You are on page 1of 10

International Journal of Mathematical Education in

Science and Technology

ISSN: 0020-739X (Print) 1464-5211 (Online) Journal homepage: https://www.tandfonline.com/loi/tmes20

Estimative abilities in probabilistic judgments of


engineering students

María Inés Cavallaro , Marta Anaya , Elsa García argiz & Patricia Aurucis

To cite this article: María Inés Cavallaro , Marta Anaya , Elsa García argiz & Patricia Aurucis
(2007) Estimative abilities in probabilistic judgments of engineering students, International
Journal of Mathematical Education in Science and Technology, 38:7, 917-925, DOI:
10.1080/00207390701579464

To link to this article: https://doi.org/10.1080/00207390701579464

Published online: 26 Sep 2007.

Submit your article to this journal

Article views: 38

View related articles

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=tmes20
International Journal of Mathematical Education in
Science and Technology, Vol. 38, No. 7, 15 October 2007, 917–925

Estimative abilities in probabilistic judgments


of engineering students

MARÍA INÉS CAVALLARO*yz, MARTA ANAYAyz,


ELSA GARCÍA ARGIZyz and PATRICIA AURUCISz
yFacultad de Ingenierı́a, Universidad de Buenos Aires, Argentina
zUniversidad Tecnológica Nacional, Argentina

(Received 9 May 2007)

The paper discusses the interaction between intuitive biases of probabilistic


thinking and mathematical knowledge. It would appear that students may answer
numerical problems correctly but falter on simple descriptive solutions. Students
appear to relinquish formal knowledge for simpler heuristics when attempting to
describe the outcome of an event. This paper explores some possible theoretical
causes of the dissociation between procedural and estimative abilities.

1. Introduction

The following recommendations were highlighted as important qualities


for Argentine engineers at the XXXVII Plenary Meeting of CONFEDI
(Consejo Federal de Decanos de Ingenierı́a, 2005):
. Capability to manage situations under uncertainty
. Aptitudes for the resolution of non-traditional problems with reference to
uncertainty, creativity, and personal initiative
. Capability of abstract thinking and critical reflection.
The aim of most basic first year courses, including mathematics, is to deliver the
theory behind specific career-orientated topics while developing abilities and
attitudes that allow future professionals to correctly assess situations and take
appropriate action. Engineers should exhibit the ability to be able to make correct
estimation of the probability of events both descriptively and numerically. Decisions
or judgments concerning uncertainty are influenced by intuitive heuristics of the
probabilistic thinking, which appear at times to deflect the student away from correct
mathematical thinking [1–3]. In a pedagogical sense, two important issues are raised
concerning the instruction of basic mathematics taught at university: first, whether
students adequately achieve the aforementioned recommendations; and second, what

*Corresponding author. Email: micavall@fi.uba.ar

International Journal of Mathematical Education in Science and Technology


ISSN 0020–739X print/ISSN 1464–5211 online ß 2007 Taylor & Francis
http://www.tandf.co.uk/journals
DOI: 10.1080/00207390701579464
918 M. I. Cavallaro et al.

help can be given to students in order to override the effect of intuitive biases of
thinking.
Several studies have been developed within a project about ‘Cognitive Problems
in Mathematics for Engineers’, especially in the area of probabilistic thinking [4–7].
The research objective was to investigate how the formal knowledge conveyed in
university courses on probability interacts with those heuristics in advanced
mathematical thinking.
More specifically, the research studies focused on:
I. The evolution of the biases of intuition through age and traditional
instruction
and
II. The interaction between procedural and estimative abilities.
Several typical heuristics associated with characteristic biases and errors were
investigated within the mentioned project. This work will present the results of two
of these biases produced by:
The fallacy of conjunction: The probability of conjunctive events is lower than the
probability of each elementary event. However, there is a tendency to overestimate
the probability of conjunctive events in comparison with the simple event.
The incidence of the causal schema: People have a tendency to follow a
normal sequence and they reason from causes to consequences. Therefore, they
‘. . . usually infer effects from causes with greater confidence than causes from
effects, even when the effect and the cause actually provide the same amount of
information about each other.’ [2, p. 118]. The interaction and differences between
procedural and estimative abilities will be discussed with reference to mathema-
tical knowledge, different components and approaches, and different modes of
thinking.

