The gender diversity–performance relationship in services and manufacturing organizations

Abstract

Empirical findings on the link between gender diversity and performance have been inconsistent. This paper presents three competing predictions of the organizational gender diversity–performance relationship: a positive linear prediction derived from the resource-based view of the firm, a negative linear prediction derived from self-categorization and social identity theories, and an inverted U-shaped curvilinear prediction derived from the integration of the resource-based view of the firm with self-categorization and social identity theories. This paper also proposes a moderating effect of industry type (services vs. manufacturing) on the gender diversity–performance relationship. The predictions were tested in publicly listed Australian organizations using archival quantitative data with a longitudinal research design. The results show partial support for the positive linear and inverted U-shaped curvilinear predictions as well as for the proposed moderating effect of industry type. The curvilinear relationship indicates that different proportions of organizational gender diversity have different effects on organizational performance, which may be attributed to different dynamics as suggested by the resource-based view and self-categorization and social identity theories. The results help reconcile the inconsistent findings of past research that focused on the linear gender diversity–performance relationship. The findings also show that industry context can strengthen or weaken the effects of organizational gender diversity on performance.

Keywords:

• competing predictions
• gender diversity
• manufacturing industry
• organizational performance
• services industry

Acknowledgments

We thank Prof. Christina Cregan and Prof. Prashant Bordia for their invaluable feedback on an earlier version of this paper.

Notes

1. Becker (2005) has cautioned that incorrect inferences may be drawn in the presence of control variables and recommended that researchers report their primary results both with and without control variables. Therefore, we repeated the regression analyses reported in Tables 2 and 3 without control variables. There was no change in the substantive results. In the absence of control variables, Model 4 in Table 2 displays significant gender diversity 2002² and gender diversity 2002² × industry type effects, whereas the corresponding effects in Model 4, Table 3 remain nonsignificant.