Collaborative testing in statistics: Group interaction, anxiety, and class performance

Abstract

Seventy-one students in two community college Statistics for the Social Sciences classes took six exams either individually or collaboratively. Assignment to test condition was randomly determined for each exam. Scores on collaboratively-taken exams were significantly higher than those for individually-taken exams, particularly for students with low GPAs and high test anxiety. Women's, but not men's, performances on the mid-term and final exams was related to the quality and quantity of their collaborative interactions. Thus, examining both quantity and quality of collaboration adds to our understanding of the underlying mechanisms of collaborative testing.