2. Mathematical knowledge

It is important to appreciate the different types of knowledge required to solve


questions where a numerical result is expected and where estimation is requested.
For numerical questions, students should evoke the appropriate procedures that
lead to a solution through strategies that are shaped by conceptual restrictions.
It is necessary that students have meaningful understanding of procedures as well as
relationships between procedures. This knowledge will ultimately bestow a
solution that is at the same level, with respect to the abstract quality, as the
information in the question. These relationships are constructed at a primary
level [8]. The main errors with these questions relate to wrong selection of formulae
or conceptual errors.
The estimative questions proposed in this study may be solved either in
a procedural manner, calculating and deciding the correct estimation, or in a more
general way as there is no available numerical data. For the latter case, the cognitive
ability required, depending on the question, with regard to mathematical knowledge
and valid reasoning, may often be higher than with procedural questions. Solving
and justifying them require building a higher level of abstract thinking than the units
Estimative abilities in probabilistic judgments of engineering students 919

of information that are being connected. These relationships are established at a


reflective level.

3. Different approaches

When students approach a problem mathematically, we find the three basic


components of mathematics as a human activity, defined by Fischbein [9]: the
algorithmic, the formal and the intuitive components. Problems where numerical
results are required have mainly an algorithmic aspect and are usual tasks proposed
by traditional teaching. However, estimative questions may raise different
approaches, according to the information provided in the question.
The algorithmic approach is where students solve numerically, following a set
rules or patterns. If there is no numerical data, students may imagine a numerical
example to explore the situation. The formal approach is where students reason
using properties and concepts in a more general and abstract manner. The intuitive
approach is where students respond according to initial thoughts, impressions that
come quickly to mind with a great degree of certainty. These components are
inseparable as an intuitive background could manipulate formal interpretation of the
use of procedures [9].

4. Different modes of thinking

The thinking processes that are provoked when facing different mathematical
problems may be analysed from the perspective of Dual Process Theory, which
proposes the existence of two modes of thinking, System 1 and System 2 ([3]).
The activities of System 1 have been characterized as being fast, highly accessible,
automatic, effortless, and unconscious. These features describe what mathematical
education literature considers intuitive reactions.
The activities of System 2 are slower, serial, effortful, and deliberately controlled.
They are also flexible and potentially rule-governed [3].
Questions requiring numerical results appear to involve System 2 but do not
exclude the involvement of System 1. That is, when facing these sorts of questions,
students plan the steps to follow, how to connect some well-known procedures to
obtain the final result. Clearly, the initial activity is a System 2 activity, and the errors
in the resolution of these sorts of problems are mostly related to conceptual
misconceptions. The students do not show any clues that they have an a priori idea
(or impression) about the possible result driving their activity. So, it may appear that
no intuitive activity is induced by these questions. However, some mathematical
activities can be achieved without effort, without interpreting or reflecting about the
results in a rather automatic way. Therefore, familiarity with procedures or
computational units may make students become experts in this skill and thus change
the activity from System 2 to System 1 [10].
The estimative questions seem to produce a very different mental activity and
interaction between systems than the procedural ones. When facing the estimative
questions, an impression [3] seems to appear immediately and intuitively in students,
so System 1 is initially activated. However, these intuitive reactions can be modified
or overridden deliberately by System 2. Although the control exerted by System 2 on
920 M. I. Cavallaro et al.

these impressions is sometimes rather loose, it still regulates and controls, producing
the different sorts of judgments and explanations [3].
Problems, whether procedural or estimative, may require different types of
mathematical knowledge while activating different components and approaches in
the mathematical activity and, significantly, different modes of thinking.

5. Methodolgy

To study the evolution of the biases of intuition through age and traditional
instruction, several groups of students of different ages without previous instruction
in probabilities and other groups of university students after taking a course in the
subject, participated in the research project.
During the first stage of the study, the same questionnaire, with problems
designed to test the incidence of causal schema and the fallacy of conjunction, was
delivered to five groups of secondary school students (150 students, aged between
13 and 18 years), all of them without traditional instruction in probabilities.
The groups constituted whole classes of different grades at the same secondary
school in Buenos Aires.
In the second stage of the study, the incidence of instruction on the biases was
studied by delivering the same questionnaire to a group of university students.
This group (U1) comprised 30 engineering students aged 19 years. All of them had
already taken a course on probabilities at the same school of engineering in
Buenos Aires.
In order to study the interaction between procedural and estimative abilities, a
questionnaire was designed testing procedural and estimative abilities in relation to
the two intuition biases – fallacy of conjunction and incidence of causal schema.
Each estimative question – which required a qualitative response – was matched with
a procedural question – which required a numerical response after calculation.
This questionnaire was delivered to a group of university students. The group
(U2) comprised 50 engineering students aged 19–25 who had taken a course in
probabilities at the same school of engineering in Buenos Aires.

