In a 2009 interview with CNBC, then-congressmen Tom Price admonished his Democratic colleagues for the use of expensive private jets, stating “this is just another example of fiscal irresponsibility run amok in Congress” (Pramuk Reference Pramuk2017). Eight years later, the media revealed that Price, acting as Donald Trump’s first Secretary of Health and Human Services, had taken at least two dozen trips with private and military jets to Europe and Asia at the cost of around $1 million dollars in taxpayer money. Left-leaning journalist Rachel Maddow gleefully reported the story, including details about Tom Price using the private jets to have lunch with his son (Maddow Reference Maddow2017). Meanwhile, a slew of Republican politicians criticized Price’s actions (Fox Reference Fox2017), and within two weeks of the first story, Price resigned from his position (Gambino Reference Gambino2017).
The journalists who uncovered Tom Price’s use of taxpayer-funded private jets created a mediated political scandal, which can be defined as an action or event that transgresses values, norms or moral codes that is made public by the news media and provokes reactions from disapproval to outrage (Esser and Hartung Reference Esser and Hartung2004; Lee Reference Lee2018; Thompson Reference Thompson2000). As the scandal unfolded, journalists who referenced Price’s prior statements on private jets highlighted his hypocrisy, or “inconsistency between claimed behavior and actual behavior” (Powell and Smith Reference Powell and Smith2013, 414). Hypocrisy is a common feature of political scandal. It can characterize misconduct from politicians of all ideological and partisan stripes from “family values” Republican and former Louisiana Senator David Vitter, who faced revelations that he had solicited prostitutes, to former Florida Democratic congresswoman and public servant Corrine Brown, who took $1 million from a fake charity meant to provide scholarships for poor children.
Drawing from two actual political scandals involving US members of congress from both major political parties, this paper tests how partisans evaluate parties after being exposed to the different stages of a hypocritical scandal – an initial scandal, revelations of hypocrisy, and when a hypocritical scandal is followed by a censuring response by same-party leaders. This article makes two primary contributions to the literature on the spillover effects of political scandals. First, I test how different features of scandal spill over on to party evaluations distinctly among co-partisans and out-partisans. Second, I assess how party evaluations change after a scandal ends in party-sanctioned resignation. I find that Republicans feel more positive toward their party after reading about the resignation of a hypocritical co-partisan politician compared to merely reading about the politician’s hypocrisy. Furthermore, I find that Republicans feel warmer about their party when reading about a hypocritical versus non-hypocritical out-party transgression. However, Democrats do not change their party evaluations after being exposed to different scandal frames involving co-partisan and out-partisan politicians. This suggests that Republicans and Democrats have different attitudes toward hypocrisy and/or they differently apply information about individuals when evaluating parties.
Political scandal, spillover effects, and hypocrisy
There is widespread evidence that political scandals may damage the evaluations of politicians involved in scandal (Carlson, Ganiel, and Hyde Reference Carlson, Ganiel and Hyde2000; Doherty, Dowling, and Miller Reference Doherty, Dowling and Miller2011, Reference Doherty, Dowling and Miller2014; Funk Reference Funk1996; McDermott, Schwartz, and Vallejo Reference McDermott, Schwartz and Vallejo2015; Żemojtel-Piotrowska et al. Reference Żemojtel-Piotrowska, Marganski, Baran and Piotrowski2017). The negative effects of scandals involving individual politicians can also have consequences for evaluations of political parties associated with scandal-ridden politicians (Lee Reference Lee2018; Maier Reference Maier2011; Schwarz and Bless Reference Schwarz and Bless1992; von Sikorski, Heiss, and Matthes Reference von Sikorski, Heiss and Matthes2020). These “spillover effects” of political scandals occur when an individual’s interconnected cognitions are activated and people associate one target with another (Lee Reference Lee2018). For example, when a politician is hypocritical, people not only evaluate the individual poorly but they also negatively appraise the politician’s party (von Sikorski and Herbst Reference von Sikorski and Herbst2020).
