Pair programming improves student retention, confidence, and program quality

Reviewer: Ann E. Fleury

This carefully designed and controlled study investigated the effects of pair programming on hundreds of students taking an introductory programming course. A significantly higher percentage of the students who had paired took the next programming course. Students who had paired expressed significantly greater confidence in their programming solutions, and produced significantly better programs. The detail of the study that I found most surprising was that lab attendance was not mandatory; I would have expected the use of required labs to provide a way of teaching and monitoring effective pair programming. The result that I found most important was that students who paired in the introductory course were more likely to take and pass the following data structures course, where all programming assignments were done individually. I strongly recommend this paper to anyone who teaches programming courses. The study described in this paper thoroughly justifies the claims made in its title. Online Computing Reviews Service

Reviewer: Silvia Teresita Acuña

There is abundant literature on experiences with extreme programming (XP) processes supporting XP's effectiveness, albeit applying a selected subset rather than all of its principles (see, for example, http://www.xpuniverse.com). This paper is another contribution to the effort of gathering empirical evidence about XP. The authors compare the effects of group and lone work on student performance, and the subsequent pursuit of computer science-related degrees, among both females and males. The authors of this paper ran a quasi-experiment to analyze gender-related student retention, pass rates, confidence, and program quality, considering both pair programming and lone programming in the development of software products. Students who programmed in pairs had higher completion and continuation rates in computer science-related majors than students who programmed independently (this continuation with the major is significant for both males and females). Students who paired in the introductory programming course were more likely to attempt the subsequent programming class, and more likely to pass than students who learned to program independently. Of the students who completed the class, the paired workers produced significantly better programs than the lone workers, and pair programming enhanced developer satisfaction and confidence. By doing a gender-related analysis, this paper contributes to the debate on collateral, though very interesting, XP-related issues, such as the idea that XP can help to increase women's representation in this field. Online Computing Reviews Service