1 GENERAL STRAIN AMID REST ORATION: AN EXAMINATION OF INSTRUMENTAL AND EXPRESSIVE OFFENSES By STEPHEN K. RICE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006
2 Copyright 2006 by Stephen K. Rice
3 In thanks to the University of Florida, my flagship.
4 ACKNOWLEDGMENTS I would like to extend my sincere apprecia tion to the members of my supervisory committee, Professors Alex Piquero, Ron Akers, Nicky Piquero, Mark F ondacaro, Brian Stults, and Chuck Peek. It is hard to believe that my first introducti on to criminology was in one of Ronâ€™s classes twenty years ago. At this point there is a ve ry satisfying internal logic to it all. The nineteen year old thanks Ron from the st art. I thank Nicky for her friendship and for providing me with a wonderful roadmap for the ear ly years ahead. I than k Brian and Chuck for the same and wish them both nothing but great success. I thank Mark for introducing me to social science (in) (of) (and) the law. I would also like to thank Lonn Lanza-Kaduce, Karen Parker, Charles Gattone, and Leonard Beeghley fo r their support and for introducing me to the habits of mind necessary to â€œkic k the tiresâ€ of good --and not so good--theory. Also, I would like to thank the Department of Sociology and the Univ ersity of Florida for the financial support in the form of the Alumni Graduate Fellowship. And finally, my sincere thanks go to Alex, a person who has somehow managed to provide that perfect piece of advice, both big and small, over what feels like every day of the past four years. I realize he will mentor many dozens of students during his stunning career, but from this one student I cannot th ank him enough for his friendship and for helping me to realize there is another way. None of this would be possible without the love and support of my family: to Karen (my heart), to Connor (my inspiration) , and to Brogan (my perspective). I thank my parents for their encouragement and love; this is a nice counter point to the second grade tutor. And I thank Martha and Jerry for their support (figuratively and liter ally) over the past year.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........8 LIST OF FIGURES................................................................................................................ .........9 ABSTRACT....................................................................................................................... ............10 CHAPTER 1 INTRODUCTION..................................................................................................................12 2 REVIEW OF THE LITERATURE........................................................................................16 General Strain: Constraint or Motivation for Crime?.............................................................16 Sequence of Topics............................................................................................................. ....18 The â€œContinuityâ€ of Strain Theory.........................................................................................18 Merton and Strain in American Society..........................................................................19 Innovators.................................................................................................................21 Conformists..............................................................................................................21 Ritualists...................................................................................................................22 Retreatists.................................................................................................................22 Rebels.......................................................................................................................22 Cohenâ€™s Delinquent Boys................................................................................................23 Cloward and Ohlinâ€™s Delinquency and Opportunity.......................................................24 Importance of community dynamics........................................................................25 Conflict subculture...................................................................................................25 Criminal subculture..................................................................................................26 Retreatist subculture.................................................................................................26 Evolution of Classic Strain Theory.................................................................................27 Agnewâ€™s Revised Strain Theory......................................................................................28 General Strain Theory.......................................................................................................... ...29 Foundational Principles and Types of Strain...................................................................30 Mediators and Conditioners............................................................................................31 Characteristics of Strain That ar e Most Likely to Cause Crime......................................32 Strain as unjust.........................................................................................................32 Strain as high in magnitude......................................................................................33 Strain associated with low social control.................................................................33 Pressure/incentive to engage in criminal coping......................................................34 Magnitude and duration/frequency..........................................................................35 Magnitude and recency............................................................................................35 Magnitude and centrality..........................................................................................36 Cumulative impact of characteristics.......................................................................36 Types of Strain That Are Mo st Likely to Cause Crime...................................................36
6 GST Samples and Areas of Inquiry.................................................................................37 Tests Related to the Strains Most Likely to Cause Crime...............................................38 Failure to achieve goals related to thrills, autonomy, stat us, acquisition of money...................................................................................................................38 Criminal victimization..............................................................................................39 Homelessness...........................................................................................................41 Discrimination based on ascribed characteristics.....................................................41 Child abuse and neglect............................................................................................42 Other empirically supported strains.........................................................................43 Restorative Justice Theory..................................................................................................... .45 Relationship to Retributism.............................................................................................46 Reintegrative Shaming....................................................................................................46 Canberra Reintegrative Shaming Experiments (RISE)...................................................49 Study hypotheses......................................................................................................50 Data collection methods...........................................................................................50 Acceptance criteria...................................................................................................51 Method of randomization.........................................................................................52 Ethical considerations..............................................................................................53 Response rates..........................................................................................................53 RISE findings...........................................................................................................54 More Recent Efforts: Jerry Lee Program........................................................................55 Departure from RISE / Re storative Justice Focus...........................................................56 3 METHODOLOGY.................................................................................................................59 Sample......................................................................................................................... ...........59 Measures....................................................................................................................... ..........61 Strain......................................................................................................................... .......61 Disjunction between expectati ons and actual achievements....................................63 Disjunction between aspira tions and expectations...................................................64 Presentation of negative stimuli...............................................................................64 Removal of positive stimuli.....................................................................................65 Empirical independence of strain categories............................................................66 Negative Affect...............................................................................................................66 Intervening Variables......................................................................................................68 Constraints to delinquent coping..............................................................................68 Facilitators of delinquent coping..............................................................................71 Outcome Measure............................................................................................................73 Sociodemographics/controls............................................................................................75 Descriptive Statistics......................................................................................................... .....75 4 ANALYSIS AND PRESENTATION OF RESULTS............................................................79 Analytic Plan.................................................................................................................. ........80 Hypothesized Relationships............................................................................................80 Measure associations................................................................................................81 Cumulative effect of strain.......................................................................................83
7 Nonlinearity..............................................................................................................84 Facilitators of delinquent coping..............................................................................84 Constraints to delinquent coping..............................................................................86 Mean-level Differences between Groups of Interest.......................................................88 Sex............................................................................................................................88 Age...........................................................................................................................89 Minority/majority.....................................................................................................90 Crime type................................................................................................................91 Regression Estimations...................................................................................................92 Study Measures and Restorative Justice..........................................................................94 5 DISCUSSION AND CONCLUSION..................................................................................104 Key Contributions.............................................................................................................. ...104 Limitations.................................................................................................................... ........108 Policy Implications............................................................................................................ ...109 Future Directions for Research.............................................................................................109 APPENDIX A INDIVIDUAL SURVEY ITEMS AND SCALE CONSTRUCTION.................................112 B CORRELATION MATRIX.................................................................................................116 LIST OF REFERENCES.............................................................................................................117 BIOGRAPHICAL SKETCH.......................................................................................................131
8 LIST OF TABLES Table page 3-1 Factor analysis (rotated) for presentation of negative stimuli (Factor 1) and removal of positive stimuli (Factor 2)..............................................................................................76 3-2 Descriptive statistics..................................................................................................... .....77 4-1 Predicted relationships, st udy measures and criminality...................................................97 4-2 Nonlinear effect, strain on criminality...............................................................................98 4-3 Mean-level differences by social/legal measures..............................................................99 4-4 Criminality regresse d on study variables.........................................................................100 4-5 Mean-level differences by RISE treatment......................................................................101
9 LIST OF FIGURES Figure page 2-1 General strain theory samples and areas of inquiry...........................................................58 3-1 Study variables............................................................................................................ .......78 4-1 Strain on criminality: Nonlinearity..................................................................................102 4-2 Strain on anger: Nonlinearity...........................................................................................103
10 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy GENERAL STRAIN AMID REST ORATION: AN EXAMINATION OF INSTRUMENTAL AND EXPRESSIVE OFFENSES By Stephen K. Rice December 2006 Chair: Alex R. Piquero Major Department: Sociology Agnewâ€™s general strain theory assumes an inte raction among three distin ct forms of social psychological strain, negative aff ect, and deviant or delinquent ad aptations. Extant research suggests that the cognitive proces ses leading to deviant or delinque nt adaptations differ markedly based on offense type, personal ch aracteristics, and indirect pro cesses such as delinquent peer exposure and injustice judgments. The present analysis extends previous work by examining general strain theory from an experimental, soci olegal perspective. More specifically, the study looks at whether standard courtroom processi ng or emotions-based reintegrative shaming conferences moderate strain, negative affect, and prospective criminality for instrumental (juvenile property offending with personal victim; juvenile shoplifting) and expressive (violent) offenses. The impact of a variety of conditi oners to include social support, self-efficacy, individual coping strategies, and external attribution is explore d, as are implications for future research and public policy. Findings indicate that individual forms of stra in differ considerably in their impact on inwardand outward-oriented affect (depression a nd anger, respectively) a nd in their relationship to instrumental and expressive offenses. Finding s also indicate important trends related to the cumulative effect of strains, of strain nonlinearity, a nd of the overarching importance of injustice
11 judgments. Finally, the present analyses lend suppor t to many, but not all, of the core precepts of restorative justice theory. In short, despite th e criminal justice systemâ€™s best intentions, many offenders are being treated in a manner that most predisposes them to future criminality.
12 CHAPTER 1 INTRODUCTION Finding its theoretical lineage in strain models by Merton (1938), Cohen (1955), and Cloward and Ohlin (1960), Agnewâ€™s (1992) ge neral strain theory (GST) emphasizes how negative relations create pressure toward crime and delinquency. Defined as relationships in which an individual is not treated as he or she would like to be treated (Agnew, 1992), strain is thought to increase negative emotions (anger primar ily, but also fear, guilt, or depression) and to increase the need for corrective action, with crime being one such response. While past tests of GST have focused on juvenile populations (e.g., Agnew and White, 1992; Baron, 2004; Mazerolle and Maahs, 2000; Piquero and Seal ock, 2004), Agnew (1995) argues that the theory can be applied to both juvenile and adult popul ations (also see deprived communities, Agnew 1999). Whereas life stressors can number in th e many hundreds (Agnew, 2001; Cohen, Kessler, and Gordon, 1995), GST is organized around three scen arios in which individuals are likely to feel they are being treated badly, namely, when ot hers (1) prevent, or th reaten to prevent, an individual from achieving positively valued goals; (2) remove, or threaten to remove, positively valued stimuli; or (3) present, or threaten to present, negativ ely valued stimuli (Agnew, 1992). As such, GST extends traditional strain theoryâ€™s focus on the disjunction between aspirations and expectations (Kornhauser, 1978). Agnew advances a range of factors that c ondition the effect of strain on delinquency-factors which influence whether an individual adapts to strain us ing delinquent or nondelinquent means. The original explication of the theory (Agnew, 1992) points to cognitive, behavioral, and emotional adaptations to strain conditioned by various personal and social resources to include deviant attitudes, deviant peers, and external attributi ons. The theory also draws attention to the
13 potential impact of magnitude, r ecency, duration, and clus tering of strainful events insomuch as stress research suggests that adverse events ar e more influential under these conditions (Thoits, 1983). Expressed succinctly, strain theory is the only major theory to principally argue that crime results from the negative affect caused by ne gative relations with others (Agnew, 1992). â€œStrainâ€ within GST can be thought of as a distal cause of crim e; it is through the processes of mediation and conditioning that crime ultimately gets explained (Pique ro and Sealock, 2004). Negative affect is presented as mediating the re lationship between strain and crime, while other variables condition strainâ€™s effect on crime.1 As what might be considered a response to critics (e.g., see Jensen (1995:152) regarding the theoryâ€™s undue breadth a nd virtual â€œunfalsifiabilityâ€), Agnew (2001, 2006a) recently completed narrative reviews of GSTâ€™s empirical record and attempted to â€œtightenâ€ any unnecessary amorphism by specifying th e types of strain that appear to be most associated with crime. In sum, . . . crime is related to verbal and physica l assaults, including assaults by parents, spouses/partners, teachers, and probably peers. Crime is also related to parental rejection, poor school performance, and work problems, including work in the secondary labor market. Crime is not related to the expected failure to achieve educational/occupational success or to unpopularity with peers. Be yond that, the relationship between various strains and crime is unclear. (Agnew, 2001:325) The 2001/2006 elaborations also provide guidance on the way forward. Outlining gaps in the literature and lessons lear ned, Agnew recommends that the next wave of research on GST eliminate cumulative measures of strain (i.e., co mbining several unrelated, and partly empirically unsupported, strains into single scal es), use GST-specific data res ources, explore the utility of 1 For example, see Mazerolle and Maahsâ€™ (2000) assessmen t of delinquent peer exposure , weak moral constraints, and behavioral propensity as cond itioners for strain on crime.
