Not practicing what they preached! Exploring negative spillover effects of news about ex-politicians’ hypocrisy on party attitudes, voting intentions, and political trust

ABSTRACT Political hypocrisy – a frequent feature of contemporary politics – oftentimes occurs when politicians resign from office and then engage in behavior that is in fundamental opposition to the standpoints they originally campaigned for as incumbents. Previous research has neglected to examine negative spillover effects of news about ex-politicians’ hypocritical behavior. Drawing from the inclusion/exclusion model and the feelings-as-information model, we conducted two experiments in two different countries and used different stimuli to increase external validity. Results suggest a dual process account of scandal spillover effects (an attitudinal and emotional mechanism) revealing that hypocrisy negatively affected both attitudes and emotions toward an ex-politician. Mediation analysis further showed that evaluations in turn negatively affected attitudes and voting intentions for the party the hypocritical politician used to belong to (attitudinal spillover process). No effects on general political trust emerged. In contrast, negative emotions had no effect on party attitudes and voting intentions but decreased political trust toward politicians in general (emotional spillover process). In line with the inclusion/exclusion model, the results help to explain inconsistent findings in previous studies that did not account for the suggested dual process account of spillover effects and underline the eroding effects of hypocrisy.

politicians frequently impose strict moral standards on other people or politicians but practice less strict moral behavior themselves (Jordan, Sommers, Bloom, & Rand, 2017). A candidate's political behavior of condemning immoral behavior that he or she in fact engages in, oftentimes serves as a starting point for a political scandal (Bhatti, Hansen, & Olsen, 2013;McDermott et al., 2015). Scandal has been defined "as the intense public communication about a real or imagined defect that is by consensus condemned, and that meets universal indignation or outrage" (Esser & Hartung, 2004, p. 1041see also von Sikorski, 2017see also von Sikorski, , 2018. Current research shows that news about scandals involving political hypocrisy frequently results in very negative emotional and attitudinal reactions in news consumers (e.g., McDermott et al., 2015) and "hypocrites are disliked because they falsely signal that they behave morally" (Jordan et al., 2017, p. 366).
Although, elected political actors frequently engage in more or less severe forms of hypocrisy (Runicam, 2018) the most severe forms of hypocrisy oftentimes occur when politicians resign from office and then engage in behavior that is in fundamental opposition to the moral values and standpoints they campaigned for when still being in office. For example, during his incumbency former US president Barack Obama repeatedly criticized Wall Street and attacked banks and bankers for their role in the international financial crisis. After resigning from office, Obama made $1.2 million from three Wall Street speeches, which triggered critical media coverage and an intensive debate (Buncombe, 2017). Former German chancellor Gerhard Schröder negotiated a pipeline project with Russia and its partially state owned enterprise Gazprom in the last weeks of his incumbency. He then resigned from all posts (he was not re-elected) and joined a Gazprom subsidiary as a chairman shortly after the elections triggering a heavily debated scandal. Schröder was accused of hypocrisy and of "concealing a conflict of interest" (Kepplinger, Geiss, & Siebert, 2012, p. 667). Similarly, former Austrian Green Party chairwoman Eva Glawischnig resigned from politics and joined the gambling company Novomatic. During her incumbency, she heavily criticized the company and stated that Novomatic promotes gambling addiction. Glawischnig's resignation and change of career resulted in a large scandal and a debate about political hypocrisy (Reisinger, 2018). Also, the question arose, if hypocrisy of former political actorsbesides negatively affecting their own reputationresults in negative spillover effects on their former political party and citizens' political trust toward other politicians more generally.
In this context, one may argue that no such negative spillover effects (e.g., on a politician's former party) should occur, since the former politician (after resigning from all posts and giving up party membership) istechnically speakingno longer part of his or her former party. Hypocritical behavior that occurs after a politician's resignation, therefore, has no official connection to his or her former job or political party. On the other hand, citizens may continue to perceive a former politician (even if he or she has already resigned) as an (typical) exemplar of the political elite and his or her former political party. It can thus be argued that hypocrisy may very well result in negative spillover effects. Such potential spillover effects would have important implications from a media psychology perspective but would also be relevant for both journalism practice and political parties (and their communication strategies) that hypocritical ex-politicians used to belong to.
To date, we lack research that has systematically examined the effects of news about political hypocrisy of former political actors. In this research we report here, we aimed at closing this research gap. The purpose of our study was twofold. First, and theoretically based on the inclusion/exclusion model (Bless & Schwarz, 2010), we examined if hypocrisy of former scandalous political actors resulted in negative spillover effects on news recipients' (a) party attitudes, and (b) voting intentions (politician's former party), and (c) political trust toward politicians more generally. Second, theoretical models in media effects research (e.g., Valkenburg & Peter, 2013) as well as evidence from political communication (e.g., Kepplinger et al., 2012) suggest that attitudinal and emotional processes should be differentiated and may mediate media effects in separate ways (Valkenburg & Peter, 2013). However, previous research has neglected to study the particular effects of emotions and its role for scandal spillover effects. Thus, we theorized and tested if negative emotions (anger, fury)in an affective transfer process (Schwarz, 2012) and in parallel to an attitudinal spillover mechanism mediate and thus help to explain potential negative consequences of political hypocrisy on party attitudes, voting intentions, and political trust. We tested our assumptions with two experimental studies in two European countries using multiple stimuli to increase external validity of our research (Reeves, Yeykelis, & Cummings, 2016). Study 1 was an online experiment using a quota-based sample in Germany, whereas Study 2, conducted to replicate Study 1, was an online experiment using a nonstudent convenience sample in Austria.

