Computational thinking tests.
Open access peer-reviewed article
This Article is part of Computer Graphics and Multimedia Section
Article metrics overview
10 Article Downloads
View Full Metrics
Article Type: Research Paper
Date of acceptance: June 2024
Date of publication: July 2024
DoI: 10.5772/acrt.36
copyright: ©2024 The Author(s), Licensee IntechOpen, License: CC BY 4.0
The recent advancement in computational thinking (CT) research has reported numerous learning benefits to school-age children. The long-standing perceived difficulty of computer programming has challenged the acquisition of CT skills from programming education. Several block-based programming environments (BBPEs) have been developed to reduce this difficulty and enhance active engagement in computational-related activities. Although numerous studies have examined students’ level of interactions during block-based programming modality (BPM) activities, a major gap in the literature is the paucity of research evidence reporting the association between these interactions and CT. This study, therefore, investigates the association between interaction patterns during BPM activities and CT skills. The present study employed a longitudinal approach where the same participants were observed over eight weeks. Thirty-five, second-year-level computer science and computer education students (mean age: 19.8; male = 23, female = 12) from a research university in Nigeria were recruited. Their computational activities over the study periods were video-recorded. The participants’ CT skills were collected using the computational thinking test and the computational thinking scale. Findings indicate four interaction patterns: learner–learner, learner–content, learner–teacher, and learner–distractor. Learner–learner and learner–content were prevalent. The interaction patterns significantly predict CT skills although significant differences exist across gender, cognitive load, spatial ability, and programming proficiency. The research has provided opportunities for educators to integrate BBPEs in learning programming and CT concepts. Although such integration is likely to occur with the help of strong educational policies, teachers are encouraged to cultivate the spirit of collaboration in students during programming activities.
computational thinking
block-based programming modality
block-based programming environments
interaction patterns
students
Author information
In recent years, the field of computer programming has assumed a critical role in enhancing diverse cognitive competencies among students [1]. In pursuit of augmenting these cognitive skills, numerous instructional modalities have emerged, broadly classified into block-based (BPM) and text-based (TPM) programming modalities [2]. The BPM presents a visual programming paradigm, employing a metaphor akin to “programming-primitive-as-puzzle-piece,” facilitating the conceptualization and execution of computer programming designs [3]. Conversely, TPM adheres to a more traditional approach, necessitating learners’ proficiency in coding across diverse text-based programming languages such as Python and Java. Proficiency in these languages holds relevance for undertaking professional programming endeavors and pursuing careers in computer science [4].
Within the programming context, there has been consistent research evidence reporting the close association between programming and computational thinking (CT) [5–9]. In the process, teachers and researchers have put continuous effort toward developing and integrating BPM and TPM into programming education to improve learners’ CT skills [2, 10, 11]. On an overarching level, CT encompasses a large intellectual foundation required to provide knowledge and understanding of the computational world, and the ability to employ such knowledge in problem-solving across different disciplines [12–14]. The recent advancement in CT research has reported numerous learning benefits to school-age children, including the ability to solve real-world problems systematically [15, 16]. There is a common agreement that CT permits students to develop creative thinking and problem-solving skills [17].
However, the acquisition of CT skills from programming education has been challenged by the long-standing perceived difficulty of computer programming [15]. To address these challenges, several block-based programming environments (BBPEs), including prominent ones such as Scratch [18] and Alice [19], have been proposed. These environments enable students to interact with programming code through drag-and-drop interfaces, allowing the creation of animations and games without the need for extensive typing. The pedagogical benefits of block-based programs lie in their technological design. For instance, Alice and Scratch minimize syntax errors—a major source of frustration for programming students—by eliminating the need for traditional code typing [20, 21]. This allows students to focus more on understanding programming concepts and logic rather than syntax. Moreover, these environments provide immediate visual feedback, which helps students quickly see the results of their actions and understand the effects of their code. This immediate feedback loop is crucial for learning and helps maintain student engagement and motivation.
However, research evidence has indicated that BPM does not always enhance students’ interaction. One explanation is that the visual and interactive elements of these programming environments can overload working memory [22]. This finding is supported by Cognitive Load Theory [23], which suggests that extraneous cognitive load can be imposed by factors such as visual clutter and the complexity of managing numerous blocks. Additionally, understanding and organizing the logical flow of programs in BBPE can be challenging, particularly for beginners. Even though these environments simplify syntax, they still require learners to grasp fundamental programming concepts like loops, conditionals, and variables. Debugging in BPM can also be cognitively demanding as identifying and correcting errors in a visual format requires careful attention and problem-solving skills. Consequently, students often need a high level of spatial ability to effectively navigate and interact with BBPE. The combination of these factors can lead to increased cognitive load, potentially hindering the learning process.
Another plausible explanation is that most BBPEs were developed for novice programmers [24, 25], and the interaction with these programs could lead to an expertise reversal effect, a concept where block-based programs are not beneficial or counterproductive to the learning outcomes of expert programmers [26, 27]. Gender difference is also another reason for the inconsistent benefits of block-based environments. Studies revealed that boys interact more in programming activities than girls [28, 29], and therefore are more likely to interact in BPM activities. However, recent studies found no gender difference in the use of programming environments [30].
Numerous studies have examined students’ CT during BPM activities. For example, Bers
However, a major gap in the literature is the paucity of research examining the association between distinct interactions in BPM activities and CT. It is proposed by the researchers that some interactions during BPM activities might foster CT more than other interactions. Nevertheless, to the best of our knowledge, previous literature has not thoroughly investigated how interaction patterns during BPM activities can represent and foster CT as demonstrated in its core principles and practices. In addition to this paucity of research evidence, limited attention is given to how students differ in their CT skills regarding the cognitive load, spatial ability, programming proficiency, and gender when interacting with these environments. A study is needed in this area to enrich the body of literature in the areas of programming and CT. Therefore, the study analyzed students’ interaction patterns when exposed to BPM and examined how these patterns predict CT. Specifically, the study addressed the following research questions (RQs):
What interaction patterns exist when students are exposed to BPM activities?
Is there a significant association between students’ interaction patterns and their profiles (gender, cognitive load, spatial ability, and programming proficiency) during BPM activities?
What is the level of students’ computational thinking skills during BPM activities?
Is there a significant association between students’ computational thinking and their profiles (gender, cognitive load, spatial ability, and programming proficiency) during BPM activities?
Do the interaction patterns predict computational thinking?
