Political Scandals and Disgust Sensitivity:
Testing the moderating effect of disgust on the public’s evaluation
of politicians involved in scandals
Nicolas Papaconstantinou 12082953
University of Amsterdam, Graduate School of Communication Master’s Thesis
Master’s programme Political Communication Science Supervisor: Bert Bakker
Date: 28th June 2019 Word Count: 7465
3 Abstract
Political scandals have always captured the attention of the public and the media, often dominating the news cycle for weeks on end or even influencing elections. However, in recent years the real impact of these scandals has come into question, as certain politicians seem unimpeded by them. This study conducts two experiments, one in which a politician is involved or not in a moral scandal and one in which he is or isn’t involved in a sex scandal, to see whether scandals affect people’s perception of those politicians. Moreover, the “disgust sensitivity” of participants was measured to see whether it has a moderating effect on this relationship. While the main effects of scandal exposure on politician’s evaluations was found to be both negative and large, this study did not find any evidence to support the moderating role of disgust sensitivity in the process. This could suggest that disgust does not have a moralizing function when politicians are being evaluated.
4 Introduction
In 2011, then-Congressman Anthony Weiner found himself as the center of a political scandal when it was discovered that he was sending sexually explicit photographs of himself to a 21-year old student, which led to his resignation from Congress that same year. More than a decade prior to this incident, President Bill Clinton was impeached by the House for lying and obstruction of justice in the Lewinsky scandal, but was able to complete his term as president. Looking at these two cases you might wonder why only some politicians have their political careers ended by scandals while others, such as President Donald Trump, seem to become increasingly unabashed with each new one. Political scandals have always captured the attention of the public and the media, often dominating the news cycle for weeks on end, but this has not always translated to a stark decline in their political support. The question that naturally arises when considering these cases is whether political scandals have the same sway over voters’ perceptions as they used to. While there is no real reason to believe this might have changed, it is worth exploring whether with time the public no longer views scandals as something that would disqualify a politician, or whether the rules are different for some well-known politicians.
Of course, forming opinions on politicians, particularly when there are scandals involved, is an incredibly complicated process, with many identified and unidentified variables each playing a small role. One potential moderator might seem a quite unlikely factor: the dispositional nature to experience disgust. Schnall, Clore, & Jordan (2008) looked at its adaptive functions and explained that while disgust started just as a way of avoiding bad food and diseases, it also became well suited as an emotion for social rejection, such as for mate selection and avoidance of individuals who are likely to impose diffuse costs to their social group. In a nutshell, disgust seems to inform people on how to make moral judgements about others and their behaviors (Pizarro & Helion, 2011). This dispositional nature to experience disgust has been labelled “disgust sensitivity” (Horberg, Oveis, Keltner, & Cohen, 2009).
There is a rich literature on the role of disgust sensitivity. One of the most important discoveries was the identification of the three sub-domains of disgust sensitivity. First, pathogen disgust, which has
5
also been treated as the “core disgust” in previous models for disgust, motivates the avoidance of infectious microorganisms and other dangerous substances. Sexual disgust developed to avoid sexual partners and behaviors that jeopardize one’s long-term reproductive success. Finally, moral disgust motivates the avoidance of social norm violators. Tybur, Lieberman, & Griskevicius, (2009) named this the Three-Domain Disgust Scale. Several studies (which will be explained in depth in the following section) also suggest that there is a relationship between disgust (either the emotion or the trait) and more negative moral judgements (Wheatley and Haidt, 2005; Gutierrez & Giner-Sorolla, 2007; Jones & Fitness, 2008; Horberg et al., 2009; Tybur., Lieberman, Kurzban, & DeScioli, 2013).
However, there is still much that remains unknown about the role of disgust sensitivity. The most relevant gap in the literature is the lack of research about the impact of disgust sensitivity on how politicians are evaluated. While there are studies which have looked at a correlation between disgust sensitivity and political conservatism (Inbar, Pizarro, Iyer, & Haidt, 2012), whether disgust sensitivity affects politics remains an unexplored field. The available literature also does not cover whether the same adaptive functions of disgust to avoid individuals who might impose social costs on the group have a similar role in the evaluation of political figures. Therefore, this study will aim to fill a gap in the academic literature by examining how political scandals affect the perceptions of politicians involved in them, and whether this relationship is moderated by how disgusting an individual might find the scandal being judged.
Research Question: Does exposure to a politician’s scandal affect how the public evaluates them and is this relationship moderated by disgust sensitivity?
The Impact of Political Scandals
Political scandals are media and political frenzies which initially attract a lot of attention and which eventually (depending on the severity of the scandal and the media’s interest) fizzle out (Tumbler & Waisbord, 2004). As one would expect, political scandals in the academic literature have been
6
demonstrated to negatively impact the perceptions of politicians. Schwarz & Bless (1992) conducted an experiment where participants were either asked or not asked to recall politicians who had been involved in a political scandal, and then told to judge the trustworthiness of politicians in general, as well as that of three specific politicians. They found that the participants who would answer the scandal question would generally rate politicians more negatively than the participants who were not asked the scandal question. Maier (2011) also tested different theories to see whether the growth of negative attitudes towards politics, politicians and democracy had been caused by the increase in political scandals. He found that exposure to political scandals decreased support for politicians and their parties, but did not affect their satisfaction with democracy.
However, the reaction of individuals to scandals of politicians seems to be the outcome of a more complex set of factors; the literature covers a wide range of variables that might push someone one way or the other. Starting with a very specific example, Miller (1999) examined what the public’s reaction was in the wake of the Monica Lewinsky scandal towards the then-President Clinton in hopes of understanding how politicians are affected by the scandals they are involved in. He found that while his behavior was deemed unacceptable, he still retained support as President. Other studies similarly found that reactions to political scandals might be overlooked depending on the perpetrators political standing; Bhatti, Hansen, & Olsen (2013) and Walter & Redlawsk (2019) looked at partisan responses to political scandals and found that citizens were far more forgiving to scandals that came from their own political party, particularly when the scandals regarded the politician’s personal life.