The negative consequences of hypocrisy in political scandals
Previous research consistently shows that hypocritical political behavior has very negative consequences and can affect individuals' attitudes toward political candidates (Bhatti et al., 2013;Jordan et al., 2017;McDermott et al., 2015). Also, political misconduct canindependently of particular cognitive effectsresult in negative emotional responses, i.e., anger (Kepplinger et al., 2012). In the following sections, we will therefore address the attitudinal and emotional consequences of hypocrisy in depth.

Attitudinal effects
Previous research clearly shows that individuals' candidate evaluations are significantly and negatively affected when political actors are involved in political scandal (Bhatti et al., 2013;Carlson, Ganiel, & Hyde, 2000;Funk, 1996;Lee, 2018;McDermott et al., 2015;Régner & Le Floch, 2005;von Sikorski & Hänelt, 2016;von Sikorski & Ludwig, 2018; for an overview see von Sikorski, 2018). However, hypocrisy in political scandals adds an extra level to scandalous behavior. By previously using condemnation of a certain behavior (e.g., "paying no taxes is wrong") a politician implies that he or she will behave morally in the future. According to Jordan et al. (2017), this form of falsely signaling moral behavioror preaching water and drinking winethen results in especially negative evaluations of a target person (e.g., a politician).
So far, two studies have specifically examined the effects of hypocrisy in political scandals. First, Bhatti et al. (2013) experimentally tested how political hypocrisy in the context of different types of political scandals (e.g., bankruptcy, drunk driving) affected individuals' evaluations of political candidates (i.e., trustworthiness). Their results revealed that hypocrisy significantly worsened candidate evaluations in political scandals, e.g., "a social democrat is punished more for using private hospitals than a Liberal, because their party position is to be skeptical towards private hospitals" (Bhatti et al., 2013, p. 424). Similarly, McDermott et al. (2015) showed that hypocritical scandals (compared with nonhypocritical scandals) decreased individuals' voting intention for hypocritical candidates and generally resulted in more negative evaluations of political actors. Based on these findings, we expected that news recipients' attitudes toward a former political actor will also be negatively affected, due to signaling moral behavior that he or she in fact does not engage in after resigning from office. Thus, our first hypothesis reads as follows: H1: Hypocrisy of a former political actor will negatively affect news recipients' attitudes toward the former politician.

Emotional effects
Furthermore, it can be argued that hypocrisy should result in an increase in negative emotions toward a former political actor. Emotions can generally be defined as "an episode of interrelated, synchronized changes in the states of all or most of the five organismic subsystems in response to the evaluation of an external or internal stimulus event" (Scherer, 2005, p. 697). In line with attribution theory (Weiner, 1985), anger and fury can be regarded as important emotions in this context because these negative emotions are regularly triggered by negatively evaluated events or behaviors (i.e., hypocrisy) in which the particular actor responsible for an event or behavior is in control . Put differently, hypocrites are usually in control when they condemn norm transgressions they in fact engage in (Jordan et al., 2017). Signaling moral behavior and then intentionally breaking own (strictly imposed) moral standards should elicit anger in individuals (Kepplinger et al., 2012). Based on this theorizing, and empirical results on the anger and fury eliciting effects of political norm transgression (Kepplinger et al., 2012), we formulated our second hypothesis: H2: Hypocrisy of a former political actor will increase news recipients' negative emotions toward the former politician.

Spillover effects of political hypocrisy
In line with previous research (e.g., Lee, 2018), spillover effects can generally be explained with the help of the accessibility-diagnosticity framework. The framework is based on spreading activation theory (Collins & Loftus, 1975) and postulates that people use implicit theories on how particular things are related in the world. Spreading activation theory assumes that specific concepts, categories, and attributes are related in a network and can activate one another (Collins & Loftus, 1975). Thus, a scandal spillover effect may occur when accessible and negative information regarding a specific category (e.g., hypocritical politician A) is processed and when the negative information is automatically perceived to be applicable to the category in general, e.g., politicians XYZ (Lee, 2018).
So far, it remains unclear if hypocritical individual behavior results in particular spillover effects and previous research that has studied spillover effects of political norm transgressions showed mixed results. For example, some studies revealed that single-politician scandals have negative effects on citizens' trust toward the government (e.g., Lee, 2018), other studies show no such spillover effects for single-politician scandals (e.g., Chanley, Rudolph, & Rahn, 2000;Maier, 2011). Furthermore, some studies show negative spillover effects on the evaluation of political parties that scandalous politicians belong to (e.g., Maier, 2011), as well as on other politicians (e.g., Bowler & Karp, 2004: Maier, 2011. In this regard, Lee (2018) has pointed out that scholars "seldom elaborate the theoretical underpinnings of scandal spillover effects" (Lee, 2018, p. 716) and this shortcoming may in parts explain the inconsistent findings. In accordance with this criticism, a different line of research shows that spillover effects only occur under very specific circumstances (Puente-Diaz, 2015;Régner & Le Floch, 2005;Schwarz & Bless, 1992).
More specifically, drawing from the inclusions/exclusion model scholars theorized that attitudinal spillover effects only occur when a specific exemplar (i.e., a hypocritical political actor) can be included in the representation formed of a certain superordinate category, e.g., the political party a politician belongs to (Puente-Diaz, 2015). Furthermore, research informed by the feelings-as-information model (Schwarz, 2012) also suggests that (negative) emotions can influence judgments and result in spillover effects (independently of attitudinal carry-over effects). In the following section, this attitudinal and emotional transfer process and the underlying theoretical models are described in detail.

