Why Do We Hate Hypocrites? Evidence for a Theory of False Signaling

Abstract

Why do people judge hypocrites, who condemn immoral behaviors that they in fact engage in, so negatively? We propose that hypocrites are disliked because their condemnation sends a false signal about their personal conduct, deceptively suggesting that they behave morally. We show that verbal condemnation signals moral goodness (Study 1) and does so even more convincingly than directly stating that one behaves morally (Study 2). We then demonstrate that people judge hypocrites negatively—even more negatively than people who directly make false statements about their morality (Study 3). Finally, we show that “honest” hypocrites—who avoid false signaling by admitting to committing the condemned transgression—are not perceived negatively even though their actions contradict their stated values (Study 4). Critically, the same is not true of hypocrites who engage in false signaling but admit to unrelated transgressions (Study 5). Together, our results support a false-signaling theory of hypocrisy.

Keywords

moral psychology condemnation vignettes deception social signaling open data open materials

Consider the hypocrite—someone who condemns the moral failings of other people but behaves badly him- or herself. Many commentators have remarked on the “peculiarly repulsive” nature of hypocrisy (Shklar, 1984, p. 57). The degree to which hypocrites are disliked cannot be explained by their transgressions alone: What makes hypocrites especially bad is that they both commit a transgression and condemn it. But why is this combination so objectionable? After all, speaking out against immorality is normally seen as laudable. It enforces norms and encourages moral behavior (Berkowitz & Walker, 1967; Feinberg, Willer, & Schultz, 2014; Feinberg, Willer, Stellar, & Keltner, 2012), such that failing to condemn transgressions has been characterized as second-order free-riding (Yamagishi, 1986). Arguably, then, people should not be so resentful toward hypocrites: They may fail to achieve their moral aspirations, but at least they oppose bad behavior.

Previous research has investigated the psychology of hypocrites, including how hypocrites justify their behavior (Batson, Thompson, Seuferling, Whitney, & Strongman, 1999) and address aversive cognitive dissonance (Aronson, Fried, & Stone, 1991). Relatively little work has examined how hypocrisy is perceived by other people (but see Alicke, Gordon, & Rose, 2013; Barden, Rucker, & Petty, 2005).

One reason hypocrisy is perceived negatively may be that it involves inconsistency between words and deeds, which people tend to dislike (Tedeschi, Schlenker, & Bonoma, 1971). Another possibility is that hypocrites may be seen as unable to resist the temptation to transgress—another negative quality (Righetti & Finkenauer, 2011). Furthermore, hypocrites may be seen as more intentionally immoral than people who behave badly without condemning such behavior (Cushman, Young, & Hauser, 2006): Their condemnation demonstrates that they understand the wrongfulness of their actions.

Here, we propose a different hypothesis, based on the idea of false signaling. We suggest that hypocrites are disliked because they use their condemnation to mislead other people about their moral behavior.

As a matter of logic, there is nothing dishonest about both taking an action and condemning it. But engaging in moral condemnation may be perceived as communicating information about one’s future behavior (Baumeister, Zhang, & Vohs, 2004). The idea that condemnation may signal moral goodness is consistent with evidence that people who punish selfish players in economic games are seen as more trustworthy than people who choose not to punish (Barclay, 2006; Horita, 2010; Jordan, Hoffman, Bloom, & Rand, 2016; Nelissen, 2008; Raihani & Bshary, 2015a, 2015b). We thus hypothesize that hypocrites inspire moral outrage because they dishonestly signal their moral goodness—that is, their condemnation of immoral behavior signals that they are morally upright, but they fail to act in accordance with these signals.

This theory of false signaling helps explain why hypocrites are often regarded as liars (McKinnon, 1991). But it also predicts that hypocrites may be seen as worse than people who falsely claim to behave morally, whom we refer to as direct liars: In cases in which moral condemnation acts as a more persuasive signal than directly claiming to behave morally would, hypocrites are actually more misleading than direct liars. Furthermore, a hypocrite’s false signals may be more destructive than a liar’s false statements (e.g., by earning the hypocrite undeserved trust or by manipulating other people into following the hypocrite’s professed standards) and may come at the expense of other people (e.g., because condemnation tarnishes the reputation of the condemned; Williams, Forgas, & von Hippel, 2005). Liars, by contrast, avoid moral condemnation and are thus less likely to malign or shame other people.

Finally, a key prediction of our false-signaling theory is that honest hypocrites, who admit to committing the acts they condemn, should not be judged negatively for behaving hypocritically because their condemnation has been stripped of any signaling function. In other words, if hypocrites are disliked because of their false signaling, a hypocrite who admits to transgressing should be forgiven—insofar as this admission cancels any false signals. In the five studies reported here, we tested these predictions.

Study 1

We began with the hypothesis that moral condemnation is treated as a signal that one will behave morally in the future. According to this theory, individuals who condemn others’ transgressions should be perceived as less likely to commit those transgressions, and as overall more moral, than individuals who have not conversed about the transgressions. But condemnation should have this signaling effect only in the absence of more direct information about the condemner’s morality. If condemners are perceived positively because their condemnation signals that they will behave morally, their condemnation should no longer matter when a more informative indicator of moral behavior is available.

Method

Design

To test these predictions, we presented subjects with vignettes and asked them to evaluate target characters in the vignettes. In a 2 × 2 between-subjects design, we manipulated whether the targets engaged in moral condemnation or not and whether subjects had direct information about the targets’ moral behavior or not. We predicted that subjects would evaluate targets who engaged in moral condemnation more positively than those who did not, but only in the absence of direct information about the targets’ moral behavior.

Subjects

We recruited subjects online using Amazon Mechanical Turk (MTurk). We predicted an interaction between our two independent variables but did not have a clear prediction for what the effect size for this interaction would be. Thus, we precommitted to recruiting 800 subjects (n = 200 per condition), which is our standard protocol for between-subjects designs on MTurk when an interaction is predicted. A total of 798 people actually completed the survey, which required them to evaluate all the target characters and answer all the comprehension questions correctly (see the Procedure section for details). However, we could not analyze the responses of the first 135 subjects who completed the survey because of a technical error in how the survey was programmed (we corrected this error before the remaining subjects participated). We analyzed the data of all remaining subjects who had unique IP (Internet protocol) addresses (to avoid duplicate respondents). Our final sample consisted of 619 subjects (mean age = 31 years, 59% male).

