You are on page 1of 13

Contents lists available at ScienceDirect

Computers in Human Behavior


journal homepage: www.elsevier.com/locate/comphumbeh

Full length article

Developing young children's computational thinking with educational


robotics: An interaction effect between gender and scaffolding strategy
Charoula Angelia,∗, Nicos Valanidesb
a
University of Cyprus, Cyprus
b
Educational Robotics and Science Organization, Cyprus

A R T I CLE IN F O ABST RACT

Keywords: The study examined the effects of learning with the Bee-Bot on young boys' and girls' computational thinking
Computational thinking within the context of two scaffolding techniques. The study reports statistically significant learning gains be-
Spatial relations tween the initial and final assessment of children's computational thinking skills. Also, according to the findings,
Scaffolding while both boys and girls benefited from the scaffolding techniques, a statistically significant interaction effect
Pre-primary school children
was detected between gender and scaffolding strategy showing that boys benefited more from the in-
Educational robotics
dividualistic, kinesthetic, spatially-oriented, and manipulative-based activity with the cards, while girls bene-
Bee-Bot
fited more from the collaborative writing activity. In regards to the children's problem-solving strategies during
debugging, the results showed that the majority of them used decomposition as a strategy to deal with the
complexity of the task. These results are important, because they show that children at this very young age are
able to cope with the complexity of a learning task by decomposing it into a number of subtasks that are easier
for them to tackle. The research contributes to the body of knowledge about the teaching of computational
thinking. In addition, the study has practical significance for curriculum developers, instructional leaders, and
classroom teachers, as they can use the results of this study to design curricula and classroom activities with a
focus on the broader set of computational thinking skills, and not only coding.

1. Introduction such as literacy, art, journalism, biology, engineering, mathematics,


science, and many more (National Research Council, 2010; Selby &
Since the publication of Jeannette Wing's seminal article Woollard, 2013). Along this line of reasoning, the Carnegie-Mellon
“Computational thinking” that appeared in Communications of the ACM University Center for Computational Thinking stresses the importance
in 2006, many other educators, scholars and researchers have strongly of computational thinking for all disciplines by stating that it is nearly
argued for the importance of integrating computational thinking as a impossible to do research in any academic field without the ability to
fundamental literacy for the 21st century in the education of students at think computationally (http://www.cs.cmu.edu/∼CompThink/).
all educational levels from pre-primary education to higher education Despite the fact that researchers have strongly argued for the im-
(Barr & Stephenson, 2011; Bundy, 2007; Grover & Pea, 2013; Guzdial, portance of integrating computational thinking in the education of
2008; Lu & Fletcher, 2009; Shute & Asbell-Clarke, 2017). Wing (2011) students starting from early childhood, the investigation of the devel-
defined computational thinking as the thought process involved in opment of young children's computational thinking remains at its in-
formulating problems and their solutions, so that the solutions are re- fancy (Bers, Flannery, Kazakoff, & Sullivan, 2014; Botički, Pivalica, &
presented in a form that can be effectively carried out by an information Seow, 2018). Contemporary research studies that have been undertaken
processing agent, i.e., a human, a computer, or a combination of both, to investigate this research area report that educational robotics con-
using tools and techniques from computer science. stitutes an attractive approach for developing young children's com-
Recognizable computational thinking skills include, among others, putational thinking, because children can directly interact with a robot
abstraction, decomposition, and algorithmic thinking (Selby & and observe the immediate effects of their interactions on the robot's
Woollard, 2013; Wing, 2008, 2011; National Research Council, 2010). behavior (Beraza, Pina, & Demo, 2010; Bers, 2010; Bers et al., 2014;
These computational thinking skills are not skills that only computer Highfield, 2010; Highfield & Mulligan, 2008, 2009; Kazakoff & Bers,
scientists value, but, also skills that can be transferred to any domain, 2012; Stoeckelmayr, Tesar, & Hofmann, 2011). In particular, research


Corresponding author. 11-13 Dramas street P.O. Box 20537 Department of Education University of Cyprus, CY-1678, Nicosia Cyprus.
E-mail address: cangeli@ucy.ac.cy (C. Angeli).

https://doi.org/10.1016/j.chb.2019.03.018
Received 3 January 2019; Received in revised form 4 March 2019; Accepted 17 March 2019
C. Angeli and N. Valanides

by Bers and her colleagues (Bers et al., 2014; Kazakoff & Bers, 2012; undertaken by The Royal Society (2014), and workshops organized by
Sullivan & Bers, 2016), showed that children as young as four years old the Computer Science Teachers Association (CSTA) and the Interna-
were able to engage in computational thinking activities using a ro- tional Society for Technology in Education (ISTE).
botics curriculum. Other researchers, such as Beraza et al. (2010), The 2010 National Research Council's report differentiated com-
Stoeckelmayr et al. (2011), Highfield (2010), and Highfield and putational thinking from computer literacy, computer programming,
Mulligan (2008) used the Bee-Bot, a small, affordable, and program- and computer applications (i.e., games and simulations), and broa-
mable floor robot in the shape of a bee, for promoting pre-primary dened the term to include core concepts from the discipline of computer
children's computational thinking. The limited programming power of science, such as abstraction, decomposition, pattern generalization,
the Bee-Bot makes it an ideal first robotic tool to use with young chil- visualization, problem-solving, and algorithmic thinking.
dren, because young children can quickly learn how to program it, and Similarly, The Royal Society (2014) offered a concise definition of
engage in rich computational thinking activities with it by touching it computational thinking as “the process of recognizing aspects of com-
and directly interacting with it (Highfield & Mulligan, 2008). In addi- putation in the world that surrounds us, and applying tools and tech-
tion, according to Kabátová, Jašková, Lecký, and Laššáková (2012), niques from computer science to understand and reason about both
learning first how to program with the Bee-Βot can facilitate young natural and artificial systems and processes” (p. 29).
children's computational thinking activities with more advanced pro- CSTA and ISTE, in collaboration with leaders from higher educa-
gramming languages in the future. tion, industry, and K-12 education, developed an operational definition
A review of the literature on the integration of the Bee-Bot and other of computational thinking as a problem-solving process that includes,
robotic tools in the education of young children showed that the ex- but is not limited to, the following elements: (a) Formulating problems
isting body of research has mainly focused, with the exception of a few in a way that enables us to use a computer and other tools to help solve
studies, on brief accounts of (a) exploring children's interactions with them; (b) Logically organizing and analyzing data; (c) Representing
the tools within the context of various content domains, such as, math, data through abstractions, such as, models and simulations; (d)
science, literacy, and engineering (Highfield, 2010; Highfield & Automating solutions through algorithmic thinking (i.e., a series of
Mulligan, 2008; Lavigne & Wolsky, 2018), (b) investigating ways of ordered steps); (e) Identifying, analyzing, and implementing possible
how educational robotics can afford learning opportunities to educate solutions with the goal of achieving the most efficient and effective
young children (Highfield & Mulligan, 2008; Kabátová et al., 2012), combination of steps and resources; and (f) Generalizing and transfer-
and (c) familiarizing children with robotics concepts and computer ring this problem-solving process to a wide variety of problems.
programming (Misirli & Komis, 2014; Selby, 2012; Stoeckelmayr et al., Despite the fact that currently there is not one unanimous definition
2011). of computational thinking, after a systematic examination of what is
While these reported efforts are considered important and worth- currently known in the literature, Grover and Pea (2013) concluded
while in terms of taking first steps toward investigating the uses of that researchers have come to accept that computational thinking is a
educational robotics in the development of young children's computa- thought process that utilizes the elements of abstraction, generalization,
tional thinking, they do not (a) provide robust empirical evidence of decomposition, algorithmic thinking, and debugging (detection and
learning gains in computational thinking, (b) describe how teachers correction of errors). Abstraction is the skill of removing characteristics
scaffolded young students' computational thinking, (c) provide detailed or attributes from an object or entity in order to reduce it to a set of
information about how young children's computational thinking skills fundamental characteristics (Wing, 2011). While abstraction reduces
were assessed, and (d) address gender differences in computational complexity by hiding irrelevant detail, generalization reduces com-
thinking between boys and girls. Hence, new research studies toward plexity by replacing multiple entities which perform similar functions
these directions will undoubtedly promote our understanding about with a single construct (Selby, 2012). For example, programming lan-
how computational thinking at this very young age can be taught. guages provide generalization through variables and parameterization.
Accordingly, the study herein sought to experimentally examine the Abstraction and generalization are often used together as abstracts are
effects of learning with the Bee-Bot on young boys' and girls’ compu- generalized through parameterization to provide greater utility. De-
tational thinking skills within the context of two scaffolding techniques composition is the skill of breaking complex problems into simpler ones
and rich problem-solving activities. In the next section, the authors (National Research Council, 2010; Wing, 2008). Algorithmic thinking is
explicitly define and describe the construct of computational thinking a problem-solving skill related to devising a step-by-step solution to a
and its dimensions. The young age of the students and related cognitive problem and differs from coding (i.e., the technical skills required to be
limitations were highly considered in terms of designing the two scaf- able to write code in a programming language) (Selby, 2012). Ad-
folding techniques and related materials. ditionally, algorithmic notions of sequencing (i.e., planning an algo-
The research contributes to the body of knowledge that can be used rithm, which involves putting actions in the correct sequence), and
to inform the teaching of computational thinking skills, and responds algorithmic notions of flow of control (i.e., the order in which in-
directly to calls for more research into how to teach computational dividual instructions or steps in an algorithm are evaluated) are also
thinking (Grover & Pea, 2013, 2018; Guzdial, 2008; Lye & Koh, 2014). considered important elements of computational thinking (Lu &
Additionally, the study has practical significance for classroom tea- Fletcher, 2009). Debugging is the skill to recognize when actions do not
chers, as they can use the results of this research to redesign their own correspond to instructions, and the skill to fix errors (Bers et al., 2014).
lessons with a focus on the broader set of computational thinking skills, For the purposes of the study reported herein, the elements of al-
and not only coding. gorithmic thinking, sequencing, decomposition, and debugging are of
particular interest and constitute the main areas of the authors’ research
2. Literature review investigation.
In regards to teaching computational thinking skills, during the last
While the concept of computational thinking in education can be decade, the research community has embraced educational robotics
traced back to the work of Seymour Papert, who strongly advocated the with genuine enthusiasm as an approach for teaching computational
idea of children developing algorithmic thinking through the LOGO thinking to pre-primary education students (Alimisis & Kynigos, 2009;
programming language (Papert, 1980, 1996), Wing's (2006) article has Benitti, 2012; Bers, 2010; Bers et al., 2014; Botički et al., 2018;
rekindled the interest for promoting computational thinking in K-12. Bredenfeld, Hofmann, & Steinbauer, 2010; Johnson, 2003; Kazakoff,
Collective efforts aiming at developing a definition of computational Sullivan, & Bers, 2013; Stoeckelmayr et al., 2011). Educators use edu-
thinking include, among others, the two National Academy of Sciences cational robotics in order to engage young students in active and
workshops (National Research Council, 2010, 2011), the initiative playful learning activities through building and programming tangible
C. Angeli and N. Valanides

