Missing persons in the study of groups

Abstract

This research examined relationships between team-level variables with various patterns of individual non-response. In the first analysis, professional basketball teams (n = 389) were used to study the relationship between team cooperation and team performance. In the second analysis, the same teams were used to study the relationship between team experience and team performance. In the third analysis, professional baseball teams (n = 1984) were used to study the relationship between team experience and team performance. Individuals were deleted from the complete data sets to simulate three different types of non-response that might be encountered in organizational group or team research. In all three analyses, team-level relationships were attenuated as individual members were deleted randomly. Team-level relationships were also generally reduced as individuals were deleted as a function of their level of participation with the team. The overall amount of variance explained, however, showed a pronounced curvilinear effect. Namely, in all three analyses, the variance explained in team performance peaked when 30–40 per cent of the low-participation members were deleted. Finally, in the first analysis, relationships were also attenuated as the least cooperative members were deleted from the data set. The results demonstrate the need for researchers to understand the diverse effects of various types of non-response in team and group research. Copyright © 2005 John Wiley & Sons, Ltd.