Procedure

In each of our vignettes, we asked subjects to imagine that they belonged to a social group in which a particular moral transgression was possible (e.g., a track team whose members might use forbidden performance-enhancing drugs). Subjects were then told about two members of the social group: the target (whom subjects would later evaluate) and the other person (whom subjects would not evaluate), both of whom were described neutrally (not using these terms).

We then manipulated whether subjects received direct positive information about the moral behavior of the target. In the no-information condition, subjects were given no information about the moral behavior of the target or the other person. In the good-information condition, subjects received direct, positive information about the moral behavior of the target (but not the other person): They were told that the target recently behaved morally (e.g., did not use drugs in his or her last athletic competition).

In addition, we manipulated whether the target engaged in moral condemnation of a wrongdoer. In the target-condemns condition, we asked subjects to imagine having a dialogue with the target in which the target mentioned that a mutual acquaintance recently behaved immorally (e.g., used drugs at an athletic competition) and expressed strong disapproval of this acquaintance’s behavior. In the other-condemns condition, subjects were told to imagine having the same dialogue, but with the other person instead of with the target. Thus, in all conditions, subjects read the same description of condemnation, but in the target-condemns condition, the target engaged in the condemnation, whereas in the other-condemns condition, the other person did (and the target was absent from the conversation).

For example, following is the full text for one scenario about performance-enhancing drugs. In this scenario, Brian is the target character, and Sam is the other person.

Imagine that you are an athlete on a track team. Recently, your coach has become concerned that members of the team are using an illegal performance-enhancing drug called Vitronil. Vitronil use threatens your team’s eligibility to compete, and gives individual athletes unfair advantages.

In the no-information condition, the scenario continued as follows:

Two of your teammates are named Brian and Sam. You know nothing about if Brian uses Vitronil. You also know nothing about if Sam uses Vitronil.

In the good-information condition, the scenario instead continued with

Two of your teammates are named Brian and Sam. You overheard another member of the track team saying that Brian did not use Vitronil at his last track competition. In contrast, you know nothing about if Sam uses Vitronil.

The scenario concluded as follows in the target-condemns condition:

One day, you are having a conversation with Brian. You tell them a story about a mutual acquaintance, Mark, who is a competitive swimmer. After you finish your story, Brian mentions that he heard that Mark got caught using Vitronil right before an important swim meet. In telling his story, Brian expresses strong disapproval of Vitronil use.

The closing passage in the other-condemns condition was the same except that all references to Brian were changed to references to Sam:

One day, you are having a conversation with Sam. You tell them a story about a mutual acquaintance, Mark, who is a competitive swimmer. After you finish your story, Sam mentions that he heard that Mark got caught using Vitronil right before an important swim meet. In telling his story, Sam expresses strong disapproval of Vitronil use.

After reading each vignette, subjects evaluated the target, using Likert scales from 1 to 7. To measure expectations of the targets’ future moral behavior, we used four items that ranged in their specificity. Subjects rated the targets on their likelihood of committing the relevant moral transgression (e.g., for the scenario just quoted, “How likely do you think Brian is to use Vitronil in the future?”), their trustworthiness in the specific domain relevant to the vignette (e.g., “How much would you trust Brian as a competitor on your team?”), their general trustworthiness (e.g., “How much would you generally trust Brian across contexts?”), and their likeability (e.g., “How much do you like Brian?”). For the question about the likelihood of transgressing, the scale ranged from very unlikely to very likely, and for the questions about trustworthiness and likeability, it ranged from very little to very much.

Each subject was presented with four vignettes, describing (a) a track team whose members could use performance-enhancing drugs (as in the vignette just quoted), (b) a chemistry course in which students could cheat on take-home exams, (c) a work organization in which employees could fail to meet deadlines on team projects, and (d) a hiking club whose members frequently dated each other and could engage in infidelity. The four vignettes were presented in random order. To reduce noise, we matched all characters in the athletic, academic, and work scenarios to the subject’s gender and made all characters in the romantic scenario of the opposite gender. The full text of all the vignettes is in the Supplemental Material available online.

Immediately after reading each vignette, subjects answered four comprehension questions. If subjects answered a question incorrectly, they were not allowed to continue participating (i.e., the evaluation questions were presented, but the survey automatically ended before the next vignette was presented). A total of 85.9% of subjects showed perfect comprehension on all the vignettes, and the percentage of subjects with perfect comprehension did not differ significantly across conditions, χ²(3, N = 747) = 2.26, p = .521.

We found high interitem reliability among the four individual dependent measures of evaluation of the targets (α = .90); thus, we averaged these ratings to create one composite scale representing positive evaluations. We report analyses of this composite measure, but we also investigated possible differences among the individual dependent measures and found that the results were largely robust across these measures (see the Supplemental Material). To compute the composite ratings, we first reverse-coded the rating of the target’s likelihood of transgressing and then took the mean rating across the four variables.

Results

To test our predictions, we conducted a 2 (condemnation condition: target condemns vs. other condemns) × 2 (information condition: good information vs. no information) analysis of variance (ANOVA) predicting mean positive evaluations of the targets across the vignettes (see Fig. 1). We found a significant main effect of information condition, F(1, 615) = 137.93, p < .001, η_p² = .183; subjects evaluated targets more positively in the good-information condition (M = 5.28, SD = 0.83) than in the no-information condition (M = 4.52, SD = 0.80). This result served as a manipulation check, demonstrating that direct positive information about the target’s moral behavior was perceived as a clear indication of moral goodness.

Fig. 1.

Results from Study 1: mean composite evaluation of the targets as a function of condemnation condition and information condition. Error bars represent 95% confidence intervals.

We also found a significant main effect of condemnation condition, F(1, 615) = 13.20, p < .001, η_p² = .021; subjects evaluated targets more positively when the target engaged in condemnation (M = 5.01, SD = 0.91) than when the other person engaged in condemnation (M = 4.81, SD = 0.87). This result confirmed our hypothesis that moral condemnation serves as a signal of moral goodness.