14 strain and its mediators in new contexts, and be more explicit about the characteristics and conditions that facilitate crime. To the last poi nt, research specifies that strains appear to be more likely to lead to crime when they are seen as unjust, high in magnit ude, associated with low social control, and create pressure or incentive to engage in criminal coping (Agnew, 2001:351). The present study contributes to the literature in several im portant ways. It employs secondary analysis of data from the Canberra Reintegrative Shaming Experiments (RISE) (19951999), a randomized controlled experiment that compares the effectiveness of standard courtroom processing with restorative justice gr oup conferences for four offense types: drunk driving, juvenile property offendi ng with personal victims, youth violent crimes, and juvenile shoplifting offenses (Sherman, Braithwaite, Stra ng, and Barnes, 2001). The restorative justiceinspired RISE study aims to test four primar y hypotheses: (1) that there will be less repeat offending after assignment to conference than after assignment to court; (2) that victims will be more satisfied with conferences than with cour t; (3) that both offenders and victims will report higher perceptions of procedural fairness in conferen ces than in court; and (4) that public costs of a conference will be lower, or at a minimum no greater, than traditional courtroom processing (Harris, 2001; Sherman et al., 2001). Toward this end, three primary data collection methods were utilized by the RISE research team: administrative police reports with arrest/offense data, observations from an offenderâ€™s respective treatment (court or conference), and detailed offender interviews following the court or conference proceeding. Interview items measure a variety of matters related to life stressors, court/conference stressors, stigma tization, social support, procedur al fairness, and delinquency. The present study focuses on the detailed offender surveys. In doing so, the present study builds on GSTâ€™s empirical record in several wa ys: (1) no known tests have examined general
15 strain from a sociolegal perspective, (2) no known tests have utilized experimental data, (3) there has not generally been as rich a set of GST-sp ecific strains and conditi oners available for study (e.g., that dealing with four distinct categories of strain, or the impact of injustice judgments, peer deviance, social support, self effi cacy, self esteem, external attribution, and cognitive/behavioral coping conditioners), (4) te sts have not traditionally utilized active offenders (although see Baron, 2004; Baron and Hartnagel, 2002; Baron, Kennedy, Forde, 2001; Piquero and Sealock, 2000, 2004), (5) tests have not generally examined international contexts (although see Bao et al., 2004; Mesch, Fishman, a nd Eisikovitz, 2003; Pinhey and Okinaka, 2004), and (6) most tests have failed to utilize su bjective measures of strain--or those types of strains stemming from events or cond itions that are dis liked by the people experiencing them (although see Agnew and White, 1992). A restorative justice context is particularly apt because it is thought to alte r the likelihood that sanctions will be viewed as unjust, viewed as high in magnitude, associated with low social control, or create pressure or incentive to engage in criminal coping--charact eristics of strain that are most thought to lead to crime (Agnew , 2001:352). Furthermore, the RISE studyâ€™s experimental design allows for the effect of treatment to be disentangled from confounding effects (Weisburd, Lum, and Petrosino, 2001). Agnew (2006a:198) calls for the use of experimental data in tests of the theory, particularly with regard to the detection of conditioners/moderators.
16 CHAPTER 2 REVIEW OF THE LITERATURE If the social sciences are to advance, researchers must adequately address the interplay between rationality and emotionality in a new paradigm of justice, one that casts emotion as linchpin for criminal justice practice and uses â€œr eason for emotion, rather than emotion against alleged reason . . . to make justice more rationa l about its effects on the emotional causes and prevention of crimeâ€ (Sherman, 2003:2) (also se e Massey, 2002). By di recting attention to predictors and intervening mechan isms absent or underemphasized in competing theories, socialpsychological strain theory provides a useful le ns through which to examine emotionality and criminality. General Strain: Constraint or Motivation for Crime? As Agnew (1995) points out, although strain theory shares characteristic s with traditions to include social control theory--that of crime resulting when indivi duals cannot get what they want or need, for example--it adds the differenti ating measure of emotionality by arguing that categories of goal blockage, removal of positive stimuli, or presentation of negative stimuli pressure a person into deviance (p. 114). Asser tions by critics of a propositional overlap between strain and social control focus on an assumption that people e xperiencing strain commit crime because they are freed from controls; in other words, rational actors become free to pursue their own self-inter est (Bernard, 1995). In essence, the debate resides in the determin ation of which families of theory align with factors that entice or pull individuals into crime on the one hand, and with those factors that pressure or push individuals into crime on the other. One such categorization places social control (Hirschi, 1969; Sampson and Laub, 1993), se lf-control (Gottfreds on and Hirschi, 1990), routine activities (Felson, 1998), rational choice (Cornish and Clar ke, 1986), and social learning
17 (Akers, 1985) in the former. Here, individuals may be pulled into crime because of factors to include the reinforcement of such behavior, e xposure to successful criminal models, the acquisition of beliefs favorable to crime, or a desire for immediate gratification. The pull/push dichotomy places strain in the la tter; as has been discussed, this pressure toward crime centers on negative treatment by ot hers, the experience of a range of negative emotions (e.g., anger, fear, depression), and cr ime as a corrective res ponse to such pressure (Agnew, 2005:25-26). Here, individuals are unable to disentangle themselves from relationships and situations in which they ar e not treated as they wish (Vold, Bernard, and Snipes, 2002:324). Of these varied affective states, anger is though t to be the most prominent mediator of the processes leading from strain to crime because anger increases felt injury and the need for adaptive, delinquent, responses (Agnew, 1992:60). Such responses are more likely when strained individualsâ€™ nondelinquent coping strategies are taxed, chronic strain lowers oneâ€™s threshold for adversity, or chronic strain increase s the likelihood of negative arousal at any given time (Agnew, 1992; Averill, 1982; Bandura, 1983). As counterpoint to criminological traditions that may lack sufficiently explicit ties to an intellectual heritage,1 contemporary advancements in strain theory are framed as building on the influential strain theories of Robert K. Merton (1938), Albert C ohen (1955), and Richard Cloward and Lloyd Ohlin (1960). As Agnew (1992, 1995) characterizes the progression, strain theories have become broader, have examined more types of strain, have begun to fold in research from medical sociology /psychology (e.g., stress; frustratio n-aggression; so cial learning) 1 For example, as Beirne (1997) character izes Beccariaâ€™s core protestations in Dei delitti e delle pene (1764): â€œ(I)n what was more than simply a casual failure to document his so urces, Beccariaâ€™s text is vi rtually bereft of intellectual bearings, which, had he openly pointed to them, would have allowed his readers . . . the privilege of observing more clearly the precise mast to which he chose to nail his colorsâ€ (Beirne, 1997:786).
18 and social psychology (e.g., equity; justice), and hope to explain cr ime in all social classes--a marked contrast to earlier traditions. Sequence of Topics To provide adequate context for the studyâ€™s use of a large-scale restor ative justice program as a platform for testing general strain theory (GST) (Agnew, 1992), the review of literature outlines the theoretical basis, and major resear ch questions, of GST while also providing an overview of restorative justice theory (Braithw aite, 1989). Charting the over half-century of scholarship that forms the basis for GST, the section begins with a discussion of Mertonâ€™s foundational work on anomie/strain, scholarship that has been characterized as â€œthe single most influential formulation in the sociology of de vianceâ€ (Clinard, 1964:10) and the â€œpurest example of a strain theoryâ€ (Hirschi, 1969:3), to be followed by an outline of the subsequent work by Cohen (1955) and Cloward and Ohlin (1960) that attempted to better understand what drives strained individuals toward deviant adaptations. The review of literature then moves on to an overview of Agnewâ€™s 1985 extension of â€œclassicâ€ strain, the 1992 formulation of GST, a nd recent empirical tests and elaborations (see Agnew, 2001, 2006a). Prior to a presentation of methodology, the review of literature then segues into an overview of restor ative justice theory and of the Canberra Reintegrative Shaming Experiments (RISE) (Sherman et al., 2001). The â€œContinuityâ€ of Strain Theory Strain theories propose that socially genera ted pressures drive people to commit crime-that if one treats another badly, s/he may get mad and engage in crime as a result (Agnew, 1995). This apparent parsimony, an important criteri on for sound theory (Akers and Sellers, 2004) but that has not escaped its own sh are of criticism (e.g., Hirschi, 1969; Jensen, 1995), belies over
19 sixty years of scholarship that is in many ways a continuity tracking Merton (1938) to Cohen (1955) to Cloward and Ohlin (1960) (Passas, 1995). Although differing in detail, all invoke a disp arity between widely shared ambitions and anticipated or real failure in realizing those ambitions as a source of high rates of crime in disadvantaged groups (Jensen, 1995). Each argues that everyone in the United States, poor as well as rich, aspires to achieve some manne r of success (monetary to Merton (1938); middleclass status to Cohen (1955)), but that significant segments of the population are prevented from doing so through legal channels such as a qua lity education and good job (Agnew, 2006a). Much of the research tradition fi nds its historical r oots in a culture-drive n â€œappetiteâ€ (i.e., the acquisition of wealth) failing to work in tandem with structuredriven opportunity to produce strain (Merton, 1938). Merton and Strain in American Society Merton (1938) posits that the socially induced disjunction be tween culturally valued goals and the means available for their attainment facilitates an anomic breakdown, or state of deregulation, leading to high rates of de viation. As opposed to Durkheimâ€™s (1951) conceptualization of anomie as an abnorma l, pathological phenomenon to be combated, Mertonâ€™s social-structural strain theory presumes that the United Statesâ€™ market economy of the 1930â€™s, a utilitarian culture that placed great emphasis on succe ss and failure, was characterized by a built-in disposition toward moral normlessn ess amid the â€œAmerican dreamâ€ ideal (Passas, 1995). Patterns of immigration during th e rise of the urban city al so provided a backdrop for a social explanation of egoism amid constricted opportunities: It appears that this highly visible tower of opportunity attracts substantial numbers of migrants who, rightly or no, believe that they have the talents that can best be developed
20 and rewarded within the ample confines of the city. But though many are called, comparatively few can be chosen. (Merton, 1968: 223) Such is Mertonâ€™s focus on the disjunction between goals and means and the centrality of money and pecuniary success in the value system of th e United States (Passas, 1995). So, to Merton, the link between strain and delinqu ency is instrumental, or utilita rian, as delinquency becomes a way of obtaining what one wants but cannot get though legitimate channels (Agnew, 1980). Put simply, â€œin the American Dream there is no fi nal stopping point . . . at each income level Americans want about twenty-five percent more (but of course this â€˜just a bit moreâ€™ continues to operate once it is obtained)â€ (Merton, 1968:190). In light of the present stu dyâ€™s focus targeting GST within a sociolegal framework, a detailed discussion on the disentanglement of Me rtonian â€œanomieâ€ from Mertonian â€œstrainâ€ has been set aside. The issue is far from clos ed, however. Some have argued that criminology should more clearly distinguish between two theo ries in the Mertonian paradigm, one focusing on a micro-level theory of deviant motivation (s train) and the other a macro-level theory of social disorganization on th e distribution of deviance (anom ie) (Messner, 1988). Similarly, Bernard (1987) argues that Mertonâ€™s paradigm contains a (conceptually distinct) â€œcultural argument that focuses on the value of monetary success and the importance of using legitimate means in pursuing that value, and a structural argument that focuses on the distribu tion of the legitimate opportunities in the so cietyâ€ (Bernard 1987:267). In terestingly, Merton himself acknowledged that his early work had not ade quately explained the link between anomie and deviant behavior, an impetus for his work on opportunity structures (Merton, 1995). Recent research has assessed whether st rain and anomie perspectives make contradictory predictions about the effects of money on deli nquent involvement. Strain, on the one hand, argues that monetary resources attenuate potential pressures in the face of goal blockage, while
21 anomie asserts that possession of money facilita tes ever-greater degrees of delinquency (Wright, Cullen, Agnew, and Brezina, 2001). Individuals facing such pressures, particularly those in the lower classes who are at a particular disadvantage, are t hought to have a number of ways to respond to the stressful situation of being unable to achieve monetary su ccess through legitimat e channels (Agnew, Cullen, Burton, Evans, and Dunaway, 1996). Merton (1938) described these possible adaptations as innovation, conf ormity, ritualism, retreat ism, and rebellion. Innovators To Merton (1968), most crime that takes place in society manifests through a process of innovation. Innovators remain allegiant to moneta ry culture goals but find they cannot succeed through education, a legitimate career, and so fo rth. Although Merton focuses primarily on the lower classes due to inadequate socialization in that stratum, innovation may take form in both new variants of white-collar crime and in more tr aditional illegal operations such as prostitution, drug dealing, or robbery. Despite the foundational wo rkâ€™s relative lack of a ttention to the precise factors that determine whether one reacts to blocked goals by i nnovating, Merton (1968) predicted that those who are le ss likely to internalize dominan t societal norms (e.g., the lower class) are more likely to violate such norms in the pursuit of goals (Agnew, 1985). Conformists Conformists generally accept the culture goals (e.g., monetary success) and institutionalized means (e.g., education; legitimat e career) and strive to achieve wealth through approved methods of middle-class values. In stable societies, most persons will choose conformity; because such behavior is aligned wi th the basic values of a society, Merton argued that conformity was critical in allowing one to â€œspeak of a human aggregate as comprising a societyâ€ (Merton, 1968).