The literature on the impact of political scandals on citizens’ evaluations of politicians shows that overall when an individual is exposed to a politician’s political scandals, they tend to judge those politicians more harshly. However, the literature also points to the fact that individuals are more likely to give the benefit of the doubt or be more forgiving with politicians who they otherwise support or are familiar with. These findings are very context specific, as they each looked at known political figures. In contrast, this study will examine how exposure to a scandal by an unknown political figure, when
7
presented with other general information about them, affects their evaluation. This evaluation will be assessed based on three different measures: how disgusting individuals find the politician, how warmly they feel towards him, and how likely they would be to vote for them. We theorize that because of the lack of contextual cues, such as their party affiliations, individuals exposed to a politician’s scandal will have very negative perceptions of the candidate. We therefore have our first hypothesis:
H1: When an individual is exposed to a politician’s scandal, their evaluation of that politician will be more negative compared to individuals who were not exposed to the same scandal.
Disgust and Moral Judgements
Arguably an even more interesting question regarding the impact of a politician’s scandal on their evaluation by the public is exactly how these opinions are formed in the first place. While scholars used to emphasize the importance of reason in human decision making and moral judgments, the role of emotions has gained increasing attention by scholars. In “The Emotional Dog and Its Rational Tail: A Social Intuitionist Approach to Moral Judgement”, Jonathan Haidt (2003) explains how the previously dominant rationalist models have been replaced by, among others, the Social Intuitionist Model, which claims that moral judgements are caused by fast moral intuitions and then rationalized after the fact. They are in other words emotional responses.
If emotions play a large role in the decision-making process, then the next question becomes what emotions are elicited in response to a politician’s scandals. A study by Molho, Tybur, Guler, Balliet, & Hofmann (2017) looked at emotional responses to moral violations, focusing specifically on anger and disgust. One of their findings found that when the “victim” of a moral violation was someone else rather than the participants themselves, disgust responses were stronger. They theorize that disgust functions to encourage avoidance of others who could prove dangerous to one’s self; anger on the other hand, being a more costly response, is only elicited when someone has been personally wronged. Otherwise, the disgust response suffices to avoid someone who has not directly harmed you.
8
A different study by Gutierrez & Giner-Sorolla (2007) examined participants disgust and anger responses to purity-violating taboo behaviors by exposing them to taboo behaviors and then asking them to report their emotions and whether they believed harm had been inflicted. They found that taboo behaviors elicited higher disgust responses than anger responses, but if harm was assumed to be part of the taboo behavior, higher anger responses were predicted, but not disgust. Another study even looked at how flexible the emotions of disgust and anger were by asking participants what contextual cues might get them to re-evaluate how a moral violation made them feel (Russell & Giner-Sorolla, 2011). They found that individuals were more willing to change their opinions in light of changing circumstances when the emotion experienced primarily was anger. Therefore, disgust is not a flexible emotion, and moral judgements made while experiencing disgust should be less prone to change as a result.
Given that we understand the circumstances in which disgust is elicited and given that it is an inflexible emotion, this study will focus on disgust. In the academic literature disgust is understood in three different “forms”, each of which interact with moral judgements differently (Horberg, Oveis, Keltner, & Cohen, 2009). Integral Disgust is disgust felt in response to an act being judged, and higher experienced disgust has been found to lead to harsher moral judgements. Incidental Disgust is disgust felt in response to stimuli that are different from the act being judged and have also been found to lead to harsher moral judgments. Finally, Trait Disgust (or Disgust Sensitivity) is a personality trait that measures the propensity to find something disgusting, which then affects the moral judgement of it. While this study is only looking at trait disgust specifically, understanding the literature on all three forms is necessary to understand why disgust sensitivity is expected to affect how individuals evaluate politicians when exposed to their scandals.
Starting with the literature on integral disgust, Horberg et al. (2009) showed that integral disgust was able to predict harsher moral judgments of acts that violate the moral principle of purity. Participants were shown scenarios containing moral violations, which violated either the principle of purity or of justice. They found that the purity violations elicited strong feelings of disgust, which then predicted
9
harsher moral judgements of the behaviors described in the scenarios, compared to feelings of disgust. Chapman, Kim, Susskind & Anderson (2006) also conducted a study on facial motor activity in response to disgusting stimuli. The study compared disgust elicited by unpleasant tastes, contaminants, and moral disgust, where the moral disgust was elicited in response to unfair treatment in an economic game. They found that the same facial motor functions were activate in response to all three types of stimuli, which further supports the adaptive interpretation that disgust is an emotion felt to encourage avoidance of dangerous individuals.
Studies on incidental disgust found that merely experiencing a feeling of disgust also led to harsher moral judgements. In the same study, Horberg et al. (2009) also found that incidental disgust would moralize purity behaviors. This means that when disgust was induced in the participants and they were asked to judge someone who had violated the moral principle of purity, judgements were harsher compared to the participants where sadness was elicited instead. Moreover, Wheatley and Haidt (2005) ran experiments in which participants were hypnotized to experience a flash of disgust when reading a specific word in a text and found that when the trigger word was present in a text describing actions that would constitute a moral transgression, participants tended to judge the action more harshly than when the word was absent.
Interestingly, when participants were asked why they would judge the actions so harshly in the condition where the trigger word was present, they could not explain why it is that they felt that way in the moment, resigning to call it a “gut-feeling”. This further supports the social intuitionist model discussed earlier. However, the literature on the effect of incidental disgust on moral judgements should be taken with a grain of salt. A meta-analytic study of experiments by Landy & Goodwin (2015) on incidental disgust and moral judgements found that there was only a small amplification effect on moral judgments, which was stronger for pathogen disgust elicitors. Moreover, the authors also accounted for what they perceived to be a publication bias and the amplification effect of incidental disgust disappeared. Therefore, the role of incidental disgust on moral judgements should not be overstated.
10
Finally, studies on trait disgust, or disgust sensitivity, have also found similar effects. In one of the more respected studies, Jones & Fitness (2008) conducted an experiment where participants were directed to act as mock jurors in a court room. They found that individuals high on disgust sensitivity tended to pass harsher convictions on the hypothetical criminals, such as longer sentences or a proclivity to attribute evilness to them. Disgust sensitivity was also found to have an impact on political life. Using large, cross country samples, Inbar, Pizarro, Iyer, & Haidt (2012) also tested the relationship between disgust sensitivity and political conservatism and found a positive correlation between the two. This could explain why political conservatives tend to be in favor of restricting immigration, as individuals with high disgust sensitivity are predisposed to be less trusting of outsiders who might impose social costs in their own group.
