Abstract
Compared to the conventional order of hypocritical actions—saying one thing and then doing another—merely reversing the order of these actions can mitigate whether an individual is judged to be a hypocrite (Barden, Rucker, & Petty, 2005). The present research examines how factors extraneous to a target’s own actions—specifically, group membership—influence hypocrisy judgments. Three experiments provided consistent evidence that reversing the order of statement and behavior mitigated hypocrisy judgments to a greater extent when observers judged ingroup targets compared to outgroup targets. This pattern was observed across two distinct groups (i.e., gender and political party). In addition, mediational evidence suggested that the greater mitigation for ingroup targets stemmed from the observer’s greater tendency to make attributions that ingroup targets had genuinely changed for the better.
Hypocrisy is socially condemned (Barden, Rucker, & Petty, 2005). In fact, people go to great lengths to avoid perceiving themselves, or having others perceive them, as hypocrites (e.g., Stone, Wiegand, Cooper, & Aronson, 1997). One need look no further than any political season to find numerous fingers pointed at candidate hypocrisy in an effort to fuel voter disdain.
Despite hypocrisy judgments being an important part of public discourse and everyday life, they have received little empirical attention. In one of only a few forays into this domain, Barden et al. (2005) examined the relationship between the order of a target’s actions and attributions of hypocrisy. The authors noted that in the archetypical case, hypocrisy follows the order of the common phrase, “saying one thing and then doing another.” Here, the saying element is a public statement establishing a standard for behavior (e.g., promoting safe sex), and the doing element is a private behavior that violates that standard (e.g., engaging in unsafe sex). Barden et al. (2005) found that when people judged actions in the reverse of the conventional order (i.e., a behavior preceded a contradictory statement), judgments of hypocrisy dropped significantly despite the equivalent inconsistency. Critically, this drop occurred because the target’s inconsistent actions were more likely to be attributed to sincere change for the better by the target. Thus, reversing the typical order served as a mitigating circumstance for hypocrisy judgments.
Mitigation and Group-Based Bias
Unlike Barden et al. (2005), the present research explores how factors extraneous to a target’s actions influence hypocrisy judgments. Specifi-cally, we examine whether an observer belonging to the same group or a different group as a target affects judgments of hypocrisy. We propose that hypocrisy judgments are subject to group-based biases that have been observed on antecedents of hypocrisy such as inconsistency judgments (Westen, Blagov, Harenski, Kilts, & Hamann, 2006), consequences of hypocrisy such as unfavorable evaluation (Beal, Ruscher, & Schnake, 2001), and other moral judgments like guilt (Feather & Deverson, 2000).
Why would one’s group membership influence hypocrisy judgments? Pettigrew’s (1979) ultimate attribution error holds that negative actions by ingroup members are more likely to produce situational attributions, whereas negative actions by outgroup members are more likely to produce dispositional attributions. This effect has been shown for a variety of intergroup categories (school, sports teams, gender, and ethnicity) and in samples around the world (Hewstone, 1990). The ultimate attribution error suggests group membership might lead people to attribute inconsistent actions to situational causes for ingroup members, but dispositional causes for outgroup members (i.e., hypocrisy). What is not clear is whether the ultimate attribution error will vary depending on mitigating circumstances.
The ultimate attribution error was proposed as a general phenomenon, expected to hold across various factors (prejudice level and intergroup conflict; Pettigrew, 1979). This and the generality of the effect suggest that it could occur even in the absence of mitigating circumstances. Within research on moral judgments, group-based bias has been shown in the absence of mitigating circumstances. Valdesolo and DeSteno (2007) had participants judge targets who, instead of assigning experiments randomly, assigned themselves to a more desirable experiment with no rationale or mitigating circumstance given for the individuals’ behavior. Observers judged ingroup or “matched” targets as having acted more fairly than outgroup or “mismatched” targets. Thus, mitigating circumstances may not be a prerequisite for ingroup bias in the domain of hypocrisy judgments. Of course, it is possible that people inferred a mitigating circumstance or that group differences without mitigating circumstances are unique to the chosen scenario.
