چرا و چه موقع حساسیت به اجرای عدالت منجر به رفتارهای ضداجتماعی و سازگار با اجتماع می شود

Abstract People differ in how injustice-sensitive they are either as victims or as observers. Whereas observer sensitivity is positively related to cooperative behavior, victim sensitivity promotes antisocial and egoistic behavior. The present article investigates the dynamics underlying these effects. Participants played an online-based public goods game and were informed about the number of people who violated a fairness rule in previous rounds of the game (no, some, or many violators). High victim-sensitive participants contributed less to the public good even in the “some violators” condition. High observer-sensitive participants contributed more to the public good even in the “many violators” condition. The findings correspond with the sensitivity to mean intentions model and cannot be explained by individual differences in general trust.

Introduction People differ with regard to how they react towards injustice. Some people are deeply concerned about injustice, they are more likely to interpret social situations in terms of justice, they experience strong negative emotions in the face of injustice, and they tend to ruminate longer about alleged injustice. Others are relatively insensitive towards injustice: Justice is not a frequent issue in their social lives, injustice is not emotionally disturbing or cognitively absorbing; those people are also less motivated to act against injustices. Justice Sensitivity has been shown to be a relatively stable and consistent personality variable that predicts how and when people react towards experienced or witnessed injustice (Mohiyeddini and Schmitt, 1997, Schmitt, 1996, Schmitt and Dörfel, 1999, Schmitt et al., 2005, Schmitt and Mohiyeddini, 1996 and Schmitt et al., 2009). People can experience justice-related situations from different perspectives (Mikula, 1994): They can be victims, beneficiaries, or observers of injustice. Accordingly, people can be justice sensitive from the victim’s perspective, the beneficiary’s perspective, and the observer’s perspective. Recent research has shown that these three justice sensitivity perspectives are positively intercorrelated, and that they all correlate positively with justice-related traits and attitudes, such as belief in a just world, belief in immanent and ultimate justice, and sense of injustice (Schmitt et al., 2005). On the other hand, the three perspectives correlate differently with other personality constructs and behavioral outcomes. More specifically, observer sensitivity and beneficiary sensitivity correlate positively with prosocial dispositions and other-related concerns such as empathy, social responsibility, modesty, or agreeableness, whereas victim sensitivity correlates positively with self-related concerns such as jealousy, neuroticism, vengeance, and paranoia (Schmitt et al., 2005). Furthermore, both vignette and experimental studies have shown that observer and beneficiary sensitivity correlate positively with prosocial behaviors such as solidarity with the disadvantaged and equal split offers in a dictator game, whereas victim sensitivity correlates positively with antisocial behaviors such as immoral choices in enticing situations and delinquent behavior in real life (Fetchenhauer and Huang, 2004 and Gollwitzer et al., 2005). The present study aims to clarify these asymmetries among justice sensitivity perspectives with regard to pro- and antisocial behavior. Specifically, we will investigate under which conditions victim sensitivity and observer sensitivity lead to pro- and antisocial behavior in a social dilemma situation. The theoretical framework underlying this research is the “sensitivity to mean intentions” (SeMI) model (Gollwitzer & Rothmund, 2009). A high sensitivity to mean intentions is defined as a particular readiness to respond to contextual cues suggesting that there is a danger of being exploited by others. The model posits that contextual meanness cues can evoke a state of suspiciousness and that people who are sensitive to mean intentions have a lower threshold for the activation of a suspicious mindset. In other words: For people who are sensitive to mean intentions, even slight cues of “meanness” can evoke a state of suspiciousness. In the present research, we assume that victim sensitivity is an indicator of sensitivity to mean intentions. In terms of regulatory focus theory (Higgins, 1997 and Higgins and Spiegel, 2004), suspiciousness involves a prevention focus, which is a regulatory state concerned with protection and safety (more generally, the absence or presence of negative outcomes). A suspicious mindset should have all social-cognitive features that are typical for a prevention focus, such as subtractive counterfactual thinking, a higher prevalence of more stable, universal, and personal attributions, a more conservative response bias, and a quicker, but also more tenacious goal-pursuit (Higgins & Spiegel, 2004). Most importantly, when a suspicious mindset is activated, people are more likely to behave uncooperatively and antisocially in order to prevent being exploited (Deutsch, 1958, Vohs et al., 2007 and Yamagishi and Sato, 1986). Since only slight contextual cues of meanness or untrustworthiness can evoke a suspicious mindset among people high in victim sensitivity, these people should be particularly likely to behave uncooperatively and antisocially when confronted with such cues. At that point, it is important to distinguish victim sensitivity from general trust. Trust