematical physics, we use the diagram shown in Figure 1 (Redish 2005,
We begin in the lower left with a particular physical system we want to describe. Critical here is
a focus on the particular aspects of the universe we want to describe. We identify measurables:
variables and parameters we can quantify through some process of measurement (at least in
principle, if not always in practice). We then decide what particular mathematical structures
are appropriate for describing the features we are interested in modelling. We then model the
physical system by mapping these measurements into mathematical symbols and expressing the
physical-causal relations between the measured quantities in terms of mathematical operations
Note that in an early era, linguists made a distinction between semantics - the lexical meaning of words, and
pragmatics - the meaning of words in contexts. The linguists we follow have come to the conclusion that this is
not a useful distinction and that all semantics is pragmatics.

Figure 1: A model of mathematical modelling.

between the symbols. From the mathematical structures we have chosen, we inherit transfor-
mational rules and methodologies for transforming relationships and solving equations. We can
then process the maths to solve problems that we were unable to see the answer to directly when
we were thinking only about the physical system. This, however, leaves us with only mathe-
matics. We then have to interpret what the result means back in the physical system. Finally,
we have to evaluate whether the result supports our original choice of model when compared to
observations or whether it indicates a need for a modification of our model.
All of these four steps, modelling, processing, interpreting and evaluating, are critical skills in
the toolbox of a scientist who uses maths to describe the behavior of the physical world. This
diagram helps a teacher focus on more than just the mathematical processing that often tends to
dominate instruction in physics. But the use of maths in physics is not so simple or sequential
as this diagram may tend to indicate. The physics and the maths get tangled.


We not only use maths in doing physics, we use physics in doing maths. We will present three

Corinne’s Shibboleth: The meaning of function attributes

Our first example is “Corinne’s Shibboleth”3 (Dray & Manogoue, 2002). The particular ex-
ample shown in Figure 2 tends to separate physicists from mathematicians. Try it for yourself
before reading the discussion that follows.
A “shibboleth” is a word or phrase that can be used to distinguish members of a group from non-members.

One of your colleagues is measuring the temperature of a plate of metal placed
above an outlet pipe that emits cool air. The result can be well described in Cartesian
coordinates by the function

T (x, y) = k(x2 + y 2 )

where k is a constant. If you were asked to give the following function, what would
you write?

T (r, θ) =?

Figure 2: A problem whose answer tends to distinguish mathematicians from physicists.

The context of the problem encourages you to think in terms of a particular physical system.
Physicists tend to think of T as a physical function - one that represents the temperature (in
whatever units) at a particular point in space (in whatever coordinates). Mathematicians tend to
consider T as a mathematical function - one that represents a particular functional dependence
relating a value to a pair of given numbers. The physicist tends to answer that T (r, θ) =
kr 2 since they interpret x2 + y 2 physically as the square of the distance from the origin. If
r and θ are the polar coordinates corresponding to the rectangular coordinates x and y, the
physicist’s answer gets the same value for the temperature at the same physical point in both
representations. In other words, the physicist assigns meaning to the variables x, y, r, and θ -
the geometry of the physical situation relating the variables to one another. Our mathematician,
on the other hand, may regard them as dummy variables denoting two independent variables
- so, (r, θ) or (x, y) do not have any geometric meaning constraining their relationship. The
mathematician focuses on the grammar of the expression rather than the physical meaning.
The function as defined instructs one to square the two independent variables, add them, and
multiply the result by k. The result should therefore be T (r, θ) = k(r 2 + θ2 ).4
Typically, a physicist will be upset at the mathematician’s result. You might hear, “You can’t add
r 2 and θ2 ! They have different units!” The mathematician is likely to be upset at the physicist’s
result. You might hear, “You can’t change the functional dependence without changing the
name of the symbol! You have to write

T (x, y) = S(r, θ) = kr 2 .” (14)

To which the physicist might respond, “You can’t write that the temperature equals the entropy!
That will be too confusing.” (Physicists often use S to represent entropy.)
Of course, there are times when physicists pay careful attention to functional form, for ex-
ample when considering the transformation from a Lagrangian to a Hamiltonian or between
thermodynamic potentials (Zia et al., 2009). But once they are experts, the context suffices to
It is to be noted that we are exaggerating the roles of the mathematician and physicist thinking to illustrate
the point that one could have starkly different interpretation of the same equation. We are not implying that a real
world mathematician would not think of Cartesian to polar coordinate transformations at all if presented with this

alert physicists to the need to pay attention to mathematical grammatical elements of functional

Novices often pay attention only to mathematical grammar

Not only do mathematicians often focus on the mathematical grammar of an equation while
ignoring physical meaning; novice students often do the same. A few years ago, one of us (EFR)
gave the example shown in Figure 3 to his second semester class in algebra-based physics.

A very small charge q is placed at a point − →

r somewhere in space. Hidden in the re-
gion are a number of electrical charges. The placing of the charge q does not result
in any change in the position of the hidden charges. The charge q feels a force, F .
We conclude that there is an electric field at the point −

r that has the value E = F/q.

If the charge q were replaced by a charge −3q, then the electric field at the
point −

r would be

a) Equal to −E
b) Equal to E
c) Equal to −E/3
d) Equal to E/3
e) Equal to some other value not given here.
f) Cannot be determined from the information given.

Figure 3: A quiz problem that students often misinterpret.

The instructor had discussed the definition of electric field extensively and he had made it “quite
clear” that the strength of the electric field was independent of the test charge. Yet on this quiz,
more than half of nearly 200 students thought that the answer should be (c). Upon discussing it
with the class after the quiz, it became clear that, although many of them could quote the result
“the electric field is independent of the test charge,” most of the students answering incorrectly
had not thought to access that knowledge. They looked at the equation E = F/q and treated
the problem as a problem in mathematical grammar. If E = F/q, then F/(−3q) = −E/3. In
this situation, however, F is the Coulomb force on the test charge due to interaction with the
source charge and thus increases proportionally as the test charge is increased, resulting in the
same field as before. Novices’ propensity to not integrate these algebraic symbols with physical
meaning leaves them vulnerable to such errors. Of course, leaving the dependence of F on q
implicit has a role in lending coherence to the novice error on this problem, but that does not
undermine our claim that students reason in terms of mathematical grammar, missing clues to
a different interpretation given by physical meaning. Writing F (q) might help a few students
realize that the algebraic symbol F is a function of q, but not that these symbols in physics
equations stand for real physical quantities.

The Photoelectric Effect Equation: Interpreting equations in the context of a physical
situation can build in implicit conditions

A third example comes from the modern physics section of the calculus-based physics class for
engineers. As usual, the instructor (EFR) tried to “twist” student expectations a bit by making
small but important modifications to standard problems. The problem shown in Figure 4 was a
surprise to some of the students but the results were more of a surprise to the instructor.

If a wavelength of light λ leads to no electrons being emitted at a zero stopping

potential, what will happen if we choose a longer wavelength? Explain your

Figure 4: A problem where the equation hides some physics.

This one is pretty easy if you keep the physics to the fore. A longer wavelength corresponds
to a smaller frequency so the photons have less energy after the modification. If the photons
did not have enough energy to knock out an electron originally, they would certainly not have
enough if we reduce their energy further. Yet nearly a quarter of the class responded that after
the modification, there would be electrons knocked out. Their reasoning relied on the Einstein
photoelectric effect equation:

eV0 = hf − Φ, (15)

where e is the charge on the electron, V0 is the stopping potential, f is the frequency of the
photon, and Φ is the work function of the metal. Their reasoning went as follows: “If hf − Φ is
zero before the change, then it will not be zero after the change. Therefore, since it’s not zero,
there will be electrons knocked out.”
This reasoning highlights the fact that Eq. 2 is not the correct equation. There is a hidden
Heaviside (theta) function that corresponds to the statement: Do not use this equation unless

the right hand side, which corresponds to the maximum kinetic energy of the knocked out
electron, is a positive number. A more mathematically correct equation is;

eV0 = (hf − Φ)θ(hf − Φ) (16)

Typically, expert physicists don’t bother to be explicit about this theta function. They simply
“check the physics” to see that the energy is positive before applying the equation.
These examples demonstrate that physicists (as well as other scientists and engineers) often use
ancillary physical knowledge, often implicit, tacit, or unstated, in their application of mathemat-
ics to physical systems. Interestingly enough, a similar idea turns out to be valuable to linguists
trying to understand how we put meaning to words: semantics.


To understand how we make sense of the language of mathematics in the context of physics,
let us consider what is known about how people make sense of language in the context of daily
life. Many cognitive linguists have studied semantics (how we make meaning with language)
for the past 100 years, and they have made considerable progress in the past 25. Although the
research community has not entirely come to an overarching synthesis, they have many ideas
that are valuable in helping us make sense of how we make meaning with maths in physics. We
offer an exceedingly brief summary of a rich and complex subject, selecting those elements that
are particularly relevant. We draw heavily on the work of (Lakoff & Johnson 1980), (Lakoff
& Nunez 2001), (Langacker 1987), (Fauconnier 1985, 1997), (Turner 1991), (Fauconnier &
Turner 2003), (Evans 2006), and (Evans & Green 2006).


In their overview of modern cognitive semantics, Evans and Green identify four core ideas
(Evans & Green 2006, p. 157):

• Encyclopedic knowledge: Ancillary knowledge is critical to determining the meaning of


• Conceptualization: Meaning is constructed dynamically.

• Embodied cognition: The meaning of words is grounded in physical experience.

• Conceptual grounding: Semantic structure expresses conceptual structure.

Encyclopedic knowledge: Ancillary knowledge is critical in the creation of meaning

The principle of encyclopedic meaning implies that we understand the meaning of words not
in terms of terse definitions provided in the dictionary but in reference to a contextual web
of concepts (represented by other words) that are themselves understood on the basis of still

other concepts. Take the example of the word “hypotenuse,” defined as the longest side of a
right-angled triangle. But to understand the meaning of “hypotenuse,” one needs to understand
the meaning of a triangle, “sides” of a triangle, right angle, and the idea of longest. These in
turn require conceptualizing a plane, shapes on a plane, lines, angles, length, and so on. In
other words, understanding and conceptualizing “hypotenuse” relies on a number of ancillary
concepts, which in turn rely on a background of other concepts in an ever-expanding web of
encyclopedic knowledge.
This idea of encyclopedic knowledge has been developed under various frameworks such as
frame-semantics (Fillmore 1976), domains (Langacker 1987), and mental spaces (Turner 1991).
The idea common to the various perspectives is that we can model knowledge as consisting of
a very large number of elements that are highly interconnected, much like the world-wide-web.
At any given moment, only part of the network is active, depending on the present context
and the history of that particular network. The meaning of a word in a particular utterance is
then determined by what part of that complex web of knowledge is accessed by that particular
utterance of that word. Modern cognitive linguists argue compellingly for these complex links
in the structure and processing of meaning:

. . . a number of scholars . . . have presented persuasive arguments for the view that
words in human language are never represented independently of context. Instead,
these linguists argue that words are always understood with respect to frames or
domains of experience . . . a frame or domain represents a schematisation of experi-
ence (a knowledge structure), which is represented at the conceptual level and held
in long-term memory, and which relates elements and entities associated with a
particular culturally-embedded science, situation or event from human experience
. . . words (and grammatical constructions) are relativised to frames and domains
so that the ‘meaning’ associated with a particular word (or grammatical construc-
tion) cannot be understood independently of the frame with which it is associated.
(Evans & Green 2006, p. 211)

The meaning associated with a word or concept could also extend to include the knowledge
we have about or in relation to the concept via our haptic, visual, auditory, olfactory, and other
senses. In Foundations of Cognitive Grammar, Langacker (1987, p. 154) argues that cogni-
tively, the meaning of the concept [BANANA] is not restricted to its linguistic specification.
Rather the meaning of [BANANA] encompasses its associations within the non-linguistics do-
mains of space (or visual domain), color, and taste and smell sensations as well as cultural
knowledge about bananas as edible, about banana trees, where they grow, and so on.
At first look, the meaning of mathematical equations could seem terse in the dictionary sense.
What does it mean to know the meaning of an equation such as y = mx + b? In the strictest
sense, it is a statement defining the value of the quantity “y” in terms of the sum of “b” and the
product of “m” and “x”.
We argue that mathematics intentionally adopts such a minimalist view. What you know about
a mathematical quantity is specified as precisely and as minimally as possible (with axioms),

and only that knowledge is to be used in constructing new knowledge5.
Though written mathematics can be terse and precise, we do tend to rely on more than just a
“dictionary meaning” in how we use and understand mathematics. For most mathematicians
(and even high school students) the equation y = mx + b carries more meaning than the literal
relation between four symbols. With our knowledge about labeling conventions, x and y are in-
terpreted as variables capable of taking on many different values, while m and b are interpreted
as constants. With this addition, the equation takes on the meaning of a relation between the in-
dependent variable (x) and the dependent variable (y). Additionally, the constancy of m implies
that the equation refers to a straight line. The constants now take on additional mathematical
meaning: m as the slope of the line and b as the intercept on the y-axis, bringing in ideas from
graphs. Thus, the meaning of the equation, understood within the domains of mathematical
conventions, straight lines, and graphs is much richer then the strict “definition” expressing the
symbol “y” in terms of other symbols.
When we use maths in physics our rich knowledge about the physical system is additionally
brought to bear in interpreting the mathematical meaning, as illustrated by the examples in
section II. As another example, think about the following equation from physics: v = v0 + at,
where v is the velocity of an object, v0 is the initial velocity, a is the constant acceleration of the
object, and t is the time. The equation is identical in grammatical structure to the mathematical
equation above. But in the context of physics, it is connected to an even richer network of ideas
involving motion, speeds, and rates of change. To understand this equation is to understand its
relation to all these stores of knowledge. In other words, the meaning of equations in physics is
encyclopedic just like the meaning of a word such as “banana” is encyclopedic.