Disgust sensitivity then, being a self-report of how disgusting one finds common disgust elicitors, should play a pivotal role in the way politician’s behaviors are judged. It should be noted however, that disgust does not always have a function in the formation of moral judgements, nor does every act that elicits disgust necessarily have to be moralized. The question therefore is whether political scandals elicit disgust in citizens, and whether this has an effect on their evaluation of politicians. While there are ongoing debates regarding whether disgust is only elicited in response to specific circumstances and whether incidental disgust plays a role in moral judgements, the bulk of the literature supports this theory.
Moreover, each of the three domains of disgust might have a different impact on politics. For example, it was found that sexual disgust was able to predict how people would vote in the 2016 US presidential elections far better than pathogen disgust did, as those high on sexual disgust were more likely to vote for Donald Trump (Billingsley, Lieberman, & Tybur, 2018). This finding disputes the belief that conservatives vote the way they do primarily because of their levels of pathogen disgust sensitivity, meaning that there is a need to study each measure independently.
Therefore, drawing on the Three-Domain Disgust Scale, this study looks at the two of the most common scandals seen in politics, moral and sex scandals, and suggests that individuals who are high on
11
moral and sexual disgust sensitivity will judge politicians more harshly when they are exposed to a political scandal in comparison to those low on those same traits. We expect that because of the adaptive nature of disgust, individuals will view the political scandals as disgusting and pass moral judgements, which will be harsher coming from those with high disgust sensitivity. Specifically, it is expected that in response to a moral scandal, those with high moral disgust sensitivity will be more prone to view the scandal as disgusting and by consequence more likely to evaluate this individual negatively. The same applies for sexual disgust, as those with high sexual disgust sensitivity will be more likely to evaluate politicians involved in a sex scandal more negatively compared to those low on the same trait. All that remains is to test whether disgust plays the same role in the passing of moral judgments in politics like it has in other experiments. We therefore formalize the following hypotheses:
H2: The negative effect of exposure to a politician’s moral scandal on their evaluation by the public is stronger for those with high moral disgust sensitivity compared to those with low moral disgust sensitivity.
H3: The negative effect of exposure to a politician’s sex scandal on their evaluation by the public is stronger for those with high sexual disgust sensitivity compared to those with low sexual disgust sensitivity.
Methodology Section
Experimental Design
This study consisted of two experiments, both of which had an experimental between-groups design with two experimental conditions. The two experiments, from here on referred to as the “moral scandal” and “sex scandal” experiments, are set up to test how a moral violation and a sexual violation by a politician (in other words, political scandals) might affect the perception of this politician. In order to experimentally test this, for each of the two experiments, participants were randomly assigned to either the scandal condition or the no-scandal condition. For the (experimental) moral scandal condition, the
12
politician is described as having blackmailed his fellow party members in order to advance his own career. In contrast, the moral no-scandal (control) condition describes the politician as having put the party interests above his own. Moving on to the sex scandal experiment, in the (experimental) scandal condition the participants are informed that the politician had a two-year-long affair with his secretary unbeknownst to his wife. On the other hand, in the (control) no-scandal experiment, the politician is described as having had a long and happy marriage with his wife.
The moderator in this experiment is the disgust sensitivity trait which was measured using the questions based on the Three-Domain Disgust scale. For the moral scandal experiment, the participants’ moral disgust sensitivity was divided into two groups: those with high moral disgust sensitivity and those with low moral disgust sensitivity. The same process was followed for the sex scandal condition, but using the sexual disgust sensitivity trait instead. Participants of this study were divided into high and low sexual disgust sensitivity groups, which were then compared with each other to test whether there was an interaction effect present between the exposure to a politician’s scandal and disgust sensitivity on the politician’s evaluation.
Procedure
Study participants received an online link to a Qualtrics survey and were offered a chance to win a 20- euro Amazon gift card for completing it. After reading the factsheet describing the basic information about the survey to follow and giving their informed consent to participate in the experiment, participants were asked to give their basic demographic information and to confirm that their native language is Dutch to establish their legibility to participate in the experiment. Following this, the participants were asked to rate how disgusting they felt several scenarios were on a 7-point Likert scale, ranging from “Not at all disgusting” to “Extremely disgusting”. Based on these answers, the participants’ Disgust Sensitivity was identified, and the participants were separated into two groups: those with low Disgust Sensitivity and those with high Disgust Sensitivity. This was done separately for sexual and moral disgust, so that each could be applied separately in the upcoming experiments.
13
Following this, study participants were informed they would now read about a hypothetical political candidate of an unknown political party running in special municipal elections taking place during the European elections and that they would be asked how they felt about this candidate based on the information provided to them. Here, the first experimental manipulation was applied as participants were randomly assigned to one out of two possible texts. Both texts included basic information about this hypothetical candidate, including information about where they were from, where and what they studied, and their political career. After answering a few filler questions about which party the participants voted for in the previous Dutch municipal elections and which party they planned to vote for in the upcoming European elections, the second experimental manipulation was applied. Just like in the previous case, participants were randomly assigned to one of the two conditions. They were told they had to read about a hypothetical candidate and then asked to rate how disgusting they thought the candidate was, how they felt about the candidate on the feeling thermometer, and how likely they would be to vote for them. The text included similar information about this candidates’ education and political career.
Finally, to complete the survey, participants were asked to complete the manipulation check, for which participants were asked to answer a question about each of the candidates. For the moral scandal experiment, participants had to answer a “Yes” or “No” question on whether the candidate had put the party interests over his own. For the sex scandal experiment, which was also answered with a “Yes” or “No”, participants were then asked whether the political candidate had an affair with his secretary. Once these questions were completed, participants were thanked for their participation and offered an opportunity to join the raffle for the gift card.
Sample
A sample of 101 Dutch adults between the ages of 20 – 61 (M = 29.2, SD = 12.24) participated in the experiment, 56% of which were female. The sample overwhelmingly represented the highly educated as 49% of the participants had attained or were in the process of attaining a Master Degree or PhD and 44% had attained or were in the process of attaining a Bachelor Degree.
14 Measures
Exposure to politician’s scandal. The main independent variable for these two experiments is whether or not the politician described in each experiment was engaged in a scandal. Participants were randomly assigned into one of the two conditions for both experiments. A pre-test was conducted to identify the sexual and moral behaviors that Dutch adults found most unacceptable and negative from their politicians to serve as the experimental conditions. A sample of (N =) 23 Dutch adults was collected, and for each behavior described participants had to rate how acceptable they found each behavior on a 7-point Likert scale, from “Totally unacceptable” (1) to “Totally acceptable” (7), and how positive or negative they found each behavior, also on a 7-point Likert scale from “Very negative” (1) to “Very Positive” (7). Also, based on the study by Molho et al. (2017), which found that disgust was elicited more strongly when the victim of a moral violation is not the participant themselves, the scenarios created for pre-test were designed to frame a third person as the victim of the various transgressions.