The concept of CT has been subject to various definitions and interpretations across the academic literature. Initially proposed by Wing [14] as involving problem-solving, system design, and understanding human behavior through fundamental computer science concepts, subsequent attempts at clarification have led to revised definitions. This definition was refined by Wing [36] to emphasize the thought processes involved in formulating problems and solutions, highlighting the importance of representation for effective execution by information-processing agents. The definition was further clarified by Aho [37], indicating that CT encompasses thought processes for formulating problem-solving steps, resulting in solutions represented as finite sets of computational steps. Other definitions, such as that by Yadav
Various frameworks have been developed to understand CT. These frameworks include those proposed by Brennan and Resnick [40] (concepts, practices, and perspectives), Repenning
Although these frameworks have provided indicators for understanding CT, they are criticized for being too limited to specific programming activities and thus insufficient for imparting a comprehensive understanding of programming and CT principles [39]. Recognizing these limitations, a more extensive framework encompassing 14 core educational principles was proposed by Fagerlund
Across the literature, the development of several CT assessment portfolios to evaluate students’ CT has been noted. A major drawback of these assessment portfolios is recognized as their closed-access nature (e.g., Fairy Assessment: Werner
Among the assessment tests, the two most widely used portfolios are the computational thinking scale (CTS) [43] and the computational thinking test (CTt) [49]. The CTt focuses extensively on “computational concepts” such as sequencing, loops, conditionals, and functions [29]. In contrast, the CTS focuses overly on “computational perspectives,” including perspectives on creative thinking, algorithmic thinking, critical thinking, cooperativity, and problem-solving. Because of their comprehensiveness in measuring CT skills, the two tests continue to receive wider applications across different disciplines, involving students of different age groups. Table 1 presents a summary of the design, grounded framework, and target audience of the CT tests.
CT test | Type | Grounded framework/constructs | Target audience | |
---|---|---|---|---|
CTS | Self-assessment test | International Society for Technology in Education framework (ISTE, 2015; creativity, algorithmic thinking, cooperativity, critical thinking, problem-solving) | Undergraduate | 29 |
CTt | Performance test | Brennan and Resnick ([40]; sequences, loops, conditionals, functions, and variables) | Middle school (K-7 and K-8) | 28 |
As previously discussed, CT is measured from a wide range of dimensions. Although it would be difficult to unify these dimensions into a single assessment protocol, several authors strongly recommend the combination of many CT tests to measure students’ CT skills more extensively [39, 50]. The idea of such unification also includes the reflection of deeper learning that “contributes to a comprehensive picture of students’ learning in CT education” [50, p. 2]. To the best of our knowledge, we found a few published empirical studies [50–53] that measure students’ CT across different dimensions. This is another gap left in the literature as many studies (e.g., [11, 54–56, 57]) measured CT skills using one dimension, and these studies tend to ignore many CT core principles. In this study, the recommendations of unifying CT dimensions were adhered to. Therefore, the CTS and CTt were used in this study.
There is common agreement that programming enhances computational thinking [58]. However, programming difficulties tend to derail the acquisition of CT skills, coupled with the fact that CT concepts are difficult to learn due to their abstract nature [59]. To reduce programming difficulties and facilitate the acquisition of CT skills, several block-based programs were developed, including the prominent ones: Alice and Scratch. Conceived as alternate programming tools during the past 10 years, BPM is generally defined as languages and tools that permit novice programmers to develop software products with little knowledge of the syntax and procedures of conventional programming language [60].
The technological advantage of these programs lies in their drag-and-drop features, which eliminate the hassles of recalling the syntax of conventional programming languages like Java. Since their emergence, they have been applied in many programming education studies to foster CT skills [9, 32, 33]. These studies have demonstrated the positive impact of using block-based programs to enhance students’ CT skills, including executive functions that help them to solve and analyze problems.
For example, it was found by Bers
Across the literature, students’ interaction in BPM activities is largely conceived as engagement. The two terms are often used interchangeably. We argue that engagement is a broader term used to represent vision (cognitive), action (behavioral), and emotions (affective). Interaction on the other hand represents specific activities that overly reflect the cognitive and behavioral components of engagement. In a typical classroom activity, authors proposed that such engagement represents three main interactions: interaction with course content, teacher, and peers [61, 62]. The first pattern is termed “learner–content interaction,” which occurs when students are watching an animated programming video or performing a programming task. The second pattern is “learner–teacher interaction,” which occurs when learners are asking or responding to teacher questions. The third pattern is “learner–learner interaction,” which occurs when learners engage in some kind of collaborative task such as think–pair–share. Because students often exhibit different attention-related behaviors in a typical classroom [63], we propose a fourth interaction pattern known as “learner–distractor interaction,” which occurs when learners are interacting with unauthorized devices such as mobile phones or engaging in off-point discussion.
Studies have explored different interactions in the context of BPM activities. A recent study by Hopcan
Despite the learning benefits of BPM, research examining their pedagogical effectiveness has shown that they are not always effective (for a review, see [22]). Thus, interactions with these programming tools are determined by external variables that include cognitive load level, spatial ability, programming proficiency, and gender.
Few studies have shown that some BBPEs can sometimes overwhelm students’ working memory, leading to cognitive overload. Although these tools aim to enhance engagement and reduce programming difficulty, the excessive use of blocks for debugging codes within the environment may distract learners and impede their ability to focus on the environment. Additionally, some studies have highlighted challenges associated with transitioning from block-based programming to text-based coding [51, 66]. While block-based environments provide a visual and intuitive entry point to programming, students may encounter difficulties when transitioning to traditional coding languages due to differences in syntax and structure.
Spatial ability is generally conceived as a type of cognitive function that is essential in processing visual information [67]. The predictive effect of spatial ability on interaction with BPM is explained by two hypotheses: the ability-as-compensator hypothesis [68, 69] and the enhancer hypothesis [68, 70]. The ability-as-compensator hypothesis assumes that BBPE can benefit students with low spatial abilities by reducing the mental effort required to work with programming blocks and illustrations. On the other hand, students with high spatial abilities do not benefit from these tools because they already have the cognitive functions required to generate sufficient mental representation regardless of the presentation formats [67].
The enhancer hypothesis suggests the opposite and claims that high spatial ability learners benefit more from dynamic visualization compared to low spatial ability learners. Recent research has confirmed the validity of the enhancer hypothesis by indicating that block-based visualization was more beneficial to learners with high spatial abilities [67]. One previous study also found that the integration of 3D program visualization enhances visual interaction only in students with high spatial ability [70].
An overwhelming number of research studies have indicated that BBPEs were specifically developed for novice programmers [71, 72]. Therefore, exposing expert programmers to these tools might lead to an “expertise reversal effect,” a concept where block-based environments become counterproductive to the learning outcomes of expert programmers [27]. However, as put forward by a recent study [26], the expertise reversal effect also occurs with a high level of expertise. In this situation, highly interactive BBPEs, including those that were designed to enhance deep exploration of concepts, may overload the memory of novice learners and benefit the expert learners.
