Novice problem solvers often fail to recognize structural similarities between problems they know and a new problem because they are more concerned with the surface features rather than the structural features of the problem. The surface features are the story line of the problem whereas the structural features involve the relationships between objects in the problem. We used an online technology to investigate whether students' self-explanations and reception of feedback influenced recognition of similarities between surface features and structural features of statistical problems. On average students in our experimental group gave 12 comments in the form of self-explanation and peer feedback. Students in this Feedback group showed statistically significantly higher problem scores over the No-Feedback group; however, the mean self-efficacy scores were lower for both groups after the problem solving experiment. The incongruence in problem scores with self-efficacy scores was attributed to students' over-rating of their abilities prior to actually performing the tasks. This process of calibration was identified as an explanation for the statistically significant positive correlation between problem solving scores and post self efficacy scores for the Feedback group (p<.01).
The CAUSE Research Group is supported in part by a member initiative grant from the American Statistical Association’s Section on Statistics and Data Science Education