Conceptualization: Meaning is constructed dynamically

One of the implications of the idea of an encyclopedic knowledge system is that the meaning of
entities not fixed but is dynamically determined based on the specific contextual clues. Seman-
ticists see linguistic units as prompts for the construction of meaning, or, conceptualization.
As described by Evans and Green (2006, p. 162), conceptualization is, “a dynamic process
whereby linguistic units serve as prompts for an array of conceptual operations and the recruit-
ment of background knowledge. It follows from this view, that meaning is a process rather
than a discrete ‘thing’ that can be ‘packaged’ by language. Meaning construction draws upon
encyclopaedic knowledge . . . and involves inferencing strategies that relate to different aspects
of conceptual structure, organization and packaging . . . ”
If utterances provide access to nodes in the network of knowledge, different parts of which are
active at different moments, then the idea of encyclopedic meaning implies that the meaning of
an utterance would depend on the local activity of the network at any given moment. It is in
this sense that the meaning is dynamic. The contexts of a particular utterance, local (what is
the conversation about, with whom, etc.) and global (personal histories, social and institutional
affordances, etc.), affect the activity of the knowledge network and, in turn, the conceptual-
ization at any given moment. Thus, the meaning of an utterance is not pre-determined but is
Although this is the ideal, many mathematicians argue that there are subtle issues that prevent this ideal from
ever being realized, even in principle (Goldstein 2006).

constructed in the moment. Take the example of the word “safe” in following sentences in the
context of a child playing on the beach (Evans & Green 2006, p.161):

1. The child is safe

2. The beach is safe

The first sentence refers to the child as free from harm. The second sentence, though identical
in structure to the first, does not imply that the beach is free from harm, rather it implies that
the beach cannot cause harm. The senses of “safe” in the two sentences (and the properties they
assign to child and beach) are slightly different depending on what the utterance is referring to.
“In order to understand what the speaker means, we draw upon our encyclopedic knowledge
relating to children [and beaches] and our knowledge relating to what it means to be safe. We
then ‘construct’ a meaning by ‘selecting’ a meaning that is appropriate in the context of the
utterance.” (Evans & Green, 2006, p. 161-162).
In a different context, for example, while referring to a beach threatened by property develop-
ers, the conceptualization and the meaning constructed for the second utterance could be quite
In the light of this, we can revisit the example of the imaginary physicist and mathematician ar-
guing about whether T (r, θ) should be “r 2 +θ2 ” or “r 2 ”. We can now understand the mathemati-
cian as attaching meaning to the equation within the domain of simple functions and variables,
while the physicist is additionally interpreting the equation within the domains of coordinate
systems adding physical meaning to the variables.
This idea ties in nicely with the resources framework being developed by the Maryland Physics
Education Research Group for analyzing student thinking in physics (Hammer 2000; Redish
2004; Hammer, Elby, Scherr, & Redish 2004). In this framework, cognitive resources (tightly
bound cognitive knowledge structures) are recruited (activated) by associations (through spread-
ing activation) and contexts. The explication of particular inferential strategies by the linguists
provide us with tools to help us begin to understand the mechanism of how contexts and asso-
ciational structures recruit or activate particular resources.

Conceptual grounding: Semantic structure expresses conceptual structure

To some extent, the cognitive dynamics of meaning construction are expressed in the structure
with which we express or communicate that meaning. Modern cognitive linguists argue that
the conceptualization in a moment (and correspondingly, the conceptual structure in that mo-
ment) is closely bound to the semantic structure. Thus, changes in semantic structure reflect
changes in conceptualization. It is important to note here that “semantic structure relates not
just to words but to all linguistic units. A linguistic unit might be a word like a cat, a bound
morpheme such as -er, as in driver or teacher, or indeed a larger conventional pattern, like the
structure of an active sentence or a passive sentence.” (Evans & Green, 2006, p. 159). Evans
and Green argue that even though the objective information or meaning provided by the active

sentence (“William Shakespeare wrote Romeo and Juliet”) is identical to that provided by the
passive sentence (“Romeo and Juliet was written by William Shakespeare”), the difference in
the semantic structure reflects a difference in conceptualization and conceptual structure. In
the active sentence, we are focusing on the actor (William Shakespeare, in this case) while
the passive formulation draws greater focus to the participant undergoing the action (the play,
Romeo and Juliet, in this case). The difference in focus could imply a different organization of
the knowledge network, which could in turn lead to differences in how we access subsequent
domains of knowledge.
Talmy (2000) suggests that the semantic representation of concepts takes place through the in-
teraction of two systems: a conceptual structuring system and a conceptual content system. The
former is highly schematic while the meaning associated with the conceptual content system
is rich and highly detailed. While semantic structure must to some extent reflect conceptual
structure, the relationship is complex and indirect.
In physics, this could mean that how we express ideas matters in determining what concep-
tualizations are activated. Mathematically, the following two equations representing Newton’s
second law convey identical information:

F = ma, (17)

a = F/m, (18)

where F is the total force on an object, m is the mass of the object, and a is the acceleration
of the object. But the conceptualization of the two equations could be quite different. Equation
(4) focuses on the force as the product of the mass and acceleration, often leading students
to think of it as the definition of force. The equal sign is typically understood as connecting
two identical entities. Equation (5), on the other hand, encourages the activation of a causal
relationship: acceleration as resulting from the forces acting on an object. The equals sign can
now carry a very different meaning of connecting effect to cause.

Embodied cognition: The meaning of words is grounded in physical experience

While encyclopedic knowledge relates to the topology of the knowledge network and cognitive
processes on that network, embodied cognition refers to the interaction of complex cognitive
functions with basic sensorimotor routines. If concepts are understood only within a network of
other concepts, how then are the first concepts derived? The thesis of embodied cognition states
that ultimately our conceptual system is grounded in our interaction with the physical world:
how we construe abstract meaning might be constrained by and often derived from our very
concrete experiences in the physical world.
Embodied cognition is not a reference to the cognitive activity that is obviously involved in
performing sensorimotor activities. The idea is that (a) our close sensorimotor interactions with

the external world strongly influence the structure and development of higher cognitive facili-
ties, and (b) the cognitive routines involved in performing basic physical actions are involved in
higher-order abstract reasoning.
There are abundant examples in everyday language and conceptualization where meaning can-
not be understood without deriving orientation from our bodily existence in the three dimen-
sional world that we experience. Lakoff & Johnson (2004) discuss extensive examples of many
spatial orientation concepts and metaphors such as “up,” “down,” “front,” and “back” that are
tied closely to our spatial experiences: “ . . . Thus UP is not understood purely in its own terms
but emerges from the collection of constantly performed motor functions having to do with our
erect position relative to the gravitational field that we live in.” (Lakoff & Johnson, 2004, p.
This then forms the basis of structuring and understanding other concepts. Thus metaphorical
statements such as: “I’m feeling up” or “He is really low these days” are conceptualized on the
basis of physically drooping when one in depressed or in a negative emotional state.
Lakoff & Johnson (2004) point out how our bodily experiences with physical objects form the
basis of conceptualizing emotions such as anger, love and events such as inflation in terms of
discrete entities or physical substances. Such conceptualization forms the basis on which we
use and understand statements such as, “Inflation is backing us into a corner,” and “You’ve got
too much hostility in you.” Many of these conceptualizations are so ingrained and automatic
that we do not even realize the metaphorical nature of them in everyday use.
The embodiment of cognition and meaning construction is what enables us to break the circu-
larity of the dictionary, where words and concepts are seemingly defined in terms of yet other
words and concepts that one might not know the meaning of. The point is that a dictionary does
not provide definitions of words in terms of other words in the dictionary. Its circular chain
implicitly expects that at some point you will come across words that you learned in a more di-
rect and perceptual context than from a dictionary and whose meaning you can discern without
further referring to the dictionary.
An infant begins to learn language by direct perceptual experience. Its parents identify multiple
objects that are clearly different from each other with the same word, “chair”. Part of the
process is epigenetic: hard-wired to permit us to learn given the appropriate experiences6 . A
toddler does not have to be taught by its parents to bind the multiple perceptions it receives
when it holds its bottle (touch, sight, and taste) into the perception of a single object. But it
does have to be taught the word to use to make that bottle appear when the child is hungry. That
association of the word with the perceived object is the part on which a dictionary definition, in
the end, relies.
The grounding of conceptualization in physical experience and actions also extends to higher
cognitive processes such as mathematical reasoning. Lakoff & Nunez (2001) argue that our
understanding of many mathematical concepts relies on everyday ideas such as spatial ori-
entations, groupings, bodily motion, physical containment, and object manipulations (such as
rotating and stretching): “Mathematical ideas are often grounded in everyday experience. Many
Joaquin Füster refers to such inherited structures as phyletic memory: memory we have as a genetic record of
our evolution as a species (Füster 1999).

mathematical ideas are ways of mathematizing ordinary ideas, as when the idea of a derivative
mathematizes the ordinary idea of instantaneous change.” (p. 29). Thus, the mathematics of
set theory can be understood as grounded in our physical experience with containers and col-
lections of objects, and conceptualizing the arithmetic of complex numbers “makes use of the
everyday concept of rotation.” (p. 29). In other words, many of the sophisticated ideas and
formulations in mathematics are intricately entwined with the physicality of our being.
In the end, all of our complex concepts must eventually come down to direct perceptual experi-
ence. This seems an almost unbelievable result, given the complexity and abstractness of con-
cepts we deal with everyday. Some cognitive linguists have begun to develop possible pathways
by which those direct perceptions can be elaborated, refined, and extended to create complex
abstractions. They include metaphor (Lakoff & Johnson 2004), polysemy (Evans 2006), and
blending (Fauconnier & Turner 2003).



One of the clear bits of evidence supporting the complex view of meaning making outlined in
the synthesis by Evans & Green (2006), is the ease with which we make sense of partial or
corrupted speech and text. Everyone is likely to recall examples when someone is speaking
and makes a slip of the tongue (a verbal typographical error) such as omitting a critical “not” or
slipping in the wrong name, but no one has any trouble in getting the intended meaning. Indeed,
the listeners may be unaware of the error until it is reviewed on a recording. We have no trouble
interpreting the verses of Lewis Carroll’s “Jabberwocky”, the second verse of which is quoted
below, despite the presence of many nonsense words. (Carroll 1871)

Beware the Jabberwock, my son!

The jaws that bite, the claws that catch!
Beware the Jubjub bird, and shun
The frumious Bandersnatch!

Despite the fact that we have no idea what a “vorpal” sword is or what a “Tumtum tree” might
look like (both from later in the poem) we make sense of it. Tenniel’s illustration of the Jab-
berwock brings in a visual non-dictionary-based element that allows the entire poem to become
realistic and interpretable.
A major goal of building expertise in physics (or, we suspect, in any other complex knowledge
structure) is creating a complex web of meaning - an encyclopedic knowledge base that adds
stability and the ability to recover from error.
In the following section we apply the principles of cognitive linguistics to students’ reasoning
in physics to argue that sophisticated use of mathematics in physics involves connecting the
mathematical equations to the rich network of knowledge and intuition about the real world,
not just at the level of variables but at the level of the interpretive basics of the mathematical
and physical processes. Our example contrasts the explanations of two students who can both
solve the physics problems presented to them. The difference in the quality of their use and in-
terpretation of equations can be understood in terms of the encyclopedic network of knowledge
that they activate in association with these equations.


Understanding an equation in physics is not limited to connecting the symbols to physical vari-
ables and being able to perform the operations in that equation. An important component is
being able to connect the mathematical operations in the equation to their physical meaning
and integrating the equation with its implications in the physical world. In this section, we
illustrate the differences in ways that meaning could be attached to an equation by analyzing
excerpts of clinical interviews (AG was the interviewer) with two students, Pat and Alex7 , in an
introductory physics class. The excerpt focuses on the students’ understanding of the equation
for the velocity of an object moving with a constant acceleration: v = v0 + at, where v is the
velocity of the object at time t, v0 is the velocity of the object at t = 0, and a is the constant
acceleration. Though both students can use the equation satisfactorily for solving problems, but
the ‘encyclopedic meaning’ ascribed to the equation by Pat is different than that for Alex.
When Alex is asked to explain the velocity equation, she focuses on the meaning of the symbols:

AG: Here’s an equation; OK, you’ve probably seen it . . .

Alex: Yeah.

AG: Right. So suppose you had to explain this equation to a friend from class;
how would you go about doing that?

Alex: Umm . . . OK . . . Well . . . umm . . . I guess, first of all, well, it’s the equation
for velocity. Umm . . . *pause* well, I would . . . I would tell them that it’s uh . . . I
mean, it’s the integral of acceleration, the derivative of *furrows brow* position,
right? So, that’s how they could figure it out . . . uh . . . I don’t know. I don’t really
*laugh* I’m not too sure what else I would say about it. You can find the velocity.
Like, I guess it’s interesting because you can find the velocity at any time if you
have the initial velocity, the acceleration, time and . . .

Her response suggests that Alex thinks about the equation as something that allows one to
calculate the velocity of an object at any moment. She does refer to the velocity as the derivative
of position and integral of acceleration but her comment does not reflect if those mathematical
operations connecting velocity to position and velocity to acceleration have any deeper physical
meaning for her.
Pat also refers to the physical meaning of the symbols in the equation, but her explanation is not
limited to that. Her explanation seems to dip into knowledge about motion, units, and processes
of change:
Pat and Alex are gender-neutral pseudonyms and throughout the paper we will refer to them by the feminine
pronoun “her”. We intentionally choose not to identify the gender of our participants so as to maintain the focus
on the substance on their reasoning and to preemptively rule out any correlations between their reasoning and their

AG: Ok. So here’s probably an equation that you have seen before, right? And um,
if you were to explain this equation to a friend from class, how would you go about
explaining this?

Pat: Well, I think that the first thing that you’d need to go over would be the defi-
nitions of each variable and what each one means, and I guess to get the intuition
part; I’m not quite sure if I would start with dimensional analysis or try to explain
each term before that. Because I mean if you look at it from the unit side, it’s clear
that acceleration times time is a velocity, but it might be easier if you think about,
you start from an initial velocity and then the acceleration for a certain period of
time increases that or decreases that velocity.

