بی عدالتی ادراک شده و بزهکاری: آزمون نظریه فشار عمومی

Introduction Early strain theories applied Merton's (1938) macro-level anomie framework to the individual level of analysis, suggesting that interpersonal variation in crime results partly from lower class individuals’ frustration with their low expectations for achieving their idealized aspirations. In particular, Cohen (1955) argued that the inability to achieve the respect that comes with middle-class status contributed to offending among working class boys. Cloward and Ohlin (1960) extended this micro-level framework by suggesting that not only could blocked legitimate opportunities promote greater crime, but that the nature of individuals’ deviant opportunities could influence the types of delinquency that they might use to cope with frustrated aspirations. Although these early versions of strain theory have received relatively little empirical support, Agnew, 1992 and Agnew, 2006 General Strain Theory (GST) has broadened the concept of strain beyond blocked economic aspirations and sparked a resurgence of research specifying the process by which crime may be used as an adaptive response to strain. Within the GST framework (Agnew, 1992 and Agnew, 2006), individuals may be pressured into crime by a variety of stressors, including the experience of negative stimuli, the loss of positive stimuli, and the disjuncture between valued/actual outcomes. Crime is therefore seen as a coping strategy, as strain elicits negative emotion and creates incentive for corrective action, although strained individuals may also cope through legitimate means. Empirical research has yielded extensive support for the GST proposition that strain promotes crime (e.g., Agnew et al., 2002, Agnew and Raskin White, 1992, Hoffman and Cerbone, 1999 and Paternoster and Mazerolle, 1994), particularly in the case of strains that lower social constraints, incentivize crime, and are perceived as high in magnitude (Hay and Evans, 2006, Manasse and Ganem, 2009, Moon et al., 2008, Ostrowsky and Messner, 2005, Piquero and Sealock, 2004 and Spano et al., 2006). Recent work on GST has therefore continued to specify the types of strain most conducive to crime, to explore the mediating role of emotion, and to examine those variables most likely to condition the relationship between strain and offending. Injustice as a form of strain While a growing body of empirical literature supports many key predictions of General Strain Theory, unfair treatment remains a theoretically important (see e.g., Agnew, 2001 and Agnew, 2006) but empirically under-researched type of strain. The perception that a strain is not just unpleasant, as all strains are, but also undeserved, may increase the likelihood of negative emotional reactions, which may increase pressure for corrective action and decrease the ability to engage in legal forms of coping (Agnew, 2001, Agnew, 2006 and Broidy, 2001). Agnew (1992) originally conceptualized unfair treatment as a distinct category of strain arising from a disjuncture between fair/just outcomes and actual outcomes. More recent elaborations of GST (Agnew, 2001 and Agnew, 2006) suggest that any strain, from the death of a loved one to the application of a curfew, will be more likely to lead to crime if it is considered unjust. Nonetheless, much of the GST research on injustice (see Eitle, 2002, Hinduja, 2007, Mazerolle and Piquero, 1998, Moon et al., 2009a, Peter et al., 2003, Rebellon et al., 2009 and Walls et al., 2007) has focused on objective, rather than subjective, strain (but see Broidy, 2001). In particular, many tests of GST examine the association between discrimination and crime, yielding mixed support for Agnew's predictions. Cernkovich and Giordano (1979), for example, examined broad, group-based discrimination but found no effect on individual-level offending. Eitle (2002), on the other hand, found that crime and substance abuse were more likely among those who reported personal experience of discrimination. Similarly, Moon et al. (2009a) found that the experience of racial discrimination was significantly related to violent, but not non-violent, offending. Mazerolle and Piquero (1998) and Mazerolle, Piquero, and Capowich (2003) found that perceptions of being unfairly graded were predictive of intention to fight, but not other offending measures. Using a randomized experimental design, Rebellon et al. (2009) found that economic inequity was associated with situational anger which, in turn, was associated with greater criminal intent. This study, however, failed to find the anticipated direct association between economic inequity and criminal intent. Although the above studies suggest a generally positive, albeit inconsistent, relationship between unjust treatment and criminal behavior, they suffer from a number of significant limitations. First, the studies often ask respondents if they have experienced a particular strain, but do not ask how “unfair” they perceive it to be. While some strains, such as discrimination, are considered objectively unjust, individuals often do not evaluate strains they experience in a negative manner (Agnew, 2001 and Froggio and Agnew, 2007) and preliminary findings suggest subjective strains to be more strongly associated with crime (Froggio & Agnew, 2007). Second, existing studies tend to use narrow and/or dichotomous measures of unfair treatment (e.g., Rebellon et al., 2009). It is therefore impossible to capture the full variation in sources of perceived injustice across respondents or to evaluate the effect of relative levels of injustice on offending. Third, existing research often fails to