instructions involving a variety of commands (i.e., move forward, move


backward, turn left, turn right) are used. It is expected that “move
forward” and “move backward” will not cause any kind of difficulty to
the children, because, as stated by Sarama and Clements (2009), chil-
dren already use these terms in their everyday communication with
others without any confusion. Nonetheless, it is speculated that this will
probably not be the case with the spatial referents “left” and “right”,
because children have difficulty using the labels “left” and “right” ap-
propriately in their everyday life (Sarama & Clements, 2009). Roberts
and Aman (1993) explained that often times young children “have dif-
ficulty determining spatial relations from perspectives that differ from their
own perspective” (p. 1258). For example, children have difficulty in
understanding how a scene would look like from somebody else's per-
Fig. 1. The Bee-Bot. spective (Piaget, 1977). Based on the results of their experimental re-
search, Roberts and Aman (1993) concluded that “determining left-right
robotic devices (Bers, 2010). directions from others' vantage points involves imagined spatial rotations
In this study, the authors used the Bee-Bot, shown in Fig. 1, to teach that result in aligning the self's left-right frame with the other's” (p. 1258).
children computational thinking skills. The Bee-Bot is a programmable This isn't the case though with front-back relations, because these are
floor robot that children can touch, hold, and interact with using six usually identifiable by distinct perceptual features, such as, for ex-
directional keys (Misirli & Komis, 2014). The directional keys are used ample, the front or back part of a person's face. It is possible though for
to enter up to 40 commands to move the Bee-Bot forward or backward, children to learn the left-right distinction by reorienting either their
as well as to turn the Bee-Bot to the left or right at a 90° angle. The own orientation or the other's orientation so that the two are in align-
execution of the commands is initiated after pressing the GO button. ment. If this is not possible, then children should carry out an imagined
There are two more buttons, namely the PAUSE and CLEAR buttons. rotation of the self or of the other, and usually children are able to do so
The PAUSE button stops the execution of an algorithm. The CLEAR by the ages of 7–11 (Roberts & Aman, 1993).
button clears the Bee-Bot's memory. If the CLEAR button is not used, An external memory system is necessary for facilitating children's
then new commands are appended to what is already stored in the Bee- problem-solving with the Bee-Bot, because it provides a reference
Bot's memory. The Bee-Bot confirms the completion of a single com- system that children can use to better facilitate the cognitive demands
mand or sequence of commands with lights and sound. It moves ac- that sequencing tasks related to spatial referents impose upon learners'
curately in 15 cm steps at a time and turns in 90°. There is no computer memory resources. An issue though that becomes pertinent at this point
programming or a computer involved during learning with the Bee-Bot. is whether both boys and girls can benefit from the same type of scaf-
Learning materials using the Bee-Bot include motivating problem- folding support.
solving scenarios that the teacher has to develop, and a variety of mats/ Research from neuroscience (Bonomo, 2010; King & Gurian, 2006;
grids, which provide great surfaces for the Bee-Bot to run on. Skills mats Sousa & Tomlinson, 2011) shows that the hippocampus, where the
are specifically used to practice particular skills in numeracy and lit- functions of memory and language exist, develops more rapidly and is
eracy, narrative mats are used for teaching mapping, sequencing, and larger in girls than in boys. As a consequence, this has an effect on the
storytelling in history and geography, and the customizable card mat is skills of sequencing, vocabulary, reading and writing. Boys, on the
appropriate for use in different teaching situations across all content other hand, have more of their cerebral cortex defined for spatial re-
domains. lationships, and because of this they learn better through movement
While the Bee-Bot constitutes an easy to use tool with young chil- and visual experience. Because of these differences, girls tend to prefer
dren, at the same time, it is imperative that teachers learn how to use it collaborative activities where they work with others to share ideas,
in appropriate pedagogical ways in order to maximize its effects on the while boys cannot concentrate for long amounts of time on writing
development of young children's computational thinking skills. As it is tasks, and, they prefer quick, individualistic, kinesthetic, spatially-or-
well documented in the literature, scaffolding children during learning iented, and manipulative-based activities (Gurian, 2003). Bonomo
with educational technology tools is important (e.g., Burke & Kafai, (2010) also found that girls have more cortical areas, thus are more
2012; Fessakis, Dimitracopoulou, & Palaiodimos, 2013; Klein, Nir-Gal, adapted to sitting still and listening to others. All in all, Krommer
& Darom, 2000; Mercer & Fisher, 1992; Shute & Miksad, 1997). (2006) suggested to use strategies in the classroom that accommodate
Accordingly, in this study, scaffolding is expected to play a sig- both genders allowing space and movement for the boys who cannot
nificant role in the development of children's computational thinking thrive in an environment where they are expected to sit down a lot and
skills during learning with the Bee-Bot, because the Bee-Bot does not get involved heavily in language and verbal-based activities, while at
provide a visual representation of the commands children use to pro- the same time fostering collaboration and verbal-emotive expression for
gram it. This weakness of Bee-Bot, along with the poor working girls.
memory skills that are commonplace in childhood (Anderson & Jeffries,
1985), impose a high extraneous cognitive load on children's limited
memory resources, and thus creates a need for finding effective ways in 3. Research questions
order to appropriately scaffold children's learning with the Bee-Bot.
Nonetheless, very little research work exists on scaffolding and inter- Undoubtedly, gender differences at the preschool age are important
ventions with educational robotics in early-childhood education for teachers to consider in order to design effective learning environ-
(Botički et al., 2018; Sullivan & Bers, 2016). Thus, the undertaking of ments and scaffolds for both boys and girls. To this end, the study re-
new efforts toward this direction is fully warranted and very much ported herein sought to investigate the effects of two techniques, which
needed. provided external memory support through scaffolds to the children, on
In this study, it is hypothesized that the use of an external memory computational thinking. In addition, the study also examined children's
system for keeping a visual record of the commands children use to debugging strategies during the problem-solving activities and whether
program the Bee-Bot is necessary for effectively scaffolding children's there were any learning gains related to children's understanding of
learning, especially, in those cases that more complex sequences of spatial relations. Specifically, the research questions are stated as fol-
lows:
C. Angeli and N. Valanides

