Two experiments using learning rate to evaluate an experimenter developed tool for splay trees

Abstract

We conducted two experiments evaluating Sketchmate, a tool used to teach the splay tree data structure and its algorithms. Learning and learning rates were compared across two groups, one using Sketchmate and the other using paperand-pencil on practice problems. Results from Experiment I showed that when students used Sketchmate with minimal feedback, there were no significant differences across learning, time spent learning, or learning rate. Experiment II used a version of Sketchmate that provided richer feedback. Results showed similar learning but Sketchmate took significantly less time. Thus when feedback was added, learning rates were significantly greater relative to the paper-andpencil condition. Discussion focuses on measuring learning rates when evaluating instructional tools.