The data showed that among the statements presented, the full list of which can be found in Appendix A, “Blackmailing party members for their own personal gain” was found to be the most morally unacceptable (M = 1.41, SD = .65) and the most negative behavior (M = 1.32, SD = .47). “Having an affair with his secretary” was deemed to be the most sexually unacceptable (M = 2.09, SD = 1.20) and negative (M = 1.95, SD = .71) behavior. These findings were then used to create the stimuli for the experimental conditions in each of the experiments.
For the moral scandal experiment, the scandal condition describes how the politician blackmailed his fellow party members for his own personal benefit, whereas in the no-scandal condition the politician was described as a loyal party member who put the party’s interests above his own. For the sex scandal experiment, the scandal condition informed the participants how the politician was caught having an
15
affair with his secretary, which had been ongoing for two years without his wife knowing. In the no-scandal condition, the politician is described as happily married to his wife and having two children.
Evaluations of Politicians. This study did not use a single measure to look at how politicians are evaluated. Instead, the study uses three different questions, each looking at a different measure with which a politician can be evaluated. The first question asked participants how disgusting they thought the politician was on a 0-100 scale, with 0 corresponding to “Not at all disgusting” and 100 corresponding to “Extremely Disgusting”. The second question was a Feeling Thermometer, which asks participants to rate how coldly or warmly they feel about the candidate on a 0-100 scale. Finally, the third questions asked participants how likely they would be to vote for him if the election were held today. This question was answered on a 7-point Likert scale ranging from “Extremely unlikely” to “Extremely likely”.
Disgust Sensitivity. To measure disgust sensitivity with the Three-Domain Disgust Scale here, Tybur (2009) developed and validated a set of 21 statements, where each one fit into one of the three domains of disgust. To measure disgust sensitivity, participants had to rate how disgusting they found each item. Because of an accidental omission, one item for the sexual disgust sensitivity factor, “Hearing strangers have sex”, was not included in the survey and was therefore excluded from the factor analysis. However, this does not severely affect the results. Because this study uses the established Three-Domain Disgust Sensitivity Scale, there was therefore no need to run an exploratory factor analysis. Instead, since the goal of the factor analysis was to confirm that the items loaded onto the previously established factors, a principal axis factoring was conducted with the 20 items. The model was restricted to fit only three factors in order to see how well the items loaded into the three factors. Tables 1-3 show all the items, their factor loadings onto each of the three factors, and the average correlation they have with the other items of the same factor. The data demonstrates that all the items included load on to their respective factors strongly and are all positively correlated with each other.
16 Table 1:
Pathogen Disgust Sensitivity Component Matrix
Label Items Component Correlation
with other items
Q81 Standing close to a person who has body odor .616 .302*
Q82 Shaking hands with a stranger who has sweaty palms .579 .277***
Q83 Stepping on dog poop .640 .308**
Q84 Accidentally touching a person’s bloody cut .725 .325*
Q85 Seeing some mold on old leftovers in your refrigerator .700 .306* Q86 Sitting next to someone who has red sores on their arm .547 .258***
Q87 Seeing a cockroach run across the floor .667 .324**
* p < .005, ** p < .01, *** p < .05
Table 2:
Sexual Disgust Sensitivity Component Matrix
Label Items Component Correlation
with other items Q88 Bringing someone you just met back to your room to have sex .802 .459*
Q89 Watching a pornographic video .811 .450***
Q810 A stranger of the opposite sex intentionally rubbing your thigh in the elevator
.393 .207
Q811 Having anal sex with someone of the opposite sex .802 .452*
Q812 Performing oral sex .661 .346***
Q813 Finding out that someone you don’t like has sexual fantasies about you
.687 .382*
17 Table 3:
Moral Disgust Sensitivity Component Matrix
Label Items Component Correlation
with other items
Q814 Forging someone’s signature on a legal document .675 .331*
Q815 Intentionally lying during a business transaction .707 .351*
Q816 Stealing from a neighbor .667 .327*
Q817 A student cheating to get good grades .655 .326*
Q818 Shoplifting a candy bar from a convenience store .581 .284*
Q819 Deceiving a friend .656 .326*
Q820 Cutting to the front of a line to purchase the last few tickets to a show .564 .274* * p < .005, ** p < .01, *** p < .05
A reliability analysis was also run to verify whether the items from the Three-Domain Disgust Sensitivity scale were internally consistent. Pathogen disgust sensitivity was based on 7 items, and produced a good Cronbach’s alpha (a = .76). The sexual disgust sensitivity trait was measured using 6 items and also produced a good Cronbach’s alpha (a = .79). Finally, the moral disgust sensitivity was based on 7 items, which produced a good Cronbach’s alpha as well (a = .74). These results show that the items used reliably describe the Three-Domain Disgust Sensitivity scale, meaning we can therefore compute the new variables.
The variables for each of the three Disgust Sensitivity scales was constructed by taking the average of the items in each factor. For pathogen disgust sensitivity, the data shows that the average participant found the scenarios described “disgusting” (M = 4.97, SD = .87), with responses ranging from “not very disgusting” (min = 2.71) to “Extremely Disgusting” (max = 7.00). Sexual disgust sensitivity had the lowest overall scores (M = 3.54, SD = 1.17), with responses ranging from “Not at all Disgusting”
18
(min = 1.00) to “Extremely Disgusting” (max = 6.50). Finally, moral disgust had the highest scores overall (M = 5.54, SD = .76), with scores ranging from “Not very disgusting” (min = 2.71) to “Extremely disgusting” (max = 7.00). In order to conduct the two-way ANOVAs necessary to test the hypotheses, the disgust sensitivity variables have to be turned into binary variables. All three variables were divided into two groups by taking the median split approach to separate them into those with low disgust sensitivity and those with high disgust sensitivity. For pathogen disgust the split was at 5.14, for sexual disgust at 2.75, and for moral disgust at 5.57. These variables were then used in the upcoming analysis.