control for variables from alternative theoretical perspectives, which may actually account for the relationship between injustice and crime. For example, while Gottfredson and Hirschi's (1990) theory of self-control might suggest that low self-control promotes both crime and perceptions of injustice, Aker's (2009) social learning theory might argue that delinquent associates promote both delinquent behavior and associated rationalizations that could invoke perceived injustice. Finally, prior research has been largely cross-sectional, thereby raising the issue of causal direction. In particular, delinquency could conceivably lead to stressful circumstances that a perpetrator could interpret as unjust. The role of negative emotion in GST While the bulk of GST research has focused on the causal relationship between strain and offending, a growing body of work has explored the central argument that strain increases the likelihood of crime through its effect on negative emotion. Agnew, 1992 and Agnew, 2006 argues that crime is an adaptive response insofar as it allows people to achieve effective, though often short-term, relief from the unpleasant negative emotions caused by strain (see Brezina, 1996 and Brezina, 2000). Anger has been given particular attention as a mediating mechanism because it tends to create the desire for corrective action and lower constraints against criminal coping strategies. Indeed, much of the GST research on unjust treatment has examined anger as an intervening mechanism, suggesting mixed support for GST predictions. Broidy (2001), for example, found that the perception of unfairness increased anger and that anger, in turn, increased illegitimate coping, although the study did not explicitly examine mediating effects. Mazerolle and Piquero (1998) found that, while perceptions of injustice were significantly related to increased anger, anger did not mediate the strain/offending relationship. Using a measure of racial discrimination, Moon et al. (2009a) also found that anger had only a minimal mediating effect. In general, while research has clearly shown that the experience of anger increases the likelihood of offending (Bao et al., 2004, Broidy, 2001, Hollist et al., 2009, Jang and Johnson, 2003, Jennings et al., 2009, Mazerolle and Piquero, 1998, Mazerolle et al., 2003, Moon et al., 2009a, Piquero and Sealock, 2004, Rebellon et al., 2009 and Sigfusdottir et al., 2004), most research on anger has shown only partial mediating effects (see Aseltine et al., 2000, Capowich et al., 2001, Hay and Evans, 2006, Hollist et al., 2009, Jennings et al., 2009, Mazerolle and Piquero, 1997, Moon et al., 2009a and Piquero and Sealock, 2004) or mediating effects for only violent/aggressive forms of offending (e.g. Aseltine et al., 2000, Bao et al., 2004 and Mazerolle et al., 2003). Although the above studies that examine mediation effects tend to yield only mixed or partial support for GST's mediation hypothesis, these findings may reflect the disjuncture between GST's claims about emotions like anger and the measures often used to reflect them. Specifically, most studies evaluating emotion as an intervening mechanism use dispositional measures of emotion, which do not focus primarily on a respondent's emotional state at the time of offense. For example, dispositional measures of anger tend to capture whether a respondent has a “bad temper” relative to other respondents, which may reflect a component of Gottfredson and Hirschi's self-control construct as much or more so than they reflect Agnew's conception of strain-induced, situational anger (see Rebellon et al., 2009). Examining these particular methodological concerns, Mazerolle et al. (2003) considered the mediating role of both situational and dispositional measures of anger. Tellingly, the authors found mediating effects for situational measures of anger, but no mediating effect for dispositional measures of anger. It is therefore likely that dispositional measures of emotion underestimate the mediating effects of emotion on offending (Agnew, 2006 and Mazerolle et al., 2003). A limited number of other studies provide further reason to believe that measures of situational emotion are critical for proper tests of GST's mediation hypotheses. For example, Jang and Johnson (2003) and Moon, Morash, McCluskey, and Hwang (2009b) found that measures of situational anger fully mediated the relationship between strain and offending, although neither of these studies looked explicitly at measures of injustice. In their examination of economic inequity, Rebellon et al. (2009) found that objective economic inequity was significantly associated with situational anger and that situational anger, in turn, was significantly associated with criminal intent. The latter study, however, failed to find a direct relationship between economic inequity and criminal intent but this may be due, at least in part, to the limitations intrinsic to a vignette study evaluating intention rather than behavior, or to the limited variance in the study's measure of economic inequity. The present study Using data from several waves of the New Hampshire Youth Study, the present study provides a more extensive test of the relationship between unjust treatment and delinquency than prior research. Specifically, it addresses several of the methodological issues limiting prior work to provide a more thorough test of (a) the extent to which perceptions of unjust treatment predict delinquency and (b) the extent to which this relationship is mediated by situational anger. In light of results, implications for GST and for crime control will be discussed.