Finally, we found a significant interaction, F(1, 615) = 8.51, p = .004, η_p² = .014; the target’s use of condemnation had a larger effect in the no-information condition than in the good-information condition. Specifically, when subjects were given no information about the target’s behavior, they evaluated the target significantly more positively when he or she condemned the transgression (M = 4.73, SD = 0.94) than when the other party condemned the transgression (M = 4.31, SD = 0.55), mean difference = 0.42, 95% confidence interval (CI) = [0.25, 0.59], t(301) = 4.77, p < .001, d = 0.55. However, in the good-information condition, we found no significant difference between the target-condemns condition (M = 5.30, SD = 0.79) and the other-condemns condition (M = 5.25, SD = 0.87), mean difference = 0.05, 95% CI = [−0.14, 0.23], t(314) = 0.49, p = .622, d = 0.06. This result confirmed our hypothesis that observers rely on a person’s statements of condemnation as a signal of moral goodness only when they lack direct information about the person’s moral behavior.

We note that the null effect of the target expressing condemnation in the good-information condition does not appear to be a ceiling effect. In the good-information condition, the mean composite evaluation (5.28) was substantially below the scale’s ceiling (7), and subjects rarely used the ceiling value (only 15.5% of responses to the evaluative questions were a “7”).

Thus, our data support our prediction that a person’s condemnation of a transgression serves as a signal of moral behavior—when direct information about the condemner’s behavior is unavailable. This suggests that condemnation is viewed positively because it signals moral behavior.

Study 2

Study 1 sheds light on why hypocrites are typically thought of as liars: If condemnation signals morality, then hypocrites mislead other people. How, then, do hypocrites, who use condemnation to imply (falsely) that they behave morally, compare with outright liars, who directly state (falsely) that they behave morally? Our theory predicts that hypocrites might be seen as worse than liars in situations in which condemnation is perceived as a stronger signal than a direct statement—and thus their deception is more misleading. Thus, in Study 2 we compared the signaling strength of condemnation of transgressions and direct statements of moral behavior.

Method

Design

The design of Study 2 was similar to that of Study 1, with just a few modifications. We again asked subjects to evaluate target characters in vignettes. In a 2 × 2 between-subjects design, we manipulated whether the target sent a signal of moral goodness or not and whether that signal was moral condemnation (of another person’s transgression) or a direct statement (that the target did not engage in that transgression). In all conditions, we provided subjects with no direct information about the target’s moral behavior (as in the no-information condition of Study 1), because this is the condition in which we found condemnation to have a significant effect on evaluations.

Subjects

We again recruited subjects using MTurk. As in Study 1, we precommitted to recruiting 800 subjects (n = 200 per condition); a total of 838 people actually completed the survey. We again analyzed the data of all subjects who had unique IP addresses, evaluated all the targets, and answered all the comprehension questions correctly. Our final sample consisted of 803 subjects (mean age = 31 years, 59% male).

Procedure

The vignettes described the same social groups as in Study 1 and again introduced the target and the other person. In each vignette, subjects were told to imagine having a conversation about which members within the social group typically engaged in the transgression in question (e.g., used drugs in athletic competitions). In the target-signals condition, subjects were told to imagine having this conversation with the target, whereas in the other-signals condition, subjects were told to imagine having it with the other person. Within this conversation, in the condemnation condition, subjects were told that the person they were talking to (the target or other person, depending on condition) emphasized his or her disapproval of the transgression. In the direct-statement condition, subjects were instead told that the person they were talking to emphasized that he or she did not engage in the transgression. Then, in all conditions, subjects evaluated the target, using the same four items as in Study 1.

To illustrate these changes, we present here the full text for the scenario about performance-enhancing drugs. Again, Brian is the target character, and Sam is the other person.

Two of your teammates are named Brian and Sam. You know nothing about if either of them use Vitronil.

After this point, the passage differed across conditions, as follows:

Target-signals/condemnation condition: One day, you are having a conversation with Brian. The two of you are discussing how different members of your team compete at meets. Specifically, you are talking about who stays clean, and who takes Vitronil. In your discussion, Brian emphasizes that he disapproves of taking Vitronil.

Other-signals/condemnation condition: One day, you are having a conversation with Sam. The two of you are discussing how different members of your team compete at meets. Specifically, you are talking about who stays clean, and who takes Vitronil. In your discussion, Sam emphasizes that he disapproves of taking Vitronil.

Target-signals/direct-statement condition: One day, you are having a conversation with Brian. The two of you are discussing how different members of your team compete at meets. Specifically, you are talking about who stays clean, and who takes Vitronil. In your discussion, Brian emphasizes that he does not take Vitronil.

Other-signals/direct-statement condition: One day, you are having a conversation with Sam. The two of you are discussing how different members of your team compete at meets. Specifically, you are talking about who stays clean, and who takes Vitronil. In your discussion, Sam emphasizes that he does not take Vitronil.

After reading each vignette, subjects answered one comprehension question to ensure that they understood the story. They were allowed to continue participating even if they answered a question incorrectly; however, as in Study 1, we analyzed only the responses of subjects who showed perfect comprehension across all the vignettes. Overall, 99.1% of the subjects met this criterion (i.e., 7 subjects were excluded because of imperfect comprehension). We found no significant differences between conditions in the percentage of subjects who showed perfect comprehension, χ²(3, N = 810) = 0.50, p = .919. As in Study 1, we found high interitem reliability among our four individual dependent measures of evaluation of the targets (α = .88), and we averaged these ratings to create a composite scale (see the Supplemental Material for analyses of the individual dependent measures).

Results

To test our predictions, we conducted a 2 (signaler: target vs. other person) × 2 (signal type: direct statement vs. condemnation) ANOVA predicting mean positive evaluations of the targets across the vignettes (see Fig. 2). We found a significant main effect of signal type, F(1, 799) = 9.92, p = .002, η_p² = .012; subjects evaluated targets more positively in the condemnation condition (M = 4.56, SD = 0.86) than in the direct-statement condition (M = 4.42, SD = 0.71). This result demonstrates that, overall, reading about a conversation in which a transgression was morally condemned led subjects to evaluate the target more positively than reading about a conversation in which a character directly stated that he or she did not engage in that behavior.

Fig. 2.

Results from Study 2: mean composite evaluation of the targets as a function of the signaler and signal type. Error bars represent 95% confidence intervals.