Randomization Check
The next step is to verify whether the participants were randomly assigned to the two conditions in each of the two experiments. To check whether the ages in the two conditions in each experiment were comparable, participants were divided into two groups using a median split approach. Participants who were 23 or younger were placed into the young group and those above the age of 23 were placed in the old group. Using these two groups two independent samples t-tests were run. For the moral scandal experiment the exposure to the politician’s scandal was the independent variable, with age as the dependent. The Levene’s test for equality of variances was insignificant, F = .232, p = .631, and therefore equal variances can be assumed between the two groups. The results show there was no significant difference between the ages of the participants in scandal condition and the no-scandal condition, t(99) = .281, p = .779. For the sex scandal experiment, the same steps were repeated, with the Levene’s test for equality of variances again being insignificant, F = .536, p - .466. Here there was also no significant difference between the ages of participants in the scandal condition and the no-scandal condition, t(99) = .509, p = .612.
To check if the gender distributions in the two conditions of each experiment were comparable with each other, independent samples t-tests were conducted. To do so, gender was treated as a binary variable, excluding the only entry which had “Other” as their gender. For the moral scandal experiment, the Levene’s test for equality of variances was insignificant, F = .716, p = .400, so equality of variances
19
can be assumed. The results show there was no significant difference between the two groups, t(99) = -.464, p = .644, as the gender distribution across the two conditions was comparable. Another t-test was conducted to compare the gender distributions of the participants in the two conditions of the sex scandal experiment. The Levene’s test was insignificant, F = .408, p = .524, and equal variances could be assumed between the two groups. Here too the results were not significant, t(99) = -.330, p = .742, showing that the gender distribution of the participants in the scandal and no-scandal conditions of the sex scandal experiment was also not significantly different. It is clear therefore that the random assignment of the participants into one of the two conditions for each experiment was successful, meaning that we can proceed with the manipulation check.
Manipulation Check
To verify whether the experimental manipulation of the two experiments worked, at the end of the survey participants were asked to complete a two-question manipulation check. For the Moral Disgust experiment the participants were asked “Did Jacob van Scheppingen put his own interests above the interests of his political party?”, for which they had to answer with “Yes” or “No”. 80.8% of the 52 participants in the scandal condition correctly stated that Jacob van Scheppingen had been putting his own interests above the party’s, and 28.6% of the 49 participants in the no-scandal condition incorrectly stated the same. An independent sample t-test was conducted in order to test whether the responses in the manipulation check for the moral scandal experiment differed significantly between its two conditions. The Levene’s test for equality of variances came back significant, so equal variances cannot be assumed between these two groups. The results show the manipulation check responses between the participants in the two experimental conditions was significantly different, t(95.35) = 6.13, p < .000, which indicates that the experimental manipulation was successful.
For the Sexual Disgust experiment the participants were asked: “Did Teun van Vuren have an affair with his secretary?”, which was also answered with “Yes” or “No”. 93.9% of the 49 participants in the scandal condition correctly stated that Teun van Vuren did have an affair, and 15.4% of the 52
20
participants in the no-scandal condition incorrectly stated that he did. Another independent samples t-test was run to verify that the responses of these two groups was significantly different. Once again the Levene’s test for equality of variances came back significant, so therefore equal variances cannot be assumed between the two groups. The results show that the responses to the sexual scandal manipulation check of the participants in the two experimental conditions was significant as well, t(89.21) = 12.82, p < .000, and we are therefore certain the manipulation was successful for the second experiment as well. With both experiments showing that the manipulation of the experimental conditions was successful, we can move on to run the analytical tests to prove (or disprove) the hypotheses.
Results Section Main effect Results
For both the moral scandal and the sexual scandal experiments participants were asked to answer three questions about the politicians they read. Study participants were asked to rate on a 0-100 scale how disgusting they found the candidate, how cold or warm they felt towards the candidate (also on a 0-100 scale), and how likely they would be to vote for this politician on 7-point Likert scale, from “Not at all likely” to “Extremely likely”. To compare the answers of the participants in the scandal condition with the no-scandal condition, two independent samples T-tests were conducted. All the results can be found in tables 4-9 in appendix B.
Starting with the disgust dependent variable, the Levene’s test for equality of variances was significant (F = 7.81, p = .006), so equality of variances between the two groups cannot be assumed. Participants who were exposed to the political scandal condition found the politician significantly more disgusting (M = 63.08, SD = 24.57) in comparison to the participants in the no-scandal condition (M = 10.84, SD = 20.16), t(97.20) = 11.705, p < .000, d = 2.32. Furthermore, the calculated effect size was extremely large, offering the piece of evidence of evidence in support of the first hypothesis.
21
Moving on to the Feeling Thermometer dependent variable, the Levene’s test for equality of variances was significant (F = 6.53, p = .012), meaning that equality of variances cannot be assumed. The results show that there was a very large, significant difference between the two groups (t(92.872) = -7.59, p < .000, d = 1.5), as participants who were in the scandal condition reported feeling significantly less warm about the political candidate (M = 34.69, SD = 27.43) compared to participants in the no-scandal condition (M = 70.67, SD = 19.82),. The results again support the first hypothesis that politicians who have their scandals exposed receive more negative evaluations.
Wrapping up the moral scandal experiment, the Levene’s test for equality of variances for the voting intention dependent variable was insignificant (F = .71, p = .402), meaning that equality of variances can be assumed. Participants in the moral scandal condition were significantly less likely to vote for the politicians (M = 2.30, SD = 1.53) compared to the participants in the moral no-scandal condition (M = 5.10, SD = 1.47), t(99) = -9.34, p < .000, d = 1.86. Moreover, the effect size was again very large. Therefore, the results of all the statistical tests run on the moral scandal experiment were not only significant, but also found very large effect sizes, supporting the claims made in the first hypothesis.
Next, we look at the sex scandal experiment, starting with the disgust dependent variable. The Levene’s test for equality of variances was insignificant (F = 3.14, p = .079), so equal variances between the two conditions can be assumed. The results show that participants in the scandal condition found the politicians as significantly more disgusting (M = 42.57, SD = 24.57) when compared to participants in the no-scandal condition (M = 24.85, SD = 29.35), t(99) = 3.28, p < .000, d = .65. The results of this experiment show there is a medium to large effect of exposure on politician’s evaluations, which providing further support for hypothesis 1.