Table 1. Ages and number of students in each group.

Group G1 G2 G3 G4 G5

Ages

13–14 14–15 15–16 16–17 17–18
N 33 33 26 33 25

6. Results

6.1. The biases of intuition with regard to age and traditional instruction
Several studies have detected persistent errors related to the causal schema and the
estimation of conjunctive events [4–7].
In this case, five groups of students aged between 13 and 18 years without
previous instruction in probability were presented with a questionnaire related to
Estimative abilities in probabilistic judgments of engineering students 921

these issues and then a group of university students with previous instruction in
probabilities was added to the study.

The case of the incidence of causal schema Here is a typical question that students
were asked in order to test for incidences of causal schema in a situation of temporal
inversion.

Problem 1 There is a box with two gold and two silver coins. Pedro picks a coin
randomly keeping it in his pocket without looking at it. Then he picks a second coin
which is a gold coin. The probability that the first coin is a gold one is smaller, equal
or bigger than the probability that it results a silver coin. Explain your answer.

In this problem, even with no instruction on conditional probability, a careful


reasoning allows one to reach the correct response of ‘Smaller’. If the second coin
is a gold one, then the first coin must be one of three coins, i.e. two silver coins
and one gold coin, so it is more probable that it had been a silver coin than a
gold coin, thus the probability of being a gold coin is smaller than the probability
of being a silver coin. However, the percentage of students who answered ‘Equal’
was significant, as it is shown in table 2.
This ‘wrong answer’ describes the use of causal schema as it correlates to the
argument that the second coin that was taken could not have had an influence on the
first coin because this one had already been removed from the box. This answer is
consistent with the idea that a ‘second event comes after the first’. The percentages
do not change significantly with age.
To investigate whether instruction had an effect on the ability to answer
correctly, a group of university students was given the same question (U1). These
students had taken a course in probability recently. The usual pattern of formal
instruction had been given to U1 students during instruction:
 
G1 PðG1 \ G2 Þ PðG2 =G1 ÞPðG1 Þ 1
P ¼ ¼ ¼
G2 PðG2 Þ PðG2 =G1 ÞPðG1 Þ þ PðG2 =S1 ÞPðS1 Þ 3
While
 
S1 PðS1 \ G2 Þ PðG2 =S1 ÞPðS1 Þ 2
P ¼ ¼ ¼
G2 PðG2 Þ PðG2 =G1 ÞPðG1 Þ þ PðG2 =S1 ÞPðS1 Þ 3
The percentage of incorrect response ‘Equal’ was 52% for this group. Knowledge
of independent events together with the statement ‘you cannot change the past’
encouraged the incorrect answer of ‘equal probability’ without the need for any
calculation. It would appear that students who had recently studied probability still
failed to avoid the bias of intuitive heuristic. The percentages of incorrect answer

Table 2. Percentages of the incorrect response


‘Equal’ for each group.

Group G1 G2 G3 G4 G5

Equal 50 47 50 44 40
922 M. I. Cavallaro et al.

‘Equal’ are similar in table 2 and the university group. It would appear that after the
emergence of formal reasoning the intuitive heuristics seem to become stable and
resistant to the influence of age and instruction.

The case of the fallacy of conjunction The following problem from the questionnaire
was given in order to investigate the heuristic, fallacy of conjunction, in relation to
age and instruction.

Problem 2 Martin is a fan of sci-fi movies. He found that his two favourite films
A and B were available in the Video Club. The probability that he takes home one
particular film, say A is smaller, equal or bigger than the probability that he takes
home both films?
Correct response: Bigger

In this case, a group of university students after taking a course in probability


(U1) was added to the study in order to investigate whether the corresponding
instruction could modify the intuitive heuristic. Percentages of judgments for each
group are shown in table 3. The correct response ‘Bigger’ is lower for the first four
grades and increases for students in G5 and U1 (55% and 56% of students
respectively). In relation to the incorrect judgments, in younger students without
instruction, the bias is to the response ‘Equal’. Their judgment is supported with the
erroneous idea of ‘equal probability’ that they assign to events when it was thought
that not enough information is given or when the decision depends on a personal
choice. University students (29%), who had received instructions, answered
’Smaller’, they had overestimated the probability of the conjunction in relation to
one of its constituents. The information that Martin’s two favourite films are
available makes the conjunction seem to be a more ’representative’ event.