1. Are there any statistically significant differences on the effects of the 15 × 15cm squares) mat was designed by the researchers to accom-
two scaffolding techniques on learners' computational thinking? modate the moves of the Bee-Bot. As shown in Fig. 2, the mat consisted
2. Are there any statistically significant differences' between boys' and of nine attractive colored pictures, showing a beehive, a blue flower, a
girls' computational thinking? red flower, a yellow flower, a brown flower, a garden with flowers, a
3. Do the two scaffolding techniques differentially affect boys' and wood, the bee Bee-Bee, and the bee Maya.
girls' computational thinking?
4. Are there any learning gains related to children's understanding of 4.2.1. First problem-solving scenario
spatial relations? The first problem-solving scenario introduced children to Rita, the
5. Are there any statistically significant differences between boys' and Bee-Bot, and encouraged children to explore and play with Rita in order
girls' understanding of spatial relations? to discover its functionality using the mat shown in Fig. 2. The children
6. Do the two scaffolding techniques differentially affect boys' and were encouraged to try Rita's POWER ON/OFF button as well as the
girls' understanding of spatial relations? SOUND ON/OFF button and to observe the results. Similarly, the
7. What problem-solving strategies do children exhibit during debug- children experimented with the four directional keys (move forward,
ging activities? move backward, turn left, turn right), and, gradually discovered that
they needed to press the GO button at the end of a sequence of com-
4. Method mands for Rita to execute the algorithm. During this time, the re-
searcher specifically asked the children to describe and explain Rita's
4.1. Participants actions every time it moved forward or backward, turned left or right,
in order to assess their understanding of the spatial referents and di-
Fifty pre-primary education children, between five to six years of agnose learner difficulties. After the first encounter with Rita, the re-
age, participated in the study after parental permission. The children searcher engaged the children in a sequence of systematic problem-
were recruited from eight urban preschools located in three different solving activities. Specifically, the scenario consisted of four subtasks
cities in a southern European country, and, were randomly selected for and one subtask at a time was read out loud by the researcher, before
the purposes of this research. The sample consisted of 26 girls and 24 proceeding to the next one. The subtasks guided the children to pro-
boys. The average age of the participants was 5.28 years and the gram Rita to move out of the beehive and collect pollen from the red,
standard deviation was 0.43 years. All children were familiar with blue, yellow, and brown flowers, as well as to visit her friend Bee-Bee,
computers, worked with computers at school as part of their everyday the wood, and the garden before returning back home (beehive). The
activities, owned a computer at home, and had basic reading and math subtasks were specifically designed to include sequences of actions in-
skills. None of the children had any prior experiences with robots in volving combinations of all four directional keys. Simple sequences
general and with the Bee-Bot in particular, either in school or outside of included tasks that made use of the keys “move forward” or “move
school. In addition, all eight teachers stated that they were unfamiliar backward”. More complex sequences included tasks that used the
with the concept of computational thinking, and, that the teaching of commands “move forward” and “move backward” in different combi-
computational thinking skills was not part of their curriculum. nations. Advanced sequences involved the commands “turn right” or
“turn left” in different combinations with the other two directional keys
4.2. Problem-solving scenarios and mat/grid (i.e., move forward, move backward). The length of a sequence varied
from one to four commands.
Three researcher-made problem-solving scenarios were used in the
study. For all three scenarios, the mat/grid shown in Fig. 2 was used to 4.2.2. Second problem-solving task
facilitate children's thinking and problem solving. The 6 × 6 (36 The second problem-solving scenario followed the same design as
the first in terms of the order, the complexity, and the length of the
programming sequences. It also made use of the same mat shown in
Fig. 2. There were eight subtasks in the second scenario. Regarding the
plot of the second scenario, the researcher initially asked the children to
wake up Rita, (i.e., to turn it on), because an important message came
for her. The important message for Rita was a birthday invitation from
Rita's friend Maya. Maya was having a birthday party and Rita was
invited to the party. Rita was delighted to receive the invitation, but at
the same time she felt sad, because she had nothing new to wear at the
party. Then, the fairy came and advised her to go from flower to flower
to collect petals and use them to make a new dress for the party. Rita
was thrilled with the idea and immediately set a plan for her journey.
The researcher scaffolded the children's (Bee-Bot's) moves from flower
to flower back to the beehive to make the dress and pick up Maya's
birthday present before setting out to Bee-Bee's house. Rita agreed with
Bee-Bee to meet at her house so they could go together to Maya's
birthday party.

4.2.3. Third problem-solving scenario


Lastly, the third scenario, which was a continuation of the second
scenario, consisted of four subtasks. The same mat as before was also
used here. Rita went to Bee-Bee's house and together they head out to
Maya's house. There, they found Maya and her other guests. Maya told
them that the party would not take place at her house, but in the flower
garden. Thus, they all went there and the party began. After many hours
of playing, Maya invited her friends back to her house to spend the
Fig. 2. The mat/grid. night.
C. Angeli and N. Valanides

Fig. 4. Type B scaffolding technique.

child told the researcher the sequence of commands and the researcher
noted down the sequence in a matrix, which was used as an external
memory system for the child. Then, the child used this external memory
system to program the real Bee-Bot and test the algorithm (i.e., se-
quence of commands) on the mat. An excerpt follows.
[Example of using Type B scaffolding]
Let's move Rita to the brown flower. Can you show me with your finger
the route you want the Bee-Bot to follow? It is far away from here, right?
Will you remember all the steps? Well, we can do this to make things
easier. I will show you a picture of the Bee-Bot. Here it is! Show me using
this picture of the Bee-Bot and the mat on the floor the route that you
want the Bee-Bot to follow. We can write it down too! Can you write it
Fig. 3. Type A scaffolding technique. down? [The researcher encourages the child to use the same symbols
as the ones found on the Bee-Bot to write down the sequence of
4.3. Scaffolding techniques commands. In many cases the child asks the researcher to write
down the sequence]. Now, you can look at the notes and use them to
The researchers used two scaffolding techniques in the form of ex- tell Rita what to do! [The child looks at the notes and programs the
ternal memory systems, namely, Type A and Type B, during the robotics Bee-Bot according to the sequence of commands the researcher or
activities with the second problem-solving scenario only. The two him/herself jotted down.]
scaffolding techniques were designed taking into consideration gender
differences, anticipating that both genders would benefit from at least 4.4. Research instruments
one of the two techniques. The aim of Type A and Type B scaffolding
was to make the children aware of their memory limitations and sen- 4.4.1. Color test
sitize them about the need to use some kind of a system or scheme for The color test was a researcher-made test and consisted of four
remembering the sequences of commands. questions. Each question presented the children with four little houses
Type A scaffolding provided each child with 5 cm × 5 cm lami- painted in different color, namely red, blue, yellow, and brown. The
nated cards of all Bee-Bot commands and each child was asked to use researcher read each question to each child individually. The children
the cards to form a sequence of commands for each problem-solving were asked to circle the house painted in a particular color. The colors
subtask. As shown in Fig. 3, the children aligned the cards in an order tested were the same colors as those used to color the Bee-Bot's mat.
one after another on the floor next to the mat to form the algorithm The test was administered in five minutes. The highest possible score on
they needed to use for solving the problem. This arrangement of cards the test that could be obtained was four points. The color test can be
was then used as an external memory system to program the Bee-Bot found in Appendix A.
and test the algorithm on the real mat. An excerpt follows.
[Example of using Type A scaffolding]
4.4.2. Spatial relations test
Let's move Rita to the brown flower. Can you show me with your finger The spatial relations test was a researcher-made test and consisted
the route you want the Bee-Bot to follow? It is far away from here, right? of four questions related to the spatial referents of forward, backward,
Will you remember all the steps? Well, we can do this to make things turn right, and turn left. Students were given a picture showing a dog,
easier. I am giving you these laminated cards. As you can see they re- his house, his toys, food, and the house of his friend. They were asked to
present the commands of Bee-Bot. You can use these cards to create the indicate the path that the dog should take in order to find his house,
sequence of commands that the Bee-Bot needs to execute to go to the toys, food, and his friend's house. The researcher read each question to
brown flower. You can place the cards in any order that you think is each child individually, explained the representation of each referent
right. Put them here on the floor next to the mat [The child puts the (i.e., forward arrow, backward arrow, left arrow, right arrow), and
cards in some sequence and lays them on the floor]. Great! Now, you children were asked to circle the correct answer. The test was ad-
can look at the cards and use them to tell Rita what to do! [The child ministered in five minutes. The highest possible score on the test that
looks at the cards and programs the Bee-Bot accordingly.] could be obtained was four points. The spatial relations test can be
found in Appendix B.
Type B scaffolding (see Fig. 4) engaged each child individually in a
collaborative activity with the researcher. The researcher used the same
4.5. Research procedures
mat and a 3D picture of the Bee-Bot and asked the children to think
about the problem-solving tasks without trying out their solutions. The
Two 40-min research sessions were administered individually for
C. Angeli and N. Valanides