Although the generality of findings suggests a robust ultimate attribution error effect, Pettigrew (1979) also speculated that certain contextual factors might attenuate the effect (category salience, social role explaining behavior). Indeed, a few studies suggest that mitigating circumstances for the target’s actions are a precondition for group-based biases. For example, Lynch and Haney (2000) found that White mock jurors reported using a mitigating circumstance (e.g., past drug abuse) when sentencing White defendants more than Black defendants. And, Westen et al. (2006) investigated the impact of political party affiliation on people’s judgments of the 2004 presidential candidates when presented with the inconsistent actions of the candidates. The candidates’ actions occurred four or more years apart, which seemed to offer a mitigating circumstance in the form of the possibility of sincere change over time. The actions of own-party politicians were judged as less inconsistent than the actions of other-party politicians.
Finally, Beal et al. (2001) varied the opportunity of participants to mitigate judgments by manipulating the order of the two explanations (internal and external) provided for a single target behavior. The order manipulation taps into a conversational norm to weigh later information more heavily. Participants selectively applied the conversational norm only to ingroup targets and only when it led them to more favorable judgments (external attribution of negative actions). Beal et al. did not report attributions or other process measures, and discussed a variety of potential mechanisms. So this research provides initial support for moderation by mitigating circumstances while leaving key questions unanswered.
Based on these past findings, it is unclear whether or not mitigating circumstances are required either for the ultimate attribution error or for the specific case of hypocrisy. The current research tests these two competing possibilities by examining ingroup versus outgroup judgments of hypocrisy in the presence or absence of a manipulated mitigating circumstance.
Experiment 1
Experiment 1 manipulated the presence of mitigating circumstances by varying the order of statements and behavior as in previous research (Barden et al., 2005). In the conventional order, the target makes a public statement establishing a standard for behavior (against drunk driving) and later commits a behavior violating that standard (driving drunk), so here the mitigating circumstance is absent. In the reversed order, the target first drives drunk and then later makes a public statement against drunk driving; this is known to mitigate hypocrisy, because people can make the inference that the person has genuinely changed.
If the mitigating circumstances and group matching effects are independent, then only two main effects should be observed on hypocrisy judgments. That is, the group bias effect should occur to the same degree regardless of order of statement and behavior and the mitigating impact of order should occur to the same degree regardless of group matching. However, if mitigating circumstances are seen as more exculpatory under group matching or are resisted more under group mismatching, this will produce an interaction instead of or in addition to the main effects. In that case the group matching effect would be larger when mitigating circumstances are present than when they are absent. The test of this interaction provides the critical analysis in each of the current experiments.
Experiment 1 included an external control condition where the statement (promoting programs to reduce adult illiteracy) is unrelated to the behavior (driving drunk), so hypocrisy is low. This condition provides an additional test of the presence of a general effect of group matching on hypocrisy.
Method
Participants and design
Participants were 188 males from a national database for paid online studies conducted through Stanford University. Participants had to be a Democrat or a Republican, and either liberal or conservative, rather than neutral. Participants ranged in age from 18 to 61 and represented a variety of ethnic backgrounds.
Participants were randomly assigned to read about either a Democratic or Republican target who was the same gender as the participant (male). Participants read that the target had made a public statement before (conventional order) or after (reversed order) behaving inconsistently with that statement, or that he had made a public statement on an unrelated issue and also committed the same behavior as in the other conditions (unrelated action control). Within the unrelated action condition, the order of the unrelated statement and the behavior was counterbalanced. This counterbalance procedure produced no effects, so the factor was collapsed. The experiment thus followed a 2 (target group matching: match, mismatch) × 2 (target actions: conventional order, reversed order) between participants design with an external control condition.
