Three studies explored the relationship between entitativity and cohesion and tested three predictions: (1) entitativity and cohesion are related but distinct concepts; (2) entitativity and cohesion are differentially driven by the same underlying cue properties; and (3) entitativity and cohesion are differentially related to performance behaviors, with cohesion serving as a better predictor of performance.

In Study 1, participants sorted a list of 40 groups based on one of three guiding principles: (1) the definition of entitativity, (2) the definition of cohesion, (3) their intuitive concept of cohesion. Results indicated that participants sorted the groups into more similar patterns when guided by the definition of entitativity and their intuitive concept of cohesion. These sorting patterns differed from the pattern they used when sorting the groups based on the definition of cohesion. These results support the hypothesis that entitativity and cohesion are distinct concepts, but they suggest that participants intuitively consider cohesion to be similar to entitativity.

In Study 2, intramural basketball teams were observed twice during a season. These observations were used to develop behavioral categories of team member interactions. Team members also provided self-report data on team entitativity, cohesion, and the presence of cue properties. In support of Prediction 1, entitativity was moderately correlated to cohesion, which was divided into social and task cohesion subtypes. In support of Prediction 2, entitativity and two cohesions were found to share different underlying cue properties. These relationships changed across time. Disconfirming Prediction 3, entitativity, cohesion, and the cue properties all predicted team performance, while seven of the eight behavioral categories failed to predict performance. Social cohesion was only marginally more related to the behavioral categories than entitativity.

In Study 3, the cue properties were manipulated via vignettes about basketball teams. Participants read the vignettes and then rated the teams' entitativity and cohesion. In support of Predictions 1 and 2, manipulating the cue properties produced a greater response in team entitativity ratings than it did in team cohesion ratings, although the ratings did move in concert.