each participant. The first problem-solving scenario was used in the first Table 1
research session, and the second and third scenarios in the second re- Holistic computational thinking assessment rubric.
search session. During the first ten minutes of the first session the re- Code Description Points received
searcher administered the color test and the spatial relations test to
each child individually. Then, for the next 15 min, each child was en- 1 Success without decomposition from first attempt 21
2 Success with decomposition into two from first attempt 20
gaged in a free play activity with the Bee-Bot and the mat. Finally,
3 Success with decomposition into three from first 19
during the last 15 min, each child was engaged in more systematic attempt
problem-solving activities with the Bee-Bot and the mat for the purpose 4 Success with decomposition into four from first attempt 18
of solving specific tasks and for the researcher to assess the child's initial 5 Success without decomposition from second attempt 17
computational thinking skills. 6 Success with decomposition into two from second 16
attempt
Two days after the first research session, the second followed. For
7 Success with decomposition into three from second 15
the purposes of the second research session, the children were randomly attempt
divided into two groups, namely, Group A and Group B. Group A was 8 Success with decomposition into four from second 14
supported with Type A scaffolding and Group B with Type B scaf- attempt
9 Success without decomposition from third attempt 13
folding. During the first 20 min of the second research session, the re-
10 Success with decomposition into two from third attempt 12
searcher scaffolded each child's problem solving with the Bee-Bot using 11 Success with decomposition into three from third 11
one of the two techniques. Then, each child had a five-min break. Upon attempt
return, each child was engaged in the problem-solving activities of the 12 Success with decomposition into four from third attempt 10
third scenario, for 20 more minutes. For the purposes of the third sce- 13 Success without decomposition from fourth attempt 9
14 Success with decomposition into two from fourth 8
nario all scaffolds were faded (removed).
attempt
15 Success with decomposition into three from fourth 7
4.6. The assessment of computational thinking attempt
16 Success without decomposition from fifth attempt 6
There is currently a dearth of research instruments in the literature 17 Success with decomposition into two from fifth attempt 5
18 Success with decomposition into four from fifth attempt 4
for assessing young children's computational thinking. In this study, the 19 Success with decomposition into three from sixth 3
researchers developed inductively a rubric for assessing children's attempt
computational thinking in a holistic way. The rubric will be presented 20 Success without decomposition from seventh attempt 2
in the results section. 21 Success with decomposition into two from seventh 1
attempt

4.7. Data analyses


allowed to convert the directional language into Bee-Bot's command
A number of data analyses were conducted in order to answer the
language and test it using the Bee-Bot and the mat. If the attempt was
research questions. For the questions concerning the effects of the two
unsuccessful, the child was allowed more attempts. A complete example
scaffolding techniques on learners' computational thinking holistic
follows:
performance, a 2 × 2 analysis of covariance (ANCOVA) was performed.
X's first attempt (unsuccessful): MOVE BACKWARD-TURN RIGHT-
In order to examine the effects of the two scaffolding techniques on
GO.
boys' and girls' understanding of spatial relations and learning gains a
X's second attempt (unsuccessful): MOVE BACKWARD-TURN
2 × 2 Repeated Measures Analysis of Variance was conducted. Lastly,
RIGHT-MOVE FORWARD-GO.
in order to examine children's problem-solving strategies during de-
X's third attempt (successful): MOVE BACKWARD-TURN RIGHT-
bugging activities, the authors used the rubric in Table 1, which pro-
TURN RIGHT-MOVE FORWARD-GO.
vided a way to evaluate children's debugging skills demonstrating not
The researchers collected all possible answers for all subtasks over
only how many failed attempts the children made before succeeding,
the two research sessions and developed a rubric as shown in Table 1.
but also whether the children used decomposition as a computational
Table 1 shows a holistic assessment of computational thinking taking
thinking strategy.
into consideration the number of attempts students made during de-
bugging, and, if decomposition was used as a problem-solving strategy.
5. Results
An interrater reliability for the rubric was calculated between two in-
dependent raters and a 97% of agreement was established.
5.1. Rubric for assessing children's computational thinking

Children's computational thinking was assessed based on their 5.2. Assessment of performances on the color and spatial relations tests
proposed solutions to the problem-solving tasks, which included se-
quences of Bee-Bot's movements expressed in directional language, such Table 2 presents the descriptive statistics of boys' and girls’ pre-test
as, MOVE FORWARD, TURN LEFT, MOVE FORWARD, and TURN and post-test scores on the test assessing knowledge of color for each
RIGHT. Children were then assessed on the extent to which they were scaffolding group.
able to convert the path expressed in directional language into algo- The results in Table 2 indicate that boys and girls in both groups
rithmic thinking composed of a series of ordered steps/commands ex- attained the same pre-test and post-test performance on the color test.
pressed in the Bee-Bot's command language. In effect, the children were The maximum possible score on the color test was four points, and, as
assessed on their sequencing skills taking into consideration previous shown in Table 2 all boys and girls received the highest possible score
attempts before succeeding. The rubric also provided a way to evaluate on the tests. As no initial statistically significant differences existed
children's debugging skills demonstrating not only how many failed between boys and girls, and between the two scaffolding groups, the
attempts the children made before succeeding, but also whether the performance on the color test was not used as a covariate in subsequent
children used decomposition as a computational thinking strategy. statistical analyses.
In more detail, the researcher first wrote down children's reasoning In regards to the participants' scores on the pre-test spatial relations
about solving the problem. For example, child X's performance on test, the highest possible score that could be obtained was four points.
Subtask 2 in Phase 2 was recorded in symbolic directional language as The descriptive statistics presented in Table 3 show that boys in Scaf-
follows: MOVE BACKWARD-TURN RIGHT-GO. Then the child was folding Group A (M = 2.67, SD = 1.05) tended to have higher scores on
C. Angeli and N. Valanides

Table 2 Table 5
Descriptive statistics of children's performance on the pre-test and post-test Scores on the four questions of the post-test spatial relations test.
assessing knowledge of color.
Question 1 Question 2 Question 3 Question 4
Pre-test Post-test
Mean SD Mean SD Mean SD Mean SD
Mean SD N Mean SD N
Girls .85 .31 .77 .29 .83 .24 .87 .30
Scaffolding Group A Boys .77 .39 .75 .29 .79 .33 .92 .25
Girls 4.00 0.0 13 4.00 0.0 13
Boys 4.00 0.0 12 4.00 0.0 12
Total 4.00 0.0 25 4.00 0.0 25 differences were detected between boys and girls and scaffolding group.
Scaffolding Group B
Table 3 also presents the descriptive statistics of children's post-test
Girls 4.00 0.0 13 4.00 0.0 13
Boys 4.00 0.0 12 4.00 0.0 12 performance on the spatial relations test, and, as it can be seen all boys
Total 4.00 0.0 25 4.00 0.0 25 and girls across the two scaffolding groups improved their scores on the
Total post-test. A 2 × 2 Repeated Measures Analysis of Variance was con-
Girls 4.00 0.0 26 4.00 0.0 26
ducted to detect possible within-subjects and between-subjects effects
Boys 4.00 0.0 24 4.00 0.0 24
Total 4.00 0.0 50 4.00 0.0 50
for students' pre-test and post-test scores on the spatial relations test.
The results showed statistically significant within-subject effects in-
dicating improvement between the first (M = 2.60, SD = 1.20) and
Table 3 final assessment (M = 3.27, SD = 0.75) of students' understanding of
Descriptive statistics of children's pre-test and post-test performance on the spatial relations, F (1, 46) = 4.90, p < .05, partial ո2 = 0.10, but no
spatial relations test. statistically significant between-subject effects were detected.
Pre-test Post-test
An analysis was then conducted to examine differences among the
four questions on the post-test. Table 5 shows learning gains for all four
Mean SD N Mean SD N questions on the test. In order to detect statistically significant learning
gains between the two administrations of the spatial relations test, re-
Scaffolding Group A
Girls 2.38 .79 13 3.42 .53 13
peated measures analyses were conducted and found statistically sig-
Boys 2.67 1.05 12 3.13 .83 12 nificant within-subjects learning gains for the first, second, and fourth
Total 2.52 .92 25 3.28 .69 25 questions only, F (1, 49) = 16.84, p < 0.01, F (1, 49) = 19.50,
Scaffolding Group B p < 0.01, F (1, 49) = 12.60, p < 0.01, respectively, but not for the
Girls 2.92 1.78 13 3.19 .99 13
third question, F (1, 49) = 2.66, p = .11.
Boys 2.42 1.00 12 3.33 .62 12
Total 2.68 1.45 25 3.26 .82 25
Total
5.3. Initial and final holistic assessment of computational thinking
Girls 2.65 1.38 26 3.31 .79 26
Boys 2.54 1.01 24 3.23 1.72 24
Total 2.60 1.20 50 3.27 .75 50 Table 6 shows, for each scaffolding group, participants’ initial score
on computational thinking as this was measured individually for each
child during the four problem-solving subtasks of the first problem-
the test than girls (M = 2.38, SD = 0.79) in the same group. This solving scenario.
however was not the case with the boys in Scaffolding Group B as they As the descriptive statistics in Table 6 show, boys and girls had very
scored lower (M = 2.42, SD = 1.00) than girls in the same group similar initial computational thinking scores. A 2 × 2 analysis of var-
(M = 2.92, SD = 1.78). It is also worth noting that while boys in iance (ANOVA) was performed to detect whether these initial differ-
Scaffolding Group A (M = 2.67, SD = 1.05) performed better than boys ences were statistically significant, and, the results showed no statisti-
in Scaffolding Group B (M = 2.42, SD = 1.00), girls in Scaffolding cally significant interaction effect between gender and scaffolding
Group A (M = 2.38, SD = 0.79) performed worse than girls in Scaf- group (F (1, 46) = 0.16, p = .26, partial ո2 = 0.003), and no statisti-
folding Group B (M = 2.92, SD = 1.78). A 2 × 2 analysis of variance cally significant differences for the simple main effects of gender
(ANOVA) was performed to detect whether the differences in partici- (p = .96) and group (p = .82).
pants’ pre-test scores on the spatial relations test were statistically Table 7 shows, for each scaffolding group, participants' final com-
significant, and, the results showed no statistically significant interac- putational thinking performance as this was measured individually
tion effect between gender and scaffolding group (F (1, 46) = 1.30, after the intervention for each child during four real-time problem-
p = .26, partial ո2 = 0.03), and no statistically significant differences solving activities. In order to detect statistically significant learning
for the simple main effects of gender (p = .75) and group (p = .68).
An analysis was then conducted to examine differences among the Table 6
four questions on the pre-test. The answer to the first question was Descriptive statistics of the initial assessment of children's computational
“TURN RIGHT”, to the second question “MOVE FORWARD”, to the thinking per group and gender.
third question “MOVE BACKWARD”, and “TURN LEFT” was the answer Mean SD N
to the fourth question. Table 4 below shows the scores on the four
questions of the pre-test spatial relations test. No statistically significant Scaffolding Group A
Girls 15.00 1.47 13
Boys 14.00 2.13 12
Table 4 Total 14.52 1.85 25
Scaffolding Group B
Scores on the four questions of the pre-test spatial relations test.
Girls 13.54 2.44 13
Question 1 Question 2 Question 3 Question 4 Boys 15.08 1.24 12
Total 14.28 2.07 25
Mean SD Mean SD Mean SD Mean SD Total
Girls 14.27 2.11 26
Girls .52 .46 .54 .42 .71 .35 .71 .35 Boys 14.54 1.79 24
Boys .56 .42 .54 .44 .79 .33 .65 .43 Total 14.40 1.95 50
C. Angeli and N. Valanides