Results Fig. 1 depicts the bivariate relationship between perceived fairness and delinquency in five separate social domains. The first two bars compare individuals who believed their mothers had generally treated them fairly over the course of the preceding six months with those who believed their mothers had generally treated them unfairly. Fig. 1 demonstrates that the former group, on average, reported engaging in fewer than four types of delinquent behavior over the course of the preceding six months whereas the latter group, on average, reported engaging in nearly seven types. Beyond its statistical significance, this difference is rather substantial in magnitude. The remaining four sets of bars in Fig. 1 suggest similar results for perceived unfairness in respondents’ relationships with their fathers, friends, other peers, and teachers. One further result is noteworthy, however. Specifically, while perceptions of unfairness among one's friends were significantly associated with more delinquency, the magnitude of this relationship is comparatively smaller than the corresponding magnitude of the relationship between delinquency and perceived unfairness among parents, teachers, or non-friend peers. This suggests that, while many youth in our sample would appear to be at greater risk for delinquency when they feel unfairly treated, they may be particularly at risk upon experiencing injustice at the hands of either authority figures, like parents and teachers, or at the hands of peers with whom they are not friends. Bivariate relationship between perceived fairness and delinquency. Fig. 1. Bivariate relationship between perceived fairness and delinquency. Figure options In light of GST's hypothesis concerning the mediating role of negative emotions like anger in promoting delinquency, Fig. 2 depicts the bivariate relationship between perceptions of injustice and person-specific, situational anger. Establishing a significant bivariate relationship between perceived injustice and anger is a necessary, though not sufficient, condition for establishing that GST is correct in predicting that anger mediates the relationship between perceived injustice and delinquency. Indeed, Fig. 2 suggests clear support for the notion that perceived injustice is associated with anger. The first two bars in Fig. 2, for example, demonstrate that respondents who believed their mothers to have been generally unfair over the course of the preceding six months reported significantly more frequent feelings of anger with their mothers. The next two bars in Fig. 2 suggest the same conclusion about respondents’ anger toward their fathers. Those respondents who reported that their fathers had generally been unfair over the course of the preceding six months tended to report higher levels of anger directed specifically at their fathers. Of note, although perceived unfairness among friends and other peers was associated with anger, Fig. 2 suggests greater emotional tolerance of unfair treatment from these groups. This conclusion is indicated by the relatively small, albeit statistically significant, gap between peer-directed anger among youth who believed their peers had treated them unfairly and the corresponding peer-directed anger among youth who believed their peers had treated them fairly. Given that the anger gap appears to once again increase in magnitude among those who perceive fair versus unfair treatment at the hands of teachers, results from Fig. 2 again suggest to us that unfair treatment is less tolerated among the youth in our sample when it comes from authority figures like parents and teachers as opposed to friends or other peers. Bivariate relationship between perceived fairness and anger. Fig. 2. Bivariate relationship between perceived fairness and anger. Figure options Table 2 extends our assessment of the relationship between perceived fairness and anger to a multivariate framework. Model 1 provides a baseline Ordinary Least Squares (OLS) assessment of the degree to which demographic characteristics are associated with situational anger. We note that our dependent variable throughout Table 2 is our situational anger scale, comprised of the average anger (ranging from 0-4) that respondents felt toward parents, peers, and teachers combined. With this in mind, Model 1 yields evidence that total anger was higher among older youth versus younger youth and among non-White versus White youth. Model 2 adds our perceived fairness scale, reflecting the average fairness that each respondent perceived in his or her relationships with parents, peers, and teachers, as an independent variable. Results reveal that average perceived fairness is associated significantly with average situational anger even after controlling for demographics. Further, although the relationship between age and anger appears to be robust against the inclusion of perceived fairness, Model 2 suggests that the relationship between race and anger from Model 1 may simply reflect minorities’ greater perceptions of unfairness. Turning to Model 3, we add our negative life events index as a predictor in order to examine whether perceived fairness is, in fact, an important type of anger-inducing strain in itself, or whether it is simply serving essentially as a proxy for more traditionally measured forms of general strain. Results reveal that, while negative life events would appear to be associated with anger, perceived unfairness remains independently associated with anger. This suggests