Pat also looks at the physical meaning of each symbol in the equation and how they are con-
nected. She also brings in knowledge about units and how the dimensions in the equation must
be consistent between terms. But she lends a deeper meaning to the equation by bringing in
additional knowledge about the physical situation of acceleration as a process that changes the
initial velocity. Further excerpts of her interview show that, for Pat, the mathematical formu-
lation of the equation is stably connected to the conceptual schema of change, where you start
with an initial quantity and then something changes it. The ‘at’ term for him reflects the total
amount by which the initial value is changed.
Later in the interview, we see that this difference in the interpretation of the equation is also
reflected in how Pat and Alex use the equation to solve a problem about differences in the speeds
of two balls thrown from a building at the same time with different initial velocities. Alex uses
the equation as a tool to generate final velocities given the initial velocity, time, and acceleration.
Pat on the other hand uses the equation much more like an expert, reaching the answer without
needing to plug in numbers and carry out arithmetic calculations, and she exhibits an expert-like
understanding of why the result should be what it is based on the structure of the equation. Pat
also spontaneously delves into multiple ways that the problem could be solved and says that if
she gets stuck, she would “switch to another one.” Alex in response to similar interview prompts
seems to think of the answer as just a result of numerical calculations rather than resulting from
the structure of the equation and says that one could not have reached the answer without doing
the calculations.

Our point here is that understanding equations in physics is not just about learning the math-
ematical operations. It is about making many connections to stores of knowledge about the
physical system; not just at the level of variable-definitions but at level of the bases of the math-
ematical operations and how those operations connect to physical meaning. Expectations that
the mathematics describing a physical system should connect to our intuitive sense about how
the physical system behaves can also provide a valuable safety net against possible errors in the
calculations and help us make sense of new or unfamiliar equations. This perspective, enhanced
by the tools developed recently by cognitive semanticists, should help give us new and deeper
insights into the use of mathematics in physics and the other sciences.

We are grateful for conversations on the topic of this paper with members of the University of
Maryland Physics Education Research Group. This material is based upon work supported by
the US National Science Foundation under Awards No. DUE 05-24987 and REC 04-40113.
Any opinions, findings, and conclusions or recommendations expressed in this publication are
those of the author(s) and do not necessarily reflect the views of the National Science Founda-

Carroll, L. (1871) Through the Looking-Glass and What Alice Found There.
Dray, T. & Manogue, C. (2003) Vector calculus bridge project website,
bridge/ideas/functions [original version posted May 2002; revised September 2003] (accessed February
15, 2008).
Evans, V. (2006) Lexical Concepts, Cognitive Models and Meaning-Construction, Cognitive Linguistics,
17:4, 491-534.
Evans, V. & Green, M. (2006) Cognitive Linguistics: An Introduction. Lawrence Erlbaum.
Fauconnier, G. (1985) Mental Spaces. Cambridge U. Press.
Fauconnier, G. (1997) Mappings in Thought and Language. Cambridge U. Press.
Fauconnier, G. & Turner, M. (2003) The Way We Think: Conceptual Blending and the Mind’s Hidden
Complexities. Basic Books.
Fillmore, C. (1976) Frame semantics and the nature of language. Annals of the NY Acad. of Sciences
280:1, 20-32.
Füster, J. (1999) Memory in the Cerebral Cortex: An empirical approach to neural networks in the human
and nonhuman primate. MIT Press.
Goldstein, R. (2006) Incompleteness: The Proof and Paradox of Kurt Gödel. Norton.
Hammer, D. (2000) Student resources for learning introductory physics, Am. J. Phys., PER Suppl., 68:7,
Hammer, D., Elby, A., Scherr, R. & Redish, E. (2004) Resources, framing, and transfer, in Transfer of
Learning: Research and Perspectives, J. Mestre, ed. (Information Age Publishing).
Langacker, R. (1987) Foundations of Cognitive Grammar, Vol 1: Theoretical Perspectives Stanford, U.
Lakoff, G. & Johnson, M. (1980/2004) Metaphors We Live By. U. of Chicago Press.
Lakoff, G. & Nuñez, R. (2001) Where Mathematics Comes From: How the Embodied Mind Brings
Mathematics into Being. Basic Books.
Poincare, H. (1905) Science and Hypothesis, translated by G. Halsted. Science Press, 1905; Dover
Publications, 1952.

Redish, E. (2004) A Theoretical Framework for Physics Education Research: Modeling student thinking,
in Proceedings of the International School of Physics, “Enrico Fermi” Course CLVI, E. F. Redish & M.
Vicentini (eds.), IOS Press.
Redish E.F. & Smith, K.A. (July, 2008) Looking beyond content: Skill development for engineers, J.
Eng. Educ. 97, 295-307.
Redish, E.F. (to be published) Problem Solving and the Use of Math in Physics Courses, in Proceedings
of the Conference, World View on Physics Education in 2005: Focusing on Change, Delhi, August 21-26,
Talmy, L. (2000) Toward a Cognitive Semantics, 2 vols. MIT Press.
Turner, M. (1991) Reading Minds: The study of English in the age of cognitive science. Princeton
University Press.

Zia, R., Redish, E., & McKay, S. (2009) Making sense of the Legendre transformation, Am. J. Phys. 77,

Emma Robertson[1,2] & Antje Kohnle[1]
[1] School of Physics and Astronomy, University of St Andrews, North Haugh, St Andrews, KY16 9SS,
[2] now at Standard Life pcl


With the aim of investigating student conceptual understanding in intermediate-
level quantum mechanics, we used internet forum discussion, published research
into student difficulties in quantum mechanics and our own teaching experience to
develop a concept test that was administered to 174 physics students in levels 2, 3
and 4 at the University of St Andrews (81 students in level 2, 49 in level 3 and 44 in
level 4, students may graduate with a BSc degree after level 4). Internet forum dis-
cussion offers insights into student thinking that may not be realized in a traditional
lecture or tutorial setting. The test covered understanding of probability, energy
and position measurement, the Heisenberg Uncertainty Principle, superposition of
states, the Pauli principle and quantum tunneling. This article discusses a subset
of test questions in detail, namely those on energy and position measurement and
quantum tunneling. The outcomes of the test showed widespread difficulties in un-
derstanding successive energy measurements, position measurement and quantum
tunneling, perhaps due to overgeneralization, false analogies and confusion of re-
lated concepts. We are currently developing and implementing visualizations and
animations to target areas of difficulty, and aim to measure their effectiveness in
improving student conceptual understanding of quantum mechanics.

Quantum mechanics is a subject that has become increasingly important in the undergradu-
ate physics curriculum, but the conceptual hurdles in this subject are perhaps greater than in
other areas of physics, due to being far from direct personal experience, often counterintuitive,
technically difficult and abstract, and fundamentally involving probabilities, in contrast to deter-
ministic classical physics. There has been substantial research done in recent years concerning
student difficulties and misconceptions in quantum mechanics (see references) which are often
seen to be similar despite different student backgrounds, teaching styles and textbooks (Singh,
2008), and reforms are under way in terms of course content, methodology and assessment, as
well as developing tools to aid conceptual understanding. Student understanding of quantum
mechanics suffers from false analogies with classical situations (Wittmann, 2005 & McKagan,

2008), difficulties with probabilistic interpretations (Bao, 2002 & Domert, 2005), confusion be-
tween related concepts (Singh, 2001) and a tendency to overgeneralize concepts learned in one
context to other areas where they are not directly applicable (Singh, 2008).
In our experience, students may become adept at doing calculations in quantum mechanics,
without having developed mental models or visual representations of fundamental concepts,
which are decisive for deep understanding and problem-solving ability. With the aim of investi-
gating student conceptual understanding at intermediate-level undergraduate quantum mechan-
ics at the University of St Andrews, we developed a diagnostic survey that was then adminis-
tered to our level 2, 3 and 4 (UK educational levels 8, 9 and 10, students graduate with a BSc
degree at the end of level 10) physics students. This survey was seen as a first step towards an
overarching aim of improving quantum mechanics instruction.


We used internet forum discussion, published research into student difficulties in quantum me-
chanics, and our own teaching experience to determine key areas that should be covered by
the test. Internet forum discussion has the potential to offer insights into student thinking that
may not be realized in a traditional lecture or tutorial setting, as students may feel less inhibited
about asking questions and contributing to discussion. As one example, consider the following
question on quantum tunneling8 “When a particle tunnels through a finite barrier, does it spon-
taneously appear on the other side, or does it follow a path through the barrier? If the latter then
does it interact with the barrier?” The way this question is phrased shows classical thinking (the
particle following a path), and seems to consider the barrier as a material barrier with which
energy could be exchanged.
We decided to focus on the topics understanding of probability, energy and position measure-
ment, the Heisenberg Uncertainty Principle, superposition of states, the Pauli principle and
quantum tunneling. We created 10 questions in total, one of these a linked question. Excepting
question 8, all questions were multiple choice (with one correct answer, excepting question 4
with two correct answers) with typically five choices. The subset of test questions discussed in
this article are shown in the Appendix. For each question, students were asked to rate their con-
fidence in their answer, and to explain their reasoning with a free-text comment. This scheme
aimed to gain more insight into student reasoning and underlying reasons for choosing answers.
The test was administered during lecture time in order to maximize participation in the mid-
dle of our summer semester in core courses at our levels 2 (UK educational level 8), 3 (UK
educational level 9) and 4 (UK educational level 10). Students were given 15-25 minutes to
complete the test. Students were told that the test was anonymous and would have no effect on
their final grade. In total, 174 students completed the test, 81 at level 2, 49 at level 3 and 44 at
level 4. About one week after the test, focus groups with a small number of student volunteers
(3-5) at each level were performed to gain additional insight. The focus group discussion was
Level 2 students had not yet started the Quantum Physics part of the course, and had not covered
any of the concepts of the test at university level. Probability, quantum tunneling and energy
from, accessed 10 August 2009.

measurement are discussed in the level 2 course. All other topics on the test are part of the
level 3 quantum mechanics courses Quantum Mechanics 1 and 2, excepting multiple position
measurements which are not part of the course. However, at the time of the test, level 3 students
had not yet covered spin (question on the Pauli principle). Excepting very few students, all
students at levels 3 and 4 have a degree intention of physics, theoretical physics or astrophysics,
some as a joint degree with other subjects.

We first looked at the percentage of correct answers and standard deviations across the different
levels. For each question excepting question 8, we first graphed the distribution of responses
for each level separately, colour-coding the different degrees of confidence. We then performed
a χ2 test on the distribution of responses to the same question for different levels (including all
degrees of certainty) to test for a significant (p < 0.01) change in the distribution of responses
from one level to the next. For each question, we transcribed the free-text comments explaining
reasoning and performed a content analysis of the comments.
The mean test score was 36.7% at level 2, 50.7% at level 3 and 48.6% at level 4. For individual
questions, as expected there is a significant change (namely an improvement, χ2 test with p <
0.01) in the distribution of responses from level 2 to level 3 excepting the two questions on
successive position measurements not discussed in this article. As stated above, none of the
concepts on the test had been covered at level 2. From level 3 to level 4, there is no significant
change in the distribution of responses excepting question 2.1 on position measurement.


Question 1.1 (see Appendix) tested students’ understanding of the effect of energy measurement
on a superposition of two energy eigenstates. The options tested misconceptions of wavefunc-
tion collapse, namely that a measurement of energy would measure one of the two energies
with certainty (A), the expectation value (B), the average of the two energy eigenvalues (C) or
a range of energy eigenvalues (E). The correct answer is (D), that the measurement gives either
energy E1 or E2 .
The results for this question are shown in Figure 1. The large majority of students (92% in
level 3 and 82% in level 4) chose the correct answer, and most of these students (98% and 87%)
were very or somewhat certain of their answer. Students choosing incorrect answers tended
to be uncertain of their answer. It seems that level 4 students have done more poorly, perhaps
forgetting quantum mechanics they had learnt in level 3, but the distributions of results is not
significantly different across the two levels (p = 0.15).
A large number of responses (63% at level 3 and 77% at level 4) contained additional comments
explaining reasoning. Misconceptions found in the comments to the correct choice D include:

• The wavefunction collapses to the energy eigenstate with the largest probability (1 com-

• The coefficients of the eigenfunctions are probabilities (2 comments)

Figure 1: Results of Question 1.1 on energy measurement. The answers are colour-coded according to
the degree of certainty (C: certain, SC: somewhat certain, SU: somewhat unsure, VU: very uncertain)
and the level of the degree programme.

• Confusion between eigenfunction and eigenenergy (“the wavefunction collapses to one

of the eigenvalues”) (2 comments)

Although the number of comments with these misconceptions is small, our lecturing experience
shows them to be not infrequent.


Student difficulties at advanced undergraduate level with the effect of prior measurements on
future measurement was studied in Singh, 2001. Question 1.2 (see Appendix) tested students’
understanding of successive energy measurements on a system that is unperturbed between
measurements. The options tested misconceptions of remaining in one energy eigenstate only
when the second measurement immediately follows the first (A), that any of the original energy
eigenvalues in the superposition state can be obtained (B), that a range of energy values can be
obtained (C), and that the collapse is to the expectation value of energy (E). The correct answer
is (D), that all subsequent energy measurements give the same result as the first measurement.
There was an ambiguity in the formulation of choice (A), so that as written, this answer could be
considered correct. We will rephrase “if they are undertaken” to “only if they are undertaken”
in the next iteration of the test.
Students were substantially less certain in their responses to this question compared with ques-
tion 1 (see Figure 2). Choices A, C and D were roughly equally common, and students choosing
all three were similarly certain of their choice. Only 35% of students in level 3 and 30% of stu-
dents in level 4 chose the correct answer, most of these students (82% at level 3 and 78% at
level 4) were certain or somewhat certain of their answer.
A substantial fraction of responses (41% at level 3 and 39% at level 4) contained additional
comments explaining reasoning. Surprisingly, none of the comments to the correct choice D (16

Figure 2: Results of Question 1.2 on successive energy measurements. The answers are colour-coded
according to the degree of certainty (C: certain, SC: somewhat certain, SU: somewhat unsure, VU: very
uncertain) and the level of the degree programme.

in total) mentioned the time-dependence of the wave function at all, instead focusing on “only
one coefficient being left in the superposition, which is now one”. In contrast, all comments (7)
to choice C and half (6 of 12) of those to choice A stated that the wave function “evolves with
time”, but apparently did not consider the nature of the time dependence, i.e. exp(−iωt). Thus,
there seems to be some confusion i) that the wave function has a time-dependence even when
in an energy eigenstate and ii) that this time-dependence exp(−iωt) is trivial in that the state is
stationary, i.e., the probability density does not depend on time.


