Gender diversity and decision-making in teams

Abstract

We study how gender composition affects the nature and quality of team decisions using a novel dataset on collegial rulings from Italian criminal courts. Exploiting the quasi-random assignment of judges and cases to panels, we find that all-female teams convict more often and issue the most appeal-proof rulings (unappealed or upheld on appeal). All-male teams convict less frequently and have the lowest appeal-proof rate, while mixed-gender teams resemble all-male teams in conviction rates but nearly match the conviction accuracy of all-female teams. Higher conviction rates and fewer reversals for all-female teams rule out differences in leniency across gender compositions as the main explanation, while mixed teams’ lower conviction rate relative to all-female teams despite similar accuracy points to coordination frictions. We formalize these patterns in a model of diagnostic decision-making where judges differ in leniency and diagnostic skills shaped by effort or ability. In line with the reduced-form evidence, model estimates and supplementary analyses of written opinions, deliberation times, and entry exam scores indicate that effort differences are the main channel and that frictions in mixed-gender teams reduce performance relative to all-female teams, without eliminating their accuracy edge over all-male teams.