6.2. The interaction between procedural and estimative abilities


The persistence of the intuitive biases, with regard to age and instruction, especially
after courses in probability at university, directed our interest to further study the
relationship and eventual interaction between procedural and estimative abilities. Our
claim was that both abilities, estimative and procedural, are somewhat dissociated.
The questionnaire, previously described in Methodology, section 5, was
administrated to 50 students (U2 aged between 19 and 25 years) who were tested
after taking a course in probability.
The case of the fallacy of conjunction To investigate the biases produced by
the fallacy of conjunction, with both procedural and estimative questions, this

Table 3. Percentage of judgments for each group.

Grade G1 G2 G3 G4 G5 U1

Bigger 21 9 8 12 56 55
Smaller 15 16 16 15 4 29
Equal 50 56 50 33 20 12
Estimative abilities in probabilistic judgments of engineering students 923

procedural question (numerical data included) was matched with Problem 2


(Martin’s Problem).

Problem 3 A football team must play two games, one game against team A and
another game against team B The probability that this team wins only the game
against A is 0.36, that it wins only against B is 0.2. And the probability that it loses
both games is 0.15.
a) What is the probability that the team wins both games?
b) What is the probability that the team wins the game against team B?

Table 4 shows that this question was answered correctly by 63% of the students
compared with only 32% answering Martin’s problem correctly as well, showing
dissociation between both types of knowledge. In other words, 31% of students who
could calculate Problem 3 correctly but could not estimate the result in Martin’s
problem. The procedural knowledge was not influential enough to overcome the bias
of the intuitive idea.
The incidence of the causal schema Problems 4 and 5 are examples of typical of
procedural and estimative questions respectively relating to the effect of the causal
schema.

Problem 4 Procedural: Numerical data are provided and numerical response is


required.In a community, there are 40 people who smoke and 60 that do not smoke.
Half of smokers have affected lungs, and a third of non-smokers have affected lungs.
A person of this community is picked at random:
a) What is the probability that this person smokes knowing that his/her lungs are
affected?
b) What is the probability that being a smoker this person has affected lungs?

Problem 5 Estimative: A qualitative response is required.


If the probability that a person has blue eyes is p. The probability that the
daughter has blue eyes if the mother has blue eyes is smaller, equal or bigger than the
probability that the mother has blue eyes if the daughter has blue eyes?

The correct response to Problem 5 is: ‘Equal’. Let M be the event ‘the mother is
blue eyed’, D the event ‘the daughter is blue eyed’

 
M PðD \ MÞ PðD=MÞ  PðMÞ
P ¼ ¼ ;
D PðDÞ PðDÞ

as P(M) ¼ P(D) ¼ p, sults that P(M/D) ¼ P(D/M)


Table 5 shows the percentages of responses to both matched questions. In
Problem 4, both parts of the problem were answered correctly by 40% of the
students.
924 M. I. Cavallaro et al.
Table 4. Percentage of responses to both matched questions,
Problem 2 and 3.

Procedural (P 3)
Estimative (P2) Correct No Correct Total

Bigger (Correct) 32% 23% 55%


Smaller 23% 6% 29%
No answer or other 8% 8% 16%
Total 63% 37% 100%

Table 5. Percentage of responses to both matched questions,


Problem 4 and 5.

Procedural (P4)
Estimative (P5) Correct No Correct Total

Bigger 6% 17% 23%


Equal (Correct) 28% 35% 63%
No answer 6% 8% 14%
Total 40% 60% 100%

In Problem 5, the problem was answered correctly by 63% of the students.


However, a dissociation between procedural and estimative abilities is observed as
only 28% solved both of them correctly.
In this case the information that the mother has blue eyes is perceived as the
cause of her daughter’s blue eyes; the explicitly mentioned probability p was
neglected by 23% of the students, who responded incorrectly ‘Bigger’.
The abilities ‘to calculate’, ‘to estimate’ or ‘make a qualitative evaluation of
a probability’ are rather dissociated. As both abilities are required abilities in the
professional development of an engineer, it is especially interesting to discuss the
possible causes of the perceived dissociation between both types of knowledge.
A noticeable phenomenon observed in our studies is that despite the fact that the
estimative question could be solved using well-known procedures, many students
avoid calculations and follow intuitive reactions. That is to say, independently they
may have the necessary knowledge to correctly solve them, but estimative questions
seem to cause another mode of thinking.