Spillover effects of scandals may not be uniformly distributed in the United States as political evaluations are increasingly viewed through a partisan lens (Iyengar, Sood, and Lelkes Reference Iyengar, Sood and Lelkes2012; Iyengar and Westwood Reference Iyengar and Westwood2015; Jacobson Reference Jacobson2013; Mason Reference Mason2015). Scandals represent actions or behavior that may be incongruent to co-partisans, who routinely engage in motivated reasoning, rejecting incongruous information, and only accepting information which conforms to their beliefs (Erisen, Redlawsk, and Erisen Reference Erisen, Redlawsk and Erisen2018; Redlawsk Reference Redlawsk2002; Taber and Lodge Reference Taber and Lodge2006). Previous literature on hypocrisy and political scandal has found that hypocritical scandals are judged more harshly than non-hypocritical scandals (McDermott, Schwartz, and Vallejo Reference McDermott, Schwartz and Vallejo2015), but it is unclear whether partisans would evaluate hypocritical and non-hypocritical scandals differently. Some suggest that scandals may erode support among a politician’s followers (von Sikorski, Heiss, and Matthes Reference von Sikorski, Heiss and Matthes2020), but others find politicians who are involved in scandal retain their support (Fischle Reference Fischle2000). The additional information of hypocrisy in a scandal may cause co-partisans to reach an affective tipping point (Redlawsk, Civettini, and Emmerson Reference Redlawsk, Civettini and Emmerson2010), or co-partisans may discount hypocrisy (Redlawsk Reference Redlawsk2002). This leads to two competing expectations: compared to a scandal without hypocritical information, exposure to a scandal with hypocritical information will result in lower party evaluations among co-partisans (H1a), or hypocritical information will be disregarded and will not significantly lower party evaluations among co-partisans (H1b).
As a scandal involving an individual politician develops, party leaders sometimes encourage her to resign. In theory, co-partisans will assimilate the congruent information about the removal of a hypocritical politician and dismiss the incongruent information about the hypocritical scandal itself (Taber and Lodge Reference Taber and Lodge2006). Therefore, I expect that co-partisans will evaluate their party more positively when they are exposed to the successful purge of a hypocritical politician from office compared to when they are merely exposed to a politician involved in a hypocritical political scandal (H2).
Exposure to information about out-party scandals, especially those that feature hypocrisy, ought to be perceived as a boon for out-partisans in the context of an affectively polarized society. Out-partisans can even experience schadenfreude – pleasure at another’s misfortune – when exposed to embarrassing mishaps involving out-party politicians (Schurtz et al. Reference Schurtz, Combs, Hoogland, Smith, van Dijk and Ouwerkerk2014), which can boost group esteem (Ouwerkerk and van Dijk Reference Ouwerkerk, van Dijk, van Dijk and Ouwerkerk2014). This suggests that, while a scandal committed by a politician from one party may increase party evaluations of a different party (Lee Reference Lee2018), a hypocritical scandal could bolster party evaluations even more. Thus, I hypothesize that exposure to information about a hypocritical scandal committed by an out-partisan politician will enhance positive feelings toward one’s party more than exposure to a scandal without hypocrisy (H3).
Research design
To test my hypotheses about scandals, hypocrisy, and resignation, I fielded two studies on Amazon Mechanical Turk (MTurk) that exposed participants to information about actual members of congress who had each been involved in a sex scandal.Footnote 1 After participants consented to taking a Qualtrics survey, they were asked their party identification and level of news consumption.Footnote 2 Then participants were randomly exposed to one of three experimental treatments. Study 1’s (N = 507) experimental stimulus focused on a scandal involving Tim Murphy, a conservative Republican congressman from Pennsylvania, and Study 2 (N = 504) exposed participants to information about a scandal involving John Conyers, a liberal Democratic congressman from Michigan.Footnote 3
Participants in each study were randomly assigned to parts or all of the following text:
Study 1:
Tim Murphy, a Republican congressman from Pennsylvania, admitted to having an extramarital affair with a personal friend in September 2017. Murphy is a member of the House Pro-Life Caucus in Congress and has voted to impose more restrictions on abortion access. News emerged that he pressured his mistress to get an abortion during a pregnancy scare. After House GOP leaders and senior Republicans put pressure on Murphy to step down, he resigned in early October.
Study 2:
John Conyers, a Democratic congressman from Michigan, was accused of sexual harassment by several women in November and December 2017. Conyers had previously cosponsored legislation which helped assist victims of sexual assault and domestic violence. The women, some of whom had worked for Conyers, say that among other things, Conyers groped and propositioned them for sex. After House Democratic leaders and other senior Democrats put pressure on Conyers to step down, he resigned in early December.