Attitudinal transfer process
The inclusion/exclusion model further specifies the aforementioned theorizing on spillover effects and rests on the premise that human judgment is context dependent and that we judge certain target objects by comparing them to a specific standard or reference category (Bless & Schwarz, 2010). For example, today's (objectively measured) temperature can feel rather cold or warm depending on yesterday's temperature. Bless and Schwarz describe context effects as assimilation or contrast effects. An assimilation effect occurs when, e.g., the negative evaluation of one politician spills over and negatively affects the evaluation of other politicians more generally. In contrast, a contrast effect occurs when the negative evaluation of one politician is used as a standard of comparison and results in more positive evaluation of another politician. In an experimental study (Schwarz & Bless, 1992), showed that when information about a scandalous politician (e.g., Richard Nixon) was made accessible it decreased trust in politicians in general; thus, "reflecting that the exemplar can be included in the representation formed of the political class (i.e., a superordinate category), resulting in an assimilation effect" (Schwarz & Bless, 2007, p. 124). In contrast, if individuals were asked to compare a scandalous politician with a specific other politician (e.g., Bill Clinton), "the primed exemplar cannot be included in the representation formed of the target (after all, Bill Clinton is not Richard Nixon)" (Schwarz & Bless, 2007, p. 124) and a contrast effect emerged. Importantly, depending on its usage the same piece of information (e.g., negative evaluation of one politician) can therefore result in both assimilation and contrast effects (Schwarz & Bless, 1992; see also Bless, Igou, Schwarz, & Wänke, 2000).
More recently, Régner and Le Floch (2005) as well as Puente-Diaz (2015) replicated these findings. For example, in Puente-Diaz's (2015) study an assimilation effect occurred (i.e., more negative evaluation of a target politician not involved in a scandal) when a non-scandalous politician was presented together with a scandalous political actor of the identical political party. Maier (2011) showed that accessible scandalous information (single-politician scandal) negatively affected individuals' political attitudes and resulted in a negative spillover effect on individuals' evaluation of the political party the scandalous politician belonged to. Transferred to the context of the present study, we expected that negative attitudes toward a former political actor (triggered by hypocritical behavior, see H1)in line with the inclusion/exclusion modelwill result in an assimilation effect negatively affecting news recipients' party attitudes as well as their general political trust toward politicians. The rationale behind this was that the exemplar (scandalous former politician) can be included in both the representation of politicians in general and the representation of the political party the former politician used to belong to. In line with this theorizing, we further expected that voting intentions for the former party of the scandalous politician will decrease. This assumption is also supported by previous research on the effects of hypocrisy (McDermott et al., 2015) as well as meta-analytical findings showing that candidate attitude is a particularly strong predictor of voting intention (Kim & Hunter, 1993). Thus, hypotheses 3a-3c read as follows: H3: Negative attitudes toward a scandalous former politician negatively predict news recipients' (a) party evaluation, (b) intention to vote for the former party of the scandalous actor, and (c) general political trust toward politicians.

Affective transfer process
In line with our aforementioned reasoning, affective reactions triggered by hypocritical behavior may also result in spillover effects. This assumption is in line with the feelings-as-information model (Schwarz, 2012). In general, the model conceptualizes the role of subjective emotional experiences (e.g., happiness, anger) and how they affect subsequent judgements. Especially, the model helps to explain under which conditions specific emotional experiences influence unrelated evaluations or decision-making processes. When individuals mis-read affective responses to contextual information (e.g., Dunn & Schweitzer, 2005;Schwarz, 2012) in evaluating a target, negative emotions like anger and fury (triggered by hypocrisy, H2) can result in negative spillover or assimilation effects when making particular evaluations, e.g., forming attitudes toward a political party (Schwarz & Bless, 2007). However, spillover effects on specific judgments are only based on feelings when these affective experiences are relevant for the specific task at hand. That is, negative emotions toward a political actor can result in more negative evaluations of his or her political party (feelings are relevant for the judgment of the target). Yet, the same emotional reaction should not affect the evaluation of, e.g., a candy bar because feelings are irrelevant for the judgement of the target (Dunn & Schweitzer, 2005;Schwarz, 2012). It follows that negative emotions toward a former politicianas relevant feelings for political judgmentswill have a negative spillover effect on political trust toward other politicians, party attitudes and voting intentions, as we formally articulated in our next set of hypotheses ( Figure 1 shows the full theoretical model): H4: Negative emotions toward a former scandalous politician negatively predict news recipients' (a) party evaluation, and (b) intention to vote for the former party of the scandalous actor, and (c) general political trust toward politicians.