Research on gender differences concerning the interaction with BPM activities has attracted controversies. There is no widely accepted evidence suggesting that interaction with BBPEs is in favor of a particular gender category. For example, several studies have reported that males interacted more (e.g., [28, 29]) and several others have reported females (e.g., [73]). There are also quite a few others that reported no difference [74, 75]. At this point, gender differences in the interaction with BPM activities remain a subject of continuing debate.
Ideally, any factor that influences students’ interaction within BBPEs should also influence the acquisition of CT skills. This is based on the assumption that students’ skills in BBPEs are significantly correlated with their CT skills [5, 6, 74]. Despite this assumption, prior studies often examined differences in gender and programming experience [11, 72, 74] with contradicting results. Only a few studies investigated the differences or relationships between computational thinking and cognitive load or spatial ability (e.g., [51, 66]).
A longitudinal approach was employed in the present study, where the same participants were observed over eight weeks to examine their patterns of interaction during BPM activities. Studies examining classroom interaction patterns often employed video data as their primary source of information. However, the main challenge of such observation is the rigorous task involved in coding a large amount of observation due to repeated viewings [76]. An alternate approach, which requires classifying classroom interactions and interpreting their educational relevance, was employed. Prior studies have shown the possibilities of generating sufficient streams of patterns using this observational approach. It is expected that our analysis will produce interaction patterns that support theories and practice and contribute to the ongoing refinement of classroom engagement. Figure 1 illustrates the overall data collection process.
Thirty-five, second-year-level computer science and computer education students (mean age: 19.8; male = 23, female = 12) from a research university in Nigeria were recruited. Across the literature, there is no numeric standard for sample size adequacy in studies involving video analysis. However, several authors recruited a minimum of 30 participants due to the strenuous task involved in repeated viewing and coding of video data [76, 77]. Recruitment of the participants was carried out using stratified random sampling. Stratification was done to ensure that the two departments (computer science and computer science education) have a representative sample, while randomization was done to ensure that each student from the two strata stands an equal chance of being selected [78].
The research benefit of this sampling technique is that individual characteristics are equally distributed by chance compared to other sampling strategies where such control is limited. Creswell [78] contends that random assignment prevents selection bias that may arise from the personal characteristics of the participants. Although the participants reported not being formally exposed to core programming concepts (evidence from preliminary self-report), initial analysis of their programming proficiency indicated that quite a few are expert and intermediate programmers while a large proportion were novice programmers. It should be noted in this research that the term “expertise” is conceived as a narrow, task-specific proficiency rather than genuine high-level professional expertise. The differences in expertise might be a result of knowledge gained from a first-year course, “Introduction to Problem-solving,” where all computer science and computer education students were exposed to theoretical computing and algorithmic thinking skills.
The study occurred in a formal classroom setting where the participants first interacted with a 20-min animated video that teaches computational concepts (see Figure 2) and later participated in a practical session that exposed them to Alice programming across eight weeks. As discussed in the literature section, learning computational thinking is crucial to solving programming and general problems. However, an important but less discussed issue is the concepts of CT to be taught in schools and the instructional environment to be used without significantly changing the existing curricula of computing education. Some authors have proposed important CT concepts to be taught. However, one useful proposal in the context of programming is that proposed by Roman-Gonzalez [49] based on Brennan and Resnick’s [40] framework, including sequencing, conditionals, looping, and functions.
Recognizing the importance of interactive multimedia instruction, a 3D animated video package that teaches CT concepts was developed using Alice (version 3). Although Alice is not commonly used in studies measuring CT, its selection was deliberate due to its unique advantages (for a review, see [22]). According to Yusuf
Across the eight weeks of the intervention, animated instructions took place in an open classroom while practical sessions took place in a computer laboratory setting where each participant interacted with the Alice software from their individual workstation. Information from the animated CT concepts and the practical tasks were delivered via an electronic board across the eight weeks. All programming activities were video-recorded to analyze the classroom interaction patterns. To obtain clear footage, three video cameras were positioned in front of the class—one at the center and two on the sides—facing the students. The cameras were mounted on tripod stands, which were adjusted to a height of 1.43 m above the ground. To ensure high-quality recordings, all cameras were set to a 0.39x wide angle, 3x digital zoom, and 4 K resolution.
The importance of obtaining explicit consent from participants during overt observation is acknowledged in this research. Therefore, before the intervention, an information sheet was sent to the participants, which included details about the video recordings. Additionally, this information was reiterated to the participants every day of the intervention. Participants were informed that their participation was voluntary and that they could withdraw at any time and request the deletion of their video data. Overall, the study was approved by an institutional review board.
Four programming tasks, related to the four CT concepts, namely, sequencing, conditionals, looping, and functions, were used for developing students’ CT skills in the Alice environment (see sample in S1–S4 in the Appendix). For each CT concept, the participants were presented with visual information that illustrates the behavior of a character in Alice and then instructed to write down the corresponding pseudocodes. Then they were instructed to implement the code in Alice to mimic the exact behavior of the character. For example, on sequencing, an Alice video containing a “human Biped” was presented to the participants. The Biped’s behavior includes rolling its head from left to right in sequence, with a delay of 0.25 s, and returning the head to its original position. Although participants performed their programming tasks on individual workstations, they could seek help from the instructor (researchers), colleagues, and the lesson notes given to them. On completion of each task, the participants shared their screens on the electronic board, where they received immediate feedback and recognition for a job well done.
Six data collection instruments were used in this study, including video recording devices that capture overall students’ programming activities, a CTt [49] that collects students’ CT skills based on concepts, and a CTS [43] that collects students’ CT skills based on perspectives. Others include cognitive load test, spatial ability test, and programming proficiency test (PPT).
The CTt is a 28-item performance test developed from the framework of Brennan and Resnick [40]. The test overly measures CT concepts and ignores CT practices and perspectives. Computational concepts addressed in the CTt include basic directions and sequences, loops, conditionals, and functions. Post-validation of the test [29] revealed a reliability coefficient of 0.793 for the entire sample: 0.721 for 5th and 6th grades, 0.762 for 7th and 8th grades, and 0.824 for 9th and 10th grades. Although the instrument was validated on middle-school students, it has been successfully applied to college and university students (e.g., [51, 79]), with substantial validity and reliability. Test samples are presented in Figures 3 and 4.
