Abstract EN:

This dissertation explores the effects of team diversity on performance variability and its organizational consequences. First, I show that organizational theory has mostly assumed variability to be detrimental to organizations, and that it is not a common dependent variable. However, effects on performance variability matter: with increasing performance variability, extreme performance becomes more likely, and—at some point—such extremes are entirely determined by variability, making predictions based on mean effect only to be flawed. I review the consequences of this logic on various organizational theories. Second, I hypothesize that team diversity has a U-shaped effect on performance variability, with both very diverse and very homogeneous teams exhibiting high performance variability, and teams of middle-range diversity limited performance variability. The reanalysis of the performance of 35 teams engaged in a business simulation confirms those effects on performance variability measured at the team population level. It illustrates how effects on variability better predicts extreme outcomes—here finishing in the top 10% of the population—and how team diversity can have such an effect. Third, I explore team performance variability at unit level, distinguishing along-time from across-tasks variability for each team. I hypothesize that diversity increases the former and decreases the latter. A longitudinal dataset of 200 student teams performing actual group-work confirms such contradictory effects. This study identifies the two hazards—social and informational—underlying the diversity-performance relationship. This dissertation contributes to team composition literature by clarifying the various effects of diversity on performance variability. For practitioners, it identifies the low risk of mid-level diversity, from higher risk at the highest and the lowest levels of diversity, driven by organizational hazards