Moving on to the Feeling Thermometer dependent variable, the Levene’s test for equality of variances was also insignificant (F = .92, p = .339) and equality of variances can be assumed. The results show that there was a large significant difference between the two groups (t(99) = -4.57, p < .000, d = .91), as participants who were exposed to the sexual scandal reported feeling colder towards the
22
politicians (M = 48.65, SD = 21.72) compared to participants who were not exposed to the scandal (M = 67.79, SD = 20.33). These findings further support the first hypothesis.
Finally, the Levene’s test for equality of variances in the voting intention dependent variable was significant (F = 4.39, p = .039), so equal variances are not assumed between the two conditions. It was found that the scandal condition participants were significantly less likely to vote for the politician (M = 3.73, SD = 1.60) than the no-scandal condition participants (M = 4.96, SD = 1.44), t(99) = -4.05, p < .000, d = .81. The observed effect size was also large, meaning that the last statistical test also supported the first hypothesis.
Significant results were found for all three dependent variables in both the moral scandal and sexual scandal experiments. Moreover, the effect sizes were found to be large for every dependent variable tested, as shown in the Cohen’s d calculation for each dependent variable. We can therefore confidently confirm hypothesis 1, that exposure to a politician’s scandals negatively affects their perception.
With the main effects established, the next step is to test whether the effect of exposure to a politician’s scandal on their perception is moderated by disgust sensitivity. To test this, six two-way ANOVAs were conducted. The first two-way ANOVA was conducted to test whether moral disgust sensitivity affects how disgusting individuals find politicians when they have been involved in a moral scandal. The results show that there was no significant interaction effect between the exposure to a political scandal and moral disgust sensitivity on how disgusting the politician was as, F(1, 97) = .743, p = .391, ηp
2
= .01.
The second two-way ANOVA on the moral scandal experiment tested whether the interaction effect of moral disgust sensitivity with the experimental condition affected how warmly participants felt about the politician. The results again show that there is no significant interaction effect between the exposure to a scandal and moral disgust sensitivity, F(1, 97) = 2.013, p = .159, ηp
2
23
Rounding out the moral scandal experiment, a two-way ANOVA was conducted to test whether voting intention is affected by the interaction effect between exposure to a scandal and moral disgust sensitivity. The results once again show that there isand how likely they were to vote for the politician, F(1, 97) = .000, p = .998, ηp
2
< .00.
None of the two-way ANOVAs conducted testing the interaction effect between exposure to a moral scandal and moral disgust sensitivity found significant results for any of the dependent variables. Moreover, the effect sizes for the interactions were small. We must therefore reject hypothesis 2, that the negative effect of exposure to a politician’s moral scandal on their perception is stronger for individuals with high moral disgust sensitivity.
Moving on to the sex scandal experiment, a two-way ANOVA was conducted to test whether sexual disgust sensitivity moderates how disgusting individuals perceive politicians when exposed to a sexual scandal. The results again show that there was no moderation effect between the exposure to sexual scandal and sexual disgust sensitivity on how disgusting the politicians was perceived as, F(1, 97) = .016, p = .900, ηp
2
< .00.
Following this, a two-way ANOVA was conducted to test whether there is an interaction effect between exposure to a sex scandal and sexual disgust sensitivity on how warmly people feel towards a political candidate. The findings found no significant interaction effect, F(1, 97) = .475, p = .492, ηp
2
< .01
Finally, a two-way ANOVA was conducted to test whether voting intention is affected by the interaction effect between exposure to a sex scandal and sexual disgust sensitivity. Once again, there was no significant interaction effect between exposure to a sex scandal and sexual disgust sensitivity, F(1, 97) = .078, p = .781, ηp
2
< .00.
Just like with the moral scandal experiment, the sexual scandal experiment did not find any significant interaction effect between exposure to a sex scandal and sexual disgust sensitivity.
24
Furthermore, the effect sizes for each interaction were even smaller than the previous comparison. Hypothesis 3, which claimed that the negative effect of exposure to a politician’s sex scandal on their perception is stronger for individuals with high sexual disgust sensitivity, is therefore also rejected.
Discussion and Conclusion
The results of the experiments confirm that scandals, be they because of a moral failing or unacceptable sexual behavior, impacts how politicians are evaluated by the public. In both experiments, participants exposed to the scandal found the politicians to be more disgusting, felt less warmly towards them, and were less likely to vote for them. The calculated effect sizes also indicate that exposure to a politician’s scandal has a large effect on how they are perceived, which suggests that for many people a scandal is a deal-breaker when it comes to a political candidate they were not familiar with previously.
As mentioned in the results section, the effect sizes observed in this experiment were unusually large, for which there are two likely explanations, the first being that the external validity of this study is not very strong due to the experimental nature of this study. Participants were exposed to fictional political candidates and then asked to give their opinions based on the limited information provided. If the politicians described in the experiment had been real people we might have seen weaker effect sizes when the information provided is taken into consideration with the information already known about them. It could also be that the results of these tests might be inflated by a social desirability bias. In the introduction to the experiment it was made clear to the participants that the study would be on the perceptions of politicians. It is therefore possible that, knowing this, participants would give overwhelmingly positive responses to the politicians in the no-scandal conditions of both experiments, since the information provided in those stimuli was overwhelmingly positive. By that same token, in the
25
scandal conditions participants might have realized that the scandals were the experimental conditions, and given very negative responses because they believed that is what the aim of the research was.
Neither of the tests on moral and sexual disgust sensitivity found a significant interaction effect with the exposure to a political scandal on any of the dependent variables, meaning that disgust sensitivity does not have a moderating effect on how politicians are evaluated when people are exposed to their scandals. One possible reason behind the insignificant results for the interaction effects is of course the sample size. While 101 participants were enough to cover the requirements needed to run the main effects, it was probably not sufficient enough to run the interaction effects. Follow-up studies might be able to shed more light on whether this is a sampling issue, or whether there really even is a moderation effect to be observed in the first place.
These findings raise more questions than they do answers, as the existing academic literature heavily suggested that disgust sensitivity had a function in encouraging avoidance of dangerous individuals by moralizing their behaviors. It was mentioned earlier disgust does not always have to have a moralizing function, and the lack of significant results could suggest that politics is a field where disgust does not have this function. We know from the results that the individuals in scandal conditions found the politicians to be much more disgusting, but this does not always mean that the scandals were necessarily moralized. This suggests that there must be a different variable effecting how different individuals react to political scandals.