The first group was randomly assigned to only receive the underlined text (scandal condition), the second group was given the underlined text and the bold text (hypocrisy condition), and the third group was given the entire text (resignation condition).Footnote 4 Thus, in Study 1, all participants were aware of Murphy’s party affiliation and extramarital affair, but only some participants saw Murphy’s incongruity of past actions and current behavior. Further, only those in the third group saw that members of his party effectively pushed him to resign.Footnote 5 These two studies were meant to complement each other, and I did not have any expectations as to differences between how co-partisans of different parties would react given the disdain hypocrites face in general (Jordan et al. Reference Jordan, Roseanna, Paul and Rand2017; Laurent et al. Reference Laurent, Clark, Walker and Wiseman2014; Powell and Smith Reference Powell and Smith2013).Footnote 6
After reading the text, all participants were asked to evaluate the political parties on a “feeling thermometer” – a 101-point scale where 0 indicates very cold feelings toward the party, 50 indicates indifference, and 100 indicates feeling very warm toward the party. Participants were first asked about the party of the target politician, so those in Study 1 were first asked about the Republican Party and those in Study 2 were first asked about the Democratic Party. Participants were then asked the same feeling thermometer question about the out-party of the target politician. These questions serve as my main dependent variables.
This study builds on previous experimental work manipulating frames from real political scandals (Cortina and Rottinghaus Reference Cortina and Rottinghaus2017; Maier Reference Maier2011), but I employ a vignette approach similar to scholars who have investigated the effects of political scandals committed by hypothetical politicians (Bhatti, Hansen, and Olsen Reference Bhatti, Hansen and Olsen2013; Botero et al. Reference Botero, Cornejo, Gamboa, Pavao and Nickerson2015, Reference Botero, Cornejo, Gamboa, Pavao and Nickerson2019a, Reference Botero, Cornejo, Gamboa, Pavao and Nickerson2019b; Doherty, Dowling, and Miller Reference Doherty, Dowling and Miller2011, Reference Doherty, Dowling and Miller2014; McDermott, Schwartz, and Vallejo Reference McDermott, Schwartz and Vallejo2015; Winters and Weitz-Shapiro Reference Winters and Weitz-Shapiro2013, Reference Winters and Weitz-Shapiro2015; Żemojtel-Piotrowska et al. Reference Żemojtel-Piotrowska, Marganski, Baran and Piotrowski2017). The vignette format exposes experimental participants to the most pertinent information, akin to headlines or sub-headlines, at different stages of scandal – the initial scandal, second-order transgressions, and resolution. Furthermore, this format may reduce recall of other details of the scandal because of the brevity of the treatment compared to a newspaper article.Footnote 7 Nevertheless, using actual political events may run the risk of pretreatment, which may bias against finding significant effects (Druckman and Leeper Reference Druckman and Leeper2012).
Results
I first test if individuals feel less warm toward their own political party after being exposed to a co-partisan politician engaging in hypocritical transgression compared to a co-partisan politician who merely transgressed without apparent hypocrisy. To assess this proposition (H1), I conducted difference-in-means tests of feeling thermometer evaluations of the target’s co-partisans in the first scandal condition to co-partisans in the hypocrisy condition.Footnote 8 In Study 1, I find that, on average, Republican identifiers and leaners feel cooler toward their own party in the hypocrisy condition (56.9 degrees) compared to the baseline scandal condition (62.3 degrees), but this difference is shy of statistical significance (t = 1.19 and p = .10 in a one-tailed test). Similarly, I find that Democrats and leaners are slightly cooler upon being exposed to the hypocrisy condition (59.3 degrees) compared to in the baseline scandal condition (62.6 degrees) in Study 2, but again this difference is not statistically significant (t = 1 and p = .16 in a one-tailed test). These results indicate that hypocrisy did not create an affective tipping among partisans of the target, suggesting that partisans discounted the additional negative information of hypocrisy.Footnote 9
To test if co-partisans feel better about their party after they are exposed to party leaders’ successful purging of the hypocritical politician (H2), I again compared the means of feeling thermometer ratings of the target’s co-partisans, but this time in the hypocrisy condition to the resignation condition. In Study 1, I find that Republicans feel 8.9 degrees more warmly toward their own party (t = −2.28 and p < .05 in a one-tailed t-test) between evaluations of the Republican Party in the hypocrisy versus the resignation treatment (see Figure 1). In Study 2, however, the resignation treatment and the hypocrisy treatment produce nearly identical feeling thermometer ratings for Democrats (t = −0.03 and p = .49 in a one-tailed t-test), suggesting that Democratic leaders’ pressure and Conyers’ resignation was not enough to make them feel more warmly toward their party (see Figure 2).
Figure 1 Predicted Values of Ratings of the Republican Party Among Republicans (Study 1).
