The effect of training subjects in self-explanation strategies on problem solving success in computer programming

Elizabeth Susan Rezel, Marquette University

Abstract

This study analyzed the effectiveness of training participants in the use of self-explanation strategies on their subsequent problem solving success in the writing of computer programs and on the participants' use of the targeted self-explanation strategies during problem solving. The study was set in four university level computer programming classes using Microsoft Visual Basic 6.0 to teach introductory programming concepts. Participants were enrolled in an accelerated undergraduate bachelor's degree program in Management of Information Technology at a private Midwestern university. The classes were divided into a training group (N = 19) that received training in the targeted self-explanation strategies and a control group (N = 20) that received training in the use of think alouds as a data gathering technique but did not receive training in any learning strategies. All participants were involved in three problem solving experiences involving the writing of computer programs to solve specified problems. During these exercises participants were audio taped while thinking aloud. Programming success of the participants in the training group was compared to that of the control group to test the first hypothesis regarding whether students who received the training would have greater success than those who did not. Results did not show a significant difference between training group and the control group in programming success. Prior programming experience was a significant predictor of programming success. The audio taped protocols were analyzed to discover the answer to two further questions. The hypothesis that participants in the training group would use the strategies during problem solving more often than students in the control group was not upheld. The third hypothesis stated that participants who were successful would use the targeted strategies more often than participants who were not successful. One of the strategies, determine both the syntax and meaning of the code and the reasons behind the given form design in the example being studied, showed significant predictive value in relation to programming success. Possible reasons for the results were discussed as were suggestions for future research.

This paper has been withdrawn.