The CTS is a 29-item self-assessment test developed from the framework of the International Society for Technology in Education (ISTE, 2015). The instrument extensively measures CT skills from students’ perspectives. Constructs assessed include
CT concepts | CT perspectives | |
---|---|---|
Sequence | X | O |
Loops | X | O |
Conditionals | X | O |
Functions | X | O |
Creative thinking | O | X |
Algorithmic thinking | O | X |
Critical thinking | O | X |
Cooperativity | O | X |
Problem-solving | O | X |
Although critics have pointed out difficulties involved in quantifying cognitive overload and have suggested impossibilities for its measurement, researchers have employed several subjective ratings in measuring cognitive load [80–82]. These ratings have provided novel approaches and have been employed in several studies. For this reason, participants’ cognitive load was measured by a combination of two subjective ratings. The first item requires the participants to indicate their level of perceived difficulty (“How easy or difficult was it for you to work on this task”; 1 = very easy, 9 = very difficult; Kalyuga
The participants’ spatial ability was measured using the paper folding test (PFT) developed by Ekstrom
The participants’ programming proficiency was measured using a PPT based on Java programming. The test is a departmental test used for formative assessment in the 2019/2020 academic session. This test was used because it had been internally validated by two experts and subsequently used as a proficiency test for students. The test consists of 10 questions. The first five questions require syntax recognition; the next four questions are on output identification; the last question requires participants to write a simple program that displays even numbers from 2 to 100.
A pilot trial was conducted to administer the computational thinking, cognitive load, and spatial ability tests to 100 undergraduate computer science students (not among the study participants). Their participation in the pilot study was voluntary. Data collected from the pilot study was subjected to reliability analysis to establish the suitability of the adopted instruments for the study. Kuder–Richardson 20 (KR-20) was conducted on the data collected from the CTt and PFT. In contrast, Kuder-Richardson 21 (KR-21) was conducted on the data collected from the CTS and cognitive load test. KR-20 is a measure of reliability for tests with binary responses (i.e., tests with right or wrong responses) having varying levels of difficulty. KR-21 on the other hand is a reliability measure for identifying the internal consistency of tests with partial credit responses, such as the Likert scale. The reliability analysis revealed a reliability coefficient of 0.846 for the CTt, 0.855 for the CTS, 0.762 for the cognitive load test, and 0.824 for the spatial ability test. Details of the reliability analysis are presented in Table 3.
No. of items | Reliability | |
---|---|---|
Concept | 28 | 0.846 |
Sequence | 4 | 0.823 |
Loops | 8 | 0.847 |
Conditionals | 12 | 0.822 |
Functions | 4 | 0.893 |
Perspective | 29 | 0.855 |
Creative thinking | 8 | 0.895 |
Algorithmic thinking | 6 | 0.838 |
Critical thinking | 5 | 0.875 |
Cooperativity | 4 | 0.793 |
Problem-solving | 6 | 0.874 |
Cognitive load | 2 | 0.762 |
Spatial ability | 20 | 0.824 |
The participants reported no prior experience with Alice. For this reason, a 1-h seminar was conducted to expose them to the software before the intervention. The rationale behind this was to make the students familiar with the Alice environment and how characters can be added to and manipulated in the virtual world. The participants’ activities were recorded after informed consent had been obtained. For every intervention, the video footage of the animated instruction and the corresponding practical session were combined, yielding eight video files. Each video file lasts approximately 50 min. During the analysis of video recordings, a manual coding approach was performed by two experts. This involved carefully observing the video footage to identify cues indicative of different interaction patterns, including gaze directions and participants’ classroom activities. To evaluate the faces of pupils, the experts closely examined the video recordings to identify instances where participants displayed behaviors such as gaze direction, engagement in classroom activities, and disruptive incidents, which served as indicators of distinct interaction patterns. This process involved annotating specific moments in the video where interactions occurred and paying close attention to the activities and participants’ language.
Self-referential evaluation techniques were not employed in this study. This evaluation typically involves participants reflecting on their own behaviors or interactions. Although this approach can provide valuable insights, it was not a focus of the analysis. Instead, the emphasis was on objectively capturing and categorizing observable interaction patterns among participants without direct involvement or input from the participants themselves. Overall, using the manual coding approach, the authors ensured thorough and detailed analysis of the video data, allowing for comprehensive identification and categorization of interaction patterns among participants.
The data coding involves a systematic procedure. First, the video data was coded using a researcher-coded approach [77]. The coding commenced by identifying distinct patterns during BPM activities [84, 85]. Four interaction patterns were coded: learner–teacher, learner–content, learner–learner, and learner–distractor interactions. Within the learner–teacher interaction domain, participants can request the teacher’s attention, initiate a talk with the teacher, or ask for more explanation of difficult programming concepts. Within the learner–content interaction, the participants can watch the animation/perform programming tasks in Alice, or read a programmed lecture note given to them. Within the domain of learner–learner interaction, the participants have many options: they can seek help from a colleague, engage in pair or collaborative activities such as meaningful discussion and pair programming, or think and share their ideas with others. Within the domain of learner–distractor interaction, the participants can engage in non-classroom activities such as checking their phones, looking outside the classroom, or engaging in off-point discussion with others (here not categorized as learner–learner interaction). More explanation of the interaction indicators is provided in Table 4. As researchers, control over the participants regarding their preferred interaction was not exercised. Instead, flexible choices were given to the participants to transition or remain within their interaction state.
Interaction patterns | Numeric code | Indicator |
---|---|---|
Learner–teacher | 1 | Student calls for teacher’s attention, asks questions, or initiates talks with the teacher. |
Learner–content | 2 | Student watches the animation, takes notes, and works with Alice software to perform programming tasks. |
Learner–learner | 3 | Voluntarily seeks help from a colleague or engages in pair or collaborative activity such as meaningful discussion, peer programming, or think–pair–share. |
Learner–distractor | 4 | Engages in activities not related to the first three categories, e.g., interacting with phones or looking outside the classroom. |
The classification of interaction patterns is exhaustive because there are no other interaction indicators present in the literature. Each of the observed patterns was categorized in a mutually exclusive way. For example, a participant cannot interact with the content and at the same with his or her peers. Goldberg
As reported earlier, the coding of the video data was done by two independent coders. The experts were requested to follow the coding template strictly. The coders could also report the limitations of the coding template in the course of coding the interaction patterns. The degree of agreement between the coders was calculated using Cohen’s kappa [86] coefficient (𝜅) in IBM SPSS version 23. Cohen’s kappa measures the degree of agreement between raters. Values of 0.4–0.6 indicate fair reliability, 0.6–0.75 indicate good reliability, and values above 0.75 indicate excellent reliability. In this study, the degree of agreement between the two coders is 𝜅 = 0.87, suggesting the suitability of the behavioral features.
Before the interventions, the PPT was administered to the participants and their scores were recorded in an Excel worksheet. The first nine questions have a maximum score of 9 points while the 10th question has a maximum score of 6 points, leading to a total of a maximum of 15 points. After the interventions, the computational thinking, cognitive load, and spatial ability tests were administered via Google Forms. To establish independent responses, the participants were placed in a formal classroom setting while intensive supervision was conducted.