The findings of this study show there are many possibilities for future research on both scandals and disgust sensitivity. While this study decided to omit any contextual factors that might impact the exposure effect of scandals, further research could examine whether the inclusion of contextual information (such as the politician’s name, their political beliefs, or their party) might moderate how a scandal affects their perception by citizens. Additionally, improving on the current experiment and conducting further tests on the role of moral and sexual disgust sensitivity could prove worthwhile, before focusing on If there is one thing that is clear from this study, it is that political scandals definitely have an
26
impact on the perception of the politicians involved, but how exactly they are affected remains a complicated and (partially) unanswered question within the field of communication science.
27 References
Aaroe, L., Petersen, M. B., & Arceneaux, K. (2017). The behavioral immune system shapes political intuitions: Why and how individual differences in disgust sensitivity underlie opposition to immigration. American Political Science Review, 111(2), 277–294. https://doi.org/10.1017/S0003055416000770
Bhatti, Y., Hansen, K. M., & Olsen, A. L. (n.d.). Political hypocrisy: The effect of political scandals on candidate evaluations, 48, 408–428. https://doi.org/10.1057/ap.2013.6
Billingsley, J., Lieberman, D., & Tybur, J. M. (2018). Sexual Disgust Trumps Pathogen Disgust in Predicting Voter Behavior During the 2016 U . S . Presidential Election, (June), 1–15. https://doi.org/10.1177/1474704918764170
Bowler, S., & Karp, J. A. (2004). Politicians, scandals, and trust in government. Political Behavior, 26(3), 271–287. https://doi.org/10.1023/B:POBE.0000043456.87303.3a
Brenner, C. J., & Inbar, Y. (2015). Disgust sensitivity predicts political ideology and policy attitudes in the Netherlands †, 38(November 2014), 27–38.
Chapman, H. A., Kim, D. A., Susskind, J. M., & Anderson, A. K. (2009). In Bad Taste : Evidence for the Oral, (February), 1222–1227.
Chapman, H. A., & Anderson, A. K. (2013). Things Rank and Gross in Nature : A Review and Synthesis of Moral Disgust, 139(2), 300–327. https://doi.org/10.1037/a0030964
Dancey, L. (2012). The Consequences of Political Cynicism: How Cynicism Shapes Citizens’ Reactions to Political Scandals. Political Behavior, 34(3), 411–423.
https://doi.org/10.1007/s11109-011-9163-z
Gerber, A. S., Karlan, D., Bergan, D., American, S., Journal, E., Economics, A., … Berg, D. (2019). Does the Media Matter ? A Field Experiment Measuring the Effect of Newspapers on Voting Behavior
28
and Political Opinions Does the Media Matter ? A Field Experiment Measuring the Effect of Newspapers on Voting Behavior and Political Opinions * media the new, 1(2).
Graham, J., Haidt, J., Koleva, S., Motyl, M., Iyer, R., Wojcik, S. P., & Ditto, P. H. (2013). Moral Foundations Theory : The Pragmatic Validity of Moral Pluralism (Vol. 47).
Gutierrez, R., & Giner-sorolla, R. (2007). Kent Academic Repository, 7, 853–868.
Haidr, J. (2003). Chapter 53 : The Emotional Dog and Its Rational Tail : A Social Intuitionist Approach to Moral Judgment. (1995), 814–834.
Haidt, J., & Graham, J. (2007). When Morality Opposes Justice : Conservatives Have Moral Intuitions that Liberals may not Recognize, 20(1), 98–116. https://doi.org/10.1007/s11211-007-0034-z
Haumer, F., & Donsbach, W. (2009). The rivalry of nonverbal cues on the perception of politicians by television viewers. Journal of Broadcasting and Electronic Media, 53(2), 262–279. https://doi.org/10.1080/08838150902907918
Hodson, G., & Costello, K. (2007). Interpersonal Disgust , Ideological Orientations , and Dehumanization as Predictors of Intergroup Attitudes, 18(8), 691–698.
Horberg, E. J., Oveis, C., & Cohen, A. B. (2009). Disgust and the Moralization of Purity, 97(6), 963–976. https://doi.org/10.1037/a0017423
Horberg, E. J., Oveis, C., Keltner, D., & Cohen, A. B. (2009). Disgust and the Moralization of Purity. Journal of Personality and Social Psychology, 97(6), 963–976. https://doi.org/10.1037/a0017423
Inbar, Y., Pizarro, D., Iyer, R., & Haidt, J. (2012). Disgust Sensitivity , Political Conservatism , and Voting. https://doi.org/10.1177/1948550611429024
Jones, A., & Fitness, J. (2008). Moral Hypervigilance : The Influence of Disgust Sensitivity in the Moral Domain. Emotion, 8(5), 613–627. https://doi.org/10.1037/a0013435
29
Landy, J. F., & Goodwin, G. P. (2015). Does incidental disgust amplify moral judgment? A meta-analytic review of experimental evidence. Perspectives on Psychological Science, 10(4), 518–536. https://doi.org/10.1177/1745691615583128
Maier, J. (2011). The impact of political scandals on political support: An experimental test of two theories. International Political Science Review, 32(3), 283–302.
https://doi.org/10.1177/0192512110378056
Miller, A. H., & Linda, T. (2019). Sex , Politics , and Public Opinion : What Political Scientists Really Learned From the Clinton-Lewinsky Scandal *, 32(4), 721–729.
Molho, C., Tybur, J. M., Güler, E., Balliet, D., & Hofmann, W. (2017). Disgust and Anger Relate to Different Aggressive Responses to Moral Violations. Psychological Science, 28(5), 609–619. https://doi.org/10.1177/0956797617692000
Olatunji, B. O., Adams, T., Ciesielski, B., David, B., Sarawgi, S., & Broman-Fulks, J. (2012). The Three
Domains of Disgust Scale. Assessment, 19(2), 205–225.
https://doi.org/10.1177/1073191111432881
Pizarro, D., & Helion, C. (2011). On Disgust and Moral Judgment, 3(3), 267–268. https://doi.org/10.1177/1754073911402394
Russell, P. S., & Giner-sorolla, R. (2011). Moral Anger Is More Flexible Than Moral Disgust. Emotion. https://doi.org/10.1177/1948550610391678
Schnall, S., Clore, G. L., & Jordan, A. H. (2008). Disgust as Embodied Moral Judgment, 34(8), 1096– 1109. https://doi.org/10.1177/0146167208317771
30
Tybur, J. M., Bryan, A. D., Lieberman, D., Caldwell Hooper, A. E., & Merriman, L. A. (2011). Sex differences and sex similarities in disgust sensitivity. Personality and Individual Differences, 51(3), 343–348. https://doi.org/10.1016/j.paid.2011.04.003
Tybur, J. M., Lieberman, D., & Griskevicius, V. (2009). Microbes, Mating, and Morality: Individual Differences in Three Functional Domains of Disgust. Journal of Personality and Social Psychology, 97(1), 103–122. https://doi.org/10.1037/a0015474
Tybur, J. M., Lieberman, D., Kurzban, R., & DeScioli, P. (2013). Disgust: Evolved function and structure. Psychological Review, 120(1), 65–84. https://doi.org/10.1037/a0030778
Wager, T. D., Barrett, L. F., Bliss-moreau, E., & Lindquist, K. (2008). Neuroimaging of emotion 1 (pp. 757–776).
