What Does Representation Represent? Disaggregating Demographic and Behavioral Components

ABSTRACT Representative bureaucracy research has often linked demographic representation directly to favorable outcomes, inferring active representation post hoc while rarely separating the behavioral component presumed to accompany such associations from the representation variable itself. This study addresses such gap by drawing on gender differences in leadership literature to operationalize supportive leadership as one behavioral component associated with female supervisory representation and examine the constituent elements of what has been interpreted as a representation effect. Using gender pay inequality across US federal agencies as the empirical context, the findings show that while both female supervisory representation and supportive leadership are each positively associated with nonsupervisory salaries when examined separately, the association attributed to representation loses significance once both are modeled together. These findings do not suggest that representation is unimportant; rather, they indicate that prior representation effects may partially reflect a behavioral component that has not been separately identified.