This study investigates differences between collaboration methods in two summer enrichment classes for students entering the sixth grade. In one treatment, students used pair programming. In the other treatment, students engaged in frequent collaboration, but worked on their own computer. Students in the two treatments did not differ significantly in their performance on daily quizzes or responses to attitudinal survey questions. However, the students who worked on their own computer completed exercises more quickly than those using pair programming. This study compares two learning environments with high levels of collaboration to isolate aspects of pair programming that are and are not responsible for the reported success of educational research focused on pair programming. This study expands our understanding of pair programming by moving beyond simplistic comparisons of learning environments with and without collaboration and by extending pair programming research to elementary school students. (Contains 6 figures and 8 tables.)