Exploring Differences in Problem Solving with Data-Driven Approach Maps
Michael John Eagle, Tiffany M. Barnes · 2014
Understanding the differences in problem solving behavior between groups of students is quite challenging. We have mined the structure of interaction traces to discover different approaches to solving logic problems. In a prior study, sig-nificant differences in performance and tutor retention were found between two groups of students, one group with ac-cess to hints, and one without. The Approach Maps we have derived help us discover differences in how students in each group explore the possible solution space for each problem. We summarize our findings across several logic problems, and present in-depth Approach analyses for two logic problems that seem to influence future performance in the tutor for each group. Our results show that the students in the hint group approach the two problems in statistically and practically different ways, when compared to the control group. Our data-driven approach maps offer a novel way to compare behaviors between groups, while providing insight into the ways students solve problems.