Analyzing data from individuals in groups: The past, the present, and the future

Research on group processes published in the social psychological literature over a 15-year period was examined to evaluate the impact of published critiques and recommendations by methodologists concerning the effects of nonindependence of observations on analyses of data from individuals in groups. The examination revealed modest progress in dealing with the nonindependence problem. Although more group researchers acknowledge the nonindependence problem than ever before, there is little evidence that they are using optimal strategies to address the problem. The most prevalent strategy is to use the group as the unit of analysis even when the data are gathered from individual group members and the research questions concern the effects of the group on its members. Although hierarchical linear modeling is a superior strategy for analyzing data from members of groups, there is no evidence of its use in the published literature during the 15 years considered.

Duke Scholars

Author Rick Hoyle Psychology & Neuroscience