Improving student learning outcomes with pair programming

This paper presents ongoing research into the use of mental model consistency (MMC) to produce more effective student programming pairs. Previous studies have found that pair programming is highly useful in improving students' enjoyment of programming as well as improving the retention rates of students enrolled in computer science programs. However, existing research provides little support that pair programming actually benefits student learning in terms of improved test or exam scores. This research focuses on evaluating the use of MMC-based student pairs to increase student performance in introductory programming courses. Empirical studies were conducted over two semesters to determine if pairings based on different levels of MMC produced more effective pairs. The results from this study indicate that MMC is a good predictor of success in a course when using pair programming and that students who migrate towards greater consistency tend to do better than those who do not migrate. However, the current results do not support that pairs based on any combination of mental models are more effective than others. Still, the authors of this paper feel that MMC is a valuable method and that if combined with other techniques to produce more compatible pairs, may yet produce substantial results. Other potential uses for MCC are also discussed.