Question 2.1 (see Appendix) tested students’ understanding of position measurements, assum-
ing a wavefunction given by an energy eigenstate in the infinite square well. The options in-
cluded collapse to a small interval ∆x without (choice A) and with (choice C) renormalization,
the collapse to a delta-function at an arbitrary position in the well (choice B) and necessarily at
the centre of the well (choice E), and the collapse to a small interval with a Gaussian probability
density centred at the middle of the interval (choice D). We considered choice C correct.
Disregarding the debate of position eigenfunctions being δ-functions (that would then spread
infinitely quickly) and the more physical collapse into a small interval ∆x, the large number
of responses choosing D was surprising (see Figure 3). At level 3, 39% of students chose D,
with 77% being very or somewhat certain, and at level 4, 30% chose D, with 85% being very or
somewhat certain. This result was particularly surprising, as level 3 students had completed a
tutorial problem on position measurement of a two-dimensional wave function just two weeks
prior to this test. We performed content analysis on the large number of 3rd and 4th year
comments explaining reasoning (55% of 3rd level and 66% of 4th level responses). For the
students who chose answer D, common reasoning included the Heisenberg uncertainty principle
(13 students), the spread in position due to the measuring apparatus (4 students) and the fact

Figure 3: Results of Question 2.1 on position measurements. The answers are colour-coded according
to the degree of certainty (C: certain, SC: somewhat certain, SU: somewhat unsure, VU: very uncertain)
and the level of the degree programme.

that the standard probabilistic distribution is Gaussian (2 students). The focus group discussion
provided additional insight into why some students chose D:

• 4S2 (level 4 student 2): “Everything statistical in real life is Gaussian. Everything’s
always fuzzy at the sides, like a laser.”

• 3S2 (level 3 student 2): “D looks more symmetric.”

• 3S4 (level 3 student 4): “D would be like redefining a new square well.”

Students who chose E commented that “the greatest probability of finding the particle lies in
the centre.” (5 students) and that “The wavefunction is symmetric, therefore measurement of
position expectation value will give a/2.” (4 students). Thus, students choosing E seem to be
confusing the expectation value, the most likely value and possible measurement outcomes.
Level 2 results are not discussed here, as position measurement had not been yet been covered
in the course. Only 54% of level 2 students answered question 2.1, and there is a large spread
in responses, with large uncertainty.


Question 8 (see Appendix) was different from the other questions in that it was not a multiple
choice question but instead asked the student to sketch the wave function for two regions depict-
ing quantum tunneling. The correct form is shown at the top of Figure 4, with an exponential
decay inside the barrier, and an oscillatory solution with identical wavelength to region I outside
the barrier. Only 12% of level 3 students and 11% of level 4 students sketched the correct wave

Figure 4: Wave function drawn in response to Question 8 on quantum tunnelling, with the number of
students quoted only referring to level 3 and 4 students. Wave functions in the figure without student
numbers were sketched only by level 2 students.

Figure 4 shows some of the wave functions sketched in response to this question, with the
number of students quoted only referring to level 3 and 4 students, as level 2 students had not
yet covered quantum tunneling. Wave functions in the figure without student numbers were
sketched only by level 2 students. The wave functions 2 and 9 are particularly common. Both
of these show the so-called “axis shift” seen in other investigations (Wittmann, 2005, McKagan,
2008). A possible reason for the axis shift given in Wittmann 2005 is that in depictions of bound
state problems, the height of the axis of oscillation is a measure of the energy of the particle,
and thus the axis shift would be consistent with students’ description of energy loss in tunneling
(see below). It is also worth noting that depictions of quantum tunneling in at least one standard
nuclear and particle physics textbook incorrectly show an axis offset. Even though the question
explicitly asked students to draw an axis into the sketch, almost no students (8% at level 3 and
14% at level 4) did so.
For the sketched wave function 6 (see Figure 4), the fact that no wave function is drawn inside
the barrier may be due to uncertainty in recalling the form of the wave function in this region.
However, one student sketching wave function 6 comments “There is low probability the par-

ticle will be found on the far side of the barrier and no probability it will be found inside the
Another well-known misconception associated with quantum tunneling is energy loss (Wittmann,
2005, McKagan, 2008). 17% of 3rd and 4th year comments (9% of students) commented that
the total energy of the electron decreases when tunneling through the barrier. This may be due to
a false analogy with classical systems such as a bullet going through a wall. It may also be due
to students considering the wave function as a classical displacement field, where amplitude is
indeed related to energy. Reasoning also showed a common confusion between amplitude and
its relation to probability and wavelength and its relation to energy. Of the 17% of students
who commented on energy loss, almost all of them draw wavefunction 2, which does not show
the wavelength becoming longer in region III. Thus, there seems to be confusion between wave
function amplitude and its relation to probability and wavelength and its relation to energy.
This was also found in the interviews:

• 3S1 (level 3 student 1): “We have exponential decay in the tunnelling then it comes back
out into a wave function but with less energy so the amplitude is decreased”.

• 4S5 (level 4 student 5): “tunnel exponentially lowers the electron’s energy so when it
emerges it has lower amplitude of oscillation”

Although wave function 9 shows an energy gain when tunneling through the barrier (the wave-
length has become shorter), one student still commented in their reasoning that the electron
loses energy when tunneling through the barrier. In contrast, the student drawing wave function
5 commented “When the particle enters the barrier, it gains energy so oscillates more quickly”,
which is consistent with the drawing. This student may be confusing the case of a particle
incident on a potential barrier of height +V0 with a potential well of height −V0 .

We have developed a preliminary diagnostic survey which aims to assess student conceptual
understanding of quantum mechanics and shed light on some of the underlying reasons for dif-
ficulties with particular quantum mechanics concepts. We found that, while students have little
difficulties with energy measurement on a superposition of energy eigenstates, some students
tend to not consider the time-dependence or its particular form when considering successive
energy measurements on the same system. For position measurement, some students tended
to overgeneralize results from other areas (Gaussian shape of the resulting wave function), or
confuse expectation value, most likely value, and possible measurement outcomes. For quan-
tum tunneling, we confirmed difficulties reported in the literature such as energy loss, and found
that difficulties arise from confusion between amplitude and its relation to probability and wave-
length and its relation to energy.
These results are preliminary, as the diagnostic survey was only used with a limited number
of students at a single institution for one year. The results of the test described here are only
a first step towards improving undergraduate quantum mechanics teaching. We plan to repeat
the test in the 2009/10 academic year, extending the test to more areas, and including more

validation of test questions. We also hope to extend the test to other institutions in order to
allow a cross-institutional comparison. Repeating the test will allow us to verify the reliability
of our results.
Visualizations can play an important role in helping students to build mental images to under-
stand quantum-mechanical concepts (McKagan, 2008). Thanks to funding from the UK Higher
Education Academy, we are developing quantum mechanics animations targeting particular ar-
eas of conceptual difficulty. The animations are being used in the 2009/10 academic year in
three quantum mechanics courses at levels 2 and 3, through incorporation into lectures and
tutorial problems. We aim to evaluate their effectiveness in improving student conceptual un-
derstanding of quantum mechanics. We are also involved in a wider debate on the reform of
undergraduate teaching of quantum mechanics.

We thank Christopher Hooley for helpful discussions and feedback, for constructive comments
and suggestions on a draft of this paper, and for his help in obtaining the level 3 data.

Questions of the quantum mechanics test discussed in this article (questions 1.1, 1.2, 2.1 and 8)


A wave function in an infinite potential well at time t = t0 is given by a superposition of the

ground state (φ1 ) and 1st excited state (φ2 ):

−iE1 t −iE2 t n2 π 2 2
ψ(x, t) = c1 φ1 e  + c2 φ 2 e  , En = , (19)
c1 and c2 are real numbers > 0.
What result do we get if we perform an energy measurement on the wave function?

A. Measurements give the ground state energy

B. Measurements give the expectation value of the energy, Ĥ

C. Measurement yields a value for energy that is the average of E1 and E2

D. Measurement yields either eigenvalue E1 or E2

E. Measurement finds the energy within an energy range ∆E


How does a measurement of energy affect the wave function?

A. The wave function collapses into an energy eigenstate. Further energy measurements will
give the same result as the first measurement if they are undertaken in rapid succession

B. The wave function collapses into an energy eigenstate. Another energy measurement
made immediately after could give any of the possible energy values

C. The wave function collapses to an energy eigenstate which evolves with time so future
energy measurements could yield different values

D. The wave function collapses into an energy eigenstate. All future energy measurements
give the same result as the first measurement

E. The wave function collapses to the expectation value of energy; Ĥ and this result is the
same for all future energy measurements.


The following diagram depicts the ground state wave function in an infinite potential well at
time t = 0. The axis values are φ(x) vs x and the walls of the well are at x = 0 and x = a:

π 2 2 2 πx
E = E1 = , φ1 (x) = sin( ). (20)
2ma2 a a

Choose which of the following diagrams depicts a possible form of the wave function at the
instant a measurement of position is made. Axes are the same scale as the diagram above.

Position measurement Position measurement at Position measurement
around 3a/4 finds the 3a/4 finds the particle to around 3a/4 finds the
particle in the range ∆x, be at this point, collapsing particle in the range ∆x,
collapsing the wave function the wave function to a delta collapsing the wavr function
to zero everywhere apart function at this point. to zero everywhere apart
from in this range. from this range.

Position measurement Position measurement will

around 3a/4 finds the always yield the position ex-
particle in the range ∆x, pectation value, x̂ = a2 ,
collapsing the wave function collapsing the wave function
to zero everywhere apart to a delta function at a/2.
from in this range.

Quantum tunnelling is used to describe the process of α decay and fission of the nucleus. Con-
sider the more general case of tunnelling for an electron with energy E incident on the potential
barrier in the plot of V (x) vs x below:

E < V0 (21)

0 x < 0, x > a,
V (x) = (22)
V0 0 ≤ x ≤ a.

Fill in the blanks in the plot of ψ(x) vs x below, sketching the x axis and the form of the
electron’s wave function in regions II and III. The electron’s wave function in region I is drawn
in already.

Bao, L. & Redish, E. F. (2002) Understanding probabilistic interpretations of physical systems: A pre-
requisite to learning quantum physics, Am. J. Phys., 70 (3), 210-217
Carr, L. D. & McKagan, S. B. (2009) Graduate quantum mechanics reform, Am. J. Phys, 77 (4), 308-319
Cataloglu, E. & Robinett, R.W. (2002) Testing the development of student conceptual and visualization
understanding in quantum mechanics through the undergraduate career, Am. J. Phys., 70 (3), 238-251
Domert, D., Linder, C. & Ingerman, A. (2005) Probability as a conceptual hurdle to understanding one-
dimensional quantum scattering and tunnelling, Eur. J. Phys., 26, 47-59

McKagan, S. B., Perkins, K. K., Dubson, M., Malley, C., Reid, S., LeMaster, R. & Wieman C. E. (2008)
Developing and Researching PhET simulations for Teaching Quantum Mechanics, Am. J. Phys., 76
(4&5), 406-417
McKagan, S. B., Perkins, K. K. & Wieman, C. E. (2008) Deeper look at student learning of quantum
mechanics: The case of tunneling, Phys. Rev. ST Phys. Educ. Res., 4, 020103-1 to 18
Singh, C. (2001) Student understanding of quantum mechanics, Am. J. Phys., 69 (8), 885-895
Singh, C. (2008) Student understanding of quantum mechanics at the beginning of graduate instruction,
Am. J. Phys., 76 (3), 277-287
Styer, D. F. (1996) Common misconceptions regarding quantum mechanics, Am. J. Phys., 64 (1), 31-34
Wittmann, M. C., Morgan, J. T. & Bao, L. (2005) Addressing student models of energy loss in quantum
tunnelling, Eur. J. Phys., 26, 939-950

Raluca Teodorescu[1] , Cornelius Bennhold[2] , Gerald Feldman[3] and Larry Medsker[4]
Department of Physics, The George Washington University, 725 21st Street NW, Washington, DC 20052,
[1] [2] Deceased [3] [4]

As part of our ongoing process of curriculum reform, we have developed a Taxonomy
of Introductory Physics Problems (TIPP), which relates physics problems to the cognitive
processes and the knowledge required to solve them. TIPP has been used in the years
2006-2009 to reform the first semester of the introductory algebra-based physics course at
The George Washington University. The reform targeted students’ cognitive development
and attitudes. The methodology employed in the course involves engaging students with
physics problems in a variety of contexts. To assess student problem-solving abilities, we
created new rubrics. To evaluate student attitudes about learning physics, we administered
the Colorado Learning Attitudes about Science Survey (CLASS) pre- and post-instruction.
Our results show clear gains in the areas targeted by our reform. This paper describes the
general features of our taxonomy of introductory physics problems, how we used it to select
and implement the physics problems in the curriculum, and the positive results that we have

Research in the cognitive sciences over the years has revealed the complex and dynamic character of
the problem-solving process (Conlin et al., 2007; Mayer and Wittrock, 2006; Ross, 2007). Forty years
of research in expert-novice problem-solving behavior have elucidated specific thinking characteristics
of the two categories of learners (Gerace and Beatty, 2005; Hsu et al., 2004; Maloney, 1994). This
research suggests that asking students to solve high-level thinking problems can help them become more
expert-like problem solvers. To achieve this goal, physics problem developers have made efforts to
create physics problems that move the novices towards a more expert-like status (Beatty et al., 2006;
McDermott et al., 2002; van Heuvelen and Etkina, 2006). However, to accomplish this objective, one
needs to understand better the relationship between the problems themselves and the thinking processes
and knowledge that they involve. Ultimately, the curriculum is determined by the type of thinking being
targeted for instruction – only then can specific problems be chosen that trigger the desired cognitive
The first part of this paper presents a basic framework (taxonomy) of physics problems which estab-
lishes a relationship between the physics problems and the cognitive processes and the knowledge that
are required for their solution. Then, using this taxonomy as a guide, the second part of the paper de-
scribes an instructional strategy designed to enhance students’ problem-solving abilities by monitoring
the cognitive load of the problems that they are asked to solve. This curriculum cannot exist without
the background framework of the taxonomy to identify the cognitive processes, so we have developed a
so-called “thinking skills” curriculum for the introductory algebra-based physics course.

