![Sample our Politics & International Relations journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5947/TOC-Subject-Sample-2021-Politics-International-Relations.jpg"})
ABSTRACT
History tells us that elected leaders who are tainted by scandal often pay a political price for their behavior. In the past few years, such allegations of sexual harassment have engulfed a large number of political (as well as entertainment, business, and even academic) figures. Many were forced to resign their positions, while others chose to end their campaigns for election or reelection. While the great majority of harassment victims are women, there are a few instances where the sex roles have been reversed – and with more women running for and winning public office (and thereby gaining positions of power and authority), it is possible that we will see more such role reversals in the future. Our study uses data from an internet-based survey of registered voters to examine citizens’ attitudes about sexual harassment and the extent to which those attitudes shape their reactions when allegations of harassment are made against a fictional member of Congress. With an innovative experimental design, we will examine whether reactions vary with (a) either the target’s or the voter’s gender or (b) the former’s response to the allegations of sexual misconduct made against him/her (denial, apology, counterframe).
Acknowledgments
An earlier version of this paper was presented at the 2019 Annual Meetings of the Western Political Science Association. We would like to thank Patricia Mitchell for her research assistance in the early stages of this project.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Supplementary material
Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/1554477X.2022.2024045
Notes
2. In 2020, presidential candidate Joe Biden was accused by a former aide of sexually assaulting her in the early 1990s (Bennett and Lerer Citation2020). Biden unequivocally denied the charges, later events raised questions about the credibility of his accuser (Phillips Citation2020), and the issue never posed a serious threat to his campaign.
3. The final vote was 52–48, split largely along party lines.
4. Not everyone agrees about the types of behavior that constitute sexual harassment (Bowman Citation2018; Craig and Cossette Citation2020; Kahn Citation2017). Also see https://www.barna.com/research/behaviors-americans-count-as-harassment/; https://www.economist.com/graphic-detail/2017/11/17/over-friendly-or-sexual-harassment-it-depends-partly-on-whom-you-ask.
5. For a detailed discussion of the scandal literature, see Cossette and Craig (2020).
6. Newmark and colleagues found that men, incumbents holding “safe” seats, those who have the support of their spouses and who are in favorable political environments are more likely to “survive a political scandal”; in contrast, early-breaking scandals, severe cases, those instances where an apology is required, and sexual misconduct decrease the probability of survival. Further, it appears that the electoral impact of scandal is often temporary for incumbents who survive an initial challenge (Praino, Stockemer, and Moscardelli 2013).
7. In this case, the probability of participants voting for a corrupt candidate varied with the former’s issue preferences (Vietnam) and belief in the importance of honesty in government.
8. The idea that voters today are willing to forgive such “political” misbehaviors as election fraud and misuse of campaign donations has been challenged by Wood and Grose (Citation2018).
9. See Schneider and Bos (Citation2014) for a different take on the stereotypical uniqueness of “female politicians” as a subgroup of women more broadly.
10. An exception is Courtemanche and Green (Citation2020). Also see Barnes, Beaulieu, and Saxton (Citation2020), who observed that some voters – namely, those with attitudes reflecting “hostile sexism” (Glick and Fiske Citation2001) – were more likely to punish women than men for sexual, but not corruption-related, transgressions.
11. Benoit (Citation2016, 7) is critical of experimental research on image repair, in part because “most studies test but a few … of the image repair strategies, making a complete understanding of image repair discourse impossible.” One study compares excuse and justification, another looks at the effects of excuse, justification, and denial, while a third examines excuse, justification, concession, and apology … and so on (p. 8). We acknowledge that our data capture only a portion of allegation-response dynamic that typically occurs in real life, but the rhetoric on both sides (accuser and target) is derived from actual cases of sexual harassment that have been reported in the news over the past few years. Even taking the inherent limitations of experimental designs into account, we believe that the analysis presented here will help us to understand how the public truly feels about the sexual (mis)behavior in which far too many of their political leaders have engaged.
12. Data were provided by Qualtrics (https://www.qualtrics.com) from panels consisting of millions of pre-screened individuals who have been recruited to participate in a variety of research studies. Respondents for our study were drawn from a national panel and self-identified as registered voters. The sample was collected to meet demographic quotas reflecting age, gender, race/ethnicity, and education level of the US population as reported by the US Census. Although Qualtrics’s panel is quite diverse, we make no claim that it is representative of registered voters nationwide.
13. The structure of our experimental design, displayed visually in our supporting materials as Figure A1, is as follows: 1 (allegation) X 4 (party affiliation + gender combinations) X 3 (response/account type). The randomization process appears to have been successful, as no statistically significant pre-exposure differences were observed among members of the various treatment groups with regard to demographics, partisanship, ideological self-identification, issue positions, or baseline candidate preferences. Thus, if differences are found after respondents read a paired attack and response, we can be confident that these were driven by exposure to the experimental stimulus.