Various analytical tools were employed to address the RQs. First, a network analysis in Gephi software was employed to investigate the interaction patterns that exist during the interventions (RQ1). Network analysis measures social structures and relationships across different entities [87]. In a typical network graph, two important entities exist. The first entity includes the points (called nodes), and the second entity includes the lines that connect the points (called edges). In this study, nodes correspond to specific patterns of interaction observed among participants during the interventions. These nodes serve as the building blocks of the network, highlighting the various forms of engagement within the classroom setting. Edges, on the other hand, represent the connections or relationships between nodes in the network. They are depicted as lines that link pairs of nodes, indicating the presence of interactions or associations between them. Edges also signify the occurrence of transitions or shifts between different interaction patterns. For example, an edge between a learner–teacher interaction node and a learner–content interaction node indicates a transition from engaging with the teacher to interacting with instructional materials.
The network graph involves a multipartite network model that depicts the co-occurrence patterns among students during the learning process. The presence of 41 nodes was noted: teacher, animated instruction (Anim), Alice software (Alice), note, phone, away, and the 35 participants. Each participant served as a unique node that interacted with the teacher, animation, Alice software, notes, phone, outside classroom environment, and other students.
To address RQ2, the interaction patterns across different factors were estimated using the network model. The significance of these interactions was examined using the chi-square test of independence. Descriptive statistics, including mean and standard deviation (SD), was employed to measure the participants’ scores from the computational thinking tests (RQ3). Differences in the CT scores were examined using independent samples
Preliminary results (Table 5) show that, on average, the participants had a low cognitive load (
Cognitive load | Spatial ability | Proficiency level | |
---|---|---|---|
Min. | 1.00 | 3.00 | 4.00 |
Max. | 7.00 | 18.00 | 13.00 |
Mean | 2.89 | 10.74 | 4.28 |
Std. dev. | 0.76 | 4.22 | 0.89 |
Skewness | 1.14 | 1.10 | 1.17 |
The network graph (Figure 5) indicates the presence of 249 parallel edges showing the interaction between the participant nodes (numbered from P1 to P35) and non-participant nodes (denoted as Anim, Alice, Note, Teacher, Phone, Away). Analysis of the interaction patterns indicates that 36.95% of the interactions were with peers (learner–learner interaction), 34.54% were with contents (learner–content interaction), 16.87% were with distractors (learner–distractor interaction), and 11.65% were with the teacher (learner–teacher interaction). It should be noted from the graph that thick lines indicate frequently occurring interaction between the participant and non-participant nodes.
To explore the association between participants’ interaction patterns during BPM activities and their gender, cognitive load, spatial ability, and programming proficiency, chi-square test of independence was conducted (see Table 6). This analysis allowed the authors to examine the relationships between these variables and identify any significant associations.
Interaction patterns ( | |||||
---|---|---|---|---|---|
Variables | Learner–teacher | Learner–content | Learner–learner | Learner–distractor | Total |
Gender | |||||
Female | 248 | 3414 | 536 | 1658 | 5856 |
27% | 34.1% | 24% | 42.3% | 34.3% | |
Male | 672 | 6593 | 1696 | 2263 | 11,224 |
73% | 65.9% | 76% | 57.7% | 65.7% | |
Total | 920 | 10,007 | 2232 | 3921 | 17,080 |
100% | 100% | 100% | 100% | 100% | |
𝜒2 = 237.94, df = 3, Cramer’s | |||||
Cognitive load | |||||
High | 120 | 1240 | 144 | 1933 | 3437 |
13% | 12.4% | 6.5% | 49.3 | 20.1% | |
Moderate | 200 | 1660 | 504 | 1052 | 3416 |
21.7% | 16.6% | 22.6% | 26.8% | 20% | |
Low | 600 | 7107 | 1584 | 936 | 10,227 |
65.2% | 71% | 71% | 23.9% | 59.9% | |
Total | 920 | 10,007 | 2232 | 3921 | 17,080 |
100% | 100% | 100% | 100% | 100% | |
𝜒2 = 745.80, df = 6, Cramer’s | |||||
Spatial ability | |||||
High | 384 | 3807 | 1096 | 672 | 5959 |
41.7% | 38% | 49.1% | 17.1 | 34.9% | |
Moderate | 376 | 3696 | 848 | 1321 | 6241 |
40.9 | 36.9% | 38% | 33.7% | 36.5% | |
Low | 160 | 2504 | 288 | 1928 | 4880 |
17.4 | 25% | 12.9% | 49.2% | 28.6% | |
Total | 920 | 10,007 | 2232 | 3921 | 17,080 |
100% | 100% | 100% | 100% | 100% | |
𝜒2 = 392, df = 6, Cramer’s | |||||
Proficiency | |||||
Expert | 88 | 1660 | 72 | 1613 | 3433 |
9.6% | 16.6% | 3.2% | 41.1% | 20.1% | |
Intermediate | 64 | 1168 | 0 | 1200 | 2432 |
7% | 11.7% | 0% | 30.6% | 14.2 | |
Novice | 768 | 7179 | 2160 | 1108 | 11,215 |
83.5% | 71.7% | 96.8% | 28.3% | 65.7% | |
Total | 920 | 10,007 | 2232 | 3921 | 17,080 |
100% | 100% | 100% | 100% | 100% | |
𝜒2 = 243.51, df = 6, Cramer’s |
The results revealed that male students exhibited higher levels of interaction with the teacher (73%), content (65.9%), and peers (76%) compared to their female counterparts. Conversely, female students showed fewer interactions with distractors (42.3%). Although the association between gender and interaction patterns was significant (𝜒2 = 237.94,
Significant associations were observed between participants’ cognitive load levels and their interaction patterns (𝜒2 = 745.80,
Significant associations were observed between participants’ level of spatial ability and their interaction patterns (𝜒2 = 392,
The results also revealed significant associations between participants’ programming proficiency levels and their interaction patterns (𝜒2 = 243.51,
Analysis of the participants’ CT scores (Table 7) shows that they obtained a mean score of 17.39 (SD = 5.60) on CT concepts and 3.48 (SD = 0.99) on CT perspectives. Out of the possible 4 points in sequences and directions, the participants obtained an average of 1.94 points (SD = 1.39). From a possible 8 points in loops, the participants obtained an average of 4.9 points (SD = 2.15). In addition, from a possible 12 points in conditionals, the participants obtained a mean score of 7.95%. A mean score of 2.59 (SD = 0.79) was also obtained in functions from a possible 4 points. From CT perspectives, the participants reported a mean perception of 3.48 (SD = 0.99): 3.33 in creative thinking, 3.88 in algorithmic thinking, 3.70 in critical thinking, 2.86 in cooperativity, and 3.49 in problem-solving. These scores suggest that the participants had moderate computational thinking skills.