We also found a significant main effect of signaler, F(1, 799) = 198.62, p < .001, η_p² = .199; subjects evaluated targets more positively when the target signaled (M = 4.85, SD = 0.84) than when the other person signaled (M = 4.15, SD = 0.56). This result demonstrates that, overall, targets’ verbal signals of their moral goodness (condemnation and direct statements) successfully conferred reputational benefits to the targets.

Finally, as predicted, we observed a significant interaction of signaler and signal type, F(1, 799) = 14.01, p < .001, η_p² = .017; reading about condemnation of a transgression rather than a direct statement about behaving morally had a larger positive effect on evaluations of the target when the target signaled than when the other person signaled. In the target-signals condition, we observed a significant simple effect of signal type, with subjects evaluating the targets more positively when they engaged in condemnation (M = 5.02, SD = 0.90) than when they gave direct statements (M = 4.68, SD = 0.74), mean difference = 0.34, 95% CI = [0.18, 0.50], t(397) = 4.15, p < .001, d = 0.42. However, in the other-signals condition, we found no significant difference between the condemnation condition (M = 4.13, SD = 0.53) and the direct-statement condition (M = 4.16, SD = 0.59), mean difference = −0.03, 95% CI = [−0.14, 0.08], t(402) = −0.53, p = .596, d = −0.05. This result demonstrates that condemnation of a transgression can act as a stronger signal of one’s own moral goodness than a direct statement of moral behavior.

Study 3

Our results in Study 2 suggest that condemnation can be a more persuasive signal of morality than a direct statement that one behaves morally. This implies that hypocrites may be judged even more negatively than straightforward liars: Their dishonesty may be more misleading, and may earn them larger undue reputation benefits. Additionally, hypocrites’ false signals may be more destructive than liars’ false statements (e.g., because their moral condemnation can malign and shame other people). In Study 3, we tested the prediction that hypocrites, who condemn transgressions they engage in, are judged more negatively than both (a) control transgressors, who engage in identical transgressions but do not condemn them, and (b) direct liars, who engage in identical transgressions but directly state that they do not.

Method

To test these predictions, we designed a new paradigm to evaluate perceptions of hypocrites. We created vignettes in which a target character discusses an acquaintance’s moral transgression and then privately goes on to engage in the same transgression. We manipulated whether, in addition to committing the transgression, the target was a hypocrite, a (direct) liar, or neither.

On the basis of pilot testing, we selected transgressions that were perceived as more mild than those used in our first two studies (e.g., illegally downloading music, rather than using performance-enhancing drugs) to avoid floor effects—that is, to prevent the wrongness of the transgressions themselves from dominating subjects’ evaluations of the targets (and consequently making it difficult to detect an effect of hypocrisy or lying). Additionally, we simplified our design by eliminating gender matching of the vignette characters and subjects and by dropping the comprehension questions. Finally, we adjusted our dependent measures. In our first two studies, we were interested in the signals that condemnation sends, so our dependent measures focused on predictions of the target’s future moral behavior. In contrast, in Study 3 (and in Studies 4 and 5), we were interested in the implications of our theory for disapproval of hypocrites, so our dependent measures focused more on evaluations of the target as a person.

Design

We again presented subjects with vignettes and asked them to evaluate the target characters in the vignettes. In a three-condition, between-subjects design, we manipulated whether, before engaging in the relevant moral violation, the target character (a) condemned the violation (hypocrisy condition), (b) directly stated that he or she did not engage in the violation (liar condition), or (c) said nothing (control-transgressor condition). We predicted that subjects would evaluate hypocrites as both worse than control transgressors and worse than liars.

Subjects

We recruited subjects online using MTurk. Because we were no longer predicting an interaction between conditions, we reduced our target cell size from 200 to 150 subjects, as per our standard protocol; thus, we precommitted to recruiting 450 subjects. A total of 461 people actually completed the survey. We analyzed the data of all subjects who had unique IP addresses and who had evaluated all the target characters. Our final sample consisted of 451 subjects (mean age = 35 years, 47% male).

Procedure

Each vignette described a conversation between two characters: a target (whom subjects would later evaluate) and a friend (whom subjects would not evaluate). In all conditions, this conversation began with the target and the friend discussing a mutual acquaintance. In this discussion, the friend mentioned that the mutual acquaintance often engaged in a particular moral transgression.

In the hypocrisy condition, the target responded to the friend by condemning the transgression. In contrast, in the liar condition, the target responded by directly stating that he or she did not engage in the relevant transgression. Finally, in the control-transgressor condition, we did not include any information about a response from the target. Shortly after this conversation ended, in all conditions, the target went on to commit the relevant violation.

For example, here is the full text for a scenario about downloading music illegally. In this scenario, Becky is the target character, and Amanda is the friend. In all conditions, the vignette began as follows:

Becky and her friend Amanda are discussing a mutual acquaintance. Amanda mentions that the acquaintance often downloads music illegally from the Internet.

In the hypocrisy condition, the scenario continued,

Becky says that she thinks it is morally wrong to download music illegally from the Internet. Shortly after their conversation, Becky goes online, and downloads music illegally.

In the liar condition, the scenario instead continued,

Becky says that she doesn’t download music illegally from the Internet. Shortly after their conversation, Becky goes online, and downloads music illegally.

Finally, in the control-transgressor condition, nothing was said about Becky’s opinion or behavior, and the scenario simply ended with

Shortly after their conversation, Becky goes online, and downloads music illegally.

Each subject was presented with four vignettes (in random order), about downloading music illegally, evading jury duty, ignoring phone calls from one’s mother, and wasting paper by printing documents single-sided.

After reading each vignette, subjects evaluated the target. We asked subjects to rate how good a person the target was, how much they liked the target, how honest they thought the target was, and how trustworthy they thought the target was. Further, as a manipulation check, and to conduct an exploratory investigation into how subjects conceptualize “hypocrisy,” we asked them to rate how hypocritical the target was. These five dependent measures were presented in random order for each vignette. Subjects responded to each item on a sliding scale, with anchors reading not at all [trait] to very [trait] (e.g., not at all trustworthy to very trustworthy). The sliding scales did not have any numerical labels, but responses were translated to scores ranging from 0 (not at all) to 100 (very).