has been defined as an “… expectancy held by an individual or a group that the word, promise, verbal or written statement of another individual or group can be relied upon” (Rotter, 1967; p. 651) or, more simply, as a general belief in human benevolence (Yamagishi & Yamagishi, 1994). The conceptual difference between general trust and victim sensitivity is that trust merely entails an expectancy regarding others’ benevolence, that is, a cognitive aspect, whereas victim sensitivity entails both a cognitive and a motivational aspect: People high in victim sensitivity are more likely to affix mean intentions to other people, but at the same time, they strongly disapprove of such intentions. For people high in victim sensitivity, the possibility that other people harbor mean intentions is highly aversive. This is what makes them more sensitive towards meanness cues. General trust, on the other hand, does not necessarily imply a motivational aspect, and therefore, no particular sensitivity towards cues of other persons’ alleged meanness ( Gollwitzer & Rothmund, 2009). The SeMI model also formulates conditions that mitigate the translation of suspiciousness into uncooperative behavior. These conditions can be features of the particular situation (such as the likelihood of being punished for behaving uncooperatively) or features of the person (such as high moral standards or a strong moral identity; cf. Aquino & Reed, 2002). Recent research suggests that moral identity (as one aspect of a person’s moral self-regulation) serves as a buffer against immoral judgments and behavior (Aquino, Reed, Thau, & Freeman, 2007). Observer sensitivity is assumed to be closely connected to one’s moral identity (Gollwitzer et al., 2005 and Schmitt et al., 2005). Thus, we expect that people high in observer sensitivity should refrain from behaving antisocially even if they have strong reason to believe that other people harbor mean intentions and that they might be exploited by others. The conceptual relations between contextual meanness cues, victim sensitivity, observer sensitivity, and uncooperative behavior are graphically depicted in Fig. 1. Moderator effects of victim and observer sensitivity according to the SeMI ... Fig. 1. Moderator effects of victim and observer sensitivity according to the SeMI model. Figure options Taken together, the present study aims to test three hypotheses. First, we expect that people high in victim sensitivity behave uncooperatively when they are confronted with only slight cues of meanness. Second, we expect that people high in observer sensitivity refrain from behaving uncooperatively even when they are confronted with strong cues of meanness. Third, general trust does not imply a heightened sensitivity towards mean intentions; thus, people low in trust should behave uncooperatively even if they are not confronted with any cues of meanness. These three hypotheses will be tested by means of social dilemma paradigm (Dawes, 1980, Komorita and Parks, 1996 and Messick and Brewer, 1983). Each individual involved in a social dilemma can decide to either cooperate or not (i.e., to “defect”). The payoff structure in a social dilemma is constructed such that the individual payoff is highest when a person defects while the others cooperate (“free-rider”), whereas the individual payoff is lowest when a person cooperates while the others defect (“sucker”). Thus, an individual’s decision whether to cooperate or to defect in a social dilemma is largely shaped by his or her expectation regarding the intentions and actions of others (Brann and Foddy, 1987, Dawes, 1980, De Cremer et al., 2001 and Kelley and Stahelski, 1970). In the present study, we manipulated to what extent individuals had reason to believe that other players in the social dilemma situation might defect. That is, we provided our participants with alleged information about the rate of “free-riders” in the game. This information was used as cues of meanness or untrustworthiness (see below).

3. Results 3.1. Average contribution and deviation from the equal division rule The three hypotheses were tested via moderated regressions. Experimental conditions were transformed into two dummy variables. Dummy_1 tested the contrast between the “no violators” and the “some violators” condition; Dummy_2 tested the contrast between the “no violators” and the “many violators” condition.2 Hypotheses 1 and 2 were tested simultaneously. Thus, victim sensitivity, observer sensitivity, and the interaction terms between the dummy variables and both justice sensitivity perspectives were included in the model. Victim sensitivity and observer sensitivity were standardized prior to computing interaction terms (Aiken & West, 1991). Table 1 lists all regression coefficients and the respective significance tests. Table 1. Results of regression analyses (Hypotheses 1 and 2). Effect Average deviation from the equal division rule Number of violations of the equal division rule β t β t Dummy_1 (contrast “no violators” vs. “some violators”) −0.14 −1.03 0.52 4.07⁎⁎ Dummy_2 (contrast “no violators” vs. “many violators”) −0.42 −3.24⁎⁎ 0.66 5.39⁎⁎ Victim sensitivity 0.04 0.36 −0.12 −1.12 Observer sensitivity −0.01 −0.09 0.10 0.96 Dummy_1 × victim sensitivity −0.30 −1.95⁎ 0.36 2.45⁎ Dummy_2 × victim sensitivity −0.22 −1.44 0.38 2.70⁎⁎ Dummy_1 × observer sensitivity 0.06 0.37 −0.25 −1.71 Dummy_2 × observer sensitivity 0.35 2.39⁎ −0.53 −3.92⁎⁎ Note: N = 324. All continuous variables (including dependent variables) are standardized. ⁎ p ⩽ .05. ⁎⁎ p ⩽ .01. Table options