Method
In Study 1, we conducted an online experiment in Germany employing a quota sample based on the demographic characteristics of the general population in Germany. We employed a between-subjects design with two experimental conditions. As a first step, we tested H1-H3b and H4a-H4b in Study 1. In Study 2, we (again) examined H1-H3b and H4a-H4b (conceptual replication) and extended our analysis also testing H3c and H4c.

Participants and procedure
The experiment was part of a larger online survey conducted by Survey Sampling International (SSI). In total, 283 participants were randomly assigned to the current study. We excluded 18 individuals with very low (below 5 minutes) or very high (above 60 minutes) response times, which led Bold lines indicate significant effects. Significant codes: *p < .05, **p < .01, ***p < .001. Study 1 = S1 and Study 2 = S2. Effects on general political trust toward politicians was examined in Study 2 only. to a final sample of 265 participants (51% female; M age = 45.71, SD = 14.82; ranging from 18 to 70 years old; 2.1% no school degree, 32.5% compulsory school; 29.3% vocational school degree, 19.4% with a high school degree; 16.6% with an academic degree). After prior informed consent, participants accessed the survey software via an online link. Exposure time was not forced. The stimulus presentation was followed by the assessment of the mediators and the dependent variables. Participants were then thanked and debriefed.

Stimulus material
Based on existing news articles reporting about scandalous political hypocrisy, news articles about the resignation of a longtime German politician were created. Similar to previous research studying the effects of political scandals (e.g., Carlson et al., 2000) and hypocrisy (e.g., McDermott et al., 2015), we used a fictitious case to ensure that individuals had no prior knowledge about it. Particular prior knowledge could have undermined the purpose of the present study, e.g., knowledge about potential consequences for the former party of the scandalous politician as discussed in the news media. The article (about 300 words) was elaborately designed (layout, typography, style, newspaper logo) to make it look like an authentic news article that had actually been published online on the platform of sueddeutsche.de. Sueddeutsche Zeitung is the largest quality newspaper in Germany. The article described the self-chosen end of a 64-year old politician's career (Green Party) who served in Germany's parliament for the previous 15 years campaigning for environmental protection. Especially, as the article stated, he campaigned against large companies' growing influence in seed business with genetically modified plants. All participants randomly assigned to group 1 (control condition, n = 136) and group 2 (experimental condition, n = 129) were exposed to the identical information. However, group 1 participants were also exposed to information stating that the politician is now retired after resigning from all posts and that he is no longer a member of the German Green Party. A political expert commented on his retirement from politics: "I regret his departure from office, we have always worked very well together in the past few years". In contrast, group 2 participants were exposed to information stating that the politician resigned from all posts and left the Green Party and joined a large and controversial international seed business company. The identical political expert (group 1) commented his retirement: "I regret his departure from office, we have always worked very well together in the past few years. However, I am shocked that he is now switching to this company contrary to his earlier beliefs".

Measures
Attitude toward the former politician Attitudes toward the former politician were gauged using nine items based on Carlson et al. (2000) and McDermott et al. (2015) with the help of a semantic differential (1 = very negative, 7 = very positive), e.g., likeableunlikeable, incompetent-competent, trustworthy-untrustworthy. The reliability of the scale was excellent (Cronbach's α = .98, M = 3.56, SD = 1.83).

Negative emotions
Negative emotions were measured with the help of two items (α = .91) based on Weiner, Graham, and Chandler (1982) as well as Forgays, Forgays and Spielberger's (1997) anger expression inventory and read as follows: "I am angry when I think of the former politician presented in the article; "I am furious when I think of the former politician presented in the article" (1 = disagree, 7 = agree; M = 3.74, SD = 2.02).
Attitude toward the green party Party attitudes were gauged using four items based on Maier (2011) on a Likerttype scale (1 = disagree, 7 = agree), e.g., "The Green Party keeps its promises to the people", "The Green Party is trustworthy (α = .97, M = 3.58, SD = 1.69).
Intention to vote for the green party Voting intention was gauged based on the following item (McDermott et al., 2015; 1 = very unlikely, 7 = very likely): "If elections were held next Sunday, how likely would you be to vote for the Green Party?" (M = 3.00, SD = 1.95).

Control variables
Before the experimental manipulation, we measured individuals' prior party identification (German Green Party) using the following item (1 = very weakly, 7 = very strongly): "How strongly do you identify with the Green Party?" (M = 2.91, SD = 1.86). Furthermore, participants completed the following item (1 = not credible, 7 = credible): "Please rate the credibility of the following news outlet: Sueddeutsche.de" (M = 4.45, SD = 1.64).