Table 7 Table 8
Descriptive statistics of the final assessment of children's computational Initial assessment of children's skill of debugging for each problem-solving task
thinking per group and gender. per group and gender.
Mean SD N Frequencies-Count (%)

Scaffolding Group A Debugging attempts Scaffolding Scaffolding Total


Girls 27.69 6.20 13 Group A Group B
Boys 31.80 3.95 12
Total 29.68 5.55 25 Task 1 FORWARD – GO
Scaffolding Group B
Girls 29.08 6.28 13 Success – First attempt - Girls 11 (84.62) 8 (61.54) 19 (73.08)
Boys 24.92 9.17 12 Without Boys 9 (75) 11 (91.67) 20 (83.33)
Total 27.07 7.92 25 decomposition Total 20 (80) 19 (76) 39 (78)
Total Success – First attempt – Girls 2 (15.38) 4 (30.77) 6 (23.08)
Girls 28.38 6.15 26 Decomposition into Boys 2 (16.67) 1 (8.33) 3 (31.25)
Boys 28.38 7.76 24 two parts Total 4 (16) 5 (20) 9 (18)
Total 28.38 6.90 50 Success – Second attempt Girls 0 (0) 0 (0) 0 (0)
– Without Boys 1 (8.33) 0 (0) 1 (4.17)
decomposition Total 1 (4) 0 (0) 1 (2)
Success – Third attempt – Girls 0 (0) 1 (7.69) 1 (3.85)
gains between the two assessments of children's computational
Without Boys 0 (0) 0 (0) 0 (0)
thinking, a repeated measures analysis was performed and found sta- decomposition Total 0 (0) 1 (4) 1 (2)
tistically significant within-subjects effects, F (1, 49) = 200.01,
p < 0.01. Task 2 TURN LEFT - FORWARD - FORWARD - GO
The descriptive statistics presented in Table 7 show that boys in
Success – First attempt – Girls 0 (0) 0 (0) 0 (0)
Scaffolding Group A (M = 31.80, SD = 3.95) tended to have higher Decomposition into Boys 1 (8.33) 0 (0) 1 (4.17)
scores than girls in the same group (M = 27.69, SD = 6.20). This two parts Total 1 (4) 0 (0) 1 (2)
however was not the case with the boys in Scaffolding Group B as they Success – First attempt – Girls 2 (15.39) 1 (7.69) 3 (11.54)
scored lower (M = 24.92, SD = 9.17) than girls in the same group Decomposition into Boys 1 (8.33) 0 (0) 1 (4.17)
three parts Total 3 (12) 1 (4) 4 (8)
(M = 29.08, SD = 6.28). In addition, based on the data presented in
Success – Second attempt Girls 8 (61.54) 8 (61.54) 16 (61.54)
Table 7, boys in Scaffolding Group A (M = 31.80, SD = 3.95) per- – Without Boys 7 (58.33) 10 (83.33) 17 (70.83)
formed better than boys in Scaffolding Group B (M = 24.92, decomposition Total 15 (60) 18 (72) 33 (66)
SD = 9.17), while girls in Scaffolding Group A (M = 27.69, SD = 6.20) Success – Second attempt Girls 1 (7.69) 0 (0) 1 (3.85)
performed worse than girls in Scaffolding Group B (M = 29.08, – Decomposition into Boys 1 (8.33) 1 (8.33) 2 (8.33)
two parts Total 2 (8) 1 (4) 3 (6)
SD = 6.28). A 2 × 2 analysis of covariance (ANCOVA) was performed Success – Second attempt Girls 2 (15.38) 2 (15.38) 4 (15.38)
to detect whether these final differences were statistically significant – Decomposition into Boys 1 (8.33) 1 (8.33) 2 (8.33)
considering participants’ initial scores on computational thinking. The three parts Total 3 (12) 3 (12) 6 (12)
results showed a statistically significant interaction effect between Success – Third attempt – Girls 0 (0) 2 (15.38) 2 (7.69)
Decomposition into Boys 1 (8.33) 0 (0) 1 (4.17)
gender and scaffolding group (F (1, 46) = 4.87, p < 0.05, partial
three parts Total 1 (4) 2 (8) 3 (6)
ո2 = 0.11) as shown in Fig. 5. No statistically significant differences for
the simple main effects of gender (p = 1.00) and group (p = .15) and Task 3 TURN RIGHT - FORWARD - GO
the covariate (p = .23) were detected.
Success – First attempt - Girls 10 (76.92) 8 (61.54) 18 (69.23)
Decomposition into Boys 7 (58.33) 11 (91.67) 18 (75)
5.4. Initial and final assessment for the skill of debugging two parts Total 17 (68) 19 (76) 36 (72)
Success – Second attempt Girls 3 (23.08) 3 (23.08) 6 (23.08)
Table 8 shows, for each scaffolding group and gender, participants' – Decomposition into Boys 4 (33.33) 1 (8.33) 5 (20.83)
two parts Total 7 (28) 4 (16) 11 (22)
Success –Third attempt – Girls 0 (0) 1 (7.69) 1 (3.85)
Decomposition into Boys 1 (8.33) 0 (0) 1 (4.16)
two parts Total 1 (4) 1 (4) 2 (4)
Success – Fourth attempt – Girls 0 (0) 1 (7.69) 1 (3.85)
Decomposition into Boys 0 (0) 0 (0) 0 (0)
two parts Total 0 (0) 1 (4) 1 (2)

Task 4 BACK - BACK - GO

Success – First attempt - Girls 0 (0) 1 (7.69) 1 (3.85)


Without Boys 3 (25) 0 (0) 3 (12.5)
decomposition Total 3 (12) 1 (4) 4 (8)
Success – First attempt – Girls 12 (92.31) 10 (76.92) 22 (44)
Decomposition into Boys 9 (75) 10 (83.33) 19 (79.17)
two parts Total 21 (84) 20 (80) 41 (82)
Success – Second attempt Girls 1 (7.69) 2 (15.38) 3 (11.54)
– Decomposition into Boys 0 (0) 2 (16.67) 2 (8.33)
two parts Total 1 (4) 4 (16) 5 (10)

initial score regarding the skill of debugging as this was measured in-
dividually for each child during the same activities that were used to
assess children's initial computational thinking score before the inter-
vention.
Based on the frequencies cross tabulated in Table 8, a Mann
Fig. 5. Interaction effect between gender and scaffolding group. Whitney U test was conducted for each problem-solving task across
C. Angeli and N. Valanides

Table 9 Table 9 (continued)


Final assessment of children's skill of debugging for each problem-solving task
per group and gender. Frequencies-Count (%)

