COS 159-10 - Group composition for team-based learning: Random versus designed groups as it affects student performance and course perceptions

Thursday, August 10, 2017: 4:40 PM
B114, Oregon Convention Center
Anna Mosser1, Brian Gibbens1, David Matthes1 and J. D. Walker2, (1)Biology Teaching and Learning, University of Minnesota, (2)Center for Educational Innovation, University of Minnesota

Team-based learning is increasingly common in the undergraduate classroom, which raises the question of how the teams should be formed. Research suggests that group composition should be balanced with respect to characteristics such as student academic ability and gender, yet this idea needs further testing in a wider variety of undergraduate classrooms. In a large introductory course for biology majors, we examined the effects of random versus designed group composition on individual and team performance as well as individual perceptions of team work and the course environment. Designed teams were balanced in terms of gender, self-reported confidence in science ability, honors student status and international student status. Throughout the course, individuals took quizzes and exams and teams engaged in graded quizzes, ungraded short-term activities, and graded long-term projects. In addition to individual and team performance on summative assessments, data on perceptions of team work and course environment were collected via team peer evaluations and an individual survey. Data were analyzed using linear statistical models with a forward stepwise approach for variable inclusion. For the individual level analyses, the team number was included as a random effect to control for multiple students within the same group.


Across two sections of the course, with 18 teams each, groups of 7-9 students were either randomly assigned or designed. We found that group type (random vs. designed) had no effect on any measure of individual or team performance. Statistical model results did show that a team’s performance was determined by the mean previous college GPA of the team members and how many of it’s members completed an introductory course survey. Teams with more female students had marginally better project scores. Individual performance was significantly influenced by previous GPA, under-represented minority status, honors student status, and gender. It was also influenced by the mean GPA of their team and again how many of the team members completed the introductory survey. The effect of survey completion appears to be a proxy for student engagement in the course. Our results suggest that our method for designed groups is no different than random assignment, yet there are other aspects of our student population that we ought to consider in group composition. Data will be presented for a current course in which teams are balanced for the number of students completing the survey and of under-represented minority groups.