Procedure
The experiment was presented online on three separate screens. On the first screen, participants read that the study involved person and group perception, and that they would be presented with two pieces of information about the target, “Mike S.”
Independent variables
Target matching
On the second page, participants saw a transcript of a radio station interview with Mike S. and a picture of him. To manipulate his party affiliation, the transcript opened with Mike saying he had been either a Democratic or Republican voter his whole life and that he volunteered for either the Democratic or Republican Party. The target match condition occurred when the participant and Mike were both Democrats or both Republicans; otherwise, the target mismatch condition occurred.
Target action manipulation
Participants then read one of three different versions of Mike’s statement and reported behavior. In the conventional order Mike made a statement against drunk driving and then 1 month later was reported to have driven drunk. In the reversed order condition, the drunk driving behavior occurred 1 month before making the statement against drunk driving, and the investigative reporting (i.e., drunk driving behavior) appeared above Mike’s statement to make the order manipulation clear (Barden et al., 2005).
In the unrelated actions condition, Mike advocated instead for reducing adult illiteracy rates. A pretest from the same population (N = 23) showed that evaluations on a 7-point scale (1 = dislike extremely, 7 = like extremely) were the same when adult illiteracy (M = 5.83, SD = 1.11) and drunk driving (M = 6.00, SD = 1.08) statements were presented by themselves, t(22) < 1. Thus any observed differences between the unrelated actions condition and the other conditions cannot be attributed to differences in the evaluation of the statements themselves. The order of the illiteracy statement and the drunk driving behavior was counterbalanced. As indicated before, this counterbalancing condition produced no effects, so it was collapsed.
Hypocrisy dependent measure
Participants were asked to write the first three traits that came to mind regarding Mike S. (Barden et al., 2005). Participants received a score of 0 if they did not mention hypocrisy, and a 1 if they wrote “hypocrisy,” “hypocrite,” or “hypocritical.”
Participant party
On a separate page, participants indicated their political party (Democratic, Republican, Green, libertarian, other, and independent). There were 107 Democrats and 81 Republicans. In addition to the 188 participants whose data are reported here, 34 participants were excluded because they reported belonging to a different party than they indicated in prescreening, a necessary exclusion criterion given the importance of accurate party identification.
Results
More participants in the conventional order (59%) mentioned hypocrisy than in the reversed order (28%), χ2(1) = 12.17, p < .001, replicating prior research (Barden et al., 2005). More important to our central question was whether order and matching had independent main effects, or interacted. A hierarchical log-linear analysis was conducted with order (conventional vs. reversed), matching, and hypocrisy entered as factors. An overall 2 (order: conventional, reversed) x 2 (target matching: matched party, mismatched party) chi-square revealed differences between the conditions on hypocrisy, χ2(1) = 4.58, p = .03. Pair-wise comparisons showed that under group matching, the reversed order caused a decrease in hypocrisy judgments (16%), compared to the conventional order (64%), χ2(1) = 15.62, p < .001, whereas under group mismatching, hypocrisy was the same for the reversed (40%), and conventional order (53%), χ2(1) = 1.07, p = .30. Thus, the reversed order mitigated hypocrisy under group matching, but not under group mismatching (see Table 1). When participant party and target party were individually added as third factors in the conventional and reversed condition, they produced nonsignificant interactions (χ2[1] = 2.33, p < .13 for participant party; χ2[1] = .03, p = .86 for target party), indicating that the effects were comparable for participants and targets of both parties. Furthermore, pair-wise comparisons showed that only in the reversed order condition did group matching elicit fewer judgments of hypocrisy (16%) than group mismatching (40%), χ2(1) = 4.32, p < .04, whereas no difference was observed in the conventional order, χ2(1) = .75, p = .39.
Experiment 1: Percentage of participants spontaneously mentioning hypocrisy as a function of target matching (match vs. mismatch) and target actions (conventional order vs. reversed order).