22 Ritualists Mertonâ€™s third adaptatio n, ritualism, might best be depict ed by Melvilleâ€™s â€œBartleby The Scrivenerâ€ (1853), a man who would â€œprefer not to â€ face life on lifeâ€™s terms. Although ritualists reject the goal of material success, they play it safe by concentrating on reta ining what little is to be gained by adhering zealously to norms such as honesty and hard work (Akers and Sellers, 2004). Retreatists Retreatists, in contrast, drop out--they withdraw from society by rejecting both the cultural goals of material success and the conventiona l means of achieving this success, often through excessive substance use as psyc hotics, autists, pariahs, out casts, and vagabonds (Merton, 1968). Rebels Mertonâ€™s final adaptation, rebe llion, is marked by individuals who replace societal values with new ones, values which can be political (e.g ., socialist renewal), spir itual (e.g., higher states of consciousness), or can take other forms. In short, while conformists strive, innovators cheat, ritualists keep playing with no hope of winning, retreatists quit playing, an d rebels attempt to get a diffe rent game going (Vold et al., 2002:140). It is important to not e that Mertonâ€™s adaptations are not static pers onality traits; instead, they are better viewed as a choice of behaviors in response to stra in. Individuals may develop patterns of behavior th at draw on several adaptations simultaneously, as in innovation, retreatism, and rebellion being brought to be ar by a drug-abusing, professional criminal espousing militant, separatist phi losophies (Vold et al., 2002). As an additional point of illustration, studentsâ€™ reactions to strains--pro cesses which came to inform Agnewâ€™s revised strain theories (1985, 1992)--may run the gamut. Consider a graduate student who experiences significant difficulty succeeding in a course of st udy: he/she may cheat (innovation), remain in
23 the program with no hope of fini shing (ritualism), quit (retreatism), or establish an entirely different set of values separating him/herself from the expectations associated with the academic â€œethosâ€ (rebellion). Using this example, it b ecomes clear why Merton ( 1968) most directly tied innovation (primarily), retreatism, and rebellion with delinquency. Cohenâ€™s Delinquent Boys As a point of contrast to Mertonâ€™s (1968) claim that the link between strain and delinquency is primarily utilitarian, Albert Cohe n (1955) argued that the â€œfactsâ€ of delinquency suggest otherwise: that most de linquency is nonutilitarian, malici ous, and negativistic (i.e., most delinquents do not hurt or steal fo r money, but do so for the â€œhell of itâ€) and that delinquency finds its foundation not in indivi dual-level, Mertonian processes but as a byproduct of lowerand working-class gang formation (Agnew, 1980). Furthermore, Cohen (1955) asserts that schools provide the testing ground for criminogenesis--that children of lower social classes are unprepared fo r the behavioral and educational expectations (â€œmeas uring rodsâ€) that dominate th e middle-class culture of the educational system (1955:88-91). As a result of the strain, some youth undergo a â€œreaction formationâ€ and commit malicious, negativistic crim es that reflect a reje ction of middle-class standards. As Cohen explains, The working-class boy, particular ly if his training and valu es be those we have here defined as working-class, is more likely than his middle-class peers to find himself at the bottom of the status hierarchy whenever he m oves in a middle-class world, whether it be of adults or of children. To the degree to which he values middle-class status, either because he values the good opinion of middle-class persons or because he has to some degree internalized middle-class standards himself, he faces a problem of adju stment and is in the market for a solution. (1955:119) To Cohen, one such solution is th e formation of a separate, deli nquent subculture in which the working-class boy can compete succ essfully. The delinquent subcu lture provides an alternative
24 avenue to status, in essence: one that explicitly repudiates th e middle-class standards that had placed the child at such a disa dvantage in the first place. Where Merton sees strain stemming from unach ieved monetary goals, Cohen sees strain very differently. Rather than strained individu als using illegitimate channels to achieve their goals, Cohen essentially argues that (to use Mertonâ€™s term) delinquents utilize rebellious adaptations in response to their inability to achieve middle-class status through either legitimate or illegitimate means; in other words, they constr uct a new set of goals an d means altogether. In sum, through their life experience, these strained juveniles learn that middle-class status is not something you can steal (Agnew, 1980), Cloward and Ohlinâ€™s Delinquency and Opportunity But what of those lower-class juveniles who app ear to largely resist both a utilitarian and nonutilitarian criminal calculus, or what contemporary research might refer to as a low-level adolescence-limited offender or abstainer (Mo ffitt, 2003)? As Cloward and Ohlin (1960) questioned, why do some strained youth engage in delinquent adap tations and/or join delinquent subcultures, and why do some youth choose a st able â€œcorner-boy responseâ€? The corner-boy response may involve non-conforming behaviors su ch as truancy, but generally does not include violent acts that indicate a re pudiation of middle-class standa rds (Cloward and Ohlin, 1960:184). In an attempt to answer this question, Cloward and Ohlin (196 0) developed an extension to strain theory focusing on the relationship between community dynamics and strain. More specifically, they argued that C ohen was only partially correct: that some delinquency is in fact based on a rejection of middle-class values, but that these delinquents tend to engage in less serious offenses. The more serious offenders, on the other hand, are more utilitarian in their outlook: they desire money more than status, an d the desire to spend money is conspicuous in nature. In short, these utilitarians desire â€œfas t cars, fancy clothes, and swell damesâ€ (Cloward
25 and Ohlin, 1960:97). Rather than wishing to live by middle-class st andards, Cloward and Ohlinâ€™s extension to strain theo ry contends that juveniles wish to achieve status on their own, lower-class, terms. Importance of community dynamics Cloward and Ohlin align with Merton by arguing that most low-class youth are blocked (or anticipate being blocked) from achieving goals of monetary success, but that their extreme, conspicuous emphasis on acquiring and spending money is uniquely lower-class in nature. Perhaps most importantly, the community dynamics an anticipated goal blockage is nested within is critical to understanding whether delinquenc y will emanate. For this reason, Cloward and Ohlin pick up on not only strain/anomie from Merton and Cohen, but also traditions to include social disorganization and di fferential association. In essence, Cloward and Ohlin assert that delinquent adaptations are best explained by oneâ€™s social location in legitimate and illegit imate opportunity structures--that â€œjust because legitimate opportunities are blocked does not necessarily mean that illegitimate opportunities are freely available . . . . Just as there is unequal access to role models and opportunities to fulfill conforming roles, there is unequal access to il legitimate roles and oppor tunitiesâ€ (Akers and Sellers, 2004:168). In light of th is distinction, Cloward and Ohlin (1960) assert that delinquent subcultures are necessary to transl ate strain into crime because soci al support acts as a form of enabler. Conflict subculture If an individualâ€™s neighborhood is highly diso rganized, a â€œconflictâ€ subculture will form that is violence-oriented (and ha rkens back to Cohenâ€™s model for a delinquent subculture). Due to the social disorganization of the neighborhood, few illegal opportunities are available to offset the lack of legal opportunities. Me ntors are few, either conventiona l or deviant. As opposed to
26 earning status through â€œconnectionsâ€ or technical sk ills (or even, for that matter, mere physical strength), status is earned by oneâ€™s willingness to risk injury or death (Cloward and Ohlin, 1960). Criminal subculture If an individualâ€™s neighborhood is more integr ated, a â€œcriminalâ€ subcu lture forms in which older criminals mentor younger ones in obtaining money and power through offenses such as theft and extortion. As such, the criminal subcultu re is â€œcomprised of pecuniary illicit acts that lead to success as conventionally defined (and) arise in organized slums consisting of stable organization between older and younger criminal s, and between criminals and conventional elements (the fence, fix, and bail bondsman)â€ (Matsueda, Drakulich, and Kubrin, 2006). This subculture corresponds with Mert onâ€™s adaptation of innovation. Retreatist subculture Related to oneâ€™s degree of une qual access (Akers and Sellers , 2004), individuals who have failed in both legitimate and illegitimate worlds (and have thereby reli nquished both goals and means, legitimate and illegitimat e) will join â€œretreatistâ€ subc ultures focused primarily on the consumption of illicit substances. These individu als are in effect â€œdouble failuresâ€ who have prospects in neither pro-social (good education, good job) nor crimin al activities. Earning status or admiration in a retreatist subculture is ex tremely limited, and can be gained by few activities other than getting high and maintaini ng a drug habit (Vold et al., 2002). In sum, Merton, Cohen, and Cloward and Ohlin provide related, yet di verging, pictures of the role of strain on delinque ncy. Merton (1968) emphasizes a universal emphasis on monetary success, tempered by special disadvantages in the lo wer classes that lead to strain and delinquent or nondelinquent adaptations. Cohen (1955) emph asizes reaction formations as youth reject middle class â€œmeasuring rods.â€ Finally, Clow ard and Ohlin (1960) ar gue that delinquent
27 adaptations are best explained by oneâ€™s social location in legiti mate and illegitimate opportunity structures. Evolution of Classic Strain Theory After many years of prominence through the 1960â€™s, strain theory rece ived strong criticism during the 1970â€™s as research began to suggest a weak relationship betw een the disjunction of aspirations and expectations for success and the theoryâ€™s inability to explain the middle-class delinquency that had been revealed by self-report data (Agnew, 1995). Ot her criticisms focused on the apparent inability of stra in theory to explain variability in delinquency, such as why many delinquents go for long periods without comm itting delinquent acts (Agnew, 1980), why many strained individuals respond in a conventional, rather than devi ant, fashion (Colvin, 2000), and how family-related variables such as marital ad justment and parental discipline relate to delinquency (Agnew, 1980). Some went so far as to call for abandonmen t of the theory altogether (Hirschi, 1969; Kornhauser, 1978) for these reasons and for a pur portedly unclear linkage between strain and social control. Bernard (1995: 83) detailed â€œconventional wisdomâ€ in criminology in the 1970â€™s regarding Merton in particular: She (Kornhauser, 1978:162-65) rejected Mertonâ€™ s argument that American culture values the goal of economic success more than the in stitutionalized means of honesty and hard work. On the one hand, she interpreted th e desire for economic success in Hobbesian terms, as a natural desire not requiring cultura l support. On the othe r hand, she argued that culture itself should be defined in terms of a design for a moral orde r that is passed as a valued heritage from generation to generation. Defined this way, economic success cannot be a component of any culture. The appropriate interpretation is that American culture, with its values of hard work and honesty, is extremely weak and is overwhel med by the Hobbesian desires for economic success. This then becomes a â€˜controlâ€™ formulation, with deviance itself requiring no positive explanation other than the weakne ss of controls. (Bernard, 1995:82-83)
28 Agnewâ€™s Revised Strain Theory Introduced partly as a thesis (Agnew, 1980) a nd then more extensively five years later (Agnew, 1985), Agnew responded to the criticism of strain/anomie theory with a patently microlevel, social psychological strain theory that attempted to address the aspirations/expectations conundrum2 and other criticisms by essentially refram ing the issue: rather than operationalizing strain as a blockage of goal-seeking like his predecessors, the revi sed theory argues that the blockage of pain avoidance more closely resonates with adoles centsâ€™ life realities as they are often in highly aversive situations from which they have no legal means of escape. Such a life can feel like a closed circle: th ere is little autonomy from the e xpectations of family and school, and this lack of power over oneâ€™s environment can produce high levels of frustration, thereby leading to illegal escape attempts and anger-bas ed delinquency. Adults, on the other hand, have legal routes of escape from such environments to include divorce, quittin g oneâ€™s job, or deciding when and if to move to another locale (Agnew, 1985:156). Other than through its generalized use of ne gative affect as mediator for strain on delinquency (a proposition, and body of literature , underutilized in earl ier iterations of anomie/strain), Agnewâ€™s revised theory (1985) wa s far better positioned to explain high rates of criminality among adolescents compared to adults , the delinquency perpetrated by more affluent individuals, why a period of time sometime s elapses between delinquent events within individuals (not easily explained, as â€œstrainâ€ had largely been held constant by Merton (1968), for example), and the impact of family-related vari ables such as parental discipline and marital adjustment (Agnew, 1980:11). 2 Empirical tests rarely find support for the prediction that delinquency is hi ghest when aspirations for success are high and expectations are low (Agnew, 1985; Kornhauser, 1978). To the contrary, evidence suggests that delinquency is highest when both aspirations and expectations are low, and delinquency is lowest when both aspirations and expectations are high (Hirschi, 1969; Liska, 1971).
