Diversity: Trendy Buzzword or Scientific Imperative? Novel Insights Using Computer Science Techniques

Abstract

Inspired by the numerous social and economic benefits of diversity [51, 67, 87, 94, 121, 125], we analyze 1,045,401 multi-authored research papers spanning 24 fields of science to understand the relationship between collaboration success—measured in terms of citation count—and four types of diversity, reflecting (i) ethnicity, (ii) fields of expertise, (iii) affiliation and (iv) years of experience. For each type, we study group diversity (i.e., the heterogeneity of the authors in the paper) and individual diversity (i.e., the heterogeneity of a scientist's entire set of collaborators). Remarkably, we find that a field's ethnic diversity—the type least reflective of the authors' competence—is by far the strongest predictor of its impact (R2 is 0.79 and 0.46 for group and individual ethnic diversity, respectively). Moreover, we show that the predictive power of individual ethnic diversity becomes insignificant in the presence of group ethnic diversity, implying that a team's composition matters, regardless of whether or not the members have participated in ethnically-diverse teams. Finally, we show that regardless of publication year or number of authors, greater ethnic diversity is invariably associated with higher academic impact.

Date of Award	Dec 2017
Original language	American English