Manipulation check
To ensure that the experimental manipulation worked as intended, we performed a manipulation check. All participants answered the following three items (1 = no information provided, 7 = a lot of information provided): (1) "Was there any information in the article about a former politician switching to a seed producer?", (2) "Was there any information on a political expert who was outraged by the current behavior of the former politician?", (3) "Was there any information in the article that the former politician used to be a member of the Green Party?". First, an independent t-test showed significant group differences for the first item, t (263) = −13.76, p < .001, Cohen's d = 1.69, indicating that individuals in the experimental condition (M = 5.50, SD = 1.83) perceived more information on the former politician joining a seed producer (control: M = 2.37, SD = 1.88). Second, we found significant group differences for the second item, t(263) = 12.34, p < .001, Cohen's d = 1.51, indicating that individuals in the experimental condition perceived more information (M = 5.38, SD = 1.58) on an outraged political expert (control: M = 2.71, SD = 1.93). Third, and as expected, no significant group differences could be detected for the third item, t(263) = 1.18, p = −0.952, Cohen's d = 0.11 (experimental: M = 5.81, SD = 1.48, control: M = 5.63, SD = 1.73) because both articles identically stated that the former politician used to be a member of the Green Party. Thus, the manipulation worked as intended.

Data analysis
To test our hypotheses, we performed an ordinary least squares path analysis using the PROCESS macro (Hayes, 2013) in Statistical Package for the Social Sciences (SPSS). Experimental condition was dummy coded with the control group as reference group. Attitude toward the politician and negative emotions were modeled as mediator variables (parallel mediation analysis). We used two separate models to examine the specific newspaper article's effect on the two outcome variables: Green Party attitude and Green Party voting intention. Also, previous research has clearly shown that partisanship affects political evaluations in the context of political scandals (Fischle, 2000;Lee, 2018). Therefore, we controlled for participants' prior Green Party identification to ensure that all effects were assessed independently of their prior party identification. Also, a participant's gender may influence evaluations of a male (female) politician's hypocritical behavior (e.g., Bhatti et al., 2013). Since participants were presented with a male former politician, we controlled for their self-reported gender to ensure that effects were assessed independently of participants' gender. Finally, we controlled for individuals' prior news source evaluation (sueddeutsche.de) to ensure that effects were assessed independently of participants' prior news source evaluations. 95% bias-corrected bootstrap confidence intervals based on 10.000 bootstrap samples were used for statistical inference of indirect effects.

Attitudinal effects
First, we investigated the effects of exposure to news about political hypocrisy on attitudes toward the former politician (H1). Results revealed a negative and significant effect of the experimental condition on participants' attitudes compared with the control group (b = -19.56, SE = 1.60, p < .001). Thus, in line with H1 political hypocrisy negatively influenced participants' attitudinal evaluation of the former political actor. Next, we examined if attitudes toward the politician affected participants' party attitude (H3a). This was the case. The results revealed that more negative attitudes toward the ex-politician resulted in significantly more negative party attitudes (b = 0.13, SE = 0.03, p < .001). This supports H3a (Table 1). Mediation analysis further revealed that negative attitudes toward the politician mediated the effects of political hypocrisy on party attitudes (indirect effect of exposure: b = -2.61, SE = 0.57), 95% confidence intervals (CIs) = [-3.81, -1.56]). The results showed no direct effect of the experimental condition on party attitude (b = 0.53, SE = 0.80, p = .657). Furthermore, we tested whether more negative attitudes toward the politician decreased participants' intention to vote for the Green Party (H3b). This was also the case (Table 2). In line with H3b the results showed that more negative attitudes toward the former politician significantly decreased individuals' voting intention for the Green Party (b = 0.02, SE = 0.01, p = .009). Also, attitudes toward the former politician mediated the effects of political hypocrisy on voting intentions (indirect effect of exposure: b = -0.45, SE = 0.21), 95% CIs = [-0.89, -0.07]). The results showed no direct effect of the experimental condition on voting intentions (b = 0.11, SE = 0.27, p = .674). Taken together, the results reveal that hypocrisy had a negative effect on attitudes toward the former politician and that more negative attitudes then resulted in a negative spillover effect on the former party of the ex-politician (evaluation/voting intention). Importantly, hypocrisy had no direct effect on the two dependent variables Table 1. Ordinary least squares path analysis, Study 1 (S1), N = 265; Study 2 (S2), N = 209. Note. *** p < 0.001; ** p < 0.01; * p < 0.05; + p < 0.10 indicating that attitudes toward the ex-politician mediated the effect of hypocrisy on party attitudes and voting intentions.

Emotional effects
As a next step, we examined whether exposure to news about political hypocrisy increased participants' negative emotions toward the ex-politician (H2). The results showed a positive and significant effect of the experimental condition on participants' negative emotions compared with the control group (b = 5.13, SE = 0.41, p < .001). Thus, (supporting H2) hypocrisy increased participants' negative emotions toward the ex-politician. Next, we examined whether negative emotions affected participants' party attitudes (H4a). In contradiction with H4a this was not the case (  Note. *** p < 0.001; ** p < 0.01; * p < 0.05