Overall, we found a main effect of our experimental manipulation with regard to the contrast between “no violators” and “many violators”, t(315) = 3.24; p = .001: Participants in the “no violators” condition contributed M = 11.95 tokens (SD = 2.99), which is 1.95 more than the equal division rule would prescribe. Participants in the “some violators” condition contributed M = 11.57 tokens (SD = 3.70), and participants in the “many violators” condition contributed M = 10.57 tokens (SD = 2.90). Thus, in all three conditions, participants contributed, on average, more than the equal division rule would have prescribed (M = 1.34; SD = 3.24). More importantly, and in line with our first hypothesis, there was a marginally significant interaction between Dummy_1 and victim sensitivity, t(315) = 1.95; p = .05: Among people high in victim sensitivity, those in the “some violators” condition contributed less than those in the “no violators” condition, whereas there was no such effect among people low in victim sensitivity. The interaction between Dummy_2 and victim sensitivity, on the other hand, was not significant, which implies that participants contributed less in the “many violators” compared to the “no violators” condition, irrespective of how victim-sensitive they were, t(315) = 1.44; p = .152. Conditional expected means for standard deviation scores of ±1 on victim sensitivity are graphically depicted in Fig. 2 (Panel A). Average deviation from the equal division rule by experimental condition and ... Fig. 2. Average deviation from the equal division rule by experimental condition and victim sensitivity (Panel A) and observer sensitivity (Panel B). Figure options In line with our second hypothesis, we found that the effect of our experimental manipulation (Dummy_2) was also moderated by observer sensitivity, t(315) = 2.39; p = .018: Among people low in observer sensitivity, those in the “many violators” contributed less than those in the “no violators” condition, whereas no such effect was found among people high in observer sensitivity. In other words: People high in observer sensitivity contributed more than the equal division rule would prescribe, even after having been confronted with a relatively large number of violators. Conditional expected means for standard deviation scores of ±1 on observer sensitivity are graphically depicted in Fig. 2 (Panel B). Hypothesis 3 was tested with an independent regression model. This model included the effects of the two dummy variables, general trust (standardized), and the interaction terms Dummy_1 × general trust and Dummy_2 × general trust. However, besides a main effect of Dummy_2 (see above), no other effects were significant (p > .33). Thus, as expected, general trust did not interact with the experimental manipulation. Unexpectedly, however, general trust did not even have a main effect on the average deviation from the equal division rule, β = .05; t(315) = 0.06; p = .96. 3.2. Number of violations The number of rounds in which participants violated the equal division rule ranged between 0 and 4 (M = 1.06; SD = 1.22). Again, Hypotheses 1 and 2 were tested simultaneously, whereas Hypothesis 3 was tested independently. The regression coefficients and respective significance tests for the model testing Hypotheses 1 and 2 are reported in the right column of Table 1. 3 Overall, we found a strong main effect of our experimental manipulation: Participants in the “some violators” (M = 1.22; SD = 1.29) and those in the “many violators” condition (M = 1.39; SD = 1.26) violated the equal division rule more often than those in the “no violators” condition (M = 0.58; SD = 0.94). Both dummy variables had significant main effects (Dummy_1: t(315) = 4.07; p < .001; Dummy_2: t(315) = 5.39; p < .001). More importantly, and in line with our first hypothesis, the interaction effect between Dummy_1 and victim sensitivity was significant, t(315) = 2.45; p = .015. People high in victim sensitivity violated the equal division rule more often in the “some violators” condition compared to those in the “no violators” condition. Among people low in victim sensitivity, the experimental manipulation did not influence the number of violations. Conditional expected means for standard deviation scores of ±1 on victim sensitivity are graphically depicted in Fig. 3 (Panel A). Number of rounds in which the equal division rule was violated by experimental ... Fig. 3. Number of rounds in which the equal division rule was violated by experimental condition and victim sensitivity (Panel A) and observer sensitivity (Panel B). Figure options In line with our second hypothesis, we found a significant interaction effect between Dummy_2 and observer sensitivity, t(315) = 3.92; p < .001: People high in observer sensitivity were less likely to violate the equal division rule independent of how many violators they were confronted with, whereas people low in observer sensitivity violated the equal division rule more often when they were confronted with “many” violators. Conditional expected means for standard deviation scores of ±1 on observer sensitivity are graphically depicted in Fig. 3 (Panel B). In line with our third hypothesis, we found that general trust only had a negative main effect on the number of violations (β = .21; t = −2.24; p = .03). Thus, the higher participants scored on general trust, the less likely they were to violate the equal division rule. More importantly, and consistent with our reasoning, general trust did not significantly interact with our experimental manipulation (p ⩾ .09).