Walter, A. S., & Redlawsk, D. P. (2019). Voters ’ Partisan Responses to Politicians ’ Immoral Behavior, 0(0), 1–23. https://doi.org/10.1111/pops.12582
Wheatley, T., & Haidt, J. (2005). Hypnotic Disgust Makes Moral Judgments More Severe, 16(10), 780– 784.
Winkielman, P., & Berridge, K. C. (2004). Unconscious Emotion, 13(3), 120–123.
Winters, M. S., & Weitz-shapiro, R. (2013). Lacking Information or Condoning Corruption When Do Voters Support Corrupt Politicians ?, 418–436.
31 Appendix A: Pre-test statements and data
Pre-test Results for Moral Scandal Experiment
Act or behavior described How morally
acceptable is this?
How negative or positive is this?
M SD M SD
Receiving debate questions in advance 4.64 1.75 3.77 1.47
Blackmailing party members for one’s own benefit
1.41 .65 1.32 .47
Plagiarizing someone else’s political speech 2.50 1.08 2.23 .79 Receiving illegal campaign donations worth over
€10,000.
1.50 .99 1.64 .77
Granting a contract in exchange for a bribe 1.59 .94 1.64 .98
Pre-test Results for Sex Scandal Experiment
Act or behavior described How sexually
acceptable is this?
How negative or positive is this?
M SD M SD
Being the offspring of first-degree cousins 4.52 1.65 3.67 .65 Having a child out of wedlock and not
supporting the mother in raising the child
2.41 1.11 2.36 .98
Going to sex parties with their partner 5.23 1.47 3.86 .55
Having an affair with their secretary 2.09 1.20 1.95 .71
32 Appendix B: Two Way ANOVA Results Tables
Table 4:
Between Subjects Effects for Disgust Evaluation Dependent Variable
Predictor Sum of Squares df Mean Square F p Partial η2
Corrected Model 69708.760 3 23236.253 45.560 .000 .585 (Intercept) 136126.454 1 136126.45 266.909 .000 .733 Exposure to Political Scandal 69129.667 1 69129.667 135.546 .000 .583 Moral Disgust 506.723 1 506.723 .994 .321 .010 Exposure to Political Scandal * Moral Disgust 379.016 1 379.016 .743 .391 .008 Error 49471.022 97 510.011 Total 262979.000 101 Corrected Total 119179.782 100
33 Table 5:
Between Subjects Effects for Feeling Thermometer Dependent Variable
Predictor Sum of Squares df Mean Square F p Partial η2
Corrected Model 33910.875 3 11303.625 19.583 .000 .377 (Intercept) 277836.242 1 277836,24 481.339 .000 .832 Exposure to Political Scandal 33382.659 1 33382.659 57.834 .000 .374 Moral Disgust 70.864 1 70.864 .123 .727 .001 Exposure to Political Scandal * Moral Disgust 1161.648 1 1161.648 2.013 .159 .020 Error 55989.897 97 577.215 Total 364567.000 101 Corrected Total 89900.772 100
34 Table 6:
Between Subject Effects for Voting Intention Dependent Variable
Predictor Sum of Squares df Mean Square F p Partial η2
Corrected Model 197.009 3 65.670 28.495 .000 .468 (Intercept) 1377.706 1 1377.706 597.809 .000 .860 Exposure to Political Scandal 196.080 1 .021 .85.082 .000 .467 Moral Disgust .021 1 .021 .009 .924 .000 Exposure to Political Scandal * Moral Disgust 1.117E-5 1 1.117E-5 .000 .998 .000 Error 223.545 97 2.305 Total 1776.000 101 Corrected Total 420.554 100
35 Table 7:
Between Subjects Effects for Disgust Evaluation Dependent Variable
Predictor Sum of Squares df Mean Square F p Partial η2
Corrected Model 11420.600 3 3806.867 5.319 .002 .141 (Intercept) 114268.568 1 114268.568 159.643 .000 .622 Exposure to sex Scandal 7184.267 1 7184.267 10.037 .002 .094 Sexual Disgust 3491.577 1 3491.577 4.878 .030 .048 Exposure to Sex Scandal * Sexual Disgust 11.397 1 11.397 .016 .900 .000 Error 69430.350 97 715.777 Total 193830.000 101 Corrected Total 80850.950 100
36 Table 8:
Between Subjects Effects for Feeling Thermometer Dependent Variable
Predictor Sum of Squares df Mean Square F p Partial η2
Corrected Model 9627.691 3 3209.230 7.182 .000 .182 (Intercept) 341459.464 1 341459.464 764.199 .000 .887 Exposure to sex Scandal 9033.503 1 9033.503 20.217 .000 .172 Sexual Disgust 189.279 1 189.279 .424 .517 .004 Exposure to Sex Scandal * Sexual Disgust 212.086 1 212.086 .475 .492 .005 Error 43341.557 97 446.820 Total 398675.000 101 Corrected Total 52969.248 100
37 Table 9:
Between Subjects Effects for Voting Intention Dependent Variable
Predictor Sum of Squares df Mean Square F p Partial η2
Corrected Model 38.850 3 12.950 5.495 .002 .145 (Intercept) 1900.671 1 1099.671 806.511 .000 .893 Exposure to sex Scandal 37.115 1 37.115 15.749 .000 .140 Sexual Disgust .715 1 .715 .303 .583 .003 Exposure to Sex Scandal * Sexual Disgust .183 1 .183 .079 .781 .001 Error 228.596 97 2.357 Total 2193.000 101 Corrected Total 267.446 100
