Fig. 1

Intersectional effects of race and gender in the space traditionally defined by the semantic differential. Effects of race or gender labels alone were evident in distinct primary dimensions (evaluative and potency, respectively). Thickness of bars for each of these effects represents two standard errors of the means derived from linear mixed effects regression models comparing each single label to the baseline condition (“none”: no mention of race or gender). Circular symbols (outer radii closely approximate one standard error in each dimension) are shown for cases where both gender and race were specified. Locations of verbal labels show two-dimensional position in the semantic space with respect to the baseline condition