Frequencies-Count (%) Debugging attempts Scaffolding Scaffolding Total


Group A Group B
Debugging attempts Scaffolding Scaffolding Total
Group A Group B
Success – Second attempt Girls 3 (23.08) 2 (15.38) 5 (19.23)
Task 1 BACK - TURN LEFT - FORWARD - FORWARD - GO – Decomposition into Boys 0 (0) 0 (0) 0 (0)
four parts Total 3 (12) 2 (8) 5 (10)
Success – First attempt – Girls 0 (0) 1 (7.69) 1 (3.85)
Decomposition into Boys 0 (0) 0 (0) 0 (0) Success – Third attempt – Girls 1 (7.69) 0 (0) 1 (3.85)
two parts Total 0 (0) 1 (4) 1 (2) Decomposition into Boys 0 (0) 0 (0) 0 (0)
four parts Total 1 (4) 0 (0) 1 (2)
Success – First attempt – Girls 1 (7.69) 1 (7.69) 2 (7.69)
Decomposition into Boys 2 (16.67) 1 (8.33) 3 (12.5) Task 3 BACK - BACK - TURN LEFT - FORWARD - GO
three parts Total 3 (12) 2 (8) 5 (10) Success – First attempt - Girls 0 (0) 1 (7.69) 1 (3.85)
Decomposition into Boys 0 (0) 0 (0) 0 (0)
Success – First attempt – Girls 4 (30.77) 2 (15.38) 6 (23.08) two parts Total 0 (0) 1 (4) 1 (2)
Decomposition into Boys 3 (25) 3 (25) 6 (25)
four parts Total 7 (28) 5 (20) 12 (24) Success – First attempt – Girls 0 (0) 1 (7.69) 1 (3.85)
Decomposition into Boys 2 (16.67) 3 (25) 5 (20.83)
Success – Second attempt Girls 3 (23.08) 3 (23.08) 6 (23.08) three parts Total 2 (8) 4 (16) 6 (12)
– Without Boys 4 (33.33) 1 (8.33) 5 (20.83)
decomposition Total 7 (28) 4 (16) 11 (22) Success – First attempt – Girls 4 (30.77) 2 (15.38) 6 (23.08)
Decomposition into Boys 4 (33.33) 1 (8.33) 5 (20.83)
Success – Second attempt Girls 0 (0) 2 (15.38) 2 (7.69) four parts Total 8 (32) 3 (12) 11 (22)
– Decomposition into Boys 0 (0) 1 (8.33) 1 (4.17)
two parts Total 0 (0) 3 (12) 3 (6) Success – Second attempt Girls 3 (23.08) 5 (28.46) 8 (30.77)
– No decomposition Boys 4 (33.33) 2 (16.67) 6 (25)
Success- Second attempt – Girls 0 (0) 2 (15.38) 2 (7.69) Total 7 (28) 7 (28) 14 (28)
Decomposition into Boys 1 (8.33) 1 (8.33) 2 (8.33)
three parts Total 1 (4) 3 (12) 4 (8) Success – Second attempt Girls 1 (7.69) 2 (15.38) 3 (11.54)
–Decomposition into Boys 1 (8.33) 1 (8.33) 2 (8.33)
Success- Second attempt – Girls 2 (15.38) 2 (15.38) 4 (15.38) two parts Total 2 (8) 3 (12) 5 (10)
Decomposition into Boys 2 (16.67) 1 (8.33) 3 (12.5)
four parts Total 4 (16) 3 (12) 7 (14) Success – Second attempt Girls 1 (7.69) 1 (7.69) 2 (7.69)
– Decomposition into Boys 0 (0) 1 (8.33) 1 (4.17)
Success- Third attempt – Girls 2 (15.38) 0 (0) 2 (7.69) three parts Total 1 (4) 2 (8) 3 (6)
Decomposition into Boys 0 (0) 0 (0) 0 (0)
three parts Total 2 (8) 0 (0) 2 (4) Success – Second attempt Girls 2 (15.38) 1 (7.69) 3 (11.54)
– Decomposition into Boys 1 (8.33) 3 (25) 4 (16.67)
Success- Third attempt – Girls 0 (0) 0 (0) 0 (0) four parts Total 3 (12) 4 (16) 7 (14)
Decomposition into Boys 0 (0) 3 (25) 3 (11.54)
four parts Total 0 (0) 3 (12) 3 (6) Success –Third attempt – Girls 1 (7.69) 0 (0) 1 (3.85)
Decomposition into Boys 0 (0) 0 (0) 0 (0)
Success- Fourth attempt – Girls 1 (7.69) 0 (0) 1 (3.85) two parts Total 1 (4) 0 (0) 1 (2)
Decomposition into Boys 0 (0) 1 (8.33) 1 (4.17)
four parts Total 1 (4) 1 (4) 2 (4) Success – Fourth attempt – Girls 1 (7.69) 0 (0) 1 (3.85)
Decomposition into Boys 0 (0) 1 (8.33) 1 (4.17)
Task 2 FORWARD - FORWARD - FORWARD - FORWARD - GO two parts Total 1 (4) 1 (4) 2 (4)
Success – First attempt – Girls 2 (15.38) 3 (23.08) 5 (19.23)
Without Boys 6 (50) 2 (16.67) 8 (33.33) Task 4 TURN RIGHT - FORWARD - FORWARD – FORWARD-GO
decomposition Total 8 (32) 5 (20) 13 (26) Success – First attempt - Girls 0 (0) 1 (7.69) 1 (3.85)
Without Boys 0 (0) 0 (0) 0 (0)
Success – First attempt – Girls 0 (0) 0 (0) 0 (0) decomposition Total 0 (0) 1 (4) 1 (2)
Decomposition into Boys 1 (8.33) 3 (25) 4 (16.67)
two parts Total 1 (4) 3 (12) 4 (8) Success – First attempt – Girls 0 (0) 0 (0) 0 (0)
Decomposition into Boys 1 (8.33) 0 (0) 1 (4.17)
Success – First attempt – Girls 3 (23.08) 4 (30.77) 7 (26.9) two parts Total 1 (4) 0 (0) 1 (2)
Decomposition into Boys 1 (8.33) 1 (8.33) 2 (8.33)
three parts Total 4 (16) 5 (20) 9 (18) Success – First attempt – Girls 5 (38.46) 1 (7.69) 6 (23.08)
Decomposition into Boys 5 (41.67) 4 (33.33) 9 (37.5)
Success – First attempt – Girls 2 (15.38) 0 (0) 2 (7.69) three parts Total 10 (40) 5 (20) 15 (30)
Decomposition into Boys 0 (0) 0 (0) 0 (0)
four parts Total 2 (8) 0 (0) 2 (4) Success – First attempt – Girls 6 (46.15) 8 (61.54) 14 (53.85)
Decomposition into Boys 4 (33.33) 2 (16.67) 6 (25)
Success – Second attempt Girls 2 (15.38) 4 (30.77) 6 (23.08) four parts Total 10 (40) 10 (40) 20 (40)
– Without Boys 4 (33.33) 4 (33.33) 8 (33.33)
decomposition Total 6 (24) 8 (32) 14 (28) Success – Second attempt Girls 1 (7.69) 0 (0) 1 (3.85)
– Without Boys 0 (0) 0 (0) 0 (0)
Success – Second attempt Girls 0 (0) 0 (0) 0 (0) decomposition Total 1 (4) 0 (0) 1 (2)
– Decomposition into Boys 0 (0) 1 (8.33) 1 (4.17)
two parts Total 0 (0) 1 (4) 1 (2) Success – Second attempt Girls 0 (0) 2 (15.38) 2 (7.69)
– Decomposition into Boys 1 (8.33) 2 (15.38) 3 (12.5)
Success – Second attempt Girls 0 (0) 0 (0) 0 (0) two parts Total 1 (4) 4 (16) 5 (10)
– Decomposition into Boys 0 (0) 1 (8.33) 1 (4.17)
three parts Total 0 (0) 1 (4) 1 (2)
(continued on next page)
C. Angeli and N. Valanides