29 In a test of the revised theory using th e â€œYouth in Transitionâ€ national sample of adolescent boys, Agnew (1985:161) found that, as expected, location in aversive family and school environments had a direct effect on deli nquency and an indirect effect through negative affect. His results also held after social contro l and subcultural variables were controlled. Taken as a whole, the revised strain theory of de linquency ushered in a new avenue of focus for criminology and more detailed specifications for th e types of strain most related to delinquency. It is to this that the presen t study next draws attention. General Strain Theory Somewhat similar to the intellectual flouris h in the late 1960â€™s and early 1970â€™s (Akers, 1973; Hirschi, 1969; Lemert, 1967; Quinne y 1970), the dynamic years surrounding 1990 provided criminology with a social-psychologi cal â€œethosâ€ that advanced a multitude of theoretical perspectives to in clude shaming and restoration (Braithwaite, 1989), self-control (Gottfredson and Hirschi, 1990), procedural ju stice (Tyler, 1990), developmental taxonomies (Moffitt, 1993), and defiance (Sherman, 1993). Th e â€œethosâ€ also influenced criminal justice research in areas to include police response to spousal assault (Paternoster, Brame, and Bachman, 1997) and choice models in corpor ate crime (Paternoster and Simpson, 1996). Such was the case in advancements related to strain. Leveraging work in medical sociology and psychology (e.g., st ress; frustration-aggression) and social psychology (e.g., equity; justice), Agnewâ€™s genera l strain theory of crime a nd delinquency (1992:48) specified three major sources of strain, explained how GST differed from longstanding traditions in differential association/social learning and social control, provided guidelines for the measurement of strain, and listed the fact ors impacting the choice of nondelinquent and delinquent adaptations.
30 Foundational Principles and Types of Strain As touched on earlier, general strain theo ry (GST) emphasizes how negative relations create pressure toward crime a nd delinquency. Defined as events or conditions that are disliked by individuals, strain is thought to increase negative emotions (anger, most importantly, but also fear, guilt, or depression) and to increase the need for correct ive action, with crime being one such response (Agnew, 2006a). Although strains number in the hundreds (Agnew, 2001), GST is organized around the proposition that individuals are like ly to feel they are being trea ted badly when they (1) fail to get something they want, (2) lose something they value, or (3) receive something bad (Agnew, 2006a). A more formalized conceptualization reads as goal blockage , removal of positively valued stimuli, and presentation of negatively valued stimuli, respectively (Agnew, 1992). Life conditions tying to the first condition may includ e having little money, autonomy, or status; the second condition, a friend or family member dyi ng, or the dissolution of a valued romantic relationship; the third condition, se xual or physical abuse by fam ily or hurtful insults by peers (Agnew, 2006a). Perhaps not surprisingly given its incorporati on of multiple threads of scholarship from multiple disciplines, GST has been criticized on the grounds of a lack of attention to strain caused by non-social factors such as illness (Agnew, 1992:75), a lack of consistency in the meaning attached to key measures (Agnew, 2001:320), a failure to fully recognize the implications of strain for social control a nd the social learning of delinquency (Agnew, 2001:348), and an unwieldy breadth.3 3 â€œIf strain can be defined in so many different ways, then strain theory is virtually unfalsifiable. There is always a new measure that might salvage the theoryâ€ (Jensen, 1995:152).
31 Mediators and Conditioners As discussed earlier, â€œstrainâ€ can be thought of as a distal cause of crime; it is through the processes of mediation and condi tioning that crime is explaine d (Piquero and Sealock, 2004). Negative affect such as anger, fear, or depres sion mediate the relations hip between strain and crime, while other variables (e.g., personal and social resources to in clude deviant peers) condition strainâ€™s effect on crime. The medi ation process can be depicted as (strain negative affect crime) while the conditioning process (with regard to criminal coping) as ((strain * coping) crime) (Piquero and Sealock, 2004:127). A lthough Agnew frames much of the theory through the lens of mediators and conditioners, one could just as easily view the conditioning processes (e.g., social control, injustice judgments) as moderators. An independent variable (in this case, strain) causes the mediator, which then causes an outcome (i.e., delinquent or nondelinquent adaptation). Moderators are associated with statistical interactions, or changes in the direction and/or stre ngth of the relations hip (Baron and Kenny, 1986). General strain theory highlights anger as it s primary mediator. Drawing on the emotions and justice literature coupling unjust treatment with anger (e.g ., Mikula, 1993; Tedeschi and Felson, 1994) and GST-specific data coupling an ger with crime (e.g., Brezina, 1998; Mazerolle and Piquero, 1998; Piquero and Seal ock, 2000), anger is thought to: . . . foster crime because it disrupts cognitive processes in ways that impede noncriminal coping; for example, it leads individuals to di sregard information that may help resolve the situation, and it reduces the abil ity to clearly express grievanc es. Anger also reduces the actual and perceived costs of crime; for exampl e, angry individuals are less likely to feel guilt for their criminal behavior because they believe that the injustice they suffered justifies crime. Finally, anger energizes the individual for action, creates a sense of power or control, and creates a desire for revenge or retr ibution--all of which lead individuals to view crime in a more favorable light. (Agnew, 2001:327)
32 Predictably, as opposed to negativ e affective states such as je alousy, depression, or fear, most empirical work on GST has highlighted anger as th e mediator to be test ed for its relationship with both strain and individualsâ€™ corrective actions (although see Aseltine, Gore, and Gordon, 2000; Bao, Haas, and Pi, 2004; Sharp, Ter ling-Watt, Atkins, and Gilliam, 2001). Efforts have also been made to better understand subcategories of this affective state. For example, Mazerolle and Piquero (1998) advance th e idea that there are two primary dimensions of anger: dispositional (trait) anger and situationa l (reactive) anger, and that situational anger is in many ways more tightly aligned with GST because it refers to anger stemming from a particular strain. The research ers do assert, however, that the tw o dimensions are not mutually exclusive: that those with a greater dispositi on toward anger may be more likely to exhibit situational anger while exposed to strain (also Mazerolle, Piquero, and Capowich, 2003). Characteristics of Strain That are Most Likely to Cause Crime General strain theory specifies a number of characteristics of strain that are thought to increase the likelihood for criminal coping. Fo r example, GSTâ€™s recent elaboration (Agnew, 2001, 2006a) draws attention to strains that are pe rceived to be unjust, high in magnitude, are associated with low social control, or create pressure or incentive to engage in criminal coping. The following section provides an overview of th ese characteristics, amo ng others, that have been outlined in GSTâ€™s theoretical and empirical work. Strain as unjust As a modification to the theoryâ€™s original pl acement of injustice j udgments under the strain category â€œfailure to achieve positively valued goals â€ (i.e., strain as a disjunction between just/fair outcomes and actual outcomes), GST specifies that unjust treatment is not a distinct type of strain but can instead be used to classify any type of st rain. In other words, any strain seen as
33 unjust should be more likely to lead to crime b ecause it is more likely to activate emotions conducive to crime such as anger (Agnew, 2001:327). Strain as high in magnitude Strain that is perceived to be high in magnitude is more likel y to be associated with crime in that it negatively influences oneâ€™s ability to accurately assess the costs of noncriminal versus criminal coping, it makes it more difficult to minimi ze the impact of high levels of strain, and it negatively impacts oneâ€™s ability to behaviorally cope in a nondelinquent nature (e.g., working harder on schoolwork in the face of very severe performance deficits/poor grades). In addition, strain that is perceived to be high in magnitude is often associated with depression--which taxes oneâ€™s overall coping ability itself--and may also generate ever-larger degrees of anger, thereby increasing the likelihood of criminal c oping per the theory (Agnew, 2001:332). Strain associated with low social control As one of several indicators that GST is in formed by psychological, social-psychological, and sociological perspectives, stra ins are also predicted to lead to crime when they are caused by or associated with low social control. Drawing on longstand ing scholarship by Hirschi (1969) and others that argues that soci al control is ineffective when an individualâ€™s bond to conventional society is weak, GST argues that low social control in the forms of (for ex ample) erratic parental discipline (low direct co ntrol), homelessness (low direct co ntrol, attachment, and commitment), parental rejection (low attachment), or work in the secondary labor market (low commitment) cause crime through an attenuation of the percei ved cost of crime and ability to cope in a noncriminal manner (Agnew, 2001:335). Conversely, a high degree of soci al control/social support is thought to enable people to cope more successfully with stressful events, it may trigger certain stress-related behaviors that function proactively to eliminat e stressful events before they
34 occur, and/or it may prevent a stressful event fr om infringing on other as pects of a personâ€™s life (Taylor and Aspinwall, 1996).4 Interestingly, there may also be a reciprocal effect between social support and stress. Those who respond to stress with high degrees of vocalized distress may drive away potential members of a support structure, while too those who respond to stress in an exceedingly passive manner may also fail to garner support (M att and Dean, 1993; Smith and Wallston, 1992).5 Pressure/incentive to engage in criminal coping Agnew draws on scholarship from social learni ng (Akers, 1985) primarily, but also routine activities (Felson, 1998), to argue that the type of strain that one experiences impacts the availability and appeal of noncriminal and criminal coping. That is, that certain types of strain are associated with exposure to individuals who effectively mode l crime, reinforce crime, and present beliefs favorable to crime (e.g., as in th e case of child abuse or bullying) (see Akers, 1985 for social learning assumptions). Furthermore, Agnew (2001:337) asserts that th e causal mechanism is strain-specific by considering Andersonâ€™s (1999) inner city Philadelphia. In that context, the presentation of negative stimuli takes the form of disrespectfu l treatment: young males learn on an almost daily basis via models, reinforcers, and the presentation of beliefs to respond to such treatment with 4 Relatedly, the psychosocial stress literature suggests that social support may moderate stress both directly and indirectly. The direct effects hypothesis specifies that social support has a generalized effect: it is beneficial in both stressful and nonstressful periods. The indirect (â€œbufferingâ€ ) hypothesis maintains that the benefits of social support are most evident during periods of high stress (Cohen and Hoberman, 1983). More specifically (and of particular import to Agnewâ€™s theses (2001)), di rect effects have been found in re search that assesses the numbers of individuals one identifies as friends or the number of organizations to which one belongs; when assessed qualitatively regarding to the degree to which an individual feels that there are those in the environment who can help him if he needs it, buffering effects have been found (Taylor and Aspinwall, 1996). 5 Whether there exists an easily measurable, and empirically supportable, â€œmiddle groundâ€ on vocalization of distress and social support remains an important question for GST; we may find that ((strain * social support) crime) (diagram based on Piquero and Sealock, 2004) is informed by within-person and without-person processes not typically assessed in past tests of GST.
