Increasingly diversity researchers call for further studies of group micro-processes and dynamics to understand the paradoxical effects of diversity on group performance. In this study, based on analyses of in-group, networked, homophilous interactions, we aim to explain further the effects of diversity on group performance in a parallel problem-solving task, both experimentally and computationally. We developed a 'whodunit' problem-solving experiment with 116 participants assigned to different-sized groups. Experimental results show that low diversity and high homophily levels are associated with lower performance while the effects of group size are not significant. To investigate this further, we developed an agent-based computational model (ABM), through which we inspected (a) the effect of different homophily and diversity strengths on performance, and (b) the robustness of such effects across group size variations. Overall, modeling results were consistent with our experimental findings, and revealed that the strength of homophily can drive diversity towards a positive or negative impact on performance. We also observed that increasing group size has a very marginal effect. Our work contributes to a better understanding of the implications of diversity in-group problem-solving by providing an integration of both experimental and computational perspectives in the analysis of group processes.
|Number of pages||26|
|Journal||Nonlinear Dynamics, Psychology, and Life Sciences|
|Publication status||Published - 1 Jan 2018|