We found high interitem reliability among our four primary individual dependent measures (i.e., excluding our hypocrisy variable, which was used as a manipulation check; α = .96). Thus, as in Studies 1 and 2, we averaged these ratings to create a composite scale representing positive evaluations of the target and used this composite as our dependent variable in the analyses reported here (analyses investigating the individual dependent measures are reported in the Supplemental Material). We note that including the hypocrisy measure in our composite variable did not qualitatively affect our conclusions; however, results from the hypocrisy measure followed a somewhat different pattern than results from the other measures, and provide insight into the ways that subjects use the term “hypocrisy” (see the Supplemental Material).

Results

To test our prediction, we conducted a one-way ANOVA investigating the effect of condition on positive evaluations of the targets across the vignettes (see Fig. 3). We found a significant effect of condition, F(2, 448) = 26.48, p < .001, η_p² = .106. We followed up with pairwise comparisons and found that hypocrites (M = 32.07, SD = 15.93) were evaluated as worse than liars (M = 36.15, SD = 15.37), mean difference = −4.08, 95% CI = [−7.63, −0.53], t(299) = −2.26, p = .025, d = −0.26, who were evaluated as worse than control transgressors (M = 44.77, SD = 14.89), mean difference = −8.62, 95% CI = [−12.04, −5.20], t(301) = −4.96, p < .001, d = −0.57. This result confirmed our prediction that hypocrites would be seen as worse than liars. It also demonstrates that people disapprove of false signalers (hypocrites and liars) more than they disapprove of people who commit the same transgressions but do not condemn other people for those transgressions or lie about engaging in them.

Fig. 3.

Results from Study 3: mean composite evaluation of the targets as a function of condition. Error bars represent 95% confidence intervals.

Study 4

In Study 4, we moved to testing our theory’s key prediction that hypocrites are perceived negatively because of their false signals. If the negative perception of hypocrisy is caused by hypocrites’ false signaling, people we refer to as honest hypocrites (who avoid sending false signals) should not be judged negatively. Honest hypocrites fail to live up to their own moral standards and criticize other people for behaviors they themselves engage in, but admit that they sometimes commit the deeds they condemn. Honest hypocrites thus condemn without signaling, and analyzing subjects’ evaluations of such characters allowed us to test our false-signaling theory against several alternatives: If traditional hypocrites are disliked because they are inconsistent, unpredictable, weak willed, or intentionally immoral, people should dislike honest hypocrites, too. However, if people dislike traditional hypocrites because they send false signals, honest hypocrites should not be judged as worse than nonhypocritical transgressors.

Method

Design

In Study 4, we compared evaluations of honest hypocrites with evaluations of traditional hypocrites and control transgressors. To this end, we used the same design as in Study 3, but replaced the liar condition with an honest-hypocrite condition. In the honest-hypocrite condition, the target responded to the friend by stating that he or she believed the behavior in question to be morally wrong but sometimes behaved that way anyway.

Thus, in a three-condition, between-subjects design, we manipulated whether, before engaging in the relevant moral transgression, the target (a) condemned the violation (traditional-hypocrite condition), (b) condemned the violation but explicitly negated any signaling value of the condemnation (honest-hypocrite condition), or (c) said nothing (control-transgressor condition). We predicted that subjects would evaluate honest hypocrites as no worse than control transgressors (and that traditional hypocrites would be seen as worse than both honest hypocrites and control transgressors).

Subjects

As in Study 3, we recruited subjects online using MTurk. We precommitted to recruiting 450 subjects (n = 150 per condition), and a total of 457 actually completed the survey. We analyzed responses from all subjects who had unique IP addresses and had evaluated all the vignettes. Our final sample consisted of 452 subjects (mean age = 35 years, 41% male).

Procedure

The procedure for presenting the vignettes and measuring evaluations of targets was identical to that in Study 3, except that we replaced the liar condition with an honest-hypocrite condition. For example, here is the full text for the scenario about downloading music illegally. Again, Becky is the target character, and Amanda is the friend. The scenario began as follows:

Becky and her friend Amanda are discussing a mutual acquaintance. Amanda mentions that the acquaintance often downloads music illegally from the Internet.

In the traditional-hypocrite condition, the scenario continued with

Becky says that she thinks it is morally wrong to download music illegally from the Internet. Shortly after their conversation, Becky goes online, and downloads music illegally.

In the honest-hypocrite condition, the passage instead continued with

Becky says that she thinks it is morally wrong to download music illegally from the Internet, but that she sometimes does it anyway. Shortly after their conversation, Becky goes online, and downloads music illegally.

Finally, in the control-transgressor condition, nothing was said about Becky’s opinion or behavior, and the passage simply ended with

Shortly after their conversation, Becky goes online, and downloads music illegally.

As in Study 3, we found a high interitem reliability among our four individual primary dependent measures (α = .94) and averaged responses to them to create a single composite measure (see the Supplemental Material for analyses investigating the individual dependent measures, including the measure of subjects’ concept of “hypocrisy”).

Results

To test our prediction, we conducted a one-way ANOVA investigating the effect of condition on positive evaluations of the targets across the vignettes (see Fig. 4). We found a significant effect of condition, F(2, 449) = 35.62, p < .001, η_p² = .137. Pairwise comparisons revealed that traditional hypocrites (M = 34.35, SD = 13.76) were evaluated more negatively than both honest hypocrites (M = 46.62, SD = 14.01), mean difference = −12.28, 95% CI = [−15.41, −9.14], t(301) = −7.69, p < .001, d = −0.88, and control transgressors (M = 45.21, SD = 13.84), mean difference = −10.87, 95% CI = [−13.99, −7.74], t(300) = −6.84, p < .001, d = −0.79. Additionally, we found that evaluations of honest hypocrites and control transgressors did not differ, mean difference = 1.41, 95% CI = [−1.76, 4.58], t(297) = 0.88, p = .382, d = 0.10. Thus, as predicted, our honest-hypocrite manipulation fully eliminated disapproval of hypocrisy; honest hypocrites received ratings that were no worse than the ratings of control transgressors.

Fig. 4.

Results from Study 4: mean composite evaluation of the targets as a function of condition. Error bars represent 95% confidence intervals.