that prior research concerning GST may sometimes have omitted an important type of strain that merits attention independent of more commonly measured strains. In particular, our results suggest preliminary support for Agnew's (2001) explicit argument that strains involving injustice may be among those most likely to promote crime-inducing anger. Table 2. OLS regression of situational anger on perceived fairness (1) (2) (3) Anger Anger Anger Age .05(.02)* .04(.02)* .04(.02)* Female .07(.06) .09(.06) .07(.06) SES -.03(.03) .00(.02) .01(.02) White -.14(.08)* -.06(.07) -.05(.07) Fairness -.60(.06)* -.46(.06)* Life Events .06(.01)* Constant 1.26(.31)* 2.61(.30)* 2.02(.32)* R-Squared .02 .16 .19 N 687 687 687 * p < .05 (one-tailed). Table options Table 3 presents a series of negative binomial analyses examining whether the bivariate relationship between perceived fairness and delinquency holds in a multivariate framework that controls for demographics and for variables derived from alternative theories. We opted to present negative binomial regression results because our delinquency measure reflects a variety score which, in turn, is distributed as a count (i.e., non-negative integers). OLS analysis would therefore not be appropriate (see Long, 1997). Further, given the over-dispersion that is suggested by the significant alpha coefficients throughout Table 3, Poisson regression would likely underestimate standard errors, thus artificially inflating the significance of each predictor variable. Table 3. Negative binomial regression of delinquency on perceived fairness (1) (2) (3) (4) Age .11(.02)* .11(.02)* .01(.02) .01(.02) Female -.32(.07)* -.32(.07)* -.09(.06) -.11(.06)* SES -.08(.03)* -.05(.03) .00(.02) .00(.02) White -.08(.09) -.01(.08) .06(.07) .07(.07) Fairness -.53(.06)* -.12(.06)* -.07(.06) Life Events .03(.01)* .03(.01)* Low SC .32(.06)* .31(.06)* Peer Del .33(.07)* .31(.07)* Attitudes -.13(.05)* -.13(.05)* Prior Del .07(.01)* .07(.01)* Anger .14(.04)* Constant .42(.33) 1.5(.34)* .82(.33)* .55(.34) α .60(.05)* .50(.04)* .21(.03)* .20(.02)* LL -1702.5 -1664.1 -1507.4 -1500.4 N 687 687 687 687 * p < .05 (one-tailed). Table options Model 1 serves as a baseline model examining the degree to which demographics are related to our delinquency measure. In line with expectations derived from innumerable studies in the criminological literature, our data reveal that the older youth in our sample, who are approaching the peak of the age-crime curve, report involvement in a greater variety of delinquent behaviors. Also in line with expectations, females tend to report involvement in fewer types of delinquency. While our data do not find evidence of a relationship between race and delinquency, they do reveal that our measure of SES is negatively associated with delinquency such that respondents whose parents have higher levels of education tend to report less delinquency. Model 2 adds our perceived fairness scale as a predictor variable, revealing that those who perceive fair treatment from parents, peers, and teachers tend to be involved in fewer types of delinquency even after adjusting for demographic variables. More strikingly, however, Model 3 yields evidence that the relationship between perceived fairness and delinquency, though lower in magnitude, remains statistically significant even after adjusting statistically for negative life events, low self-control, perceived peer delinquency, negative attitudes toward delinquency, and prior delinquency. The importance of this result is difficult to over-state. In particular, it would appear that the relationship between perceived fairness and delinquency among youth in our sample is robust against the inclusion of those variables that represent the strongest and most reliable predictors of delinquent behavior in the existing quantitative literature. Moreover, in keeping with GST's predictions, Model 4 suggests that situational anger is significantly associated with delinquency net of these same controls and that it appears to mediate the remaining relationship between perceived fairness and delinquency. In sum, OLS and negative binomial analyses suggest strong support for GST's assertions that perceived fairness is an important predictor of delinquency and that its influence on delinquency is mediated by negative emotion. Fig. 3 depicts a path analysis estimated via Lisrel 8 as a means of both replicating and summarizing the above findings. Recalling that our “propensity measure” combines our low self-control, perceived peer delinquency, and attitudes toward delinquency scales with our prior delinquency index, Fig. 3 reveals the expected relationship between propensity and delinquency. Further, it reveals that propensity is associated strongly with both negative life events and perceptions of fairness, which themselves are significantly related. Independent of these relationships, Fig. 3 reveals that negative life events and perceived fairness are associated with anger and that anger, in turn, is associated with delinquency. We note that, in an alternative model (not shown), we added estimates of the path coefficients linking fairness and life events directly to delinquency. Neither estimate yielded a statistically significant coefficient. Perhaps as a result, fit statistics for this alternative model were poorer than the fit statistics depicted in Fig. 3, which all indicate a very good fit to the data. Replication and summary model (N=746)ab. Fig. 3. Replication and summary model (N = 746)ab.