Note. The external control condition elicited a uniformly low percentage of spontaneous hypocrisy mentions, P = 7% (n = 61).
The external control condition elicited fewer hypocrisy mentions than even the reversed order condition (7%), χ2(1) = 10.29, p = .001, which is as expected since the statement and behavior were unrelated. In this condition, group matching produced no effect on hypocrisy judgments, χ2(1) = .10, p = .76, and no effects were observed when either participant party or target party were added as third factors, both χ2(1) = .31, p = .58.
Discussion
Experiment 1 found that the reversed order provided a mitigating circumstance for hypocrisy when the observer and target were from matched but not mismatched groups. The larger impact of the mitigating circumstance under group matching produced group bias on hypocrisy, but only in the reversed order condition. In contrast, when the mitigating circumstance was absent (conventional order), and when inconsistency was entirely absent (unrelated actions control), no group bias was observed. These results provide the first evidence that group bias is enhanced on hypocrisy when there is a specific combination of circumstances involving mitigating conditions.
Experiment 2
Experiment 2 sought to generalize the matching effects observed in Experiment 1 from political party to gender. To make this change for Experiment 2, all mentions of political party were removed and instead gender of the target person (Mike or Michelle) was varied and made salient. To focus on conditions that elicited at least a moderate level of hypocrisy, Experiment 2 included only conventional and reversed order conditions. Based on Experiment 1, it was predicted that the impact of mitigating circumstances (order) would be greater for targets that matched the observer’s gender compared to targets that were a mismatch. Finally, to increase generalizability of our dependent measures, Experiment 2 used scale items rather than spontaneous items to measure hypocrisy (Barden et al., 2005).
Method
Participants and design
Introductory psychology students (73 women and 73 men) at Ohio State University participated in partial fulfillment of a course requirement. The design was a 2 (order: conventional, reversed) × 2 (target matching: match, mismatch) between participants factorial.
Independent variables
Participant gender
A single item, at the end of the questionnaire, assessed whether participants were male or female.
Target matching
The drunk driving scenario from Experiment 1 was adapted to eliminate references to politics. Target gender was manipulated by referring to the target as Michelle or Mike and by using clearly gendered pronouns (his, he or her, she) 10 times throughout the scenario. Thus, participants judged targets who were matched or mismatched to them in gender.
Order manipulation
This manipulation was handled the same way as in Experiment 1.
Dependent Measure
On a page following the target information, hypocrisy judgments were assessed on two items using the same 7-point scale (1 = not at all, 7 = absolutely). One asked participants how hypocritical the target was and the other asked the extent to which the target was a hypocrite (α = .82).
Results
Hypocrisy judgments were analyzed using a two-way, 2 (order: conventional, reversed) × 2 (target matching: matched gender, mismatched gender) ANOVA. The conventional order elicited greater hypocrisy judgments (M = 6.38, SD = 0.98) than the reversed order (M = 4.46, SD = 1.94), F(1, 142) = 47.91, p < .001, η2 = .252. Target matching also produced a main effect, with targets of mismatched gender (M = 6.00, SD = 1.46) eliciting greater hypocrisy than matched gender ones (M = 5.33, SD = 1.84), F(1, 142) = 7.31, p < .01, η2 = .044.
Critically, these main effects were qualified by a significant Target Order × Matching interaction, F(1, 142) = 7.31, p < .01, η2 = .049, indicating that the impact of order depended on group membership (see Figure 1). Investigating the simple effects, the matched condition showed a large drop in hypocrisy judgments from the conventional (M = 6.36, SD = 1.02), to the reversed order (M = 4.03, SD = 1.84), F(1, 87) = 61.31, p < .01, η2 = .117, whereas in the mismatched condition, there was a smaller, but reliable, drop in hypocrisy judgments from the conventional (M = 6.40, SD = .96) to the reversed order (M = 5.33, SD = 1.90), F(1, 55) = 7.28, p < .01, η2 = .413. Thus, the presence of mitigation (reversed order) produced lower hypocrisy ratings for matched than mismatched targets, F(1, 58) = 6.52, p < .05, η2 = .101, which was not observed in the absence of that mitigation (conventional order), F < 1. When participant gender and target gender were individually added as third factors, they produced no main effects or interactions (all ps > .13) indicating that the effects were comparable for participants and targets of both genders. 1
Experiment 2: Judgments of hypocrisy as a function of whether information was presented in conventional order for hypocrisy or in the reversed order and whether the target’s gender was matched (ingroup) or mismatched (outgroup) to the participant’s gender.