Study 5

In Study 4, honest hypocrites, whose condemnation was stripped of its signaling function, were not judged more negatively than control transgressors, who engaged in transgressions without condemnation. However, it is possible that honest hypocrites are in fact judged negatively for their hypocrisy, but are given additional credit for voluntarily disclosing their transgressions, which offsets the negative evaluation of their hypocrisy. In Study 5, we tested this alternative explanation by investigating evaluations of hypocrites who disclosed transgressions that were unrelated to their condemnation, and thus did not negate the false signals implied by their condemnation.

Method

Design

In Study 5, we modified our Study 4 design to include a disclosure-hypocrite condition that involved hypocrisy (condemnation followed by transgression) and a disclosure about a transgression unrelated to the condemnation. To this end, we altered our vignettes so that the targets in all conditions committed two moral transgressions, rather than one, and we presented our vignettes in a way that naturally introduced these two transgressions. Then, in a four-condition, between-subjects design, we manipulated whether, before engaging in these two transgressions, the targets (a) condemned one transgression (traditional hypocrite), (b) condemned one transgression and admitted to engaging in the other transgression (disclosure hypocrite), (c) condemned one transgression and admitted to engaging in that same transgression (honest hypocrite), or (d) said nothing (control transgressor).

Thus, both the disclosure hypocrite and the honest hypocrite ultimately committed the same two violations. They also each condemned one of the violations and admitted to committing one of the violations. However, only the honest hypocrite admitted to committing the same violation he or she condemned, and thus negated the false signal implied by that condemnation. We predicted that whereas honest hypocrites would be seen as no worse than control transgressors (because their hypocrisy did not involve false signaling), disclosure hypocrites would be seen as no better than traditional hypocrites (because their hypocrisy still involved false signaling, despite also involving disclosure).

Subjects

As in Study 4, we recruited subjects online using MTurk. We precommitted to recruiting 150 subjects per condition (i.e., a total of 600 subjects). A total of 612 subjects actually completed the survey. All of these subjects had unique IP addresses and had evaluated all the vignettes, so none were excluded from analyses. Thus, our final sample consisted of 612 subjects (mean age = 34 years, 48% male).

Procedure

To implement our design, we collapsed our four vignettes (in which each target committed one violation) into two vignettes (in which each target committed two violations). Specifically, in one vignette, the target downloaded music illegally and ignored his or her mother’s phone calls, and in the other, the target tried to get out of jury duty and wasted paper by printing single-sided. We presented subjects with both vignettes in a random order.

We modified each vignette to structure the target’s conversation around the two moral transgressions at hand. Specifically, we introduced the two moral issues by explaining that the target and the friend were discussing issues in their lives and then listing the relevant topics as examples (e.g., downloading music illegally and answering parents’ phone calls). Then, depending on the condition, the vignette presented any relevant condemnation and disclosure information. Finally, the vignette indicated that the target went on to commit both transgressions.

For example, here is the full text for the scenario about downloading music illegally and answering parents’ phone calls. Again, Becky is the target character, and Amanda is the friend. In all conditions, the vignette began as follows:

Becky and her friend Amanda are discussing issues in their lives, like downloading music and answering their parents’ phone calls.

In the traditional-hypocrite condition, the vignette continued,

Becky tells Amanda that she thinks it is morally wrong when people download music illegally from the Internet. Shortly after their conversation, Becky goes online, and downloads music illegally. She also notices that her mother is calling, and ignores the call.

In the disclosure-hypocrite condition, the passage instead read,

Becky tells Amanda that she thinks it is morally wrong when people download music illegally from the Internet, but that she sometimes ignores her mother’s phone calls. Shortly after their conversation, Becky goes online, and downloads music illegally. She also notices that her mother is calling, and ignores the call.

In the honest-hypocrite condition, the passage read,

Becky tells Amanda that she thinks it is morally wrong when people download music illegally from the Internet, but that she sometimes does it anyway. Shortly after their conversation, Becky goes online, and downloads music illegally. She also notices that her mother is calling, and ignores the call.

Finally, in the control-transgressor condition, there was no mention of Becky’s opinion or behavior, and the passage simply ended with

Shortly after their conversation, Becky goes online, and downloads music illegally. She also notices that her mother is calling, and ignores the call.

We orthogonally counterbalanced both (a) which of the two transgressions we listed first (when introducing them and explaining that the target engaged in them) and (b) which of the two transgressions was condemned by the targets in the three hypocrite conditions (and consequently, which transgression was disclosed by the targets in the disclosure-hypocrite condition—as this was always the noncondemned transgression).

With the exception of these modifications, Study 5 was identical to Study 4, and thus used the same dependent measures. We again found a high interitem reliability among our four individual primary dependent measures (α = .93) and averaged them to create a single composite measure (see the Supplemental Material for analyses investigating the individual dependent measures, including the measure of subject’s concept of “hypocrisy”).

Results

To test our predictions, we conducted a one-way ANOVA investigating the effect of condition on positive evaluations of the targets across the vignettes (see Fig. 5). We found a significant effect of condition, F(3, 608) = 28.60, p < .001, η_p² = .124. As in Study 4, pairwise comparisons revealed that honest hypocrites (M = 41.94, SD = 14.51) were evaluated more positively than traditional hypocrites (M = 28.22, SD = 15.30), mean difference = 13.73, 95% CI = [10.38, 17.08], t(305) = −8.06, p < .001, d = −0.92, and were seen as no worse than control transgressors (M = 38.85, SD = 15.94). In fact, honest hypocrites were seen as marginally better than control transgressors in Study 5, mean difference = 3.09, 95% CI = [−0.34, 6.53], t(303) = 1.77, p = .077, d = 0.20.

Fig. 5.

Results from Study 5: mean composite evaluation of the targets as a function of condition. Error bars represent 95% confidence intervals.

Critically, pairwise comparisons also revealed that disclosure hypocrites, who merely admitted to committing a moral transgression but did not negate the false signal implied by their condemnation, did not receive similarly positive evaluations. Disclosure hypocrites (M = 30.03, SD = 15.95) were seen as significantly worse than honest hypocrites, mean difference = −11.91, 95% CI = [−15.34, −8.48], t(304) = −6.83, p < .001, d = −0.78, and were not seen as significantly better than traditional hypocrites, mean difference = 1.81, 95% CI = [−1.70, 5.32], t(305) = 1.02, p = .310, d = 0.12. This result demonstrates that, as predicted, mere disclosure is insufficient to eliminate subjects’ disapproval of hypocrites: Hypocritical targets must use disclosure that negates the false signals implied by their condemnation in order to eliminate this disapproval.