Discussion
Study 1 corroborates and extends previous results on the effects of hypocrisy (Bhatti et al., 2013;McDermott et al., 2015) and negative spillover effects set off by political scandals (e.g., Maier, 2011;Puente-Diaz, 2015;Schwarz & Bless, 1992). In line with the inclusion/exclusion model (Schwarz & Bless, 1992), the results reveal that news about an ex-politician's hypocritical behavior negatively affected participants' attitudes toward him. Individuals obviously tended to perceive the scandalous former politician as a typical exemplar of the Green Party, although, he had resigned from all posts and was no longer a member of the party. This, in turn, resulted in more negative party attitudes and decreased individuals' voting intentions. In contrast, the emotional route resulted in no negative spillover effects on the political party. This is an unexpected finding. However, the feelings-asinformation model (Schwarz, 2012) offers an explanation. According to the model, one necessary precondition for affective spillover effects is that the angereliciting source (i.e., former Green Party politician) is not salient when making subsequent judgments of an unrelated target (for empirical support see, e.g., Dunn & Schweitzer, 2005; for an overview see Schwarz & Clore, 2003). However, when individuals evaluated the Green Party (target) the former Green Party affiliation of the hypocritical ex-politician (source) may have been activated and rendered salient helping individuals to identify the anger-eliciting source (hypocritical expolitician). As predicted by the feelings-as-information model this "source awareness" may have inhibited negative affective spillover effects on the Green Party because individuals may have realized that anger elicited by the hypocritical behavior of the ex-politician is irrelevant information for judging the Green Party (Dunn & Schweitzer, 2005;Schwarz, 2012;Schwarz & Clore, 2003). In summary, the results of Study 1 support the negative attitudinal spillover process on the party of the former political actor. At the same time, negative emotions toward the politician did not result in negative carry-over effects on the political party. To further validate our findings and to additionally test the effects of hypocrisy on political trust toward other politicians (H3c and H4c), we conducted a second study in a different European country: Austria.

Method
Study 2 was an experiment involving an online survey (N = 209) conducted to validate the results of Study 1 and to further investigate the impact of hypocrisy on political trust. Also, we used a different stimulus to increase the external validity of our results (Reeves et al., 2016).

Participants and procedure
Participants were 209 Austrian citizens (nonstudent convenience sample) living in different geographical regions of Austria (all Austrian states were represented) with a mean age of 30 years (age range: 16 to 65; 58.9 % female; level of education: 1.4% no school degree, 18% compulsory/vocational school degree, 49% high school degree; 31.6% academic degree). Individuals were invited by email and posts on social media (e.g., Facebook) andafter prior informed consentaccessed the survey software via an online link. The procedure was identical to that of Study 1. Again, we used a control condition (news without hypocrisy, n = 105) and an experimental condition (identical news with hypocrisy, n = 104).

Stimulus material
Participants were exposed to a different, yet, similar news article reporting about hypocrisy, as in Study 1 (Reeves et al., 2016). Again, the article was about 300 words long and was elaborately designed (layout, typography, style) to make it look like an authentic news article. For reasons of comparability, the case was similar to that in Study 1. Yet, in Study 2, we did not assign the article to a particular news source to prevent for potential news source effects on our dependent variables (see Study 1). Also, the case was adapted to the Austrian context.

Measures
All items used in Study 2 were identical to the items used in Study 1; yet, we measured the mediators and dependent variables with help of 5-point scales.

General political trust toward politicians
We used the following three items (1 = disagree, 5 = agree) based on Bless et al. (2000) and Maier (2011), e.g., "I can trust the Austrian politicians to make the right decisions for the people", "The politicians in Austria keep their promises to the citizens (α = .82, M = 2.34, SD = 0.71).

Data analysis
As in Study 1, we performed an ordinary least squares path analysis using the PROCESS macro (Hayes, 2013) in SPSS. Experimental condition was dummy coded with the control group as reference group. Attitude toward the politician and negative emotions served as mediator variables. We used three separate models to examine effects on the three outcome variables: party attitude, voting intention, political trust. Again, we controlled for participants' prior Green Party identification and gender. 95% bias-corrected bootstrap confidence intervals based on 10.000 bootstrap samples were used for statistical inference of indirect effects.

Attitudinal effects
First, we investigated the effects of exposure to hypocrisy on attitudes toward the former politician (H1). As in Study 1, results revealed a negative and significant effect of the experimental condition on participants' attitudes compared with the control group (b = -12.07, SE = 0.91, p < .001) revealing that hypocrisy negatively influenced participants' evaluation of the ex-politician. Next, we analyzed if attitudes toward the former politician affected participants' party evaluation (H3a). Indeed, more negative attitudes toward the ex-politician resulted in significantly more negative party attitudes (b = 0.08, SE = 0.03, p = .014) (Table 1). Also, the results revealed that negative attitudes toward the former politician mediated the effects of political hypocrisy on party attitudes (indirect effect of exposure: b = -0.99, SE = 0.48), 95% confidence intervals (CIs) = [-2.02, -0.12]). No direct effect of the condition on party attitude could be detected (b = -0.05, SE = 0.55, p = .928). As a next step, we tested whether attitudes toward the politician decreased participants' voting intentions (H3b). Supporting H3b and in line with Study 1, the results showed that more negative attitudes toward the former politician significantly decreased individuals' voting intentions (b = 0.03, SE = 0.01, p = .020) ( Table 2). Also, negative attitudes toward the ex-politician mediated the effects of hypocrisy on voting intentions (indirect effect: b = -0.30, SE = 0.14), 95% CIs = [-0.61, -0.05]), but showed no direct effect (condition on voting intentions; b = -0.07, SE = 0.18, p = .678).
However, the results showed that negative emotions resulted in a negative spillover effect on political trust toward politicians in general (H4c). Our findings revealed that negative emotions toward the ex-politician resulted in a significantly reduced level of political trust in individuals (b = -0.17, SE = 0.08 p = .031). As predicted, the findings showed that negative emotions mediated the effects of hypocrisy on political trust (indirect effect: b = -0.59, SE = 0.27), 95% CIs = [-1.13, -0.08]). Overall, we found no direct effects of hypocrisy on any of the three outcome variables with the mediators in the model. This indicates that attitudes toward the ex-politician and negative emotions served as relevant mediators for the effect of hypocrisy on party attitudes, voting intentions and political trust (Figure 1, Table 4). 1