As collections of research-based physics problems became more numerous and diverse, Physics Edu-
cation Research (PER) groups have undertaken attempts to classify the problems. This has been done
explicitly (Campbell, 2008; Harper et al., 2006) by establishing categories of existing physics problems
or implicitly (van Heuvelen and Etkina, 2006) by labeling the problems created by the group.
The uniqueness of our taxonomy derives from the fact that it attempts to put together elements from cog-
nitive science, educational psychology and expert-novice research. By taxonomy we mean “an orderly
classification of items according to their presumed natural relationship” (Hauenstein, 1998).


Our framework seeks to group physics problems according to the cognitive processes that are needed
to solve them in a hierarchical order, such that lower-level thinking processes correspond to appropriate
low-level problems while more complex problems are associated with higher-level processes. We rely
on the following resources: a) cognitive research findings adopted by PER, b) expert-novice research
discoveries acknowledged by PER, c) a taxonomy for educational objectives and d) collections of physics
problems created by Physics Education researchers or developed by textbook authors.
In developing the Taxonomy of Introductory Physics Problems (TIPP), we applied the following proce-

1. We considered cognitive processes and knowledge domains that PER researchers (Beatty et al.,
2006; Ross, 2007; Tuminaro and Redish, 2007) identified as relevant for physics problem solving.

2. We searched among taxonomies of educational objectives (Bloom, 1956; Haladyna, 1997; Hannah
and Michaelis, 1977; Marzano and Kendall, 2007) and found one (Marzano and Kendall, 2007)
that presents the problem-solving process in accord with the criteria presented above.

3. We developed an algorithm that uses Marzano‘s taxonomy (Marzano and Kendall, 2007) to clas-
sify the physics problems according to the cognitive processes and knowledge they involve.

4. We built a Taxonomy of Introductory Physics Problems (TIPP) which is a database that contains
text-based and research-based physics problems and explains their relationship with cognitive
processes and knowledge.

5. We assessed TIPP validity and reliability.

2.1.1 Marzano’s New Taxonomy of Educational Objectives In a previous paper (Teodorescu

et al., 2008a), we described in detail Marzano’s New Taxonomy (MNT). Here we will only remind
the reader that it is a two-dimensional framework having three systems as one dimension and three
domains of knowledge as the other dimension. The actions of the three systems (Self, Meta-Cognitive,
and Cognitive) upon the three knowledge domains (information, mental procedures and psychomotor
procedures) are driven by the levels of the student’s self-awareness. Marzano establishes the following
six levels of knowledge processing that take place in one’s mind:

• Level 6: Self-System
• Level 5: Meta-Cognitive System

• Level 4: Knowledge Utilization (Cognitive System)
• Level 3: Analysis (Cognitive System)
• Level 2: Comprehension (Cognitive System)
• Level 1: Retrieval (Cognitive System)

For the purpose of this project we will restrict ourselves to the lowest four levels that pertain to the
Cognitive System. Table 1) presents a summary of Marzano’s New Taxonomy.

Level 1: a) Recalling
RETRIEVAL b) Executing
Level 2: a) Integrating
COMPREHENSION b) Representing (Symbolizing)
a) Ranking (Matching)
Level 3: b) Classifying
ANALYSIS c) Analyzing errors
d) Generalizing
e) Specifying
a) Decision Making
Level 4: b) Overcoming Obstacles
KNOWLEDGE c) Experimenting
UTILIZATION d) Investigating

Table 1: Summary of Marzano’s New Taxonomy of Educational Objectives (Marzano and Kendall,

2.1.2 Classification of Physics Problems in TIPP We consider several ways of classifying physics
problems. We will describe below the most suitable classification for instructional purposes, where a
problem is classified according to:

• the type of knowledge involved in it. The knowledge we imply here is information (vocabulary
terms, facts, time sequences, generalization and principles) and mental procedures (single rules,
algorithms, tactics and macro-procedures).
• the highest complex cognitive process that is necessary to solve it (for both information and mental

To illustrate our procedure we present below an example problem and explain the algorithm we use to
classify it.

Example Problem: A car of mass m moves with a speed v on a frictionless air track and collides with
an identical car that is stationary. If the two cars stick together after the collision, what is the final
kinetic energy of the system?

A student required to solve this problem would process the knowledge (information and mental proce-
dures) in different ways. Concerning information, the student has to:

• recall the concepts of mass, velocity, momentum and kinetic energy;

• decide what are the key elements that need to be taken into account (integrate the facts);
• represent the information (symbolize the facts).

Regarding mental procedures, the student needs to execute the algorithms of writing the conservation of
momentum law, solving for velocity and calculating the kinetic energy.
In conclusion, in TIPP, based on Table 1, this problem would have the designation (I:2, MP:1), indicating
that the highest cognitive process required with regard to information is symbolizing (level 2) and with
regard to mental procedures is executing (level 1). The interested reader can find more details about TIPP
in Teodorescu et al. (2008b).


To test TIPP validity (Morris and Fitz-Gibbon, 1978; Wiersma and Jurs, 1990), we initiated discussions
about Section 2.1.2 of this document between one graduate student and six professors (five teaching in
universities and one in a community college). The graduate student and the professors have been asked
about issues related to TIPP relevance, clarity and comprehensiveness. Unanimously, they found TIPP
To test the reliability, we asked one graduate student and three instructors to categorize problems accord-
ing to the algorithm presented in Section 2.1.2 at different moments in time. The problems were selected
from seven textbooks and several PER collections including (Beatty et al., 2006; McDermott et al., 2002;
van Heuvelen and Etkina, 2006). The Cohen’s kappa coefficient of inter-rater reliability agreement coef-
ficient (Cohen, 1960) ranged between 0.70-0.83. According to the literature, the values of this coefficient
suggest that there is a good level of agreement between the raters and that the taxonomy appears to be
Note: Training is necessary for achieving these values of agreement. In addition, clear assumptions
about the material taught in class have to be made before the classification. Unless specified, different
instructors usually assume different levels of coverage of the same topic. Without training and clear
assumptions, the Cohen’s kappa coefficient can be below 0.50.


Our pedagogy relies on didactic constructs such as the GW-ACCESS problem-solving protocol, classifi-
cation schemes and learning progressions that we have developed and implemented in our introductory
physics course. In Teodorescu et al. (2008a) we presented descriptions of these strategies and the struc-
ture of our curriculum. Here we will elaborate more on the aspects that target the enhancement of
cognitive development through physics problem solving.

The key elements of our pedagogy seek to progressively engage students with problems of increasing
cognitive complexity and, at the same time, to continually offer them the “big picture” of the covered
material. To accomplish the first objective, we used the GW-ACCESS protocol (Table 2) taught in steps
on a weekly basis and learning progressions. Table 2 shows the correspondence between the cognitive
processes used in our problem-solving protocol and the ones described in MNT.

ACCESS Cognitive Processes (According to MNT)

Assess the problem Classifying, Ranking (Level 3a, 3b)
Create a drawing Representing information (Level 2b)
Conceptualize the strategy Integrating, generalizing, specifying, decision making,
overcoming obstacles (Level 2a, 3d, 3e, 4a, 4b)
Execute the solution Executing mental procedures (Level 1b)
Scrutinize your results Analyzing errors, checking reasonableness (Level 3c)
Sum up your learning Meta-learning (Level 5)

Table 2: Cognitive processes from MNT involved in the GW-ACCESS protocol.

The learning progressions are entities that combine traditional and research-based physics problems
such that, during each week as students learn a new topic, they also gradually experience a new type
of thinking that involves certain contexts, problem features, and cognitive processes. By context we
mean two things: abstract scenario versus real-world scenario and symbolic versus numeric. By problem
features we mean surface features (e.g. inclined plane, bungee jumper, etc.) versus deep features (e.g.
Newton’s 2nd law, energy conservation, momentum conservation, etc.). By cognitive processes we mean
the thinking processes featured in Table 1.
It should be mentioned that the term “learning progression” has been previously used by Anderson et al.
(2005) with a different meaning. By “learning progression”, they mean a different instructional approach
that extends over several years of instruction, which is not our present context.
Four types of learning progressions have been created, and are explained below:

• Learning Progression 1 (Figure 1), used in homework and in recitations, blends deep features and
contexts maintaining the same surface features. The key insight here is that from week 2 to week
6 we gave students the same problem involving a skier and an inclined plane which was altered
to deal with a different physics principle. Also, in weeks 2 and 4 the problems were in symbolic
context, while in week 6 in numeric context.
• Learning Progression 2 (Figure 2), used in recitation and homework, combines problems having
the same surface features and contexts but different deep features. Figure 2 shows that the system
presented (surface feature) is always a bungee jumper and the context is always numerical, but the
physics principle (deep feature) involved in each of the problems is different.
• Learning Progression 3 (Figure 3), used to link two or more curriculum units, brings together
problems having the same deep features but different surface features and contexts. Note how
a symbolic problem regarding Newton’s 2nd Law given in week 2 in recitation is followed by

Figure 1: An example of Learning Progression 1 taken from recitation. Note the same surface features
(inclined plane) but different contexts and deep features.

numeric problems (involving Newton’s 2nd Law and the same free-body diagram) given in the
homework. All these problems require similar free-body diagrams.
• Learning Progression 4 shown previously in Teodorescu et al. (2008a) was designed to have
students practice specific cognitive processes in addition to the thinking triggered by the above
progressions. The cognitive processes targeted are: filtering the relevant information, creating a
strategy, problem classification, ranking of different physical quantities, comparison and analy-
sis of different phenomena, analyzing procedural errors (i.e., identify errors in wrong solutions
offered) or informational errors (i.e. numbers, premises). Table 3 summarizes all the learning
progressions we used and explains their parameters.

To continually offer the “big picture” aspect of the studied material, we used classification schemes that
organize the different types of problems from one or more units that have been covered. One of these
schemes used for the unit of Fluids was shown in Teodorescu et al. (2008a). Figure 4 shows the scheme
we use for the unit of Energy.

Figure 2: An example of Learning Progression 2 taken from homework. Note the same surface features
(bungee jumper) and context (numerical) but different deep features.

Learning Progression Surface Deep Contexts Curriculum Units

Features Features
Learning Progression 1 same different different recitation + homework
Learning Progression 2 same different same homework
Learning Progression 3 different same different all
Learning Progression 4 targets specific cognitive processes recitation

Table 3: Summary description of the blending of surface features, deep features and contexts in the
learning progressions.


By teaching students thinking abilities according to the methods described in Section 3, we aimed to help
them develop the cognitive abilities necessary for physics problem solving. The question then becomes:
What is the best method for testing the effectiveness of this new approach to teaching physics? In other
words, how do we evaluate the cognitive abilities that students have? For the purpose of this project, we

Figure 3: An example of Learning Progression 3 that combines problems from recitation and home-
work. Note the same deep features (Newton’s 2nd Law, the same free-body diagram), but different
surface features (helicopter and train) and contexts (symbolic and numeric, abstract and real-world).

Figure 4: Problem classification scheme used for the unit of Energy.

have chosen rubrics as a straightforward method of measuring the improvement of student performance
in cognitive processes as applied to solving a physics problem. One of the main advantages of rubrics is
that they allow for the evaluation of students’ performance on specific component processes independent
of their performance on other component processes of problem solving.
In our reform approach, we aimed to pay special attention to student attitudes and dispositions towards
learning physics. We believe that each of the methods presented in Section 3 contributed to improvement
in student attitudes. In addition, we spent some class time discussing why physics is necessary and
when, where and how students can use the physics they learned and the abilities they acquired. To
evaluate student dispositions we used the Colorado Learning Attitudes about Science Survey (CLASS)
(Adams et al., 2006).


In designing our rubrics, we follow the approach developed by Etkina et al. (2006). The assessment
problem featured below has been given to students after they learned Newton’s 2nd Law (week 3) and in
the final exam (week 15). Separate rubrics were designed to grade the main cognitive processes needed
to solve this problem: Assess the problem, Create a drawing, Conceptualize the strategy and Execute
the solution parts. Each rubric has a range of scores from 0 to 3, with 0 (missing), 1 (inadequate), 2
(needs improvement), or 3 (adequate).

Figure 5: Assessment problem: Two blocks of masses M1 = 8 kg and M2 = 20 kg are connected

like in the picture shown (fig 5). The two inclined planes are frictionless. The string and the pulley have
negligible mass. Knowing that θ = 60o and β = 30o , find the acceleration of each block and the tension
in the string.

Figure 6: Students’ scores for Assess the problem ability.