Dimension | Mean | Std. dev. | |
---|---|---|---|
Concept | 35 | 17.39 | 5.60 |
Sequences | 35 | 1.94 | 1.39 |
Loops | 35 | 4.90 | 2.15 |
Conditionals | 35 | 7.95 | 2.85 |
Functions | 35 | 2.59 | 0.79 |
Perspective | 35 | 3.48 | 0.99 |
Creative thinking | 35 | 3.33 | 0.85 |
Algorithmic thinking | 35 | 3.88 | 1.16 |
Critical thinking | 35 | 3.70 | 1.47 |
Cooperativity | 35 | 2.86 | 0.94 |
Problem-solving | 35 | 3.49 | 1.02 |
An independent samples
CT concepts | CT perspective | ||||||||
---|---|---|---|---|---|---|---|---|---|
Grouping | Sequence | Loops | Conditionals | Functions | Creative | Algorithm | Critical | Cooperative | Problem-solving |
Gender | 21.21∗∗∗ | 5.24∗∗ | 34.07∗∗∗ | 3.41∗∗ | 14.55∗∗∗ | 9.36∗∗∗ | 7.67∗∗∗ | 17.36∗∗∗ | 12.67∗∗∗ |
[1.37] | [2.14] | [2.75] | [0.78] | [0.84] | [1.15] | [1.47] | [0.93] | [1.01] | |
Cognitive load | 50.54∗∗∗ | 517.68∗∗∗ | 123.001∗∗∗ | 288.81∗∗∗ | 251.62∗∗∗ | 322.57∗∗∗ | 295.21∗∗∗ | 357.78∗∗∗ | 348.04∗∗∗ |
[0.01] | [0.05] | [0.14] | [0.03] | [0.03] | [0.04] | [0.03] | [0.04] | [0.04] | |
Spatial ability | 143.86∗∗∗ | 132∗∗∗ | 375∗∗∗ | 371∗∗∗ | 751.34∗∗∗ | 494.55∗∗∗ | 397.01∗∗∗ | 932.30∗∗∗ | 671.61∗∗∗ |
[0.02] | [0.14] | [0.30] | [0.26] | [0.08] | [0.06] | [0.04] | [0.09] | [0.07] | |
Proficiency | 544.42∗∗∗ | 346.46∗∗∗ | 574.04∗∗∗ | 361.74∗∗∗ | 127.82∗∗∗ | 744.08∗∗∗ | 561.92∗∗∗ | 185.15∗∗∗ | 115.63∗∗∗ |
[0.99] | [0.04] | [0.06] | [0.29] | [0.13] | [0.08] | [0.06] | [0.18] | [0.12] |
CT concepts | CT perspective | ||||||||
---|---|---|---|---|---|---|---|---|---|
Grouping | Sequence | Loops | Conditionals | Functions | Creative | Algorithm | Critical | Cooperative | Problem-solving |
M ± SD | M ± SD | M ± SD | M ± SD | M ± SD | M ± SD | M ± SD | M ± SD | M ± SD | |
Gender | |||||||||
Male | 1.78 ± 1.31 | 4.96 ± 2.16 | 8.47 ± 2.77 | 3.61 ± 0.79 | 4.39 ± 0.83 | 4.94 ± 1.10 | 4.76 ± 1.41 | 3.94 ± 0.94 | 4.56 ± 0.98 |
Female | 2.25 ± 1.47 | 3.78 ± 2.12 | 6.95 ± 2.72 | 1.56 ± 0.76 | 2.19 ± 0.85 | 2.76 ± 1.24 | 2.58 ± 1.57 | 1.68 ± 0.91 | 2.35 ± 1.07 |
Cognitive load | |||||||||
High | 1.91 ± 1.37 | 3.07 ± 2.53 | 5.74 ± 3.15 | 1.36 ± 0.63 | 2.22 ± 0.93 | 3.49 ± 1.33 | 2.25 ± 1.68 | 1.44 ± 0.87 | 2.10 ± 1.14 |
Moderate | 1.85 ± 1.35 | 4.27 ± 2.21 | 7.17 ± 2.75 | 2.41 ± 0.80 | 3.22 ± 0.93 | 3.63 ± 1.24 | 3.39 ± 1.56 | 2.75 ± 1.04 | 3.31 ± 1.12 |
Low | 2.20 ± 1.37 | 6.27 ± 1.93 | 8.67 ± 2.47 | 3.69 ± 0.78 | 4.42 ± 0.79 | 4.03 ± 1.04 | 4.89 ± 1.35 | 3.97 ± 0.89 | 4.63 ± 0.92 |
Spatial ability | |||||||||
High | 2.90 ± 1.46 | 5.53 ± 1.50 | 9.23 ± 1.70 | 2.81 ± 0.65 | 3.58 ± 0.74 | 4.15 ± 0.89 | 4.02 ± 1.18 | 3.17 ± 0.88 | 4.78 ± 0.87 |
Moderate | 1.75 ± 1.29 | 5.26 ± 1.97 | 8.65 ± 2.45 | 2.87 ± 0.67 | 3.37 ± 0.82 | 3.93 ± 1.11 | 3.76 ± 1.43 | 2.90 ± 0.91 | 3.54 ± 0.98 |
Low | 1.22 ± 1.38 | 3.66 ± 2.49 | 5.50 ± 2.91 | 1.96 ± 0.69 | 2.97 ± 0.86 | 3.47 ± 1.34 | 3.24 ± 1.70 | 2.42 ± 0.88 | 2.08 ± 1.15 |
Proficiency | |||||||||
Novice | 1.01 ± 0.66 | 4.68 ± 1.92 | 7.94 ± 2.55 | 2.31 ± 0.65 | 3.53 ± 0.77 | 4.09 ± 0.96 | 3.94 ± 1.25 | 3.12 ± 0.90 | 3.72 ± 0.86 |
Intermediate | 1.70 ± 0.01 | 4.85 ± 2.47 | 6.57 ± 3.23 | 2.96 ± 0.69 | 2.75 ± 0.80 | 3.24 ± 1.40 | 2.98 ± 1.77 | 2.13 ± 0.67 | 2.83 ± 1.15 |
Expert | 3.00 ± 0.01 | 5.82 ± 2.45 | 9.14 ± 3.09 | 3.39 ± 0.61 | 2.99 ± 0.82 | 3.54 ± 1.35 | 3.35 ± 1.71 | 2.39 ± 0.80 | 3.11 ± 1.13 |
To understand the factors predicting CT skills, OLR and BLR were conducted using the interaction patterns as factors (Table 10). Except for the learner–distractor interaction, all the interaction patterns significantly predict students’ CT skills with significant ORs. However, learner–content and learner–learner interaction had high chances of the prediction. For example, learner–learner interaction had a higher chance to predict students’ CT skills related to sequence and directions (OR = 2.20, 95% CI [3.41–3.87],
CT concepts | CT perspective | ||||||||
---|---|---|---|---|---|---|---|---|---|
Interaction patterns | Sequence | Loops | Conditionals | Functions | Creative | Algorithm | Critical | Cooperative | Problem-solving |
Learner–teacher | Est. = 1.97∗∗ | Est. = 3.03∗∗ | Est. = 6.59∗∗∗ | Est. = 1.67∗∗ | Est. = 1.93∗∗ | Est. = 16.53∗∗∗ | Est. = 15.34∗∗∗ | Est. = 18.86∗∗∗ | Est. = 15.06∗∗∗ |
OR = 1.66 | OR = 1.03 | OR = 3.91 | OR = 1.32 | OR = 1.77 | OR = 6.93 | OR = 6.40 | OR = 7.61 | OR = 5.96 | |
CI [1.78–2.16] | CI [2.82–3.24] | CI [6.27–6.92] | CL [1.42–1.91] | CI [1.86–2.01] | CI [11.85–21.22] | CL [10.67–20.01] | CI [15.25–22.48] | CI [12.42–17.71] | |
Learner–content | Est. = 1.28∗∗ | Est. = 8.30∗∗∗ | Est. = 7.96∗∗∗ | Est. = 1.68∗∗ | Est. = 2.89∗∗ | Est. = 48.23∗∗∗ | Est. = 47.61∗∗∗ | Est. = 18.50∗∗∗ | Est. = 30.50∗∗∗ |