General Discussion

We have sought to explain why hypocrites—who condemn transgressions they engage in—are seen as worse than individuals who commit the same transgressions without condemning them. The puzzle, as we see it, is this: Condemnation of bad behavior is typically seen as virtuous (because it discourages that behavior), and people who do not condemn bad behavior can be seen as second-order free riders (Yamagishi, 1986). So why do hypocrites get moral blame—not credit—for their condemnation?

Our experiments provide an answer: Hypocrites are disliked because they falsely signal that they behave morally. This theory explains that hypocrites do in fact free-ride; they do so not by refusing to condemn bad behavior, but by using condemnation to imply that they will behave morally—without incurring the costs of actually doing so.

Our findings, by elucidating the conditions under which hypocrisy is perceived negatively by other people, may shed light on previous work showing that hypocrites themselves experience hypocrisy as aversive, and may explain why a fear of hypocrisy in public contexts is especially effective at motivating virtuous behavior (Aronson et al., 1991).

Our results support our theory of hypocrisy by demonstrating that condemnation of immoral behavior is perceived as a signal of moral behavior (Study 1) that can be more convincing than directly stating that one behaves morally (Study 2). These results are consistent with theories that moral language conveys much beyond its literal meaning (Strandberg, 2012), and that condemnation communicates information about one’s values and behaviors (Baumeister et al., 2004). Condemnation may also be interpreted as a more sincere signal of morality compared with direct statements because, at first blush, condemning other people is less obviously self-promotional than stating that one behaves morally. Because people actively monitor social information for its veracity (Barasch, Levine, Berman, & Small, 2014; Fein, Hilton, & Miller, 1990; Hess & Hagen, 2006; Lin-Healy & Small, 2012), and overt self-promotion can thus backfire (Gordon, 1996), condemnation may be a more persuasive signal.

These results also build on the finding that people who punish selfishness in economic-game experiments are trusted not to act selfishly themselves (Barclay, 2006; Horita, 2010; Jordan et al., 2016; Nelissen, 2008; Raihani & Bshary, 2015a, 2015b). Verbal condemnation can function as costly punishment (Fehr & Fischbacher, 2004): It harms the transgressor’s reputation and is also risky for the condemner—because the transgressor might retaliate (e.g., by publicizing the condemner’s misdeeds). Previous research has shown that punishment can function as a costly signal (Zahavi, 1975) of morality, so long as punishing is less costly for people who typically behave morally than for those who behave immorally (Jordan et al., 2016; Jordan & Rand, 2016). Perhaps, then, verbal condemnation of immoral behavior is perceived as a strong signal because it acts as a costly signal—in ways that direct statements that one behaves morally (which carry few costs) do not. Future research should explicitly investigate the effect of cost on perceptions of condemnation of immorality, direct statements of one’s own morality, and other signals of morality. Moreover, future research should investigate whether praising good behavior signals morality. Praising an action one does not engage in may inspire less outrage than condemning behavior one does engage in, if praise serves as a weaker signal of morality because it is less costly (i.e., the target of praise is unlikely to retaliate).

Studies 1 and 2 also extend theories about credibility-enhancing displays (Henrich, 2009)—costly indicators that one holds a particular belief (e.g., eating a mushroom to signal that one believes it is healthy). Much as actions can undermine credibility, hypocrisy negates a signal implied by condemnation. Whereas credibility-enhancing displays signal beliefs about states of the world, we have shown that condemnation signals future moral behavior.

Our theory is further supported by Studies 3 through 5, which showed that people dislike hypocrites more than direct liars, and that this is because hypocrites falsely signal. One straightforward explanation for why hypocrites’ false signals inspire moral outrage is that misleading other people is generally regarded as wrong (Bell & Whaley, 1991)—and hypocrites are especially misleading, because condemnation is an especially convincing signal.

A hypocrite’s false signals may rouse further disapproval, moreover, because they lead to negative outcomes, such as unfairly boosting the hypocrite’s reputation or shaming other people into changing their behavior while the hypocrite carries on. Furthermore, unlike direct statements that one behaves morally, condemnation can harm other people by maligning the condemned—which may make hypocrisy seem particularly wrong (Crockett, Kurth-Nelson, Siegel, Dayan, & Dolan, 2014; Eber, 2007; Smith, Parrott, Ozer, & Moniz, 1994). Consistent with the hypothesis that hypocrites are judged as worse than liars for reasons beyond their being more misleading, our supplementary analyses showed that hypocrites were rated especially negatively relative to liars on measures of being likeable and a good person; the difference between ratings of hypocrites and liars was smaller on measures of trust and honesty (see the Supplemental Material).

An important future direction is to investigate perceptions of condemnation and hypocrisy across cultures. Our data are limited to American MTurk samples, which raises questions about the generalizability of our results (Henrich, Heine, & Norenzayan, 2010). (We note, though, that within our samples, results were robust across demographic variables; see the Supplemental Material.) Although condemnation appears to be widespread across cultures, its prevalence does vary substantially (Henrich et al., 2006). How does this variance correlate with the signal value of condemnation and with disapproval of hypocrites? Future research should address this question, and also investigate hypocrisy in less contrived contexts (e.g., reported examples from daily life) and other culturally relevant domains (e.g., religious hypocrisy).

In conclusion, we propose that hypocrites are disliked because their condemnation falsely signals moral goodness. We have supported this theory with evidence that when condemnation’s signaling value is negated, hypocrisy is forgiven.

Footnotes

Acknowledgements

We gratefully acknowledge helpful comments and assistance from Adam Bear, Adam Chekroud, Fabian Schellhaas, and Amanda Zheutlin.

Action Editor

Leaf Van Boven served as action editor for this article.

Declaration of Conflicting Interests

The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

Funding

This research was funded by a grant from the John Templeton Foundation.

Supplemental Material

Additional supporting information can be found at

Open Practices

All data and materials have been made publicly available via the Open Science Framework and can be accessed at https://osf.io/pszjz/. The complete Open Practices Disclosure for this article can be found at https://journals-sagepub-com.web.bisu.edu.cn/doi/suppl/10.1177/0956797616685771. This article has received badges for Open Data and Open Materials. More information about the Open Practices badges can be found at .

