Expanding the horizons of educational pair programming: A methodological review of pair programming in computer science education research

Educators and researchers continue to explore the benefits, real or imagined, of implementing pair programming as part of the computer science pedagogy. Current reviews of computer science educational research practices do not focus on educational pair programming. This thesis presents a review of the research methods used in recent educational pair programming research. The primary purpose of this review is to inform the ongoing dialogue about and to provide evidence-based recommendations for improving educational pair programming research.;Replicating the design of a previous computer science education methodological review, this study inspected a sample of 108 articles from a population of 129 of articles related to educational pair programming published from 2000 to 2008. Articles were classified using a 112-variable taxonomy, identifying report elements, research methodology, research design, kinds of variables inspected, and statistical practices.;Major findings include several differences between the methodological characteristics of educational pair programming research when compared to general computer science education research, including: (a) an increased proportion of studies involving human participants, (b) a decreased proportion of quantitative methodologies, and (c) an increased proportion of controlled research designs. There exists some minor evidence that researchers affiliated with institutions in the United States are more likely than their counterparts outside of the United States to inspect only student attitudes and implement a posttest-only research design, and less likely to implement an experimental or quasi-experimental methodology.