35 violence6. Efforts to cognitively reinterpret oneâ€™s strain or to cope in a noncriminal manner are ineffective due to ever-increasing levels of a buse by perpetrators and regular reminders from members of the community that such abuse requires â€œactionâ€ (Anderson, 1999; Agnew, 2001). Taken within the context of the broade r discussion, Andersonâ€™ s ethnography (also see Bernard, 1990) vividly illustrates th e other characteristics of strain that are most likely to lead to criminal adaptations as well: that of strain (here: overt, severe disrespect) being seen as violating justice rules, being perceived to be high in ma gnitude (in degree, duration, frequency, recency, and centrality) (see below), and in being associ ated with low social control (e.g., as in an inadequate police presence). Magnitude and dur ation/frequency Features of strains to incl ude their duration, frequency, recency, and centrality (i.e., relationship to oneâ€™s values/ac tivities/goals/needs) ar e thought to influence an individualâ€™s perception of magnitude (Agnew, 2001). With rega rd to duration and freq uency, chronic strains and/or those of high fre quency are thought to be more deleteri ous than those that are short-lived and/or of low frequency. This prediction is consistent with re search that models a nonlinear relationship between chronic/cumulative stress a nd psychological distress (Wheaton, Roszell, and Hall, 1997).7 Magnitude and recency Recent strains are thought to be more injuri ous than older ones. As with duration and frequency, recency is an important concept in attempts to disaggregate the effects of acute stressors from chronic strains. There, â€œlife even ts checklistsâ€ are oftent imes designed to not only 6 Also see Bandura (1983:128): â€œPeople are most frequently rewarded for following the behavior of models who . . . possess certain social and technical competencies, command so cial power, and who, by their adroitness, occupy high positions in various social hierarchies.â€ 7 But see Jang and Johnson (2003), who argue that chronicity may have an inverse, â€œdullingâ€ effect.
36 estimate the effect of chronic strain on psycholog ical distress separately from the effect of acute stressors, but also to estimate the timing of discrete stressors to gauge the effect of more recent stressors from those that occurred earlier (Avison and Turner, 1988). Magnitude and centrality Centrality, on the other hand, refers to the rela tionship between a strain and the identities, activities, needs, values, or goals of an individual (Agnew, 2001). GST predicts that strains that are highly â€œcentralâ€ to an individual are more likely to be associated with negative affective states and the need for correctiv e action. Consider, for example, the case of three sisters who experience an identical objective strain: the deat h of a parent. Resear ch indicates that the subjective evaluation of that stra in can differ markedly between such sisters--that factors to include its centrality (a byproduct of the perceived role of the parent in oneâ€™s life vis--vis needs, activities, and identities), personal and social resources, and individua l traits impact the perceived magnitude of the death. Cumulative impact of characteristics The four characteristics of strain outline d earlier (re: unjust/high magnitude/social control/social learning ) are â€œroughly equal in importance an d . . . the absence of any one characteristic substantially reduces the likeli hood that strains will result in crime--although strains with only some of thes e characteristics may still incr ease the likelih ood of crimeâ€ (Agnew, 2006a:70). At present, th e empirical record is quite limited on the cumulative impact of these characteristics. Types of Strain That Are Most Likely to Cause Crime As a complement to the characteristics of strain that are thought to be most likely to lead to crime, the GST elaborations (Agnew, 2001, 2006a) al so review the types of strain that have received the strongest degrees of direct empirical support or, in the event of limited tests, are
37 most likely to be conditioned by injustice judgmen ts, perceptions of high magnitude, low social control, or pressure or incentive to engage in criminal coping: . . . certain types of strain are related to cr ime whereas others are not. At this point, it seems safe to conclude that crime is relate d to verbal and physical assaults, including assaults by parents, spouses/partners, teachers, and probably peers. Crime is also related to parental rejection, poor school performance, and work problems, including work in the secondary labor market. Crime is not rela ted to the expected failure to achieve educational/occupational success or to unpopul arity with peers. Beyond that, the relationship between various strains and crime is unclear. (Agnew, 2001:325) Agnew does not intend for the list of strains to be considered closed, or altogether definitive. Some positive evidence may even be consider ed â€œad hocâ€ (Agnew, 2001:325), as in predictors that are less driven by GST itself than by compleme ntary research that holds strong promise; for example, in predictors related to oneâ€™s status , or in the blockage of autonomy goals (Anderson, 1999; Moffitt, 1993). As â€œempirical research is starti ng to point to those type s of strains that are and are not related to crime, it is not wise to de pend on such research to fully resolve the issue. There are hundreds of specific types of strain; it will take empirical resear chers a long while to determine their relative importan ce . . . . â€ (Agnew, 2001: 326). GST Samples and Areas of Inquiry This is not to say that empiricists have faile d to better specify GST. There have been over thirty direct tests of individual-level GST to-d ate not to include, for example, complementary studies that examine macro-level strain (Agne w, 1999), delinquent problem solving (Brezina, 2000), life trajectories on stra in and delinquent adaptations (Hoffmann and Cerbone, 1999), the influence of money on delinquent involvement (Wright, Cullen, Agnew and Brezina, 2001), and the impact of opportunity structures (Hoffma nn and Ireland, 2004). Tests have been conducted with a diverse array of samples, methodologies , and operationalizations and is dynamic with regard to specification (e.g., Agnew, 2001).
38 As outlined in Figure 2-1, the array of studies includes national-level samples of adolescents (Agnew, Brezina, Wright and Cullen, 2002; Brezin a, 1999; Jang and Johnson, 2003; Mazerolle, 1998; Paternoster and Mazerolle, 1994), state, city, and community-level samples of adolescents (Agnew and White, 1992; Aseltine , Gore and Gordon, 2000; Bao, Haas and Pi, 2004; Baron, 2004; Hay, 2003; Piquero and S ealock, 2000, 2004), university samples (Broidy, 2001; Mazerolle et al., 2003; Sharp et al., 2001), and adult community samples (Burton, Cullen, Evans and Dunaway, 1994; Eitle, 2002). Researchers have also explored th e theoryâ€™s ability to account for demographic differences in crime w ith regard to gender (Broidy and Agnew, 1997; Eitle, 2002; Piquero and Sealock, 2004; Slocum, Simpson, and Smith, 2005) and race (Eitle and Turner, 2003; Jang and Johnson, 2003; Kaufman, Rebellon, Thaxton, and Agnew, 2005) and have begun to test the theory among youth in di fferent cultures and economic systems (Bao et al., 2004; Mesch, Fishman, and Eisikovitz, 2003; Pinhey and Okinaka, 2004). Tests Related to the Strains Most Likely to Cause Crime GST is a propositionally dense, multifaceted theore tical effort. Rather than continue to focus on research questions and samples th at have received little support since 1992,8 future specifications should continue to build on what work s. To this end, an overview of evidence for several of the most promisi ng types of strain follows. Failure to achieve goals related to thrills, autonomy, status, acquisition of money The pursuit of the goals of th rills/excitement, high autonom y, masculine status, and the immediate acquisition of money relates more to indi vidual traits such as th rill seeking, exposure 8 Agnew places the following strains in this category: (1) failure to achieve goals such as educational success, occupational success, and middle-class status; (2) supervision/discipline by authority figures (e.g., parents, teachers) that is consistent, not excessive, and not abusive; (3) burdens associated with the care of conventional others; (4) strong demands associated with conventional pursuits that provide external rewards or intrinsic satisfaction; (5) unpopularity with peers and isolation from unsupervised peer activities (note: Agnew explains that the lack of empirical support for these strains is for typical , non-extreme cases) (Agnew, 2006a: 75-78).
39 to subcultural groups, or structural conditions (e.g., poverty) than it does to conventional socialization or social control (Agnew, 2001:343) . The pressure underlying each of these types of strain is more easily alleviat ed by crime than what one finds with occupational or educational strain.9 Further, despite the relative paucity of research in crim inology focusing on economic goals as compared to educational or occupational ones, evidence suggests that crime (as an adaptation) is frequently used to acquire money. Farnwort h and Leiber (1989), for ex ample, found that the failure to achieve monetary goals was a more significant predictor of crime than were educational expectations. Relatedly, Cer nkovich, Giordano, and Rudol ph (2000:131) attempted to assess the impact of economic goals (the â€œAmerican dreamâ€) on delinquent behavior. Their results provided a level of support for the pred iction of economic strain on income-generating offense involvement. Moreover, in a test of GST with homeless street youth, Baron (2004) found that monetary dissatisfac tion was related to the studyâ€™s measure of total crime and of property offending. Another concept strongly asso ciated with monetary dissatisfaction, that of relative deprivation (i.e., whether an individual feels that s/he is worse off monetarily than his/her comparative reference group) (Passas, 1995) , was found to be related to property, violent, and total crime. In general, assessments of mo netary dissatisfaction may be particularly useful indicators of strain because they tap into â€œthe reality of the momentâ€ (Agnew et al., 1996). Criminal victimization A burgeoning body of literature suggests that cr iminal victimization is strongly related to criminal offending. In addition to non-strain-s pecific research which provides a degree of 9 As discussed earlier, see Anderson (1999) regarding stat us frustration; also Moffitt (1993) regarding desire for autonomy and Katz (1988) regarding thrills / excitement. Katz (1988) argues that individualsâ€™ experiences and immediate social environments facilitate the construction of crime as sensually compelling, while Moffittâ€™s (1993) taxonomy charts adolescence-limited delinquency as a byproduct of the interaction between early maturity and the autonomy aspects of peer activities.