Note: Confidence intervals for means are at the 90% level.
Figure 2 Predicted Values of Ratings of the Democratic Party Among Democrats (Study 2).
Note: Confidence intervals for means are at the 90% level.
Why were Republicans more sensitive to the treatments than Democrats? Various studies have found that conservatives and Republicans value loyalty and authority more than liberals and Democrats (Graham, Haidt, and Nosek Reference Graham, Haidt and Nosek2009; Rempala, Okdie, and Garvey Reference Rempala, Okdie and Garvey2016; Severson Reference Severson2018). Thus, Republicans may have felt a strong sense of betrayal after exposure to group norm violations when reading about hypocrisy, but a sense of justice after observing co-partisan leaders “correcting” the violation. Differences in the stimulus may have also contributed to diverging results in the two studies. The hypocrisy treatment in Study 1 included information about an affair and novel information about the congressmen pressuring his mistress to have an abortion. However, in Study 2, the hypocrisy treatment included information about harassment, but the additional information was about the severity of the transgression. Furthermore, going against the congressional caucus of which one is a member (Study 1) may be seen as a greater betrayal than behaving inconsistently with a single piece of legislation (Study 2).
My final hypothesis (H3) anticipates that out-party partisans evaluate their own party more positively when observing a hypocritical scandal versus a non-hypocritical scandal by an out-partisan politician. I compared feeling thermometers of the out-party for out-partisans of the target in the scandal and hypocrisy conditions using difference-in-means tests. Figures 3 and 4 present the results for Study 1 and Study 2, respectively. While in both studies, out-partisans move in a positive direction, only in Study 2 do Republicans experience a statistically significant boost in feelings of their party, moving from 59.3 to 68 degrees on the feeling thermometer (t = −1.87 and p < .05 in a one-tailed t-test). Again, I find that Republicans appear more sensitive to the treatment than Democrats. Despite the evidence that both Democrats and Republicans “loathe” their opponents (Iyengar, Sood, and Lelkes Reference Iyengar, Sood and Lelkes2012), some work has found that Republicans have stronger bias against Democrats and feel more positive when seeing the Democratic Party fail than vice versa (Iyengar and Westwood Reference Iyengar and Westwood2015; Zillmann and Knobloch Reference Zillmann and Knobloch2001).Footnote 10
Figure 3 Predicted Values of Ratings of the Democratic Party Among Democrats (Study 1).
Note: Confidence intervals for means are at the 90% level.
Figure 4 Predicted Values of Ratings of the Republican Party Among Republicans (Study 2).
Note: Confidence intervals for means are at the 90% level.
Finally, I explore two additional tests of my theory. First, I test how strength of partisanship conditions reactions to scandal, hypocrisy, and resignation of scandalous politicians. Those with strong priors (i.e. strong partisans) have been shown to be stronger motivated reasoners than those with weaker priors (i.e. weak partisans and partisan leaners) and more resilient to change in attitudes toward their party after experiencing incongruent information (Taber and Lodge Reference Taber and Lodge2006). I anticipate that strong partisans will discount a hypocritical political scandal compared to a non-hypocritical political scandal, leading to no difference in party evaluations between the two conditions (extending H1b). However, weak partisans and leaners will be more likely to change and reach an affective tipping point (Bisgaard Reference Bisgaard2015; Gaines et al. Reference Gaines, Kuklinski, Quirk, Buddy and Verkuilen2007; Redlawsk, Civettini, and Emmerson Reference Redlawsk, Civettini and Emmerson2010) (extending H1a). I expect that strong partisans will boost their party evaluations more than weak partisans and leaners in the resignation condition compared to the hypocrisy condition (extending H2), and as out-partisans in the hypocrisy condition compared to the scandal condition (extending H3) because strong partisans prioritize directional goals more than weak partisans and leaners (Schaffner and Roche Reference Schaffner and Roche2017).
