Observing students' behavior during problem solving: Determinants of success

Students working on their own, or in small groups, on tasks given to them by the teacher is a typical element of programming lessons. The task, if suitably complex, is meant to go beyond merely applying learned programming concepts to (computational) problem solving. There have been some attempts to classify problem solving processes and to identify how students can succeed or fail in such a scenario, however, comparatively little is known on this part of programming education when compared to other research on programming. This article presents the results of an exploratory, observational study of students in grade 7 (N = 28) that work on a task. The behavior was recorded and qualitatively coded using the CCPS model of Chevalier et al. [6] as a deductive starting point. During coding we inductively refined this model to better explain what happens during that time students spend coding. The newly introduced subcategories were then used to help explain whether students successfully solved the task. The results show that out of a number of variables, the distinction between a more arbitrary and a more purposeful behavior when interacting with the programming environment was the only significant predictor and the different groups are also picked up in a clustering of the various approaches of students.