Collaborative versus individual problem solving in computational thinking through programming: A meta‐analysis

Abstract

AbstractComputational thinking (CT), which is a cognitive skill used to solve problems with computational solutions, has drawn increasing attention among researchers and practitioners due to the growing recognition of CT competence as a 21st century skill. Collaboration is commonly integrated into CT education to facilitate novice learning, but there is inadequate knowledge regarding the influences of collaboration in CT education. This meta‐analysis examined the overall effects on the cognitive, social and affective competencies of collaborative versus individual problem solving in CT through programming. We identified 33 publications involving 4717 learners, which allowed for 220 effect size comparisons. We found a medium effect size (Hedges' g = 0.562; p < 0.001) in favour of collaborative problem solving on cognitive learning outcomes and a small effect size (Hedges' g = 0.316; p < 0.01) on affective learning outcomes using a random‐effects model. Categorical moderator analysis revealed the moderating effects of educational level, programming environment, study duration, grouping method and group size. The competency model that was generated from the synthesized literature on collaborative computational problem solving yielded insights into the learning design and assessment of relevant activities. Practitioner notesWhat is already known about the topic? Although professional programming usually requires collaboration to accomplish common goals in the work environment, programming is predominantly taught on an individual basis in educational settings. Some intervention studies have reported inconsistent learning outcomes (positive and negative) from integrating collaborative problem solving (CPS) into programming education. There is a need for further scrutiny and exploration of the role of collaboration in the process of learning to program. What does this paper add? Our meta‐analysis showed that CPS is an effective pedagogical tool for developing students' cognitive, affective and social competencies. The moderator analysis revealed that CPS in programming has cognitive benefits irrespective of grouping method, group size and educational levels; however, there is a need for caution in regard to designing collaborative problem‐solving activities for different programming environments and durations. The model generated from the literature synthesis can yield insights into better assessing and designing collaborative problem‐solving activities. Implications for practice and/or policy? Integrating CPS into programming education can prepare students with 21st century skills. Social and affective competencies are crucial to students' learning capacity, which is an issue that needs to be better assessed. Future studies could look more rigorously into the effects of CPS in visual programming and robotics programming among younger age groups, such as primary or middle school students.