Discussion
By systematically varying mitigating circumstances, Experiments 1 (political party) and 2 (gender) showed that the reversed order had a greater impact on hypocrisy when the observer and target were from matched compared to mismatched groups. This differential impact of the reversed order resulted in a bias in favor of ingroup members over outgroup members on hypocrisy in the reversed order, which was absent in the conventional order condition.
Experiment 3
Past work demonstrated that presenting hypocrisy information in an order opposite to the conventional order mitigates hypocrisy judgments because the inconsistency between statement and behavior is attributed to the target being a changed person (Barden et al., 2005). If individuals are more likely to look for, find, and use an attribution to allow leniency for matched targets or resist finding and applying the same attribution for mismatched targets, that could underlie a group matching effect on hypocrisy judgments. That is, attribution of sincere change that is differential based on group matching could explain how ingroup bias is justified and operates in the mind of observers. Experiment 3 tested the hypothesis that observers are more likely to see a target as having changed when an inconsistent statement follows behavior than when the statement precedes the inconsistent behavior, but this change attribution is more likely under group matching than mismatching.
Importantly, parallel effects on trait judgment and attributions to change would provide an opportunity to directly establish the role of attribution on ingroup bias. Though long anticipated by theory (Pettigrew, 1979), such mediation evidence is surprisingly rare. Some studies find effects on attributions but fail to find effects on traits (Weber, 1994), other studies find effects on either attributions or traits but do not measure the other (Beal et al., 2001; Rosenberg & Wolfsfeld, 1977), and finally some studies show the same effects on traits and attributions but do not test the correlation between them (Duncan, 1976; Taylor & Jaggi, 1974). In a rare exception, Chatman and von Hippel (2001) found that attributions mediated the main effect of group matching on trait judgments in response to a target portrayed as out of work. Thus, in addition to testing our mechanism for ingroup bias in hypocrisy judgments, Experiment 3 offers a rare opportunity for direct evidence of an attribution mechanism for a group-based judgment, and the possibility of the attribution influencing the joint effects of group matching and the mitigating factor of order on judgments of the trait of hypocrisy.
Method
Participants and design
Fifty-one Ohio State University undergraduates (37 women, 14 men; 33 Democrats, 18 Republicans) received course credit for participation. All participants indicated they had voted for the candidate from their own political party in the 2008 U.S. presidential election. Participants were randomly assigned to read about a Democratic or Republican target. For simplicity, participants were presented targets of the same gender. The experiment followed a 2 (target matching: match, mismatch) × 2 (order: conventional, reversed) between participants design.
Procedure
The experiment was conducted using the same procedures, instructions, vignette, and order conditions as Experiment 1, with the following changes. Male participants were directed to a website with a male target “Mike S.” that used male pronouns, and female participants were directed to a website with a female target “Michelle S.” that used female pronouns. Additional measures were added as indicated next.
Independent variables
Target matching
Observer and target matching on political party was handled as it was in Experiment 1.
Order manipulation
The drunk driving vignette appeared as in the conventional and reversed order in Experiment 1, except that female participants read about a target named Michelle instead of Mike.
Dependent variables
Hypocrisy index
Spontaneous items from Experiment 1 and the scale items from Experiment 2 were both used with the only change being that they referred to either Mike or Michelle depending on participant gender. The measures were standardized and averaged to form a hypocrisy index (α = .81).
