Demographic Diversity, Conflict, and Work Group Outcomes: An Intervening Process Theory

Research has suggested that, within a work group, diversity with respect to members' demographic backgrounds can have a powerful effect on both turnover from the group and on the group's performance on cognitive tasks (i.e., “thinking” tasks that involve generating plans or ideas, solving problems, or making decisions). While such diversity tends to increase turnover, its effects on cognitive task performance are more mixed, sometimes enhancing performance and sometimes impairing it. An understanding of how diversity leads to these outcomes may help managers enhance work group effectiveness. Thus, in this paper I develop a theoretical model to explain the turnover and mixed performance consequences of demographic diversity in work groups.The proposed model suggests that each demographic diversity variable (e.g., diversity with respect to age, gender, race, group tenure, organization tenure, education, or functional background) can be classified according to its level of visibility and its level of job-relatedness. Visibility is the extent to which the variable is easily observed by group members, and job-relatedness is the extent to which the variable directly shapes perspectives and skills related to cognitive tasks. The model then suggests that the visibility and job-relatedness of a diversity variable indirectly influence how much turnover and/or performance enhancement the variable yields. More specifically, the visibility and job-relatedness of a diversity variable influence the levels of affective (emotional) and substantive (task) conflict in the group, and the levels affective and substantive conflict, in turn, influence the amount of turnover from the group and the group's performance.After generating six research propositions based on the model, including three addressing demographic diversity variable- conflict linkages and three addressing conflict-turnover and conflict-performance linkages, I discuss boundary conditions of the model and offer recommendations for future research.