Table 9 (continued) In regards to the teaching techniques the authors used to teach
computational thinking, two scaffolding strategies were developed
Frequencies-Count (%)
taking into consideration the technical affordances of the Bee-Bot. In
Debugging attempts Scaffolding Scaffolding Total detail, the two strategies were conceived by studying the limitations of
Group A Group B the interface of the Bee-Bot, which does not really keep a record of the
sequence of commands. Thus, the authors hypothesized that the use of
Success – Second attempt Girls 0 (0) 0 (0) 0 (0)
an external memory system for keeping a visual record of the com-
– Decomposition into Boys 0 (0) 1 (8.33) 1 (4.17)
three parts Total 0 (0) 1 (4) 1 (2) mands used to program the Bee-Bot would be necessary for effectively
scaffolding children's learning. Before deciding on the kind of external
Success – Second attempt Girls 1 (7.69) 0 (0) 1 (3.85) memory systems, the authors considered gender differences in order to
– Decomposition into Boys 0 (0) 0 (0) 0 (0) ensure that both boys and girls would benefit from the scaffolds.
four parts Total 1 (4) 0 (0) 1 (2)
Consequently, based on findings from neuroscience (Bonomo, 2010;
Success – Third attempt – Girls 0 (0) 0 (0) 0 (0) Gurian, 2003; King & Gurian, 2006; Sousa & Tomlinson, 2011), the
No decomposition Boys 0 (0) 1 (8.33) 1 (4.17) authors came up with two scaffolding techniques for providing external
Total 0 (0) 1 (4) 1 (2) reference support to the children. One technique used individualistic,
kinesthetic, spatially-oriented, and manipulative-based activities with
Success – Third attempt Girls 0 (0) 0 (0) 0 (0)
–Decomposition into Boys 1 (8.33) 1 (8.33) 2 (8.33) cards, and, it was hypothesized that boys would benefit from learning
two parts Total 1 (4) 1 (4) 2 (4) with this technique more than girls, while the other technique was a
collaborative writing task activity, and, it was hypothesized that girls
Success – Fourth attempt – Girls 0 (0) 1 (7.69) 1 (3.85) would benefit from this technique more than boys.
Decomposition into Boys 0 (0) 0 (0) 0 (0)
three parts Total 0 (0) 1 (4) 1 (2)
According to the findings, while both genders benefited from both
scaffolding techniques, a statistically significant interaction effect be-
Success – Fifth attempt Girls 0 (0) 0 (0) 0 (0) tween gender and scaffolding technique was detected showing that
Decomposition into Boys 0 (0) 1 (8.33) 1 (4.17) boys benefited more from the individualistic, kinesthetic spatially or-
four parts Total 0 (0) 1 (4) 1 (2)
iented, and manipulative-based activity with cards (scaffolding Type
A), while girls benefited more from the collaborative writing activity
(scaffolding Type B). The implications of these findings are very im-
gender and scaffolding group in order to examine initial differences
portant, because they show that while learning gains were obtained
regarding the skill of debugging. All Mann Whitney U tests performed
from learning with both types of scaffolding, the design of a gender-
showed statistically insignificant differences between the dependent
friendly learning environment differentially affected children's perfor-
variable, thus, the null hypotheses about having the same distribution
mance on the computational thinking tasks. This is essential, because, it
of the dependent variable across gender and scaffolding groups were all
shows that a learning environment must be appropriately designed to
retained.
provide both genders with activities that meet their distinctive needs at
Table 9 shows, for each scaffolding group and gender, participants’
this point in their development. A developmentally-appropriate gender-
final assessment regarding the skill of debugging.
friendly robotics learning environment will allow both boys and girls to
Based on the frequencies cross tabulated in Table 9, for each pro-
experience a positive flow of experience, and to endure and persevere in
blem-solving task a Mann Whitney U test was conducted for gender and
learning.
scaffolding group in order to examine differences regarding the skill of
Regarding children's understanding of spatial relations, it was ex-
debugging. All Mann Whitney U tests performed showed statistically
pected that “move forward” and “move backward” would not cause any
insignificant differences between the dependent variable, thus the null
kind of difficulty to the children, because children already use these
hypotheses about having the same distribution of the dependent vari-
terms in their everyday communication with others without any diffi-
able across gender and scaffolding groups were all retained.
culty, but it was also speculated that this would probably not be the
case with the spatial referents “left” and “right”, because children have
6. Discussion and implications difficulty with using these terms in their daily life (Sarama and
Clements (2009)). Repeated measures analyses showed statistically
In this study, it was hypothesized that the use of robotics activities significant differences between pre-test and post-test scores on the
with the use of a small programmable floor robot named Bee-Bot would questions related to the spatial referents “move right”, “move forward”,
be an effective way for developing young children's computational and “move left”, while the scores related to the referent “move back”
thinking. The hypothesis was confirmed based on the results of the were high from the beginning and no significant changes occurred that
study reporting statistically significant learning gains between the in- could be attributed to the intervention. Based on these findings, the
itial and final assessment of children's computational thinking skills. In authors conclude that at this very young age, all students can benefit
addition, the authors hypothesized that learning with robotics activities from instruction in learning the spatial referents, and, that the referent
would be an effective way to teach children about spatial relations. This that does not cause any difficulty to the children at this age is “move
hypothesis was also confirmed as the findings showed statistically sig- back”.
nificant within-subjects learning gains between the initial and final The rubric also enabled the authors to investigate children's pro-
assessment of children's understanding of spatial relations, while no blem-solving techniques during the debugging activities. Based on the
differences were identified for gender and group. The implications of results, no differences were identified between the two scaffolding
these findings are important, as they provide robust support for the groups, gender, and number of commands in an algorithm. The findings
integration of robotic devices in the curriculum of early-childhood showed that a number of children used some kind of an external
education. In essence, robotics can be used as an educational tech- memory system to first jot down the sequences of commands before
nology to reform and enhance the traditional early-childhood curri- programming the Bee-Bot to test them. In the case of failure, these
culum so that children can be taught about computational thinking children recognized the error and fixed the algorithm before testing it
skills that are so much needed for surviving in the 21st century. In again. The majority of the children though used the strategy of de-
addition, these findings show that robotics can be used to teach young composition to deal with the complexity of the task. Essentially, these
students about spatial relations, which constitutes an important com- children decomposed the task in a number of subtasks equal to the
ponent of the math curriculum in preschool education. number of commands in the task, and, chose to execute one subtask at a
C. Angeli and N. Valanides

time. So, for example, if the task was “turn left-move forward-move are used to situate their thinking.
forward”, the majority of the children solved the task by running the Based on the experiences gained from the undertaking of the study
following sequences of commands – “turn left-go”, “clear”, “move for- presented herein, the authors concur with Gibson and believe that a
ward”, “go”, “clear”, “move forward”, “go”. These results are im- future research investigation about the teaching of abstraction in the
portant, because they show that children at this very young age are able context of early-childhood education will be worthwhile in terms of
to cope with the complexity of a learning task by decomposing it into a understanding better the teaching of the skill of abstraction in young
number of subtasks that are easier for them to tackle. children. In more detail, future research studies can examine how
within the context of robotics activities the skill of abstraction can be
7. Future research directions taught to young children by using teaching strategies that make use of
external reference systems, such as the ones used in the study herein.
The empirical findings of the study lead to two future research di- Moreover, from a theoretical point of view, it would be interesting to
rections. The first area of research identified is whether the skill of invest research effort and time in examining how the development of
abstraction can be taught to young children and how it can be in- the skill of abstraction can be understood from a developmental per-
corporated in a computational thinking curriculum for early-childhood spective. According to the tenets of the neo-Piagetian theory, people,
education. A number of computer science education researchers have regardless of their age, are able to develop abstract thinking (Lister,
written about their concerns in regards to teaching abstraction in K-6 2011). Accordingly, computational thinking researchers can investigate
(e.g., Armoni, 2012, 2013). These concerns are primarily linked to the the development of children's, aged from 3 to 11 years old, abstraction
incompatibility between abstraction, an essential process in computer skills so that more insight can be gained about the teaching of these
science and computational thinking, and children's weakness to un- skills in young children.
derstand abstraction because of their very young age. Armoni (2012, The second research area is related to the need to undertake qua-
2013) explained that abstraction is an inherent component of compu- litative studies in the area of computational thinking in order to un-
tational thinking that is always encapsulated during the process of derstand the strategies children use to think computationally about
thinking about and automating a solution to a problem. From a Pia- various problems taking into consideration learners' individual cogni-
getian perspective, children before the age of seven cannot really un- tive differences, such as, for example, differences in processing various
derstand concrete logic, whereas children between seven and eleven symbol systems and representations. In the context of this line of re-
years old can solve problems that apply to concrete objects, but not search, researchers may also want to consider the length of the algo-
problems that apply to abstract concepts or phenomena. Conversely, rithm (i.e., number of commands in a sequence) and how this affects
Gibson (2012) argued that high school is too late for exposing students children's problem solving. Studies in this research area will be of ut-
to abstraction for the first time, and stated that early exposure during most importance, because, they can provide insights about how all
kindergarten is necessary. In his research, Gibson (2012) found that children can learn to think computationally.
young children can think abstractly when concrete reference systems

Appendix A

Please circle the house with the blue color.

Please circle the house with the red color.

Please circle the house with the brown color.

Please circle the house with the yellow color.

Appendix B

Please help Doggie find a few things:


C. Angeli and N. Valanides

Which way does Doggie need to look to find the blue house? Circle one of the four arrows.

Which way does Doggie need to look to find his toys? Circle one of the four arrows.

Which way does Doggie need to look to find his food? Circle one of the four arrows.

Which way does Doggie need to look to find the yellow house? Circle one of the four arrows.

References thinking education 2018 (CTE2018).