Discussion
Study 2 was conducted to validate the results of Study 1 and to investigate the spillover effects of hypocrisy on political trust toward other politicians. First, we Note. *** p < 0.001; ** p < 0.01; * p < 0.05; + p < 0.10 Table 4. Mean values of mediator and dependent variables, Study 1 (S1), N = 265; Study 2 (S2), N = 209. Note. General political trust toward other politicians was measured in Study 2 only. Constructs in Study 1 were measured on 7-point scales, constructs in Study 2 were measured on 5-point scales.
were able to fully replicate the findings of Study 1. That is, the findings again showed negative spillover effects on the party of the ex-politician via the attitudinal route but not the emotional route. The fact that both studies were conducted in different countries further underlines the robustness of this effect. In Study 2, we furthermore examined negative spillover effects on political trust. Interestingly, negative attitudes toward the former politician did not affect political trust toward other politicians more generally. One explanation for this non-significant finding may be the circumstance that individuals were able to integrate the scandalous politician into their representation of the superordinate category "Green Party", but not into their representation of the category "politicians in general" (Schwarz & Bless, 2007). One may speculate that the politician's former affiliation to the Green Party was still accessible to news recipients (this is also supported by the findings of our manipulation check mentioned above). This mental connection may have triggered an assimilation effect and affected individuals' party attitudes and voting intentions in negative ways. This assumption is actually in line with current findings on scandal spillover effects showing that scandalous information about one politician can negatively affect another politician (not involved in scandal) when both are members of the identical party (Puente-Diaz, 2015). In contrast, and exactly reversed compared with the attitudinal process, the emotional route resulted in a negative spillover effect on political trust but no negative carry-over effects on the political party could be detected ( Figure 1). First, the non-existing spillover effect on the Green Party is in line with the feelings-as-information model (Schwarz, 2012) and corroborates the findings of Study 1. Again, evaluating the Green Party may have rendered the ex-politician's former party affiliation accessible and individuals may have realized that the anger eliciting source (expolitician) is irrelevant information for judging the Green Party (Dunn & Schweitzer, 2005;Schwarz, 2012;Schwarz & Clore, 2003). In contrast, the results showed a negative affective spillover effect on political trust toward other politicians. This target category (politicians in general) offered no specific cue to the anger eliciting source (ex-politician) and thus led to the affective transfer process (Schwarz & Clore, 2003). In summary, our results suggest an attitudinal and emotional spillover mechanism.

