Sex and ideology: liberal and conservative responses to scandal

ABSTRACT

Research finds citizens are less likely to penalize politicians implicated in sex scandals compared to corruption. Still, observational data reveals that some politicians have better luck surviving sex scandals than others. Do voters punish politicians for sex scandals? We argue yes – some do. Whereas liberals are inclined to view sex scandals as personal matters – unrelated to a politician’s job performance – conservatives are more likely to view sex scandals as moral outrages that disregard traditional values and threaten the social order. Conservatives are thus less forgiving of sex scandals than liberals, especially when women politicians are implicated. Using evidence from a survey experiment in the US designed to isolate the effect of scandal type (corruption vs. sex) and candidate sex, we investigate heterogeneous effects by political ideology. We find that liberals tend to be forgiving of sex scandals, but not corruption. Conservatives, by contrast, punish men’s sex scandals on par with men’s involvement in corruption. And, conservatives assign women a penalty bonus for either type of scandal. That is, they are significantly more likely than liberals to punish women for involvement in either type of scandal – sex or corruption.

Acknowledgments

We confirm that IRB approval was granted for the survey used in this manuscript. We would like to thank Emily Beaulieu Bacchus. Conversations with her about this work, as well as our other collaborative work, brought valuable insights to this project. We also thank Catherine Reyes-Housholder for helpful comments, as well as anonymous reviewers. This article benefitted from a pilot grant from the University of Kentucky College of Arts and Sciences.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Studies using MTurk consistently replicate results from other representative samples (Mullinix et al. 2015).

2 We used Qualtrics’ “randomizer” function to assign treatments. See Appendix Table B1 for sample characteristics. Table B2 shows a multinomial logit predicting treatment assignment.

3 See Appendix Table B3 for the exact wording of each vignette.

4 We exclude partisan treatments because co-partisans are less likely to suspect (Barnes and Beaulieu 2014) or punish (Anduzia, Gallego, and Muñoz 2013; Cossette and Craig 2019) corruption among co-partisans. Still saying “your representative” might prime readers to think about their own representatives’ partisanship. Since this possibility would be applied equally across treatment groups, thinking about “your representative” when you have shared partisanship would bias the results toward the null.

5 See Table A2–A3 for OLS models. We estimate generalized ordered logistic models in Tables A4–A5 to account for the possibility we violate the proportional odds assumption: see Tables A4–5, Figures A1–A2.

6 https://wapo.st/1ixpBGM?tid=ss_tw&utm_term=.4ec6bf726ff9

Additional information

Funding

This article benefitted from a pilot grant from the University of Kentucky College of Arts and Sciences.