OR = 1.43 | OR = 5.28 | OR = 4.20 | OR = 1.72 | OR = 2.73 | OR = 10.35 | OR = 10.09 | OR = 7.96 | OR = 8.78 | |
CI [1.20–1.36] | CI [8.02–8.62] | CI [7.64–8.28] | CI [1.53–1.82] | CI [2.79–2.99] | CI [41.89–54.58] | CI [41.65–53.57] | CI [14.91–22.13] | CI [26.86–34.13] | |
Learner–learner | Est. = 3.64∗∗∗ | Est. = 13.49∗∗∗ | Est. = 9.46∗∗∗ | Est. = 3.55∗∗ | Est. = 2.54∗∗ | Est. = 30.51∗∗∗ | Est. = 28.17∗∗∗ | Est. = 36.09∗∗∗ | Est. = 44.87∗∗∗ |
OR = 2.20 | OR = 6.13 | OR = 5.10 | OR = 2.16 | OR = 2.22 | OR = 8.01 | OR = 8.89 | OR = 8.13 | OR = 10.22 | |
CI [3.41–3.87] | CI [13.11–13.87] | CI [9.50–9.78] | CI [1.12–3.98] | CI [2.41–2.68] | CI [25.25–35.79] | CI [23.14–33.21] | CI [31.23–40.96] | CI [41.01–48.74] | |
Learner–distractor | Est. = −0.27∗ | Est. = −0.29∗ | Est. = −0.62∗ | Est. = −0.71∗ | Est. = −0.31∗ | Est. = −1.28∗∗ | Est. = −1.96∗∗ | Est. = −1.57∗∗ | Est. = −1.32∗∗ |
OR = 0.12 | OR = 0.65 | OR = 0.73 | OR = 0.82 | OR = 0.67 | OR = 0.24 | OR = 1.25 | OR = 1.50 | OR = 1.62 | |
CI [−1.56–1.02] | CI [−0.82–0.24] | CI [−1.66–0.43] | CI [−1.73–0.31] | CI [−1.73–1.11] | CI [−3.22–0.66] | CI [−4.16–0.25] | CI [−3.55–0.42] | CI [−3.18–0.54] |
The results indicate that learner–content interaction had higher chances of predicting CT skills related to creative thinking (OR = 2.73, 95% CI [2.79–2.99],
The present study investigated students’ interaction patterns during BPM activities and predicted their CT skills using the interaction patterns. Participants include 35 second-year computer science and computer education students whose classroom interactions were observed across eight interventions. The analysis revealed three important findings. First, learner–learner and learner–content interactions were the prevalent interaction patterns during BPM activities. Quite a few clusters of students engage in learner–teacher and learner–distractor patterns. From the perspective of the instructional quality model [88, 89], the prevalent incidence of learner–learner and learner–content interactions demonstrates the quality of BBPEs. The findings suggest the importance of these interaction patterns to students. At this point, the authors were compelled to believe that BBPEs have the potential to support self and collaborative programming activity. This position has been supported by considerable empirical evidence [90, 91]. However, the presence of students who often interact with distractors suggests that BPM does not always support meaningful learning. In a recent review, Yusuf and Noor [22] found that BPM is an important programming teaching tool, but it is not always effective in many experimental conditions.
The second finding revealed that interaction patterns during BPM activities differ significantly across gender, cognitive load, spatial ability, and programming proficiency. Students’ CT skills also differ across these factors. With regard to gender, the study found that male students interacted more with the teacher, content, and peers compared to their female counterparts. Male students also obtained higher CT scores except for the facets of sequences and directions. Recent and previous studies have revealed that male students engage more actively in classroom activities that require critical and creative thinking as well as problem-solving skills [92] while significant differences exist in favor of females for verbal fluency [93]. Nevertheless, there is no widely accepted empirical evidence indicating that interaction with BBPEs favors a particular gender category. Several studies have supported these findings (e.g., [28, 29, 94–96]), yet several others have reported no difference [74, 75, 97]. Concerning gender differences in CT skills, results are also mixed across the literature. For example, Niousha
A plausible explanation for the gender difference is rooted in gender theory [99]. This theory conceptualizes gender as categories of social expectations, roles, and behaviors. Although this view remains controversial in the literature, it does infer that some social activities are gender-stereotyped. Proponents of this view argue that disciplines such as mathematics and computing are perceived to be masculine and, therefore, females are more likely to struggle in these disciplines. In their previous study, Espino and Gonzalez [94] raised the issue of stereotypical gender preferences in the context of BPM and argued that the reported gender differences in computing and CT skills might be due to the compatibility of certain computing activities to a specific gender category. They further explained that males generally preferred programming constructs, which are reflected more in their CT scores.