40 support (e.g., Lauritsen, Sampson and Laub, 1991; Sampson and Lauritse n, 1993), direct tests suggest that victimization is a particularly acute type of stra in that holds up under scrutiny. Victimization is typically perceived as high in ma gnitude, perceived as unjust, is likely to occur in environments typified by low social control, and because it is more common in delinquent peer groups, it likely activates mechanisms inhe rent to the social l earning of crime (Agnew, 2001). In Baronâ€™s (2004) study of ho meless street youth, property victimization and violent victimization (i.e., personal attack by another) were related to e ngaging in all measures of crime; robbery victimization (force used, or threatened to be used, against the individual to get their money or other possessions) was re lated to engaging in violent o ffending. In Eitle and Turnerâ€™s (2002:217) study of risk and prot ective factors associated wi th young adult substance use and mental health, the strain characteristics of victimization were embellished; instead of only operationalizing â€œvictimizationâ€ via personal experi ence as victim, the researchers broadened the concept to include witnessing community and do mestic violence and hearing about the violent victimization of significant others . That is, their study included th e possible strain of â€œvicariousâ€ victimization. Findings indicated that proximal exposure to violen ce in the community, a history of receiving traumatic news, and di rect victimizations in the community increased the risk for young adult criminal offending (Eitle and Turn er, 2002:231-2). Relatedl y, Agnewâ€™s (2002:629) examination of a national sample of adolesce nt boys indicated that experienced physical victimization and certain types of anticipated and vicar ious victimization were associated with delinquency. He cautions, however, that additiona l research is required to better understand the victimization/offender relationship and the possible impact of pers onality traits, coping skills, physical size, and associati on with delinquent peers.
41 Homelessness Agnewâ€™s review (2001) also identif ies a particularly intense form of experienced strain that represents â€œa major challenge to a broad range of (an individualâ€™s) goals, needs, values, activities, and identiti esâ€ (p. 345): homelessness itself. In Baronâ€™s (2004) study, homelessness was found to be related to the studyâ€™s measure of total crime, while an earlier study (Baron and Hartnagel, 1997) of homeless youths found familial and school factors to have minimal influence on criminal behavior; instead criminal behavior was found to be more influenced by immediate factors to include homelessness, lengthy unemploym ent, lack of income, and criminal peers. Relatedly, Hagan and McCarthy (1992, 1997) found life on the street to be an important intervening variable of social control and strain on crime. Discrimination based on ascribed characteristics Despite the centrality of matters of race, ethnicity, gender, and discrimination within the social sciences, until recently scant attention has been paid to these subpopulations vis--vis intra-group processes and the im pact of stress on offending (C ernkovich and Giordano, 1992). Utilizing a sample related to ri sk and protective factors associ ated with young adult substance use and mental health, Eitle and Turner (2003) ex amined whether racial differences in criminal involvement were explained by grou p differences in exposure to st rains, group differences in the strength of coping mechanisms, a combination of the two, or group differences in the strength of relationship between stress exposure and crime. Their findings reveal ed that racial differences in criminal involvement were explained by differences in exposure to strain. In other words, race was found to be a marker for increased risk of stress exposure. In related work utilizing a sample of African American children, Simons, Chen, Stewart, and Brody (2003) set out to assess an associa tion between exposure to discrimination and delinquent behavior. For African American boys, they found that an association between
42 discrimination and delinquency was mediated by feelings of anger and depression and by the belief that aggression is a necessary interpersonal tactic. For African Amer ican girls, anger and depression were found to mediate part of th e effect of discrimi nation on delinquency, but discrimination continued to disp lay a small, direct effect. And finally, in a test featuri ng the same population as Eitle and Turner (2003) but split by gender, Eitle (2002) explored whether discrimi nation experiences were related to criminal activity and/or substance use disorders. Discri mination was found to be a significant predictor of crime and substance use disorders. Child abuse and neglect Extant GST scholarship (also see Colvin, 2000) suggest that physical, emotional, and sexual abuse of children are related to stress and crime and are of particular salience when they have been experienced recently (Agnew, 2006a). As in the case of homelessness, child abuse and neglect threatens a childâ€™s id entities, needs, activities, and goa ls and is oftentimes perceived as unjust. Long-term impacts of abuse and ne glect include posttraumatic effects, cognitive distortions, altered emotional ity (e.g., depression, anxiety) , dissociation (e.g., detachment, numbing), and disturbed relatedne ss (e.g., altered sexuality, aggre ssion) (Briere, 1992). Such treatment undermines attachments and commitment s (thereby resulting in low social control), oftentimes leads children to model the behavior of their parents (Fag an and Wexler, 1987), and may lead children to associate with stigmatized others, including deviant peers. Patterson (1975), for example, illustrates that parent-child interaction is a predictor of deviant behavior later in life via social learning mechanisms. There may also be a temporal matter at hand. In a longitudinal analysis of the varying impact of childhood and adolescent maltreatme nt on multiple problem outcomes, Thornberry, Ireland, and Smith (2001:957) found that adolesce nt and persistent maltreatment had stronger
43 and more consistent negative consequences during adolescence than did maltreatment experienced only in childhood. Individuals who experience chronic strains develop negative emotionality because chronicity reduces a person â€™s ability to cope in a legal manner (Agnew, 2006a). Using interview data with youths who had b een identified as chemically abusing or dependent and had been adjudicated for a vari ety of offenses (e.g., major property felonies, misdemeanors, felonious assaults), Piquero and Sealock (2000) examined the amount of physical and emotional abuse directed toward youth in their household and the am ount of physical abuse directed toward others in the household (i.e ., vicarious strain) (Agnew, 2002). Analyses indicated that exposure to abus e was predictive of property crim e and interpersonal aggression. Further, in building on their earlier work, Pique ro and Sealock (2004) found that males within their sample were more likely to cope with emo tional and physical abuse with crime. Relatedly, Baron (2004) found emotional, se xual, and physical abuse to be associated with criminal involvement among his sample of homeless street youth. Overall, the many types of possible maltreatment within a family is a fruitful area for research in that emotionally abused children ofte ntimes come to perceive the world as a hostile place (Nicholas and Bieber, 1996). As such, GS Tâ€™s inclusion of a host of mediators and conditioners allows for detailed specification of what might constitute a â€œhostile place.â€ For example, does such a determination center on perceptions of injustice, a compounding of perceptions related to strain magnitude, low social control, social learni ng-inspired incentives to engage in criminal coping, or some other mechanism altogether? Other empirically supported strains As strains number in the hundreds (Agnew, 200 1) and the types of strains most strongly related to crime number in the dozens, only a sa mple of empirically s upported predictors has
44 been outlined above. In addition to these st rains, evidence also provides support for the following: (1) supervision/discip line that is erratic, excessive , and/or harsh, (2) negative secondary school experiences, (3) ab usive peer relations, (4) work in the secondary labor market, (5) unemployment â€“ especially when it is persiste nt and blamed on others, (6) marital problems, (7) residence in economically disadvantaged co mmunities, and (8) parental rejection (Agnew, 2006a:71-74). Although child abuse is separated from parental rejection in the types of strain most likely to lead to crime (Agnew, 2006a), the empirical re cord suggests that rejection and abuse may be different sides of the same coin. In one su ch study, Weatherburn a nd Lind (1997:7) found that parental neglect accounted for most of the explaine d variation in juvenile participation in crime when included in a model with poverty, single parent household, and crowded dwelling. Similar results were found when the measure for abuse replaced that of neglect, but the researchers cautioned that this was likely due to the high correlation between abuse and neglect. The possible confluence raises important questions ab out overarching strain categories. While child abuse no doubt falls into the cate gory of the presentation of a nega tive stimulus, what of parental neglect? It would appe ar that the perceptual component is paramount: would a parent who fails to express love or affection for a child be better framed as nega tive stimulus or the removal of a positive stimulus? The theoryâ€™s original explic ation anticipated this issue, as the major categories of strain were presente d as â€œideal typesâ€ (Agnew, 1992:50). While many of the categories above pick up on general strain theory-specific scholarship (e.g., Agnew and Brezina, 1997 on relational probl ems with peers; Agnew, 1999 on community differences in crime rates), a great many ar e underdeveloped or rely on criminology-wide research such as Sampson and Laub (1993) on pare ntal rejection, excessive discipline, negative
45 secondary school experiences, and marital problems. In short, despite the approximately fifteen years that have elapsed since GS Tâ€™s introduction, a great many questi ons remain as to the strains that are likely to lead to crime and perhaps mo re centrally, the relationship between the strains, mediators, conditioners, and outcome variables. Other key, and yet to be unanswered, questions include how general strain theory â€œoperatesâ€ in sociolegal and experimental contexts--questions that the present study works to address. Moving forward, prior to a presentation of study methodology the literature review segues into an overview of restorative justice th eory and the Canberra Reintegrative Shaming Experiments (RISE) (Sherman et al., 2001). Restorative Justice Theory As one of the largest and most methodologically sophisticated restorative justice efforts to date, the Canberra Reintegrative Shaming Experi ments (RISE) project provides a singularly unique, practical application of the theory. At its core, restorative justice â€œis about the notion that because crime (or any other kind of injust ice) hurts, justice should heal. This is an alternative to the view that justice must be punitive--responding to hurt with hurt that is the wrongdoerâ€™s just deserts. So rest orative justice is about hurt bege tting healing as an alternative to hurt begetting hurtâ€ (Braithwaite and Braith waite, 2001:4-5). Standi ng as a complement to efforts which also â€œinvertâ€ longstanding emphases on punitiveness (e.g., defiance theory) (Sherman, 1993), restorative justice theory esta blishes its voice partic ularly well in â€œmicro arenasâ€ such as criminal justice by emphasizing th e restoration of harm by offenders to victims as well as reintegration of the offender back into the community (Braithwaite and Braithwaite, 2001; Kurki, 2000).
46 Relationship to Retributism Rather than viewing â€œjusticeâ€ retributively--that of classic, utilitarian standards of punishment and deterrence, or as crime being prim arily a violation of the law and act against the state--restorative justice presumes that crime is, in no small measure, also a violation of one person by another. According to the theory, crime causes harm and suffering to both victims, offenders, and the larger community. Retributio n increases suffering and alienates parties from one another and from the community, healing and reparation fail to occur, while social capital and informal social controls lose strengt h (Christie, 1977; Tosouni, 2004; Zehr, 1990). Restorative approaches, to the contrary, fo cus on problem solving, obligations, dialogue, negotiation, restitution and reconciliation (Zehr, 1990). Restorat ive justice interventions directly involve offenders, victims and members of the co mmunity in the problem solving process, and offenders are held accountable for repairing the vi ctimâ€™s loss (physical and/or emotional) without being stigmatized as a criminal (Kurki, 2000; Tosouni, 2004). Reintegrative Shaming Instead of labeling the person as deviant (s ee Lemert, 1967), reintegr ative shaming focuses on labeling the harmful (criminal) act as evil, treating the offender with respect, while at the same time criticizing his or her behavior (Braith waite, 1989). In short, the crime is rejected while the person retains a level of respect. As opposed to passive responsibility underpinning the retributive paradigm, reintegrative shaming emphasizes active responsibility--a wholesale shift in orientation that is thought to change an o ffenderâ€™s attitude toward the law and facilitate conformity through emotional motivation and conf ormity (Kurki, 2000). The more traditional criminal justice process, that of stigmatization, is thought to increase reoffe nding through its abject disrespect and humiliation of the individual; s/he becomes cast as pariah instead of recipient of just and right punishment (B raithwaite, 1989; also see Sherman, 1993).