Attribution of change
Next, participants responded to two questions, one being: “To what extent do you think that any inconsistency between what Mike [Michelle] said about drunk driving and his [her] behavior regarding drunk driving was the result of Mike [Michelle] becoming a different kind of person?” The second question was identical except that it ended with “learning a lesson.” Each question was presented with a 7-point scale (1 = not at all, 7 = absolutely). The two items were averaged to form an index (α = .68).
Participant party
Finally, participants indicated their political party (Democratic, Republican, Green, libertarian, other, or independent) and which presidential candidate they voted for in 2008 (Barack Obama, John McCain, other).
Results
Dependent measures were submitted to a 2 (target matching: match, mismatch) x 2 (order: conventional, reversed) analysis of variance (ANOVA).
Hypocrisy index
The target was rated as more hypocritical in the conventional order condition (M = .49, SD = .65) than in the reversed order condition (M = −.44, SD = .86), F(1, 47) = 19.1, p < .001, η2 = .26. The main effect of matching was not significant, F(1, 47) = 1.43, p = .24, η2 = .02. Critically, the interaction between order and matching was again significant, F(1, 47) = 3.99, p = .05, η2 = .05, indicating that the impact of order depended on group matching (see Figure 2). Simple effects tests revealed that the matched condition showed a reliable drop in hypocrisy from the conventional (M = .57, SD = .53), to the reversed order (M = −.76, SD = .64), F(1, 47) = 21.57, p < .001, η2 = .30, whereas in the mismatched condition, the conventional (M = .40, SD = .78) and reversed orders (M = −.09, SD = .95), F(1, 47) = 2.65, p = .11, η2 = .04, were seen as equally hypocritical.
Experiment 3: Judgments of hypocrisy as a function of whether information was presented in conventional order for hypocrisy or in the reversed order and whether the target’s political party matched (ingroup) or mismatched (outgroup) the participant’s political party.
In addition, when the dichotomous measure (from Experiment 1) and the scale measure (from Experiment 2) were analyzed separately, the same pattern of results was observed. 2 And, when participant party, target party, and participant gender were individually added as third factors, they produced no main effects or interactions (all ps > .25) indicating that the effects were comparable across these factors. As in Experiments 1 and 2, the reversed order mitigated hypocrisy to a greater extent for matched than mismatched targets. As a result, the presence of mitigation produced a lower hypocrisy rating for matched than mismatched targets, F(1, 47) = 5.43, p = .02, η2 = .07, which was not observed in the absence of that mitigation, F < 1, ns.
Attribution of change
A main effect indicated that participants in the reversed order condition were more likely to believe that the target had changed (M = 4.19, SD = 1.35), than participants in the conventional order condition (M = 2.65, SD = 1.32), F(1, 47) = 17.40, p < .001, η2 = .24. The main effect of matching was not significant, F < 1, p > .3. Importantly, the predicted Order x Matching interaction emerged, F(1, 47) = 5.34, p < .03, η2 = .07. Decomposition of the interaction revealed that the matched target condition showed greater attributions of change in the reversed order (M = 4.75, SD = 1.38) than the conventional order (M = 2.42, SD = 1.29), F(1, 47) = 22.39, p < .01, η2 = .31, whereas the mismatched condition showed the same attributions in the reversed order (M = 3.58, SD = 1.04) and conventional order (M = 2.91, SD = 1.38), F(1, 47) = 1.63, p = .21, η2 = .02. Thus, the pattern for attribution of change was consistent with hypocrisy judgments.