Figures 6-9 show the rubrics used to grade each of the abilities and the positive shifts in student perfor-
mance for each of the abilities targeted. The gain indicated in the figures is simply the difference between
the post- and pre- distribution means.
The results for Assess the problem ability (Figure 6) indicate the highest gain in students’ performance
as compared with the other abilities. This may suggest that, among the abilities taught, this one seems to
be the best learned by the students.
The results for Create a drawing ability (Figure 7) indicate the lowest gain in students’ performance as
compared with the other abilities. This may suggest that, among the abilities taught, this one seem to be
the hardest for students to learn.
The results for Conceptualize the strategy ability (Figure 8) indicate a high gain in students’ perfor-
mance. This gain, associated with the low pre-test distribution mode (which is 1), may imply that stu-
dents made good progress in learning this ability.
The results for Execute the solution ability (Figure 9) indicate a moderate gain in students’ performance.
It should be noted that the assessment problem we used is considered a difficult problem for algebra-
based physics students. The moderate gain can be due to the difficulty of this problem.
The positive shifts obtained after the instruction for all these abilities suggest that students’ problem-
solving behavior has improved and our teaching strategy seemed to be effective. In addition, what we

Figure 7: Students’ scores for Create a Drawing ability.

appreciate as extremely important is the interesting dynamics of students’ thinking – the fact that the
improvement depended on the nature of the cognitive processes exercised. Students did not perform in
the same way on all the cognitive processes. Their pre-instruction scores for Assess the problem were the
highest, as was the gain for this ability. For Execute the solution, students had the lowest pre-instruction
scores, yet the gain was moderate. An interesting case is Create a drawing – while students scored
relatively highly before instruction, they did not improve much. Student post-instruction performance
was highest on the Assess the problem ability and lowest on the Execute the solution ability. The highest
gain was achieved for the Assess the problem ability and the lowest for the Create a drawing ability.
We note, however, that these variations could also possibly reflect influences of the instructor or the
curriculum design, in addition to the abilities of the students.


One of the recently developed instruments to measure students’ beliefs and attitudes is the Colorado
Learning Attitudes about Science Survey (CLASS) (Adams et al., 2006). The survey consists of 42 ques-
tions that are scored in eight categories: 1) real-world connections, 2) personal interest, 3) sense making
effort, 4) conceptual understanding, 5) applied conceptual understanding, 6) problem solving (general),
7) problem solving (confidence), and 8) problem solving (sophistication). The research by Adams et
al. (2006) on student attitudes before and after traditional introductory physics instruction revealed that
student beliefs and attitudes deteriorate during the course of a standard physics class, meaning that they

Figure 8: Students’ scores for Conceptualizt the strategy ability.

have moved towards views held by novices and away from the views held by experts. Our results show
improvement of student attitudes in seven out of eight categories. Figure 10 shows a comparison between
the typical negative normalized gains found in a calculus-based physics course (as reported by Adams
et al. (2006)) and the positive normalized gains we obtained in our Phys 11 Fall 2008 course. This
comparison is not a perfect match because we compare our algebra-based physics course with a typical
calculus-based physics course. We are in contact with the developers of this instrument, and we plan to
use data related to algebra-based courses as soon as they become available. These data will allow us to
make a better evaluation of our achievements. Nevertheless, this comparison is still meaningful since it
is known that calculus-based students have fewer difficulties with physics than algebra-based students,
which suggests that they have better attitudes about learning physics.

In this article, we explain the Taxonomy of Introductory Physics Problems (TIPP) that we created. We
also describe how we used TIPP to implement problems in our curriculum. The assessment we admin-
istered used rubrics to evaluate students’ thinking while solving physics problems, and we characterized
their attitudes with CLASS. The rubrics we designed made possible the independent evaluation of the
different types of thinking involved in the problem-solving process. The results we obtained show gains
in students’ thinking and also improvement in their attitudes.

Figure 9: Students’ scores for Execute the solution ability.

Further work on TIPP will extend the taxonomy to address meta-cognitive aspects of student thinking.
Regarding the curriculum, we will continue to refine our methodology and assessment while transferring
the course into a SCALE-UP collaborative format (Beichner et al., 2007) using a grant that we recently
received from the National Science Foundation.

Adams, W. K., Perkins, K. K., Podolefsky, N., Dubson, M., Finkelstein, N. & Weiman, C. E. (2006). A
new instrument for measuring student beliefs about physics and learning physics: the Colorado Learning
Attitudes about Science Survey. Physical Review Special Topics: Physics Education Research 2(1),
Anderson, C. W., Mohan, L., & Sharma, A. (2005). Developing a learning progression for carbon cycling
in environmental systems. Paper presented at the Annual Meeting of The Ecological Society of America,
Montreal, Canada.
Beatty, I.D., Gerace, W.J.L. & Dufresne, R.J. (2006). Designing effective questions for classroom re-
sponse system teaching. American Journal of Physics, 74(1), 31-39.
Beichner, R. J., Saul, J. M., Abbott, D. S., Morse, J. J., Deardorff, D. L., Allain, R. J., Bonham, S. W.,
Dancy, M. H. & Risley, J. S., (2007). The Student-Centered Activities for Large Enrollment Under-

Figure 10: The positive CLASS normalized gains achieved in Phys 11 Fall 2008 course as compared to
the negative ones represented by Adams et al. (2006).

graduate Programs (SCALE-UP) Project. In E. F. Redish (Ed.), Research-Based Reform of Introductory

Physics; available online at
Bloom, B.S. (1956). Taxonomy of Educational Objectives, Handbook I: The Cognitive Domain. NY:
David McKay.
Campbell, R. W. (2008). A conceptual analysis of introductory physics textbooks. MS Thesis, University
of Arkansas.
Cohen, J. (1960). A coefficient of agreement for nominal scales. Educational and Psychological Mea-
surement, 20(1), 37-46.
Conlin, L. D., Gupta, A., Scherr R. E. & Hammer, D. (2007). The dynamics of students-behaviors and
reasoning during collaborative physics tutorial sessions. Proceedings of the 2007 Physics Education
Research Conference, AIP.
Etkina, E., van Heuvelen, A., White-Brahmia, S., Brookes, D. T., Gentile, M., Murthy, S., Rosengrant, D.
& Warren, A. (2006). Scientific abilities and their assessment. Physical Review Special Topics: Physics
Education Research 2(2), 020103.
Gerace, W. J. & Beatty, I. D. (2005). Teaching vs. Learning: Changing Perspectives on Problem Solving

in Physics Education. Invited talk at the 9th Conference of the Cyprus Physics Association and Greek
Physics Association, Nicosia, Cyprus.
Haladyna, T.M. (1997). Writing test items to evaluate higher order thinking. MA: Allyn & Bacon.
Hannah, L.S. & Michaelis, J.U. (1977). A Comprehensive Framework for Instructional Objectives: A
Guide to Systematic Planning and Evaluation. MA: Addison-Wesley.
Harper, K.A., Freuler, R.J. & Demel, J.T. (2006). Cultivating Problem Solving Skills via a New Problem
Categorization Scheme. In Proceedings of the 2006 Physics Education Research Conference, Syracuse,
Hauenstein, A. D. (1998). A Conceptual Framework for Educational Objectives: A Holistic Approach
to Traditional Taxonomies. MD: Univ. Press of America.
van Heuvelen, A. & Etkina, E. (2006). The Physics Active Learning Guide, CA: Addison-Wesley.
Hsu, L., Brewe, E., Foster, T. M. & Harper, K.A. (2004). Resource Letter RPS-1: Research in problem
solving. American Journal of Physics, 72(9), 1147-1156.
Maloney, D. P. (1994). Research on problem solving. In D. L. Gabel (Ed.), Handbook of Research on
Science Teaching and Learning. NY: Simon & Schuster MacMillan.
Marzano, R. J. & Kendall, J. S. (2007). The New Taxonomy of Educational Objectives, 2nd Ed. CA:
Corwin Press.
Mayer, R.E. & Wittrock, M.C. (2006). Problem Solving. In H. Alexander and R. Winnie (Eds.), Hand-
book of Educational Psychology, 2nd Ed., NY: Wiley.
McDermott, L., Shaffer P. & University of Washington Physics Education Group (2002). Tutorials in
Introductory Physics, NJ: Prentice Hall.
Morris, L. L. & Fitz-Gibbon, C. T. (1978). How to Measure Program Implementation. CA: Sage Publi-
Ross, B. H. (2007). Cognitive Science: Problem Solving and Learning for Physics Education. In Pro-
ceedings of the 2007 Physics Education Research Conference, Greensboro, NC.
Teodorescu, R., Bennhold, C. & Feldman, G. (2008a). Enhancing cognitive development through physics
problem solving: An example of a thinking-skills curriculum. In Proceedings of the 2008 Groupe Inter-
national de Reserche de L’Enseignement de la Physique Conference, Nicosia, Cyprus.
Teodorescu, R., Bennhold, C. & Feldman, G. (2008b). Enhancing cognitive development through
physics problem solving: A taxonomy of introductory physics problems. In Proceedings of the 2008
Physics Education Research Conference, Edmonton, Alberta, Canada.
Tuminaro, J & Redish, E.F., (2007). Elements of a cognitive model of physics problem solving: Epis-
temic games, Physical Review Special Topics- Physics Education Research 3(2), 020101.
Wiersma, W. & Jurs, S. G. (1990). Educational Measurement and Testing. MA: Allyn & Bacon.

Thongloon Vilaythong[1], Sune Pettersson[2] & Oleg Popov[3]
[1] Department of Physics, National University of Laos, P.O.BOX: 7322, Dongdok Campus, Vientiane
Lao PDR.
[2] Department of Physics, Umeå University, SE-901 87, Umeå, Sweden.
[3] Department of Mathematics, Science and Technology Education, Umeå University, SE-901 87, Umeå,


In this study we have analysed students’ collaborative activities when they per-
formed an experimental task with a pendulum. Five groups, with three to seven first
year science students in each group, were videotaped. We continuously recorded
each group during their practical work session with an average duration of 55 min-
utes. Some students from each group were interviewed directly after the experi-
mental session. When the video recordings were analysed, students’ activities and
communication patterns were classified into eleven categories. The analysis evi-
dences that the students collaborated actively in the groups helping each other han-
dle the equipment and discussing how to perform the measurements. However in
the largest working group, two students did not touch the measurement equipment
and they did not take part in solving the experimental tasks. Other students in this
and other groups collaborated well, but they revealed poor experimental skills and
not all groups were able to complete the task. The students’ group discussions were
mainly focused on understanding the experimental procedures and collecting data
for the report rather than on the physics content. This revealed their shared priority
of experimental activity: to pass the compulsory experimental part of the course.
Just a few students tried to raise the issues about the goals stated in the laboratory
work sheet concerning understanding of relations between different pendulum pa-
rameters and verification of the value of g (acceleration due to gravity). Evidence
collected from the analysis of the video recordings and the follow-up interviews
with the students showed that difficulty in understanding the laboratory instruction,
faulty equipment, and insufficient help from the laboratory instructor constituted
the main factors that led to a non -successful completion of the experimental task
by some groups.

In this paper a concept of collaborative activity is used in accordance with the definition sug-
gested by Dillenbourg (1999) and Hron (2003) that it is a situation in which two or more people

learn or attempt to learn something together. Learners work together on a topic, exchange their
opinions about a subject matter, clarify the meaning of concepts or aim at a joint problem so-
lution. This definition also coincides with traditional understanding of collaborative activity
in the Lao educational system, which includes individual students taking different roles in the
process of achieving a common purpose. For instance, in the case of laboratory tasks in physics,
students help each other with reading the laboratory instructions, handling measurement equip-
ment, observing phenomena and taking notes.
The educational physics laboratory is a complex activity involving making predictions, manip-
ulating apparatus, making measurements, and talking about physics (Niedderer, 2002). Ac-
cording to AAPT (1997), the goals of the laboratory are to engage students with experimental
processes, help students to develop a broad array of basic skills and tools of experimental and
data analysis, master basic physics concepts, understand the role of direct observation in physics
and to distinguish between inferences based on theory and the outcomes of experiments, and to
develop collaborative learning skills (AAPT Committee). Furthermore, Högström et al (2009)
found that the objectives of the laboratory were to link theory to practice, to learn experimen-
tal skills and to get to know the methods of scientific thinking. Theyßen (2007) indicated that
laboratory work is an integral part of physics education at university level, and with physics as
a subsidiary subject the major objective should be that students learn physics - related content
that is relevant for their future studies and professional life.
Learning physics with experiments is a new form of work for many first year science students
in Laos as their previous experience of practical work is almost non-existent (Vilaythong &
Popov, 2008). Science students coming to the physics course are from the Faculties of Science,
Medicine, Engineering, and Forestry. The physics curriculum for first year science students at
the National University of Laos (NUOL) includes an experimental component which at present
consists of six experimental tasks. One of them is an experimental task with a pendulum. The
simple pendulum is a physical system which is easy to make and to study and it is often used
to teach investigative skills and skills of measurement (Gauld, 2005; Matthews, 2005). In Laos,
the simple pendulum has been taught as a demonstration of vibrational motion.
However, the experimental activities did not seem to encourage physics students as they were
supposed to do. Many students got low scores or did not pass the experimental part of the
course. This situation inspired us to study this topic. This study was focused on analysing the
students’ collaborative activities during the experimental pendulum task such as manipulation of
materials, apparatus, tools, measuring instruments, and communication. The following research
questions were formulated.

1. How do the students collaborate during the laboratory work session?

2. How do the students carry out measurements and handle equipment?

3. What factors affect the outcome of the students’ work?


The experiment groups were formed by the students themselves. The compositions of the work-
ing groups were unequal with respect to the number of group members and the abilities of the
students. Each group consisted of four to seven students working with one set of equipment.
Students come to experiment class every other week. The duration of an experimental ses-
sion is scheduled to be 90 minutes. The laboratory instructions were written in detail in the
Lao language: presenting the exact procedure of how to make the necessary measurements, the
technique of using the equipment, and the method of verifying the value of g (acceleration due
to gravity). Normally, the same physics lecturer conducts lectures, problem solving exercises,
and experimental sessions for the first year science students during a semester. However, be-
cause of planning problems, it might happen that students do experiments before lectures. For
instance, students in this study carried out the pendulum experiment during the second or third
week of the course, while the corresponding theory was lectured in the sixth or seventh week.
The purpose of the pendulum experiment was that students should learn to master fundamental
skills of measurement and be able to calculate physical quantities from measured data. Courses
of action in this task included deciding the proper numbers of cycles the pendulum should swing
for the measurement, finding the dependence of the period on angles, length of string, and mass.
Finally, students had to verify the value of g.