However, as science educators who always understand and appreciate the need for gender inclusivity, the authors maintain a neutral stand on these justifications but believe that such differences might partly manifest from the lens of geographical context. For example, in Nigeria, enrollment into computing courses is generally skewed in favor of boys (85.87%), thereby creating a gender disparity [100]. While the authors believe that BPM is for everyone, it is worth noting that girls lagged far behind boys in computing courses in most African countries including Nigeria. For this reason, expanding programming tools to meet the expectations of the girl child remains the authors’ top priority.
For cognitive load, the study found that participants with low cognitive load interacted more with the teacher, contents, and peers compared to their counterparts with high and moderate cognitive load. These differences also hold for CT skills. The effect of cognitive load when working with BBPEs has been widely reported. In a recent review, Yusuf and Noor [22] reported the pedagogical effectiveness of BBPEs in promoting CT skills but highlighted its cognitive load effects. The authors reported that learners often struggle to interact with these tools due to the transient attribute associated with their virtual realities. Other studies also reported this problem when students are exposed to coding using BBPEs [51, 66].
Although cognitive load was found to affect CT skills and meaningful interaction in the BPM classroom, this effect is counterbalanced by students’ spatial ability. The study found evidence of the enhancer hypothesis [68], which assumes that high spatial ability learners benefit more from dynamic visualization compared to low spatial ability learners. Prior studies also reported evidence of the enhancer hypothesis in the context of BPM [67, 70]. Literature examining the effect of cognitive load and spatial ability on CT skills and BPM activities is scant. Few available studies found that CT is a factor of spatial ability and cognitive load [51, 66], thus supporting the findings.
The study also found some evidence of the expertise reversal effect because novice programmers significantly benefited from the BPM activities as indicated by their meaningful interaction than the intermediates and experts. Novice programmers were also reported to have higher scores on CT perspectives while expert programmers obtained high scores on CT concepts. The evidence of expertise reversal effect in this study is an indication that most BBPEs are not inclusive; they are largely beneficial to a typical expert category but become counterproductive to others and vice versa. In support of this finding, Spanjers
Explanations have been offered for the reason of the expertise reversal effect in most block-based environments. Kalyuga [27] explains that when information presented in these environments is familiar to the learners, they easily process its transience and ignore the content because of their potential and anticipation for higher mental objects. Prior research has also shown that measures to improve students’ learning outcomes using block-based environments as additional instructional guidance are often more beneficial to novices and counterproductive for expert learners, who do not need additional instructional guidance [102]. The experts have to reconcile their guidance in their schema with the additional guidance, which might further induce extraneous load.
It is quite surprising that expert programmers performed better in CT concepts than novice programmers despite having low meaningful interaction during the BPM activities. This difference is also explained in the context of expertise reversal effect. For instance, Aysolmaz and Reijers [26] explain that expertise reversal effect also occurs with a high level of expertise. In this situation, highly interactive programming environments, including those that were designed to enhance deep exploration of CT concepts, may overload the memory of novice learners and benefit the expert learners. This argument confirms the validity of the present finding, which indicates that participants with different types of expertise benefited more from the programming environment depending on the concerned CT dimension.
The third finding revealed that interaction patterns during BPM activities significantly predict CT skills. Specifically, two interaction patterns were found to significantly predict CT. Learner–learner interaction had a higher chance of predicting students’ CT skills related to sequence and directions, loops, conditionals, functions, cooperativity, and problem-solving skills. On the other hand, learner–content interaction had higher chances of predicting CT skills related to creative thinking, algorithmic thinking, and critical thinking. Although learner–distractor interaction had lower chances of predicting CT skills, such interaction posed a negative effect. There is a consensus in the literature claiming that students’ skills in BBPEs significantly correlate with their CT skills [5, 6, 74]. A practical implication of this finding points to the need for active participation of students in BPM activities. The finding also suggests the need for collaborative activities along with the integration of interactive animation for effective learning of programming and improvement of CT skills. To this end, although the acquisition of CT skills may differ across factors, engaging students in collaborative programming activities using interactive BPM would provide an enabling environment for students to acquire more CT skills.
Overall, the above findings suggest the crucial role of Alice in this research. Firstly, it provides a user-friendly interface that lowers the barriers for students new to programming. Its drag-and-drop functionality and visual representation of code enable learners to grasp fundamental programming concepts more easily, fostering engagement and participation. Secondly, by immersing students in programming tasks within the Alice environment, the authors were able to observe and analyze their interaction patterns comprehensively. From how they navigate instructional materials to how they collaborate with peers, Alice serves as the context for capturing these behaviors. This allows the authors to gain insights into the dynamics of student engagement and the impact of different interaction modalities on learning outcomes. Moreover, Alice facilitates the integration of computational thinking assessment into the study. Through the BPM activities, we assessed students’ CT skills along with factors contributing to these skills.
This study has shown the possibilities of classifying students’ interaction patterns using data obtained from time-dependent distribution. These interaction patterns are of significant importance because they appear to predict meaningful learning. By employing statistical models, the study found that students’ CT skills and their interaction with BBPEs are factors of gender, cognitive load, spatial ability, and programming proficiency. Despite these factors, interaction patterns during BPM activities predict students’ CT skills. The research has provided opportunities for educators to integrate BBPEs in learning programming and CT concepts. Although such integration is likely to occur with the help of strong educational policies, teachers are encouraged to cultivate the spirit of collaboration in students as collaborative activities in this research were found to predict CT skills more than other interaction patterns. To advance research on CT and BPM, it is imperative to always consider learners’ demographic profiles as they play important roles in meaningful learning. More importantly, students’ gender should be given more consideration due to its sensitivity to computing research particularly in sub-Saharan Africa, where girl-child education in computing is overly neglected.
While acknowledging the validity of the study findings, various limitations could affect such validity. First is the limited sample size. Although similar interaction patterns are expected in other studies that employ larger samples, more valid findings are predicted due to more diversity. However, larger samples could also pose difficulty in coding larger amounts of video data. Second, the video data was collected by a researcher-coded approach, which is prone to errors due to the strenuous task of coding a large amount of information. Although the researchers believe in the potential of technology to automatically identify interaction features using computer vision and deep learning algorithms, such an approach is expensive and could also lead to coding errors because some false faces might be captured as true faces, especially when there are blurred images. Despite these limitations, the study still retains substantial validity.
The authors declare no conflict of interest.
Source data (raw scientific data accompanying the research) for this article is available on Figshare: https://doi.org/10.5772/acrt.deposit.26317015
Supplementary Data
Written by
Article Type: Research Paper
Date of acceptance: June 2024
Date of publication: July 2024
DOI: 10.5772/acrt.36
Copyright: The Author(s), Licensee IntechOpen, License: CC BY 4.0
© The Author(s) 2024. Licensee IntechOpen. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
Impact of this article
10
Downloads
52
Views
Join us today!
Submit your Article