Mediation analysis
A mediated moderation analysis was conducted on the attribution of change measure (see Muller, Judd, & Yzerbyt, 2005). In a series of multiple regressions, order (0 = reversed, 1 = conventional) and matching (0 = mismatched, 1 = matched) interacted to predict hypocrisy judgments (the proposed dependent variable; b = .41, t[47] = 2.00, p = .05), and change attributions (the proposed mediator; b = −.48, t[47] = −2.31, p < .03, see Figure 3). Furthermore, hypocrisy ratings were negatively related to change attributions (b = −.66), t(47) = −6.21, p < .001. Controlling for change attributions, the Order x Matching interaction on hypocrisy was eliminated (b = .28), t(47) = 1.46, p = .15, whereas the relationship between change attributions and hypocrisy judgments remained reliable (b = −.49), t(47) = −3.75, p < .001. A A follow-up analysis, using macros provided by Preacher, Rucker, and Hayes (2007), revealed that change attributions mediated the relationship between order and hypocrisy in the matched condition (z = 2.90, p < .005), but not in the mismatched condition (z = 1.17, p = .24).
The association between the Political Party Matching x Order interaction and judgments of hypocrisy, as fully mediated through the change attribution.
General Discussion
Across three experiments, compared to the conventional order, the reversed order (behavior then statement) mitigated hypocrisy judgments when the observer and target were from matched groups to a greater extent than mismatched groups (Experiments 1, 2, and 3). The results demonstrate that factors extraneous to the target’s actions, namely whether a perceiver matches or mismatches the target’s group membership, can significantly affect the amount of hypocrisy ascribed to the target. Second, group matching did not produce a uniform group-based bias on either trait judgments or attributions to change. Instead, group-based bias was only observed when mitigating circumstances were systematically added while holding target actions constant (i.e., reversed order).
Although prior research provided some evidence of group-based bias on antecedents of hypocrisy (e.g., inconsistency judgments; Westen et al., 2006), much of the prior work did not manipulate the presence or absence of mitigating circumstances, nor did it demonstrate why mitigating circumstances produce group-based bias. We found that the mitigating circumstances, in the form of reversed order, increased attributions to change, but to a greater extent under group matching. Critically, these attributions to change mediated the impact of order and matching on hypocrisy judgments. This suggests a more general mechanism—that greater mitigation of moral judgments (e.g., guilt for a car crash) under group matching can occur because the mitigating circumstance (e.g., rainy weather) introduces another attribution for behavior.
Like the current research, Beal et al. (2001) varied circumstances to produce mitigation and this interacted with group matching. However they did not assess process variables. Only Chatman and von Hippel (2001) provided mediational evidence to support the role of attribution in group-based judgment, however this was based on a main effect of ingroup bias. Thus, the current work represents a rare case of direct evidence for the attribution mechanism underlying group-based bias on trait judgments. And critically, the current findings are distinct in that they show direct evidence of an attribution explaining the joint impact of group-matching and mitigating circumstances on judgment.
The current research indicates that the impact of mitigating circumstances can be eliminated for outgroup targets (Experiments 1 and 3), or it can be of smaller magnitude (Experiment 2) than for ingroup targets. In both cases, group matching produced greater mitigation than mismatching regardless of target group membership (e.g., mitigation was not greater for male or female targets). This is consistent with the target action (drunk driving) being no more stereotypic of one group than another (political party, gender). In contrast, some prior research on mitigating circumstances (Bodenhausen & Wyer, 1985) used behaviors that were stereotypic of one target group but not another (e.g., Hispanic and blue-collar crime), and effects were driven by stereotypicality rather than group matching. Stereotype-consistent behavior produced more stereotype-consistent judgments (parole recommendation) and eliminated the impact of mitigating circumstances. This raises the intriguing question of when the impact of mitigating circumstances might reflect the combination of matching and target group stereotypicality (e.g., moderate group identification, milder stereotypes). Such speculations await future research.
In sum, group matching effects on hypocrisy judgments can be based on differential mitigation, can occur independent of stereotype relevance, and can be explained by attributions that the target has changed. This significantly expands our understanding of when and how ingroup and outgroup members, when presented with the same target actions, can reach such different determinations of the level of hypocrisy.
Footnotes
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.