I test these extensions of my hypotheses by regressing the feeling thermometers on the interaction between the treatment and strength of partisanship.Footnote 11 The results are underpowered, but reveal some interesting differences between strong and weak partisans and leaners. First, the results of H2 in Study 1 appear to be driven by weak partisans (increasing from 53.8 to 60.9 degrees, t = −1.36 and p < .1 in a one-tailed test). Next, although I found no evidence for H3 in Study 1 among all Democrats, leaners experience a statistically significant increase in party evaluations from 45.7 to 56.4 degrees (t = 1.62 and p < .1 in a one-tailed test). In Study 2, I found no evidence for H1a for all Democrats, but breaking the results down by strength of partisanship, I find that those leaning Democratic experience a huge drop in party evaluations from the scandal condition to the hypocrisy condition (moving from 55.5 to 36.3, t = 3.4 and p < .01 in a one-tailed test). Finally, I find that the boost out-group members feel toward their party after exposure to a hypocritical scandal compared to a non-hypocritical scandal of an out-party politician is driven by weak Republicans (increasing from 51.3 to 64.5 degrees, t = 1.5 and p < .05 in a one-tailed test) in Study 2. Altogether this suggests that those with weaker attachments are more sensitive to congruent and incongruent information than strong partisans.
Second, I probe how news exposure affects how partisans interpret the different treatments. The most politically interested and knowledgeable tend to double down on their attitudes and are more likely to engage in motivated reasoning (Erisen, Redlawsk, and Erisen Reference Erisen, Redlawsk and Erisen2018; Taber and Lodge Reference Taber and Lodge2006).Footnote 12 Thus, I expect that co-partisans who pay closer attention to the news are more likely to discount information on hypocrisy than those who pay less attention (H1b), are more likely to embrace information about a co-partisan hypocrite’s resignation (compared to hypocritical scandal) (H2), and are more likely to boost party evaluations when exposed to a hypocritical (versus non-hypocritical) out-party politician (H3). To test this, first I created a dummy variable of news exposure where those who say they follow the news very closely are coded as 1, and those who follow the news somewhat, not very or not at all closely, are coded as 0. Then I regressed the feeling thermometer evaluations for each hypothesis on the interaction of news exposure and the treatments. Only one of the interactions is statistically significant in a one-tailed test (p < .05). For H1 in Study 1, Republicans who follow the news very closely had a higher feeling thermometer of the Republican Party in the non-hypocritical scandal condition (70.8) compared to the hypocrisy condition (56.1), but those who did not follow the news very closely had nearly identical feeling thermometer ratings for the two treatments (57.3 and 57.5). This suggests that politically aware individuals were more sensitive to treatments, but this pattern does not hold for the other hypotheses (see Supplementary Appendix Figures 7A–12A and Tables 11A–16A).
Discussion and conclusions
Political scandals involving individual politicians have the potential to damage evaluations of a politicians’ party (Lee Reference Lee2018; von Sikorski, Heiss, and Matthes Reference von Sikorski, Heiss and Matthes2020). This study employs vignette experiments that draw on actual political scandals. Although this may enhance external validity, it also has its limitations. First, respondents may be pretreated with information about the political scandal, thus attenuating treatment effects (Druckman and Leeper Reference Druckman and Leeper2012). Second, vignette experiments may not replicate how respondents would gain information about scandals in the real world despite the use of a real-world scandal. Future work could overcome these limitations by employing an article format (cf. Carlson, Ganiel, and Hyde Reference Carlson, Ganiel and Hyde2000; Cortina and Rottinghaus Reference Cortina and Rottinghaus2017; Maier Reference Maier2011; von Sikorski and Herbst Reference von Sikorski and Herbst2020). Moreover, these two studies recruited via Amazon MTurk, whose pool is more diverse than other convenience samples, but tends to be younger, more male, less religious, more educated, and more liberal than the US population (Boas, Christenson, and Glick Reference Boas, Christenson and Glick2020; Levay, Freese, and Druckman Reference Levay, Jeremy and Druckman2016).Footnote 13 While an MTurk sample produces a valid Sample Average Treatment Effect (SATE) (Coppock and McClellan Reference Coppock and McClellan2019), future work should test the hypotheses on a more representative sample.
In the polarized two-party system of the United States, spillover effects of scandals differ significantly depending not only on the politician involved but also who is evaluating the politician. There was a large gap in party evaluations between Republicans who were exposed to the hypocritical transgression and those who read about the resignation of the hypocritical co-partisan politician. Furthermore, Republicans felt significantly warmer toward their own party when reading about a hypocritical transgression than non-hypocritical scandal committed by a Democratic politician. Yet, Democrats did not feel significantly different about their party after exposure to both in-party and out-party hypocritical scandals. Future work should investigate if attitudes toward hypocrisy differ by partisans from the two parties, how strength of partisanship shapes evaluations of hypocritical behavior from individual politicians, and how partisans from different parties may assimilate attitudes toward individuals into groups.
Supplementary Material
To view supplementary material for this article, please visit https://doi.org/10.1017/XPS.2020.36.