The participants were 25 (11 females and 14 males) first year science students at NUOL enrolled
in the physics course with compulsory laboratory work. The pendulum experiment was carried
out by students in groups of three to seven, and five of these groups were studied. These groups
were marked as G1, G2, G3, G4, and G5, and each group member by A, B, C, . . ., respectively.
Groups G1 and G2 consisted of three students each, G1 with all males and G2 all females.
These two groups were formed for comparison between females and males groups. They were
also formed for comparison between small and large groups. Groups G3 and G4 consisted of
six students each with equal number of females and males. Group G5 had seven students, two
females and five males. At the beginning of each laboratory session, students that were sched-
uled to work with the pendulum were asked their permission for videotape recording. During
the experimental sessions the video camera, placed on a tripod, was moved to different posi-
tions to ensure that most of the student’s collaborative activities were observed. The videotape
recording was begun when students started experimental activities. The videotape recordings
were continuous with an average duration of 55 minutes for each group. One or two students in
each group were interviewed directly after the experiment session. Nine students in total were
interviewed and each interview lasted about five minutes. The interview questions were about
the objective of the experimental task, collaborative activities during the experiment session,
laboratory instruction procedure, and the organisation of the groups.


The development of the categories for data analysis was based on the CBAV (category-based
analysis of videotape) model, which was developed in the EU project “Labwork in Science”
and described in detail by Niedderer (2002) and Scharfenberg (2007). The categories were
also based on the students’ actual collaborative activities during the experiment session. These
categories are shown in table 1.

Categories Description
Topics for activities
Interaction with third person. 3P Asking/listening to the lab instructors.
Lab instruction. LI Reading/activities related to lab instruction.
Manipulation of apparatus. MA Handling apparatus and devices. Reset equip-
Paper note. PN Taking note/drawing table.
Measurement. ME Using measurement equipment to collect data.
Observation. OB Look at what other students do or look at exper-
iment occurrence.
Topics for discussion
Variables discussion. VD Using words referring to pendulum parameters
and process of data collection.
Steps of data collection discussion. SD Talk about the handling of apparatus.
Results discussion. RD Discussing the results obtained.
Physics formula discussion. PD Describing relationship between variables and
Other O Activities not at all related to the experiment.

Table 1: Categories for description of the students’ activities during the pendulum experiment.

The categories such as interaction with a third person (3P), laboratory instruction (LI), manip-
ulation of apparatus (MA), paper note (PN), measurement (ME), and observation (OB) were
used for analyse of the students’ activities. The categories such as variables discussion (VD),
steps of data collection discussion (SD), results discussion (RD), and physics formula discus-
sion (PD) were used for analyse of the students’ verbalisation. The category of other (O) was
used for the students’ activities that were not at all related to the experiment. Each recording
was watched several times. In a first step the whole videotape was watched to get a general view
of the students’ activities. In the next step each student was assigned a symbol for easy marking
of their activities. After that the first five minutes of the recording was played several times
until the activities of all individual students had been marked. This process was then repeated
for each 5 minute section of the recording. At the same time errors with handling equipment
and faulty reasoning by each individual were also noted.

In presenting the result of our analysis of the videotape recordings, we use a simplified aggre-
gated category scheme. We added all categories under the topic activities (3P, LI, MA, PN, ME,
OB) into a broader category Activities (Act) which includes everything the students do to solve
the experimental task except for their talk. All types of talk that is related to the experimental
task are added in the new aggregated category Discussion (Disc) which includes four categories
from table 1 (VD, SD, RD, PD). Furthermore, adding Activities and Discussion (Act & Disc)
we obtain a category that includes everything that can be said to be a participation in solving
the laboratory task. We also formed a special category, MA & ME that are a measure of the
students’ actual handling of the experimental equipment. Students’ activities during the exper-
imental session developed continuously, but for the sake of analysis we quantified the activities
in five minutes intervals thus extracting the main content of activities during this period of time.
The results are presented for each of the groups in the figures below.


3.1.1 G ROUP G1

All the members of this group collaborated actively during the experiment session (figure 1).
At the beginning, students discussed the number of cycles the pendulum should swing, values
of angles, length of the string and setting up of the equipment. Usually two students took
part in handling the equipment and doing measurements; one student was observing and taking
notes. After thirty minutes they were all involved in changing the length of the string, and they
continued the measurements under this new condition. During the experiment session, students
often asked for help from the laboratory instructor. However, students experienced difficulties
with the release of the pendulum and obtaining good accuracy when pressing the stop-start
button of the stop-watch. Sometimes, the string of the pendulum touched the protractor but the
students did not know how to deal with that problem.

3.1.2 G ROUP G2

The collaborative activities of the students in group G2 during the experimental session are
shown in figure 2. Generally, students have good solidarity in group work. At the beginning of
the experiment session, students were excited to use the equipment. They started manipulation
with the pendulum immediately, without discussing the laboratory procedure and how to use
the equipment, so they were releasing the pendulum in an incorrect way. They then turned to
discuss the use of the equipment. Their discussions were mainly focused on the technique of
releasing the pendulum, the setting of angles, the method of reading the stop-watch scale, and
reading the laboratory instructions. The students seemed hardly to understand the procedure
despite spending quite a lot of time reading laboratory instructions. When one student read
the laboratory instructions, she asked a student in another group about procedure for doing the
experiment, and sometimes she briefly touched the equipment. Only one student attempted to
handle the equipment to carry-out measurements but she did not succeed. She performed the
following actions: one hand catches a string close to the protractor and other hand holds up

Figure 1: The number of students in group G1 taking part in different kind of activities and
discussion during an experiment session. Group G1 consists of three male students.

the ball of the pendulum and releases it with force. Other students just observed the pendu-
lum movement and did measurements, and sometimes took notes. Generally, students in this
group showed very poor skill at handling measurement equipment. They did not know how to
adjust the angle on the protractor. Students also misunderstood reading the scale of the stop-
watch. The experimental task was not completed at the end of experimental session and no
measurements were done, because they did not know how to continue.

3.1.3 G ROUP G3

This is the first of the large groups consisting of six students, three males and three females,
working with one experimental set. There were only a few students that had opportunity to
handle the equipment. The other students just counted the number of cycles the pendulum
swang, observed the pendulum moving, read laboratory instructions, and took notes. Accord-
ing to results shown in figure 3, at the start, few students discussed the procedure for doing the
experiment and tried to handle the equipment, while the other students read laboratory instruc-
tions, practiced the use of the stop-watch, and observed equipment. After five minutes, students
in this group collaborated well. However, all students seem to be problematic with following
laboratory instructions and the use of measurement equipment. Students performed very poorly
with handling measurement equipment: they adjusted the angle of the pendulum by turning
the protractor upward. Students were confused when taking numerical readings on the scale of
the stop-watch. For about ten minutes, from the fifteenth to twenty-fifth minutes, student did
not do any measurements; they just looked at each other and then smiled. After that, only two
students attempted to use the experimental equipment to collect data. At the end of experiment
session, students did not have any activities, although this group had not succeeded with the
experimental task.

Figure 2: The number of students in group G2 that take part in different kind of activities and
discussion during an experiment session. Group G2 consists of three female students.

Figure 3: Variation with time of the number of students in group G3 that take part in differ-
ent kind of activities and discussion during an experiment session. Group G3 consists of six

Figure 4: Variation with time of the number of students in group G4 that take part in differ-
ent kind of activities and discussion during an experiment session. Group G4 consists of six

3.1.4 G ROUP G4

This is the second of the large groups, in this study. As shown in figure 4, in the beginning
students in this group worked well with collaborative activities. Students discussed the use of
the measurement equipment. The most common discussion was about how to handle equipment
to carry out measurements, such as the length of string and angles to be used, numbers of cycles
the pendulum should swing, and the technique of releasing the pendulum. Some of them tried to
release the pendulum to collect data, while other students read the laboratory instructions. Due
to lack of experience in handling equipment, they hardly understood the procedure for the ex-
periment described in the laboratory instructions; therefore students revealed poor experimental
skills in handling equipment and carrying out measurements. After ten minutes, the students fo-
cused on reading the laboratory instructions and observing the equipment. Only a few students
tried to use the equipment to carry out measurements, but they still did not succeed. Students
sometime had differences in understanding the experimental task. One student wanted to follow
the exact procedure as it was described in the laboratory instruction; and relate the next mea-
surement to previously obtained data and also repeat each measurement once or twice. While
another student wanted to finish the experimental task quickly. After about thirty-five minutes,
although the experimental task was not yet completed, students did not continue the measure-
ments. They talked with the laboratory instructor and offered to repeat this experimental task
again on some other occasion.

3.1.5 G ROUP G5

According to figure 5, all students took part in doing the experiment. However, most students
seldom took part in handling equipment to carry out measurements. Many students showed little

Figure 5: Variation with time of the number of students in group G5 that take part in different
kind of activities and discussion during an experiment session. Group G5 consists of seven

self-confidence in handling equipment. Some of them just took part in adjusting the angle of
pendulum, counting the number of cycles of the pendulum, reading laboratory instructions, or
taking notes. Although all students in the group had equal opportunities to handle the equipment
to obtain data, two students did not touch the equipment at all. They took very little part in doing
the experiment according to the videotape recording. However, this was the only group that
showed good skills in handling the equipment to obtain data. Usually two or three students, who
seemed to have the best understanding in the group of the process of carrying out measurements,
used the equipment to obtain data and they also formulated the conclusion that the period of the
pendulum depends on the length of the string; it does not depend on the numbers of cycles
of the pendulum and the angles. At the end of the experiment session students did not do
any measurement. They planned to determine if the period would vary with the mass of the
pendulum but it just led to a discussion.


In the interviews, students were asked to describe how they understood the purpose of the
experimental task with the pendulum. Almost all interviewees said that the purpose of the
experimental task was to determine relations between periods and parameters of the pendulum
(numbers of cycles of the pendulum, angles, mass, and length of string), and then calculate the
value of acceleration due to gravity. These purposes were defined identically to the way it was
written in the laboratory instructions. All interviewees complained that the main factors that
led to a non-successful completion of the experimental task were that they lacked experience of
using measurement equipment, the difficulty in understanding the laboratory instructions, faulty
equipment, and too little help from the laboratory instructor. In our observation, students had
problems with faulty equipment where the stop-watch did not work regularly; students usually

used weights of the balance scale instead of the mass of the pendulum, and used string with
many knots. Concerning the organisation of the groups, students commented that working in a
small group gave them more opportunity to handle equipment. A few students mentioned that
when working in a large group they can help and share experiences with each other.

Focusing on research questions, this study found that generally students had intensive collabora-
tion during experimental sessions. In quantitative term students’ cooperation was active, but in
qualitative terms it was of rather poor content, focusing mainly on technical issues of measure-
ment procedures. Many students took an active part in handling equipment, but showed poor
abilities and skills in obtaining data. These results correspond to the results of Theyßen (2007)
who found that most students have very little experience in handling experimental equipment.
Students could not read the stop-watch scale, and had difficulty in following the laboratory
instruction procedure.
Other factors affecting the outcome of the students’ work, according to interviews and our
observations, were difficulties in understanding the laboratory instructions, group sizes (6-7
students in group), and insufficient help from the laboratory instructor. Different problems
were observed in students’ experimental technique. A general tendency was that students spent
time on doing measurements without proper organisation of the experiment and handling of the
Some students spent a lot of time on activities not related to the laboratory. At the beginning of
the experimental session, the students spent about 15 to 20 minutes preparing their written report
for the previous experimental session by copying from each other. For about ten minutes at the
end of experiment session, they leisurely conversed about other courses and issues of student
life. From the main author’s personal experience, Lao students take less initiative in talking and
expressing their ability during studies and laboratory activities compared with students at the
same levels in neighbouring countries. This “modest behaviour” is typical in traditional Lao
culture, particularly for women. This demands special attention to gender differences during
laboratory activities.
Comparing the two small groups with the three larger groups we find that all students were
active during the whole session in the small groups, whereas several students in the larger groups
did not make any contribution for large parts of the session. This is specially pronounced at the
end of the session in which 2-3 students in each group did not take part in the group work. We
conclude that it is difficult to engage all students in laboratory work when the group size is six
or more students.
The only group in this study that formulated any conclusions about the pendulum experiment
was the largest group, G5. This seems to confirm the opinion of some students in the interviews
that in large groups you can help each other. However, a closer examination of our observations
reveals that it is a subgroup of students in G5 that do all the measurements and data collection
as well as formulating the physical conclusions at the end. We see no sign that the three absent
students in the end ever noted the correct physical discussion in the rest of the group. We suspect

that these three students will just quickly copy the data and conclusions from the others without
taking time to understand what has been done.
Approximately half of the time students spent doing hands-on activities. Mainly male students
showed high engagement in doing the experiment but both female and male students had meagre
skills in handling equipment. Nearly half of the students had problems in using a stop-watch
and protractor. Most of the activities were accompanied by discussions among the students in
the workgroups. The students’ group discussions were mainly focused on understanding the
experimental procedures and collecting data for the report rather than on the physics content.
This study suggests three changes to improve the laboratory sessions. First, students need
special teaching on the use of basic measurement equipment. Second, the procedure in the
laboratory instructions should be revised so that it will be easier to follow. Third, the labo-
ratory instructors should encourage female students to take an active part in the experimental

This study was supported by the Swedish International Development Agency (Sida).