47 It is important to note, however, that ther e are voices who reject a strict â€œoppositional contrastâ€ between retributive and restorative just ice. Daly, for example, argues that there are parallels between retributive, re habilitative, and restorative jus tice that should be recognized by scholars: Specifically, restorative justice practices do fo cus on the offence and the offender; they are concerned with censuring past behaviour a nd with changing future behaviour; they are concerned with sanctions or outcomes that ar e proportionate and that also "make things right" in individual cases. Restorative justi ce practices assume mentally competent and hence morally culpable actors, who are expected to take responsibility for their actions, not only to the parties directly injured, but pe rhaps also to a wider community. (Daly, 2001) Nonetheless, although retributive principles ar e still dominant in publ ic opinion, there is evidence of a burgeoning interest in punishment that is less vindi ctive and more concerned with strengthening social bonds and empowering fa milies (Umbreit, 1994). In some ways public sentiments are punitive and progressive at once, wishing the correctional system to do justice, protect the public, and reform the wayw ard (Cullen, Fisher, and Applegate, 2000). A number of justice programs have been desi gned with restorative justice/reintegrative shaming principles at their core. As outlined by Bazemore and Umbreit (2001) and the Office of Juvenile Justice and Delinquency (1998), such programs include victim-offender mediation, peacemaking circles, restorative community servi ce, financial restitution and personal services, victim impact panels, victim empathy classes, and family group conferencing. As family group conferencing is the program design that ties mo st closely to the Canberra Reintegrative Shaming Experiment (RISE), it is to this that attention is drawn. In family group conferences (FGC), offenders face victims in the presence of family and friends (of both offender and victim), discuss their misconduct, and attempt to restore the material and emotional harm the misconduct cause d to society and to the victim (Morris and Maxwell, 1998). FGCâ€™s rationale focuses on the inherent strength and integrity of the family
48 unit, the importance of community support, the o pportunity for parents to be intimately involved in misconduct by their children, and the fundamental ri ght for victims of crime to have a role in the resolution of offenses (Hudson, Galaway, Morris, and Maxwell, 1996; Tosouni, 2004). Braithwaite and Mugford (1994) describe conferencing as a reintegration ceremony where emotional disapproval is communicat ed within a continuum of respect for the offender. They suggest that the confrontation wi th the victim contribu tes to the shaming, wh ile the participation of caring supporters contributes to the reintegration. On balance, such forums are thought to be applicable to minor as well as serious crimes and to both adult and youthful offenders (Cunneen, 2002). In practice, however, most programs cen ter on youthful non-violent offenders (Kurki, 2000). Consistent with the detection, interpretation, and analysis of program effects in the social sciences (see Rossi, Lipsey, and Freeman, 2004), pa st research has explored specific elements within these programs that, comp ared to longstanding methods root ed in retributive justice, appear to be associated with crime reduction and greater satisfaction by the parties involved. Considerable scholarship suggests that crime re duction and satisfaction ar e inextricably linked-that strategies of regulation based on the way in which members of the public are treated by legal authorities can enhance their willingness to cooper ate with and defer to those legal authorities (Tyler, 1990). Analogous with research in procedural just ice (Tyler, 1990), for example, most program evaluations have tended to focus on a whole range of positive outcomes in addition to reoffending to include participant satisfaction, per ceptions of fairness and justice, participation rates, and agreement completion rates (Morri s and Maxwell, 1998). Levels of perceived satisfaction and procedural fair ness are almost universally hi gh regardless of participant
49 characteristics (see Tosouni, 2004 for discussion), a nd these perceptions tend to be higher than what one finds in traditional courtroom processi ng (Latimer, Dowden, and Muise, 2001). With regard to participation, while o ffender attendance has been found to be high in most initiatives, victim participation typically ranges between forty and sixty percent (Bonta, Wallace-Capretta, and Rooney, 1998; Maxwell and Morris, 1996).10 In general, high participation rates do not necessarily correlate with part icipation by the offenderâ€™s family ; instead, they more commonly relate to being provided an opportunity to have a voice in the process (Robertson, 1996).11 Regarding agreement completion, agreements are co mpleted in the vast majority of restorative cases and tend to exceed rates of completion in traditional programs (Latimer et al., 2001). With regard to recidivism, meta-analytic tech niques have been used as a counterpart to comparative, narrative, assessments. Reviewing f ourteen restorative justi ce evaluations (most of which were not randomized, nor were many easily comparable), Bonta et al. (1998) found a small reduction (8%) in reoffending among participants in restorative justi ce interventions. In a separate analysis of more methodologically ri gorous programs, Latimer et al. (2001) found a small, positive effect size (+.07) (see Tosouni, 2004). Canberra Reintegrative Shaming Experiments (RISE) A large experimental study carried out in Canberra, Australia between 1995 and 1999, The Reintegrative Shaming Experiments (RISE) projec t (Sherman et al., 2001) sought to test core tenets of restorative justice theory primarily, wh ile also including research questions which tap into related efforts that â€œinvertâ€ longstanding em phases on punitiveness such as defiance theory 10 Analyses suggest that a â€œfear of offe nderâ€ or â€œsympathy for juvenileâ€ effect may be at play as victims are more likely to want to participate when the offender is a juvenile (Umbreit and Bradshaw, 1997). 11 Other dynamics may also be at play, as in research that suggests that when victims have no supporters with them, or when the offendersâ€™ supporters outnumber their own, the likelihood of victim revictimization (i.e., feeling upset, perceiving that s/he is being disrespected) is significantly higher (Daly, 2001).
50 (Sherman, 1993) and procedural ju stice (Tyler, 1990). Reflective of the spread of family group conferencing in many other parts of the world (e.g., New Zealand, USA, Britain, South Africa, Canada), the Canberra RISE project is comprised of four separate experi ments that compare the effectiveness of standard cour troom processes and restorative justice conferences for four offense types: drunk driving (over .08 blood alcohol content), juvenile property offending with personal victims, juvenile shoplifting offe nses, and youth violent crimes (Harris, 2001).12 Study hypotheses The study aimed to test four primary hypot heses (Sherman et al., 2001). The first hypothesis, drawing on Braithwait eâ€™s (1989) assertion that form al court justice stigmatizes offenders and makes it difficult for them to lead fully productive lives, predicted less repeat offending after assignment to conference than after assignment to court. Also consistent with relevant research (e.g., Bazemore and Umbreit, 2001; Strang, 2004; Tyler, 2003), the second and third hypotheses predicted that vi ctims who participated in confer ences would be more satisfied than court participants, and that offenders and victims would report higher perceptions of procedural fairness in conferences than in court. The fourth a nd final hypothesis predicted that public costs of a conference woul d be lower, or at a minimum no greater, than traditional courtroom processing. Data collection methods Three methods were used: (1) Administrative police reports of the offenderâ€™s attitude, the police station and officer that referred the case, blood alcohol content (drunk driving only), 12 It is important to note that while there are four major categories for which RISE organizes its reports and findings, there is more than one offense type within the property an d violence offending categories. Offenses in the property experiment include â€œtheft, burglary, car theft, shoplifting in owner-operated shops (not big stores), criminal damage, fraud, vehicle break-in and receiving or possession of stolen goods,â€ while offenses in the violence category include â€œmostly assault occasioning actual bodily harm, common assault, fighting and arsonâ€ (Sherman, Strang, Angel, Woods, Barnes, et al., 2006).
51 offense type, RISE assigned treatment, and re-offending data. (2) Observations from both treatments on the duration of the treatment, violent incidents, pr esence of supporters, levels of expressed reintegrative shaming, offender guilt admission, apology, defiance, and contribution to the outcome. (3) Participant interview surveys following the court or conference proceedings, with items measuring a variety of variables to include restor ation and reintegrative shaming, procedural justice, and self-reported pre-tr eatment delinquency (She rman et al., 2001). Participant interviews were sought between two a nd four weeks after final treatment of the case (Harris, 2001). Acceptance criteria Consistent with the Wagga-Wagga commun itarian model of justice (Moore and Oâ€™Connell, 1994), conferences were facilitated by police officers. Because the aim of the RISE experiments was to compare cases that were assi gned to court with equa lly serious cases that were assigned to conference, a case could be accepted into th e experiments only if it would normally be dealt with by a court. The research pr otocol also required that eligible cases not be so serious that in the estimati on of the apprehending police officer they could only be dealt with in court. Thus, the aim was to include "middlerange" offenses, neither so serious that officers would be reluctant to bypass tr aditional processing, nor so triv ial that they would normally warrant a warning (Sherman et al., 2001). Cases were randomly assigned (see below for protocol) based on police referral to RISE and eligibility criteria to include: (1) the crime was one of the target offenses, (2) the offender(s) had made a full admission of responsibility for the offense, (3) a sergeant had approved the case being sent to RISE, (4) there was no reason to believe the offender woul d object to a conference if the case was assigned to a conference, (5) the offender did not have any outstanding warrants or bonds requiring him/her to go to court, (6 ) the offender lived in the Australian Capital
52 Territory (ACT), and (7) the police officer re ferring the case agreed to accept the RISE recommendation for all co-offenders in the case. A dditional drunk driving criteria also applied. All individuals who fit the profile were eligible for referral. Eligible offenders for juvenile personal property and juvenile property securi ty offenses had to be under 18 years old and violent crime offenders had to be under 30 years old.13 Method of randomization When police officers had a matter that was RI SE eligible, they called a 24-hour mobile phone number that was in the custody of a member of the research team. Officers were then asked a series of ten screening questions to veri fy the eligibility of the case. Once case details were entered into a log, an e nvelope containing the random assi gnment (court or conference) was opened and the officer was informed of the disposition. Prior to commencement of data collection, treatments were randomly assigned us ing a sequence of quasi-random numbers. After being informed of the treatment (court or conference) by the RISE staffer, the apprehending officer processed the case accordingl y. Victim data were li mited to the juvenile personal property and youth violence experiment s. No drunk driving offenses included direct victims, so consequently there was no drunk dr iving data on victims. Files were created consisting of data from reports by police officer s, data from observations by trained RISE research staff of court and conference treatmen ts, and offender interview data from RISE staff after the court or conference proc eedings (Sherman et al., 2001). There were a number of reasons why cases assi gned to conference went to court: (a) the offender rejected conferencing when it was assigned and asserted the right to have his/her case processed in the normal way thr ough court, (b) the offender reo ffended immediately after being 13 Although the violent crime category includes offenders up to the age of twenty-nine, the RISE study refers to it as â€œyouth violent crime.â€
53 assigned to conference and assignm ent was altered to court, (c) the offender withdrew his/her full admission of the offense, requiring a dete rmination of guilt by a court, (d) the offender persistently failed to show up for the conference, or did so in an intoxicated state, (e) the offender's attitude or behavior was such that the officer decided to send the case to court, (f) because of an administrative error on the part of the police, the case was sent to court when it had been assigned to conference, (g) the conferen ce failed to reach an outcome accepted by all parties and the facilitator referred the matter to court (She rman et al., 2001). Ethical considerations Consistent with guidelines for successful interventions, all conferences were voluntary in nature in that participants could choose to withdr aw and have their case hear d in court. Informed consent was gained from participan ts at several points in the pro cess to include consent from the offender immediately after randomization, from a ll those in attendance (e .g., offender, victim, offenderâ€™s family) prior to observation of c onference cases (not required for courtroom proceedings, as was a public venue), and from the offender prior to follow-up interview. As Harris (2001) explains, â€œRandomi zation simply involved offering some offenders the option of choosing an alternative to the nor mal intervention and thus did not require explicit consent to the randomization. This was important because it pr ovided some protection against knowledge of being randomly assigned affecting offenders â€™ behavior or perceptionsâ€ (p. 96). Response rates Eighty-six percent of drunk drivers, 66% of juvenile property offending with personal victims, 75% of juvenile shoplifting, and 68% of youth violent offenders were the subject of at least one observation. Offender interviews were reported at 85%, 76%, 76%, and 72%, respectively.
