Data Aggregation in Team Research: Theoretical Considerations and Practical Recommendations

Abstract

Studying teams comes with notable theoretical and analytical challenges. One major challenge is data aggregation, the process of bringing a variable that was measured at the individual level to the team level of analysis. We provide an accessible resource about data aggregation that integrates theoretical and empirical insights, dispels common misconceptions, and provides practical guidelines for researchers. We stress that decisions about aggregation should be based on theoretical considerations about the nature of variables and their interrelations. Based on the logic of latent variable models, we distinguish reflective and formative team constructs. For reflective team constructs, we discuss issues of within-team agreement and the notion of the conceptual mean. For formative team constructs, we distinguish between well-defined and ill-defined constructs, and offer guidelines for their aggregation. This discussion results in twelve concrete guidelines that can be used by researchers, reviewers, and editors in the design and evaluation of team studies.

Plain Language Summary Title

Creating team properties out of individual data: Guidelines for researchers.

Plain Language Summary Title

Plain Language Summary: Researchers who study teams, often want to measure a property of the team. Often, such a property (for example: team cohesion) cannot be directly observed, and researchers therefore ask team members to provide information about this (for example through survey questions). In those cases, researchers have to create one team variable out of the information provided by multiple team members, and this is called data aggregation. This is a very common process in team research, but it is also tricky. In this paper, we make an important distinction between reflective team constructs and formative team constructs. Reflective team constructs are properties of a team for which it is assumed that team members can experience them and that they experience them all in the same way (for example: team cohesion). Formative team constructs do not assume this, but are constructed by the researcher as a summary of individual properties. An example is age diversity, which is directly computed by the researcher based on the ages of the different members. The assumptions for these two types of constructs are very different, and this has implications for how researchers should deal with them. This paper provides a discussion of this and guidelines for researchers about what they should do.