References

Alicke

Gordon

Rose

(2013). Hypocrisy: What counts? Philosophical Psychology, 26, 673–701.

Aronson

Fried

Stone

(1991). Overcoming denial and increasing the intention to use condoms through the induction of hypocrisy. American Journal of Public Health, 81, 1636–1638.

Barasch

Levine

E. E.

Berman

J. Z.

Small

D. A.

(2014). Selfish or selfless? On the signal value of emotion in altruistic behavior. Journal of Personality and Social Psychology, 107, 393–413.

Barclay

(2006). Reputational benefits for altruistic punishment. Evolution & Human Behavior, 27, 325–344.

Barden

Rucker

D. D.

Petty

R. E.

(2005). “Saying one thing and doing another”: Examining the impact of event order on hypocrisy judgments of others. Personality and Social Psychology Bulletin, 31, 1463–1474.

Batson

C. D.

Thompson

E. R.

Seuferling

Whitney

Strongman

J. A.

(1999). Moral hypocrisy: Appearing moral to oneself without being so. Journal of Personality and Social Psychology, 77, 525–537.

Baumeister

R. F.

Zhang

Vohs

K. D.

(2004). Gossip as cultural learning. Review of General Psychology, 8, 111–121.

Bell

J. B.

Whaley

(1991). Cheating and deception. New Brunswick, NJ: Transaction.

Berkowitz

Walker

(1967). Laws and moral judgments. Sociometry, 30, 410–422.

10.

Crockett

M. J.

Kurth-Nelson

Siegel

J. Z.

Dayan

Dolan

R. J.

(2014). Harm to others outweighs harm to self in moral decision making. Proceedings of the National Academy of Sciences, USA, 111, 17320–17325.

11.

Cushman

Young

Hauser

(2006). The role of conscious reasoning and intuition in moral judgment: Testing three principles of harm. Psychological Science, 17, 1082–1089.

12.

Eber

(2007). The performance-enhancing drug game reconsidered: A fair play approach. Journal of Sports Economics, 9, 318–327.

13.

Fehr

Fischbacher

(2004). Third-party punishment and social norms. Evolution & Human Behavior, 25, 63–87.

14.

Fein

Hilton

J. L.

Miller

D. T.

(1990). Suspicion of ulterior motivation and the correspondence bias. Journal of Personality and Social Psychology, 58, 753–764.

15.

Feinberg

Willer

Schultz

(2014). Gossip and ostracism promote cooperation in groups. Psychological Science, 25, 656–664.

16.

Feinberg

Willer

Stellar

Keltner

(2012). The virtues of gossip: Reputational information sharing as prosocial behavior. Journal of Personality and Social Psychology, 102, 1015–1030.

17.

Gordon

R. A.

(1996). Impact of ingratiation on judgments and evaluations: A meta-analytic investigation. Journal of Personality and Social Psychology, 71, 54–70.

18.

Henrich

(2009). The evolution of costly displays, cooperation and religion: Credibility enhancing displays and their implications for cultural evolution. Evolution & Human Behavior, 30, 244–260.

19.

Henrich

Heine

S. J.

Norenzayan

(2010). The weirdest people in the world? Behavioral & Brain Sciences, 33, 61–83.

20.

Henrich

McElreath

Barr

Ensminger

Barrett

Bolyanatz

. . . Ziker

(2006). Costly punishment across human societies. Science, 312, 1767–1770.

21.

Hess

N. H.

Hagen

E. H.

(2006). Psychological adaptations for assessing gossip veracity. Human Nature, 17, 337–354.

22.

Horita

(2010). Punishers may be chosen as providers but not as recipients. Letters on Evolutionary Behavioral Science, 1(1), 6–9.

23.

Jordan

J. J.

Hoffman

Bloom

Rand

D. G.

(2016). Third-party punishment as a costly signal of trustworthiness. Nature, 530, 473–476.

24.

Jordan

J. J.

Rand

D. G.

(2016). Building costly signaling from the ground up: A model of third-party punishment as a costly signal of exposure to repeated interactions. Retrieved from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2794084

25.

Lin-Healy

Small

D. A.

(2012). Cheapened altruism: Discounting personally affected prosocial actors. Organizational Behavior and Human Decision Processes, 117, 269–274.

26.

McKinnon

(1991). Hypocrisy, with a note on integrity. American Philosophical Quarterly, 28, 321–330.

27.

Nelissen

R. M. A.

(2008). The price you pay: Cost-dependent reputation effects of altruistic punishment. Evolution & Human Behavior, 29, 242–248.

28.

Raihani

N. J.

Bshary

(2015a). The reputation of punishers. Trends in Ecology & Evolution, 30, 98–103.

29.

Raihani

N. J.

Bshary

(2015b). Third-party punishers are rewarded, but third-party helpers even more so. Evolution, 69, 993–1003.

30.

Righetti

Finkenauer

(2011). If you are able to control yourself, I will trust you: The role of perceived self-control in interpersonal trust. Journal of Personality and Social Psychology, 100, 874–886.

31.

Shklar

J. N.

(1984). Ordinary vices. Cambridge, MA: Harvard University Press.

32.

Smith

R. H.

Parrott

W. G.

Ozer

Moniz

(1994). Subjective injustice and inferiority as predictors of hostile and depressive feelings in envy. Personality and Social Psychology Bulletin, 20, 705–711.

33.

Strandberg

(2012). A dual aspect account of moral language. Philosophy and Phenomenological Research, 84, 87–122.

34.

Tedeschi

J. T.

Schlenker

B. R.

Bonoma

T. V.

(1971). Cognitive dissonance: Private ratiocination or public spectacle? American Psychologist, 26, 685–695.

35.

Williams

K. D.

Forgas

J. P.

von Hippel

(Eds.). (2005). The social outcast: Ostracism, social exclusion, rejection, and bullying. New York, NY: Psychology Press.

36.

Yamagishi

(1986). The provision of a sanctioning system as a public good. Journal of Personality and Social Psychology, 51, 110–116.

37.

Zahavi

(1975). Mate selection—a selection for a handicap. Journal of Theoretical Biology, 53, 205–214.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.58 MB

1.26 MB