14. Each candidate’s party affiliation and status as either challenger or incumbent was specified, but otherwise the biosketches (see online Appendix C) were crafted in such a way as to ensure that the two portrayals were essentially equivalent.
15. The content of both allegations and responses was based on actual cases reported in the news, mostly during the period after the Weinstein scandal broke in October 2017.
16. Our survey included a validity check at each stage of the experiment (post-allegation and post-response) asking respondents to identify the content of the news story they had just read, i.e., the nature of the charge that was being made against the representative (inappropriate sexual behavior) and the accused’s response (denial, apology, counterimaging). Anyone who answered incorrectly – and very few did – was dropped from the survey.
17. Also see current figures provided by the Center for American Women and Politics at Rutgers University (https://www.cawp.rutgers.edu/).
18. This was a question asking respondents to indicate which of two statements was closer to their own opinion (the alternative was that “I am willing to tolerate an elected official who has been accused of sexual harassment if he or she has the potential to pass laws that I support”); see online Appendix B for exact wording and format.
19. In a study of negative campaign ads, however, Craig and Rippere (Citation2016) found that denying the opponent’s charges worked better for male politicians.
20. We acknowledge the difficulty of re-creating political reality in a laboratory or survey setting. It is nonetheless important for future work on related topics to employ a healthy mix of research designs, including aggregate-level and survey studies that examine how voters react to real-life politicians about whom they may already have developed strong feelings one way or the other.
21. For example, see https://news.gallup.com/poll/221216/concerns-sexual-harassment-higher-1998.aspx; http://cdn.cnn.com/cnn/2018/images/10/11/rel9c.-.trump.and.russia.pdf; http://cdn.cnn.com/cnn/2017/images/12/21/rel12d.-.sexual.harassment.pdf; https://www.ipsos.com/en-us/news-polls/NPR-Sexual-Harassment-and-Assault; https://www.ipsos.com/sites/default/files/ct/news/documents/2018-10/ipsos_npr_sexual_harass ment_topline_103118_final.pdf.
22. Estimates of victimization have varied considerably over the years, largely (we suspect) due to differences in question wording and sampling methodology from one survey to the next; see Craig and Cossette (Citation2020).
23. The “all else equal” caveat is important. Whatever general preferences someone might have (and express in response to a nonspecific survey question), the impact of candidate gender in a particular election is often conditioned by other factors. For example, see Bauer (Citation2015); Brooks (Citation2013); Dolan (Citation2014b); Martin (Citation2019); Teele, Kalla, and Rosenbluth (Citation2018).
24. Details regarding partisan differences can be found in online Appendix A.
25. We look at change in this instance using OLS because, unlike vote preference, the favorability variable is not dichotomous. When this model is replicated using favorability at T3 as our dependent variable (and including favorability at T1 as a predictor), results are very similar to those portrayed in below. The same is true when we run the favorability model using ordered logit, with favorability at T3 as dependent (and including favorability at T1 as a predictor).
26. Each of these declines is statistically significant at p < 0.001.
27. See Caldwell and Moe (Citation2017); Garber (Citation2017); Shabad (Citation2018); Tumulty (Citation2021); Viebeck and Weigel (Citation2017).
28. We also have only considered a mixed-gender contest. Future research might focus on male-male or female-female matchups, as well as situations involving same-sex allegations of sexual misconduct.
Additional information
Notes on contributors
Stephen C. Craig
Stephen C. Craig (Ph.D., Northwestern University) is the author of The Malevolent Leaders: Popular Discontent in America (1993), coauthor of Politicians Behaving Badly: Men, Women, and the Politics of Sexual Harassment (2020), and editor or co-editor of several books including After the Boom: The Politics of Generation X (1997), Ambivalence and the Structure of Political Opinion (2005), and The Electoral Challenge: Theory Meets Practice (2006, 2011). His articles have appeared in American Political Science Review, American Journal of Political Science, and various other academic journals.
Paulina S. Cossette
Paulina S. Cossette (Ph.D., University of Florida) is coauthor of Politicians Behaving Badly: Men, Women, and the Politics of Sexual Harassment (2020). As a 2014-2015 APSA Congressional Fellow, she served as a legislative aide for US Senator Jack Reed (RI), focusing on environmental and energy policy. Her research on congressional cosponsorship networks and campaign advertising has appeared in Polity, Politics and Policy, Political Communication, American Politics Review, and other journals. She now owns a business, Acadia Editing Services, in which she conducts copy editing and survey questionnaire design for academic authors.
Angela Farizo McCarthy
Angela Farizo McCarthy is a lecturer in the Department of Political Science at the University of Florida. Primarily, her research focuses on general trends in public opinion. Her work is published in Political Research Quarterly, The Journal for the Scientific Study of Religion, and Geographical Review. She is currently working on a book project entitled: Religious Believing: Understanding the Impacts of Religious Beliefs on Policy Opinions.