General discussion
The aim of this study was to examine whether hypocritical behavior of former political actors results in negative spillover effects and influences political judgment and behavior. Evidence from two experimental studies based on samples from two different European countries reveals thateven if a political actor has resigned all offices and is no longer member of a political partyan ex-politician's hypocritical behavior can significantly damage news recipients' party attitudes and their intention to vote for the party the scandalous ex-politician used to belong to. Furthermore, the findings also showed that hypocrisy of an ex-politician eroded individuals' trust toward other politicians more generally. Our findings corroborate and extend previous results on the effects of hypocrisy (Bhatti et al., 2013;Jordan et al., 2017;McDermott et al., 2015) and negative spillover effects set off by political scandals (e.g., Bowler & Karp, 2004;Puente-Diaz, 2015;Schwarz & Bless, 1992). Yet, previous research examining spillover effects has generated rather inconsistent findings and it has been argued that scholars (e.g., in political science) have, so far, neglected to "elaborate the theoretical underpinnings of scandal spillover effects" (Lee, 2018, p. 716;see also von Sikorski, 2018). This shortcoming may partly explain the inconsistencies in previous research results.
Our theorizing in the present study was based on the inclusion/exclusion model (Bless & Schwarz, 2010;Schwarz & Bless, 1992) and the feelings-as-information model (Schwarz, 2012). The findings nicely blend in with the models' assumptions and with previous results (Bless et al., 2000;Dunn & Schweitzer, 2005;Puente-Diaz, 2015;Schwarz & Bless, 1992) showing that an ex-politician's hypocrisy negatively affected participants' attitudes toward him and that individuals obviously tended to perceive the scandalous ex-politician as a typical exemplar of the Green Party. This, in turn, resulted in more negative party attitudes and decreased individuals' voting intentions. In line with these findings, the results also show that anger toward the ex-politician resulted in an assimilation effect and decreased individuals' trust toward the political elite more generally. This truly innovative finding adds to the literature showing that news about hypocrisy result in negative spillover effects not only via an attitudinal process but via an emotional route as well (Dunn & Schweitzer, 2005;Schwarz, 2012).
However, not all of our hypotheses were supported. Although, news including political hypocrisy affected both of the mediators in the expected way the attitudinal process resulted in negative spillover effects on the political party only and had no effect on political trust more generally. As already discussed above, individuals were obviously able to integrate the scandalous politician into their representation of the superordinate category "Green Party", but not into their representation of the category "politicians in general" (Schwarz & Bless, 2007), which is in line with previous research (Puente-Diaz, 2015). In contrast, and exactly reversed compared with the attitudinal process, the emotional route resulted in a negative spillover effect on political trust but no negative carry-over effects on the political party could be detected. The feelings-as-emotion model offers an explanation for this finding (Schwarz, 2012). According to the model, emotions only affect subsequent judgments when individuals are unaware of the source of their emotions. When individuals become aware of, e.g., an angereliciting source (e.g., a hypocritical politician) due to certain cues (former Green Party affiliation of a politician) that serve as a source reminder, individuals may come to the conclusion that emotions they experience are irrelevant for judging a target (e.g., Green Party) thus inhibiting the affective spillover effect (Dunn & Schweitzer, 2005). Furthermore, the resultsin line with previous scandal research (Bhatti et al., 2013;Lee, 2018)show that individuals prior Green Party identification predicted Green Party evaluation (more positively) and voting intentions for the Green Party (higher). Also, source evaluation (sueddeutsche.de) in Study 1 positively predicted party evaluation. One explanation for this finding is that persons who generally evaluate this quality news outlet as positive might be more likely to identify with the Green Party in Germany, since supporters of the Green Party in Germany are rather educated and often use (and positively evaluate) liberal news media such as sueddeutsche.de. Also, gender positively predicted Green Party voting intentions. That is, female participants were generally more likely to vote for the Green Party. One explanation for this finding is the fact that the Green Party in Germany and Austria (similar to Green Parties in other countries) campaigns for topics such as female equality, a topic that is of particular importance to female voters thus affecting particular voting intentions (Kaufmann, 2002).
In summary, our findings suggest that scandal spillover effects can be explained with the help of two separate mechanisms: an attitudinal and an emotional scandal spillover process. Shedding light on these processes can be regarded particularly important and a key contribution of the present study. Also, this dual process account helps to explain some of the inconsistent findings in previous research revealing that scholarsby all meansshould more precisely regard for the particular theoretical foundations of scandal spillover effects. Not accounting for discrete emotions may result in spurious findings and an incomplete understanding of scandal spillover effects.

Limitations
This study has some noteworthy limitations. To begin with, we examined a specific type of hypocrisy; working for a controversial company and engaging in behavior that is in fundamental opposition to the standpoints, a politician originally campaigned for. Several examples (Obama, Schröder, Glawischnig) show, this type of career change is a typical form of hypocritical behavior. Yet, future studies should test whether the results extend to other types of ex-politicians' hypocrisy. To that end, however, researchers have pointed out that it is difficult, if not impossible, to experimentally examine real cases of political hypocrisy (McDermott et al., 2015). Particular knowledge about a case may undermine the actual purpose of the study resulting in ceiling effects because most respondents will have formed opinions prior to the experiment. Thus, future research should use panel study designs to examine the effects of hypocrisy as a natural event in-between different panel waves (e.g., Fischle, 2000;von Sikorski, Heiss, & Matthes, 2019); this would be a valuable methodological extension. Also, future studies should examine how particular reactions of other political actors (e.g., former colleagues, the speaker of the party a hypocritical ex-politician used to belong to) affect negative spillover effects. Finally, the role of certain news reporting styles should be examined. For example, does it matter whether a news article makes an explicit distinction between a hypocritical ex-politician and his or her former political party? Can negative spillover effects be dampened? Those limitations notwithstanding, our findings pose important implications for understanding spillover effects triggered by hypocrisy.

Conclusion
News about political hypocrisy of an ex-politician can result in significant negative spillover effects on the ex-politician's political party, news recipients' voting intentions, and political trust toward other politicians more generally. Two experimental studies conducted in two different countries robustly show that effects can be explained by two mechanisms operating in parallel. First, hypocrisy negatively affects the attitudinal evaluation of ex-politicians (attitudinal spillover process) and in turn negatively influences individuals' party attitudes and voting intentions but not their general political trust. Second, hypocrisy increases anger toward expoliticians (emotional spillover process) and negatively affects general political trust but has no effect on party attitudes and voting intentions. Thus, ex-politicians' double standards are damaging on two accounts hurting the former political party and political trust in general.
Note 1. We also tested an alternative model in which negative emotions toward the former politician and attitudes toward the former politician were causally linked. The results reveal that hypocrisy results in more negative emotions toward the politician which, in turn, result in more negative attitudes toward the politician and eventually result in less positive party attitudes. Yet, hypocrisy in both studies also had a strong and negative direct effect on attitudes toward the former politician that cannot be explained by negative emotions (Study 1: b = -12.57, SE = 1.83, p < 0.001; Study 2: b = -7.06, SE = 1.04, p < 0.001). Thus, we modelled both negative emotions and attitudes toward the former politician in a parallel mediation model (see Figure 1).

Disclosure statement
No potential conflict of interest was reported by the authors.