Task performance in small group settings: the role of group members’ self-efficacy and collective efficacy and group’s characteristics

Abstract

The present study extends the literature by investigating the relative salience of self- and collective efficacy in predicting group performance among early adolescents in Indonesia. A total of 435 early adolescents (mean age 11.70 years, 53% female) were randomly assigned to groups of three to four and completed three group tasks (task 1: puzzles; task 2: math; task 3: puzzles). Results indicated that collective efficacy was a more salient predictor of group performance than self-efficacy. Examination of the interaction effects suggested that the positive effects of collective efficacy on task performance were moderated by the levels of individual members’ self-efficacy, that is, when group members’ self-efficacy was high, task performance was lower. However, the debilitating effect disappears in the last task, where self-efficacy no longer undermines task performance when group members’ collective efficacy is high. Interestingly, heterogeneity of the group in terms of gender and ethnicity composition significantly predicted task performance of the group: groups comprising more boys and groups comprising members of different ethnic backgrounds had lower task performance. Findings are discussed in relation to their theoretical contributions and applied implications for classroom teaching and learning.

Keywords:

• Self-efficacy
• academic performance
• academic achievement
• collective efficacy
• Indonesia
• Asia

Notes

1. Following a reviewer’s comment, we conducted further analyses using groups (k = 116) instead of students (n = 435) at the unit of analysis and obtained similar findings. While all predicted relationships between variables were identical to the individual-level analyses, some variables such as our interaction term were no longer significant at p < .05 level. One plausible reason is due to the decrease in sample size from 435 to 116. In view of the number of variables we have included in our regression analyses (nine variables), the substantially smaller sample size may have decreased the power of the test, thus affecting the statistical significance of the findings. We recognised this as a limitation of the current study and have listed it as such under the limitations section.