Implications for teaching

The results of the studies carried out within the research project, ‘Cognitive Problems
in Mathematics for Engineers’, support the idea that many of the intuitive biases,
which affect the competencies required to estimate correctly when a qualitative
judgment is required, are not overridden by traditional teaching and training, which
concentrates on mainly procedural problems.
The resolution of traditional probabilistic problems involves operations like
selection of data, counting, and resolution of sequences of procedures, which should
be directed by conceptual abilities. However, as previously mentioned, it is also
important for an engineer’s professional performance to make decisions and to make
Estimative abilities in probabilistic judgments of engineering students 925

correct estimations. These estimations do not always follow the procedural


characteristics that students have met in traditional courses.
According to the results of the study, both abilities are rather dissociated.
Therefore, it would be convenient to improve both of them with specific problems.
The incorporation of qualitative problems in the students’ training, in which
qualitative analysis and estimation is required, will be necessary to improve this
competency.
It would be convenient to present to the students problems involving some of the
well-known heuristics so as to generate the conscience of bias produced by ‘own
intuition’. It would also be convenient that they involve different contexts.
These meta-cognitive activities may help the students to adopt a more careful and
scientific posture before problem solving.
Some of these specific questions should be treated during lessons, and
instructional interventions should be designed in a way that generates cognitive
conflict, with debate among peers, thus permitting the emergence of both correct
conceptions and meta-cognitive abilities.

References

[1] Fischbein, E. and Grossman, A., 1997, Schematas and intuitions in combinatorial
reasoning. Educational Studies in Mathematics, 34, 27–47.
[2] Tversky, A. and Kahneman, D., 1982, Judgment under uncertainty: heuristic and biases.
In: D. Kahneman, P. Slovic and A. Tversky (Eds) Judgment Under Uncertainty: Heuristic
and Biases (Cambridge: Cambridge University Press).
[3] Kahneman, D., 2002, Maps of bounded rationality: a perspective on intuitive judgment
and choice. Nobel Prize Lecture. http://nobelprize.org/economics/laureates/2002/kahne-
mann-lecture.pdf.
[4] Cavallaro, M.I and Garcı́a Argiz, E., 2005, Problemas de inversión temporal y estimación
de eventos conjuntos en probabilidades. Proceedings del III INMAT. Buenos Aires.
[5] Cavallaro, M.I., Garcı́a Argiz, E. and Aurucis, P., 2005, Resolución vs. Estimación.
Sesgos del pensamiento probabilı́stico. XII EMCI. Encuentro Internacional Sobre la
Enseñanza de la Matemática en las Carreras de Ingenierı´a. San Juan.
[6] Cavallaro, M.I. and Garcı́a Argiz, E., 2003, Heurı́sticas intuitivas en el pensamiento
probabilı́stico y conocimiento formal. Proceedings del II Inmat, Buenos Aires.
[7] Garcı́a Argiz, E. and Cavallaro, M.I., 2003, Dificultades en el desarrollo del pensamiento
probabilı́stico de alumnos de ingenierı́a. Proceedings del XI EMCI (Encuentro
Internacional sobre la Enseñanza de la Matemática en Carreras de Ingenierı́a).
San Miguel de Tucumán.
[8] Hiebert, J. and Lefevre, P., 1986, Conceptual and procedural knowledge in mathematics:
an introductory analysis. In: Conceptual and Procedural Knowledge: The Case of
Mathematics (Hillsdale, NJ: Lawrence, Erlbaum Associates), pp. 1–23.
[9] Fischbein, E., 1993, The interaction between the formal, the algorithmic, and the intuitive
components in a mathematical activity. In: R. Biehler, R. Scholz, R. Strasser and
B. Winkelmann (Eds) Didactics of Mathematics as a Scientific Discipline (Dordrecht:
Kluwer Academic Publishers), pp. 231–245.
[10] Leron, U. and Hazzan, O., 2006, The rationality debate: applications of cognitive
psychology to mathematics education. Educational Studies in Mathematics, 62, 105–126.

You might also like