Abstract
Given the focus of the media on female candidate appearance in the 2008 presidential election, this research investigates the effects on voter evaluations of calling attention to female candidate attractiveness. The current research hypothesizes that pointing out candidate attractiveness likely has a negative effect on subsequent evaluations and reports of vote likelihood, particularly for female candidates. Role congruity theory, which argues that evidence of prejudice against female leaders is the result of a discrepancy between people’s stereotypes of women and their stereotypes of leaders, provides an explanation for these findings. This study establishes the negative influence of calling attention to a candidate’s attractiveness. In particular, a female candidate described as attractive are evaluated more negatively than a male candidate described as attractive and compared to male and female candidates, who are not described as attractive.
Notes
Physical attractiveness conveyed through photographs or other visual materials at times is not always beneficial. There is a line of research investigating the beauty is beastly effect with mixed results. The beauty is beastly effect emerges when physically attractive women are being considered for masculine-stereotyped jobs including managerial positions, for which physically attractive women are rated less favorably than unattractive women (Heilman and Saruwatari Citation1979). A review of the physical attractiveness literature argues that on a whole attractiveness is always advantageous for men, but only for women when applying for feminine jobs (Morrow Citation1990). Johnson et al. (Citation2010) find that attractiveness is most beneficial when sex and a sex-typed job match meaning that attractive women are advantaged when seeking feminine jobs and attractive men are advantaged when seeking masculine jobs. Therefore, the beauty is beastly effect for women may only arise when the position being sought is a masculine job for which physical attractiveness is viewed as irrelevant. Two meta-analyses, however, conclude that the beauty is beastly effect is not robust with attractiveness being advantageous for men and women by having a consistent and moderately positive effect on subsequent evaluations (Eagly et al. Citation1991; Hosoda, Stone-Romero, and Coats Citation2003).
Originally, we excluded a single item from the Benevolent Sexism scale to improve the Cronbach’s alpha. The excluded item, “in a disaster, women ought not necessarily to be rescued before men”, may have been difficult for students to answer because of the wording. Unfortunately, item analysis does not indicated that the reliability for this measure can be improved by dropping additional items. Factor analysis revealed that two of the items were not loading well on the same factor as the other 8 items. Therefore, the analysis includes a scale constructed of 8 items. The results do not vary when we include 10 item scale or the 11 item scale.
Please see Mutz (Citation2011) for an in-depth discussion of why ANOVA rather than OLS regression is the appropriate method for analysis when analyzing experimental data with interactions and for why control variables are not necessary for analysis of experimental data with randomization into conditions. Nevertheless, OLS regression does not alter the results nor does the inclusion of control variables such as ideological perceptions of the candidates, participant sex, participant party identification, and participant race, all of which are not significant.
Although the STATA package does not support ANOVA, we investigated whether Benevolent Sexism or Hostile Sexism are mediators using the Hicks and Tingley (Citation2011) mediation package (for more information on mediational analyses see Imai et al. Citation2011). The percent mediated for Benevolent Sexism is 0.00% and for Hostile Sexism is 0.02%.
For a discussion of marginal effects and their superiority for understanding interactions please see Brambor, Clark, and Golder (Citation2006).