AAPT (American Association of Physics Teachers). (1997). The Goals of the Introductory Physics
Laboratory ( Visited September 9, 2009.
Dillenbourg, P. (1999). Introduction: What do you mean by “collaborative learning”? In P. Dillenbourg
(Ed.), Collaborative Learning: Cognitive and computational approaches (pp. 1-19) Amsterdam: Perg-
amon, Elsevier Science. Visited Septem-
ber 9, 2009.
Gauld, C. (2004). Pendulums in the Physics Education Literature: A Bibliography, Science & Education,
13, 505-526.
Hron, A. & Friedrich, H.F. (2003). A review of web-based collaborative learning: factors beyond tech-
nology. Journal of Computer Assisted Learning 19, 70-79.
Högström, P., Ottander, C., & Benckert, S. Implementation of Objective for Laboratory Work in Sec-
ondary School Science. Paper present at the 2008 NARST Annual International Conference, Impact of
Science Education Research on Public Policy, March 30-April 2. Baltimore Marriott Waterfront Hotel,
Baltimore, MD. Visited September
9, 2009.
Matthews, M.R., Gauld, C. & Stinner, A. (2004). Introduction. Science & Education, 13, 609-611.
Matthews, M.R. (2004). Idealisation and Galileo’s Pendulum Discoveries: Historical, Philosophical and
Pedagogical Considerations. Science & Education, 13, 689-715.
Niedderer, H., Tiberghien, A., Buty, C., Haller, K., Hucke, L., Sander, F. & Fischer H. (2002). Talking
Physics in Labwork Contexts - A Category Based Analysis of Videotapes. In: D. Psillos & H. Niedderer
(Eds.): Teaching and Learning in the Science Laboratory. (Dordrecht: Kluwer), 31-39.

Peter, D.R. (2008), Teacher perception of different modes of conducting physics experiment. SEAMEO
(Regional Centre for Education in Science & Mathematics) Conference paper number 11. Visited September 9, 2009.
Scharfenberg, F.J., Bogner, F.X. & Klautke, S. (2007). A Category-based Video Analysis of Students’
Activities in an Out-of-school Hand-on Gene Technology Lesson. International Journal of Science Ed-
ucation, first article, 1-17.
Theyßen, H. (2007). Towards targeted labwork in physics as a subsidiary subject: enhancing the learning
efficiency by new didactical concepts and media. European Journal of Physics, 28, S105-S114.
Vilaythong, T. & Popov, O. (2008). The situation with practical work in physics education in Laos In the
Proceedings of the XIII. IOSTE Symposium on The Use of Science and Technology Education for Peace
and Sustainable Development, pp.559-566.


Dean Zollman[1] & Johannes vd Wirjawan[2]

[1] Kansas State University, Manhattan, KS, USA.
[2] Widya Mandala Catholic University, Surabaya, Indonesia.

When U.S. President James Garfield was shot in 1881, physicians were unable
to determine if the bullet had entered a vital organ. This knowledge was needed
quickly to decide the medical treatment required to save Garfield’s life. Alexander
Graham Bell believed that he could use his recent invention, the telephone, with
another relatively new development, the induction balance, to detect the location
of the bullet. Over a decade before the discovery of x-rays, this attempt at non-
intrusive medical imaging failed. However, the apparatus provides students with
a way to learn several aspects of electromagnetism and AC circuits in a context
that should be motivating to medical students who are studying physics. Even the
reasons for the failure are directly related to understanding magnetic fields. We are
in the process of developing such a lesson.

This paper connects well with a plenary lecture at this conference on the state of the art and the
use of magnetism for non-invasive bodies (Swithenby, 2009). We discuss the first attempt to
try to use magnetism for non-invasive medical imaging and our progress on preparing teaching-
learning materials about this rather unique (and somewhat forgotten) incident in the application
of physics to medicine.
This effort is part of a larger project which we have called Modern Miracle Medical Machines.
It is a series of lessons that are context-based and aimed at students who will want to study either
human or veterinary medicine. We have been creating relatively short activities that could fit
into a standard physics course for these students. We are not trying to build a curriculum at this
point but trying to create some instructional materials that emphasize the relevance of physics
that they are learning to their careers.
We have developed or are developing lessons on several topics. The most complete set of
lessons focuses on vision defects and using wavefront aberrometry to diagnose those defects.
The positron emission tomography, which emphasizes tomography much more than the positron,
is essentially done. The lesson on Magnetic Resonance Imaging (MRI) will be completed soon,
while one on Computer Aided Tomography (CT Scans) still needs some work. All of them will
be available on our website,, relatively soon.

Now to the story, James Garfield was President of the United States in the 1880s. Much different
from today, when the president wanted to travel at that time he took the train. He was standing
apparently by himself on a train platform in Buffalo, New York, when a person who was upset
because he had not been appointed as an ambassador, approached the President and shot him.
The bullet lodged fairly deeply in Garfield who was a relatively large man. The doctors wanted
to know how close it was to vital organs so they would know whether they could remove it with
the available surgical techniques. At that time no non-intrusive ways to determine the location
were available to the physicians. The common method of locating a bullet in a body at that time
was for the physician to stick his finger into the hole created as the bullet entered. (They were
essentially all men in those days.) And, of course, they did not have surgical gloves. Nor did
they did understand the need for sterilization. This invasive probing led to infection and to an
end to the story only 80 days later.

While the physicians were using their rather crude methods, Alexander Graham Bell, inventor
of the telephone, had a different idea about locating the bullet. We think that this early attempt
at medical imaging is a rather interesting attempt to apply physics to medicine. And it connects
physics to history and social relevance. Thus, our long-term goal is to have students be able to
replicate what Bell attempted, understand the underlying physics and understand why it did not
We originally learned about this event in an opening of a book on the history of medical imaging,
but the physics was not described very well there (Kelves, 1997). See also (Kuhfeld, 1991).
Fortunately for this purpose, Google has digitized the American Journal of Science magazine
back to the beginning. As is shown in Figure 1, Alexander Graham Bell presented this paper at
a meeting of the American Association for the Advancement of Science in Montreal in 1882,
and it later appeared in several printed formats (Bell, 1882a, 1882b, 1882c, 1883). So, the paper
was presented and printed very shortly after Garfield had died. (All of the papers are available
from Google Books. The URLs are very long and not included here. They will be available on
our web site.)
In this paper he described his attempt to try to find the bullet in Garfield. His basic scheme was
to construct some inductance coils which he moved over Garfield’s body. These coils would be
connected to some oscillating source which would generate an audio frequency and would be
connected to Bell’s relatively new invention, the telephone. If a coil came close to the metal
bullet, the inductance would change and, thus, the frequency which was driving the telephone
would also change. Then, Bell would have devised a great new application for the telephone.

The first difficulty is that the inductance change will be extremely small. Because this change is
small, any change in frequency will be difficult to hear. But Bell had a solution to this problem:
use a bridge circuit. Then, he starts with a balanced bridge and no sound coming through the
telephone. When the induction changes in one part of the bridge, sound occurs in the telephone.

Figure 1: A reproduction of the first page of Bell’s paper which appeared in the American
Journal of Science in 1882.

It is very easy (or much easier) to hear a change from no sound to a little sound than it is to hear
a slight change in sound.
This idea was not totally new. Bell had already started using similar circuits because of what
he called disturbing noises in telephones. Telephone wires and telegraph wires were frequently
run side by side. The currents in the telegraph wires were causing “disturbing noises” in the
telephone. So, he had set up some bridge circuits to decrease the noise. The basic system
had been invented by D. E. Hughes a few years earlier (Hughes, 1879). The Hughes’ paper
states, “ . . . two separate induction-coils, each having its primary and secondary coils, were
joined together in such a manner that the induced current in one coil was made to neutralize the
induced current in the opposite coil, thus forming an induction balance.”
The general apparatus is shown in Figure 2 which is also Figure 2 from Bell’s paper. It contains
two sets of induction coils and a telephone earpiece. We are still trying to understand the details.
However, it seems that Bell obtained audio frequency alternating current using the microphone

and the clock. Each time the clock ticks the microphone closes a bit and chopped current goes
through the whole circuit.

Figure 2: Bell’s initial design for an apparatus to find the location of the bullet.

Figure 3 shows a schematic of an improved version. Bell’s caption which is in the middle of
this figure says that this was the arrangement that he used with Garfield on July 26, 1881. This
version has two sets of coils and a small chopper that creates the oscillating current. Some
capacitors which are not well explained are apparently needed to do some filtering.

Figure 3: An improved version which was used on President Garfield on July 26, 1881.

The statement in the paper is that for some unknown cause he really could not get it to work.
So he went back to the drawing board and refined his apparatus again. As Figure 4 shows he
now used sliding coils. They could slide back and forth in order to get his bridge to balance. He

tried again on Garfield. This time, all of the circuitry was placed in a neighboring room so that
the physical noise of the electromagnet bouncing back and forth would not be heard. A drawing
(Figure 5) shows his assistant moving the coils around while Bell is listening.

Figure 4: Another refinement which was used on August 1, 1881.


The failure is described in Bell’s words from his American Journal of Science paper. He noted
an “area of feeble sound [that] was due to some external cause. . . . I was by no means satis-
fied however, with the results obtained, for no such effects had been observed before in our
experiments with bullets.”
These results were not what Bell was expecting. Because the results did not match his expecta-
tions, he returned to a previous experiment, locating a bullet that was buried in a piece of meat,
and found that he could detect the location. Even with the first apparatus, before he tried it on
Garfield, he had tried it with bullets buried in sand, shooting bullets into carcasses of meat and
even with veterans of the Civil War who had shrapnel. His system worked with all of them, but
it did not work with Garfield.
Now by this time Bell was rather perplexed. And so he went on August 2 to make sure that no
extraneous metal was lying around. Of course, when he ran his experiment to try to find the
bullet, he did indeed ask that all the metal be removed. The surgeons had told him that they
were sure that no metal was in the room. However, another new invention was present: the
innerspring mattress. Garfield being President had the latest and greatest mattress. He had a
mattress underneath him that was made out of horse hair but underneath the horse hair was a set

Figure 5: A drawing of the final attempt to detect the bullet. Note that the part of the apparatus
which makes noise is in an adjoining room. This drawing appeared in Frank Leslie’s illustrated
newspaper, v. 52, no. 1351 (1881 August 20), pp. 412-413. Original caption: “The attempted
assassination of the president - the discovery of the location of the bullet by means of Professor
Bell’s induction-balance / from a sketch by William A. Skinkle” (Downloaded from the U.S.
Library of Congress.)

of springs, just as most American mattresses have these days. However, then such a mattress
was a new idea; Bell did not know about them.
Thus, under the President was an evenly distributed layer of metal completely destroying Bell’s
attempt at medical imaging. At about the time of this discovery Bell had to leave Washington.
He left an assistant in charge to try to figure out how they were going to deal with this issue.
Moving Garfield apparently was not an option at that point. He had been lying there for quite
awhile. He was a very ill and relatively large person. He died before Bell and his associates
were able to figure out what to do.


We would like to create a set of experiments to have students conduct measurements similar to
the ones Bell completed. We recognize that our students will have significantly better equipment
than was available to Bell and his assistants. So far we have experimented with both telephone
receivers and oscilloscope detectors. Because the signal can be rather weak, the oscilloscope
has some advantage but, of course, it is not historically accurate. We have also learned that with
modern day detection, we can get some results with just one pair of coils. We do not need the
bridge circuit. But we would like the students to work with the bridge as well. We do plan to
use an audio oscillator for convenience in most experiments. However, it would be good for
students to understand how one could create an oscillating current with a chopping device.
In following our usual procedure we will conduct some research on how students use their
existing knowledge to understand the physics underlying Bell’s efforts. Then we will create the

lessons and test them to be sure that they are meeting our goals.

Alexander Graham Bell’s attempt to locate a bullet in President James Garfield was an in-
teresting application of the basic properties of electromagnetism. Almost 20 years before the
discovery of X-rays, he applied physics in an attempt to conduct non-intrusive medical imaging.
He was foiled only because of another recent advance in technology, the innerspring mattress.
The basic physics of Bell’s concept is accessible to today’s undergraduates. Building a set of
lessons to help students understand this event, why it might succeed, and why it did not in this
case, can help students learn the physics of electromagnetism. In addition the lessons can show
connections between basic science and applications to other societal issues.

Modern Miracle Medical Machines was supported by the National Science Foundation Grant
04-27645. Dr. Wirjawan’s visit to KSU was supported by his university.

Bell, A.G. (1882a). Upon the Electrical Experiments to Determine the Location of the Bullet in the
Body of the Late President Garfield and Upon the Successful Form of Induction Balance for the Painless
Detection of Metallic Masses in the Human Body. Washington, DC: Gibson Brothers, Printers.
Bell, A.G. (1882b). Upon the Electrical Experiments to Determine the Location of the Bullet in the
Body of the Late President Garfield and Upon the Successful Form of Induction Balance for the Painless
Detection of Metallic Masses in the Human Body. Paper presented at the American Association for the
Advancement of Science, Montreal.
Bell, A.G. (1882c). Upon the Electrical Experiments to Determine the Location of the Bullet in the
Body of the Late President Garfield and Upon the Successful Form of Induction Balance for the Painless
Detection of Metallic Masses in the Human Body. The Telegraphic Review and Electrical Review, 11,
Bell, A.G. (1883). Upon the Electrical Experiments to Determine the Location of the Bullet in the
Body of the Late President Garfield and Upon the Successful Form of Induction Balance for the Painless
Detection of Metallic Masses in the Human Body. American Journal of Science, 25(145-150), 22-61.
Hughes, D.E. (1879). Induction-balance and Experimental Results Therewith. Proceedings of the Phys-
ical Society, 3, 81-89.
Kelves, B.H. (1997). Naked to the Bone: Medical Imaging in the Twentieth Century. New Brunswick,
NJ: Rutgers University Press.
Kuhfeld, A.W. (1991). For Whom Bell Toils: Medial Imaging by Telephone. IEEE Engineering in
Medicine and Biology, 88-89.
Swithenby, S. (2009). Brain imaging and learning physics. Paper presented at the GIREP-EPEC 2009
International Conference: Physics Community and Cooperation.


Dorothy Langley[1,2,3] & Rami Arieli[1]

[1] Department of Science Teac