Bredenfeld, A., Hofmann, A., & Steinbauer, G. (2010). Robotics in education initiatives in
Europe-status, shortcomings and open questions. Workshop proceedings of international
Alimisis, D., & Kynigos, C. (2009). Constructionism and robotics in education. Teacher conference on simulation, modeling and programming for autonomous robots (pp. 568–
Education on Robotic-Enhanced Constructivist Pedagogical Methods, 11–26. 574). Darmstadt: Germany.
Anderson, J. R., & Jeffries, R. (1985). Novice LISP errors: Undetected losses of informa- Bundy, A. (2007). Computational thinking is pervasive. Journal of Scientific and Practical
tion from working memory. Human-Computer Interaction, 1(2), 107–131. Computing, 1(2), 67–69.
Armoni, M. (2012). Teaching CS in kindergarten: How early can the pipeline begin? ACM Burke, Q., & Kafai, Y. B. (2012). The writers' workshop for youth programmers: Digital
Inroads, 3(4), 18–19. storytelling with scratch in middle school classrooms. Proceedings of the 43rd ACM
Armoni, M. (2013). On teaching abstraction in computer science to novices. Journal of technical symposium on computer science education (pp. 433–438). New York: ACM.
Computers in Mathematics and Science Teaching, 32(3), 265–284. Fessakis, G., Dimitracopoulou, A., & Palaiodimos, A. (2013). Graphical interaction ana-
Barr, V., & Stephenson, C. (2011). Bringing computational thinking to K-12: What is in- lysis impact on groups collaborating through blogs. Journal of Educational Technology
volved and what is the role of the computer science education community? ACM & Society, 16(1), 243–253.
Inroads, 2(1), 48–54. Gibson, J. P. (2012). Teaching graph algorithms to children of all ages. Proceedings of the
Benitti, F. B. V. (2012). Exploring the educational potential of robotics in schools: A 17th ACM annual conference on Innovation and technology in computer science education
systematic review. Computers & Education, 58(3), 978–988. (pp. 34–39). ACM.
Beraza, I., Pina, A., & Demo, B. (2010). Soft & hard ideas to improve interaction with Grover, S., & Pea, R. (2013). Computational thinking in K–12: A review of the state of the
robots for kids & teachers. Proceedings of SIMPAR 2010 international conference on field. Educational Researcher, 42(1), 38–43.
simulation, modeling and programming for autonomous robots (pp. 549–557). . Grover, S., & Pea, R. (2018). Computational thinking: A competency whose time has
Bers, M. U. (2010). The TangibleK robotics program: Applied computational thinking for come. Computer Science Education: Perspectives on Teaching and Learning in School, 19.
young children. Early Childhood Research & Practice, 12(2), 1–20. Gurian, M. (2003). What could he be thinking? How a man's mind really works. London:
Bers, M. U., Flannery, L., Kazakoff, E. R., & Sullivan, A. (2014). Computational thinking Macmillan.
and tinkering: Exploration of an early childhood robotics curriculum. Computers & Guzdial, M. (2008). Education: Paving the way for computational thinking.
Education, 72, 145–157. Communications of the ACM, 51(8), 25–27.
Bonomo, V. (2010). Gender matters in elementary education research-based strategies to Highfield, K. (2010). Robotic toys as a catalyst for mathematical problem solving.
meet the distinctive learning needs of boys and girls. Educational Horizons, 88(4), Australian Primary Mathematics Classroom, 15(2), 22–27.
257–264. Highfield, K., & Mulligan, J. (2008). Young children's engagement with technological
Botički, I., Pivalica, D., & Seow, P. (2018). The use of computational thinking concepts in tools: The impact on mathematics learning. Proceedings of international congress in
early primary school. Proceedings of the international conference of computational mathematical education: Vol. 11, (pp. 6–13). Mexico, USA.
C. Angeli and N. Valanides

Highfield, K., & Mulligan, J. (2009). Young children's embodied action in problem-solving National Research Council (2010). Report of a workshop on the scope and nature of com-
tasks using robotic toys. In M. Tzekaki, M. Kaldrimidou, & H. Sakonidis (Vol. Eds.), putational thinkingWashington, DC: The National Academies Press.
Proceedings of the 33rd conference of the international group for the psychology of National Research Council (2011). Report of a workshop of pedagogical aspects of compu-
mathematics education. Vol. 2. Proceedings of the 33rd conference of the international tational thinkingWashington, DC: The National Academies Press.
group for the psychology of mathematics education (pp. 273–280). Thessaloniki, Greece: Papert, S. (1980). Mindstorms: Children, computers, and powerful ideas. New York: Basic
PME. Books.
Johnson, J. (2003). Children, robotics, and education. Artificial Life and Robotics, 7(1–2), Papert, S. (1996). An exploration in the space of mathematics education. International
16–21. Journal of Computers for Mathematical Learning, 1(1), 95–123.
Kabátová, M., Jašková, L., Lecký, P., & Laššáková, V. (2012). Robotic activities for vi- Piaget, J. (1977). The development of thought: Equilibration of cognitive structures. New York:
sually impaired secondary school children. 3rd international workshop, teaching ro- Viking.
botics, teaching with robotics (pp. 22–31). . Roberts, R. J., & Aman, C. J. (1993). Developmental differences in giving directions:
Kazakoff, E., & Bers, M. (2012). Programming in a robotics context in the pre-primary Spatial frames of reference and mental rotation. Child Development, 64(4),
classroom: The impact on sequencing skills. Journal of Educational Multimedia and 1258–1270.
Hypermedia, 21(4), 371–391. Sarama, J., & Clements, D. H. (2009). Early childhood mathematics education research:
Kazakoff, E. R., Sullivan, A., & Bers, M. U. (2013). The effect of a classroom-based in- Learning trajectories for young children. New York: Routledge.
tensive robotics and programming workshop on sequencing ability in early child- Selby, C. C. (2012). Promoting computational thinking with programming. Proceedings of
hood. Early Childhood Education Journal, 41(4), 245–255. the 7th workshop in primary and secondary computing education (pp. 74–77). New York:
King, K., & Gurian, M. (2006). The brain–his and hers. Educational Leadership, 64(1), 59. ACM.
Klein, P. S., Nir-Gal, O., & Darom, E. (2000). The use of computers in pre-primary, with or Selby, C., & Woollard, J. (2013). Computational thinking: The developing definitions.
without adult mediation? Effects on children's cognitive performance and behavior. Proceedings of the 45th ACM technical symposium on computer science education.
Computers in Human Behavior, 16(6), 591–608. Canterbury: ACM.
Krommer, D. (2006). Boys and girls together: A case for creating gender-friendly middle Shute, R., & Miksad, J. (1997). Computer assisted instruction and cognitive development
school classrooms. Clearinghouse, 79(6), 247–251. in preschoolers. Child Study Journal, 27(3), 237–253.
Lavigne, H., Orr, J., & Wolsky, M. (2018). Exploring computational thinking in preschool Shute, V. J., Sun, C., & Asbell-Clarke, J. (2017). Demystifying computational thinking.
math learning environments. Society for information technology & teacher education Educational Research Review, 22, 142–158.
international conference (pp. 38–43). Association for the Advancement of Computing Sousa, D. A., & Tomlinson, C. A. (2011). Differentiation and the brain: How neuroscience
in Education (AACE). supports the learner-friendly classroom. Bloomington: Solution Tree Press.
Lister, R. (2011). Concrete and other neo-Piagetian forms of reasoning in the novice Stoeckelmayr, K., Tesar, M., & Hofmann, A. (2011). Pre-primary children programming
programmer. Proceedings of the thirteenth australasian computing education conference: robots: A first attempt. Proceedings of 2nd international conference on robotics in edu-
Vol. 114, (pp. 9–18). Australian Computer Society, Inc. cation (pp. 185–192). Vienna, Austria: INNOC - Austrian Society for Innovative
Lu, J. J., & Fletcher, G. H. (2009). Thinking about computational thinking. ACM SIGCSE Computer Sciences.
Bulletin. Vol. 41. ACM SIGCSE Bulletin (pp. 260–264). Chattanooga, TN: ACM. (1) Sullivan, A., & Bers, M. U. (2016). Robotics in the early childhood classroom: Learning
Retrieved 24 June, 2015 from URL http://portal.acm.org/citation.cfm?id=1508959 outcomes from an 8-week robotics curriculum in pre-kindergarten through second
&dl=ACM&coll=portal. grade. International Journal of Technology and Design Education, 26(1), 3–20.
Lye, S. Y., & Koh, J. H. L. (2014). Review on teaching and learning of computational The Royal Society (2014). Vision for science and mathematics education. London: The Royal
thinking through programming: What is next for K-12? Computers in Human Behavior, Society.
41, 51–61. Wing, J. M. (2006). Computational thinking. Communications of the ACM, 49(3), 33–35.
Mercer, N., & Fisher, E. (1992). How do teachers help children to learn? An analysis of Wing, J. M. (2008). Computational thinking and thinking about computing. Philosophical
teachers' interventions in computer-based activities. Learning and Instruction, 2(4), Transactions of the Royal Society A, 366(1881), 3717–3725.
339–355. Wing, J. (2011). Research notebook: Computational thinking - what and why? Retrieved 3
Misirli, A., & Komis, V. (2014). Robotics and programming concepts in early childhood March, 2015 from URL http://www.cs.cmu.edu/link/research-notebook-
education: A conceptual framework for designing educational scenarios. Research on computational-thinking-what-and-why.
e-Learning and ICT in Education (pp. 99–118). New York: Springer.

You might also like