Measurement of the effect of member performances on team outcomes in two-stage exams

Abstract: Two-stage exams have been used in physics classrooms, yet no empirical study has examined the effect of individual member performances on team performance. Using direct measures of individual and team performances in two-stage physics exams, we study the effect of member performances on team outcomes. Our data reveal three results of interest. First, either the best or the mean of member performances significantly affects team performance, whereas the standard deviation of member performances has a weak effect. Second, member performances explain less than half the variance in team performances, indicating that team outcomes in two-stage exams are not determined simply by individual performances. Third, cross-validation shows that the best predictor of team performance is the best member's performance, highlighting the need to form balanced teams with high-scoring individuals distributed evenly throughout teams before team exams are conducted. As the first empirical analysis built on the input-process-output framework, this study provides researchers of two-stage exams with insight into how to investigate factors that can make team problem-solving more productive in the team round.