54 RISE findings The RISE research team has released a vari ety of process and outcome evaluations; early results from interviews with the first 548 offende rs comprising all four experimental groups and from observations of 597 drunk driving and 133 youth crime cases provid e a number of useful insights (Sherman and Barnes, 1997; Sherman and Strang, 1997). Conferen ces were likely to last far longer than court proceedings (confer encing lasted approximately ninety minutes as compared to ten minutes for court processing (H arris, 2001)). There were higher levels of reintegrative shaming observed during conferences as compared to court (Sherman and Strang, 1997). Contrary to prediction, there was no less--and at times, more--stigmatizing shaming during conferences than during c ourt processing. Young offenders in conferences were twice as likely to apologize and to receive forgiveness and were more emotional and remorseful than their court counterparts. There was al so a higher perceived level of pr ocedural justice in conference as compared to court (Sherman and Barnes, 1997). There was also a reported improvement in attitudes toward the justice system overall (cou rt and police). Feelings of bitterness and anger between the groups were found not to differ w ith the exception of dr unk drivers, who were significantly more likely to be angry after court (Sherman and Barnes, 1997). Interestingly, and perhaps not altogether of primary concern to those who suggest that â€œsuccessâ€ in restorative justice can, and should, be measured both directly (via reoffending) and indirectly (through measures of satisfaction and perceptions of fa irness) (Makkai and Braithwaite, 1994; Morris and Maxwell, 1998), the RISE recidivism findings report (Sherman, Strang, and Woods, 2000) offers a mixed picture. Across the four experiments that make up RISE, very different results were found to have emerged across the different offense categories. Among the youth violence participants, offenders w ho were assigned to conference subsequently offended at substantially lower leve ls than did those assigned to c ourt: thirty-eight fewer offenses
55 per year per one hundred offenders. In contrast , this trend was not found for any of the other experiments. For drunk driving offenders, for ex ample, a very small increase in detected reoffending was found for the conferenced offenders (relative to court) : approximately four offenses per year per one hundred offenders (Sherman et al., 2000). In the report of findings, the research team r ecommends more nuanced analysis to explore the reasons for the differences across offense t ype. Despite the fact that conference-assigned offenders reported their treatment to be more procedurally fair than did conference-assigned offenders, that trend did not appear to translate into higher levels of compliance with the law in three of the four offense categories.14 As opposed to combining disparate offense types together in subsequent research, the res earchers recommend that testing c ontinue to delineate different effects for different offenses as opposed to con centrating on vari ability in offender background. In brief, according to Sherman et al. (2000), â€œT he substantive conclusion of RISE is that restorative justice can work, and can even redu ce crime by violent offenders. But there is no guarantee that it will work for all offense types. Caution and more research are needed before rapid expansion of any new approach to treati ng crime. Less caution is needed, however, in testing restorative justice on more serious types of violent offences. The findings in this report provide firm ground for repeating the violence ex periment in many other venues and with more refined types of violent offenses, including r obbery, assault, and grie vous bodily harmâ€ (pp. 1819). More Recent Efforts: Jerry Lee Program It is important to note that RISE is but one effort. Over the past several years, Sherman, Strang and colleagues have been designing and imple menting a variety of tests of face-to-face 14 The final recidivism report was based on one-year before and after standardized periods of comparison.
56 restorative justice programs in the United States , England, and Australia under the auspices of the Jerrry Lee Program in Randomized Controlled Trials of Restorativ e Justice (Sherman and Strang, 2004).15 Taken together, the research projects have yielded the following conclusions: (1) there are substantial benefits for crime vic tims, as measured by interview; (2) there are substantial reductions in repeat offending in some tests, in some settings, with some demographic groups, as measured by arrest or charge da ta; (3) there ar e substantial increases in repeat offending in other tests, in some settings, with some demographic groups, as measured by arrest or charge data; (4) there are consistent a nd substantial reductions in crime victimsâ€™ stated desires for violent revenge against offenders; and (5 ) in several tests, restorative justice makes no significant difference in repe at offending across demographi c groups (Sherman and Strang, 2004). Departure from RISE / Restorative Justice Focus While there are certain commonalities between gene ral strain theory and restorative justice theory, most notably regarding the importance of social support (see Agnew, 2006:180-1), the present study departs from a re storative justice-centered asse ssment in several ways. The restorative justice approach, on the one hand, ge nerally centers on decision points such as a formal plan to repair harm (e.g., through restitution), commun ity mentoring through volunteer efforts or facilitated employment opportunities, the differentiation of shame from guilt and embarrassment, and an overarching focus of the consequences of stigmatizing versus reintegrative shaming (Braithwaite, 1989).16 In practice, however, 15 As of August, 2004, the Jerry Lee Program had almost thirty randomized, controlled trials in deliberative restorative justice completed or underway; structural dimensions of the trials include with or without courts (e.g., diversion), with or without deliberation, and with personal versus collective vi ctims (Sherman and Strang, 2004). 16 For example, Harrisâ€™ (2001) and Tosouniâ€™s (2004) tests of restorative justice theory utilize measures to include reintegration, stigmatization, shame-guilt, emba rrassment-exposure, and unresolved shame.
57 . . . the research literature on restorative jus tice has not risen to the challenge of capturing (collective emotional dynamics) in the research reports . . . . The re sult of this failing is that even the most literate of criminologists and criminal lawyers understa nd restorativ e justice in terms of material reparation to victims rath er than in terms of the symbolic reparation which all evidence to date s uggests is more important. (Bra ithwaite and Braithwaite, 2001:59) The present general strain theory approach, on the other hand, works to disentangle four distinct sources of strain and two measures of negative affect among a host of understudied GST conditioners to include cognitive avoidance, in justice judgments, self-efficacy, and external attribution. While it is alt ogether possible that restora tive justice proponents would place Agnewâ€™s (1992) categories of stra in under an umbrella term â€œs tigmatization,â€ only GST provides a fully adequate framework to map the varied sources of strain that manife st in social life (four, here) onto specific forms of negative affect. It is important to note, however, that both theories presume that there are consequences for sanctions th at are administered in ways that are disliked (Agnew, 2001:352), and both theories have been ch aracterized as â€œconsensusâ€ theories that presume that compliance with the law is an important social goal (Braithwaite, 1989:38).
58 Figure 2-1. General strain theory samples and areas of inquiry. Representative Scholarship: (1) Agnew and White (1992); Broidy (2001); M azerolle and Piquero (1998); Burton, Cullen, Evans, and Dunaway (1994). (2) Broidy and Agnew (1997); Mazerolle (1998); Piquero and Sealock (2004); Eitle (2002); Hoffmann and Su (1997); Landau (1997). (3) Jang and Johnson (2003); Eitle and Turner (2003); Landau, 1998. (4) Agnew and White (1992); Sharp, Terling-Wa tt, Atkins, and Gilliam (2001); Landau, 1997; Landau, 1998. (5) Broidy and Agnew (1997). (6) Agnew (2002); Eitle and Turner (2003). (7) Mazerolle, Piquero and Capowich (2003); As eltine, Gore and Gor don (2000); Baron (2004). (8) Agnew (1999). (9) Agnew, Brezina, Wright and Cullen (2002). Note: These areas of inquiry, and examples of scholarship, are by no means exhaustive. Other important efforts include the assessment of delinquent problem solving (Brezina, 2000), opportunity structures (Hoffma nn and Ireland (2004), and life-cour se trajectories on strain and delinquent adaptations (e.g., Agnew, 1997; Hoffmann and Cerbone, 1999). Juvenile Male Racial / ethnic majority Objective strain Experienced strain Subjective strain Situational anger Organized community Adult (1) Female (2) Racial / ethnic minority (3) Subjective strain (4) Vicarious/anticipated strain (6) Dispositional anger (7) Deprived community (8) Emot react to subjective strain (5) Positive emotionality / high constraint Negative emotionality / low constraint (9)
59 CHAPTER 3 METHODOLOGY The present study employs secondary analysis of data from the Canberra Reintegrative Shaming Experiments (RISE) (1995-1999), a randomized controlled experiment that set out to compare the effectiveness of standard courtr oom processing with restorative justice group conferences for four offense types: drunk drivi ng, juvenile property offending with personal victims, youth violent crimes, and juvenile s hoplifting offenses (Sherman et al., 2000; Sherman et al., 2001).1 2 The RISE study concentrates on repeat offending after conference or court, victim satisfaction with the overall process, offender and victim perceptions of procedural fairness, and financial cost to the public (Sherman et al., 2001; Harris, 2001). Toward this end, the research team captured administrative polic e reports, third-party observations of both treatments, and participant interviews following the court or conference with items measuring a variety of variables to include re storation and reintegrative shami ng, procedural justice, and selfreported pre-treatment delinquency (Sherman et al., 2001). Sample In order to minimize data loss across collect ion points, the presen t study is organized around cross-sectional analysis of the interview data. The RISE research team conducted the interviews approximately two to six weeks after final treatment (Harris, 2001:98). In the interviews, offenders were asked qu estions related to initial contac t with police, the experience in 1 The administrative, observational, and interview data are available through the Inter-university Consortium for Political and Social Research (ICPSR 2993). 2 Cases were randomly assigned based on police referral to RISE and eligibility criteria. Among the criteria is that the offender had made a full admission of responsibility for the offense. The research protocol also required that eligible cases not be so serious that in the estimation of th e apprehending police officer th ey could only be dealt with in court. Thus, the aim was to include only "middle-rang e" offenses, neither so seri ous that officers would be reluctant to bypass traditional processing, nor so trivial that they would normally warrant a warning (Sherman et al., 2001). All conferences were facilitated by police officers and in most cases all known offenders were in the room along with victims, victim supporters, and offender supporters.
60 court or conference, and reflecti ons and experiences in the period of time that elapsed since the treatment. This design lends support to proper te mporal order being established, despite the fact that all data were collected during the same session (also s ee Piquero and Sealock, 2000, 2004).3 The survey instrument features approxim ately two hundred individual questions with additional (branch) questions ba sed on offender response. Interviews took approximately one hour and twenty minutes to administer (Harris, 2001).4 Post-project assessment indicates that conferences took approximately ni nety minutes to complete comp ared to ten minutes for court (Sherman et al., 2001). The present analysis re lies on a subset of the full RISE sample. Despite the RISE studyâ€™s inclusion of drunk driving offende rs (N = 737 interviews), the pr esent study solely focuses on survey data from juvenile pers onal property offenders (N = 111 interviews), juvenile property (shoplifting) offenders (N = 86 interviews), and youth violent crime offenders (N = 52 interviews) (i.e., N = 249 total respondents).5 Despite the possibly deleterious effect on stat istical power by disregarding several hundred offenders, it was determined that analysis of rein tegrative shaming-centered survey data that is devoid of its â€œsymbolic heartâ€ (Sarat, 2001)--namel y, direct victims--would too seriously impact the efficacy of study variables and would likely render court/conference split-sample analysis 3 Reliance on retrospective reports about psychosocial variables should be approached with caution, however (Henry, Moffitt, Caspi, Langley, et al., 1994). 4 The instrument underwent extensive pre-testing with offenders from a range of backgrounds and offense types. All survey questions involved responding to questions that were read word-for-word and answered on preestablished scales. During some sections of the interview, the intervie wer captured all responses while in others (particularly for potentially sensitive issues) the respondent recorded answ ers in a separate booklet (Harris, 2001:98-9). 5 Interviews are based upon completion rates of 76 percent of juvenile personal property offenders, 73 percent of juvenile property security offenders, and 72 percent of youth violent crime offenders.
61 fatally flawed.6 By its very nature, reintegrative shaming wishes to advance a different model for â€œdoing justiceâ€ by reintroducing into the system those most di rectly harmed by crime (i.e., victims) (Strang, 2004:1). By designing an e xperimental treatment facilitated by police and lacking direct victim involvement, a question of â€œapples to applesâ€ co mparisons (drunk driving court versus drunk driv ing conference) may have been warranted. As discussed, offenders from the three crime types were randomly assigned to either standard courtroom processing (N = 125) or conferencing circle (N = 124). Due to the relatively few cases available in the juvenile violent crim e category, the RISE research team amended its selection criteria mid-project to include violent offenders up to the age of twenty-nine (Sherman et al., 2001). Offenders eligible for placement in to the juvenile persona l property and juvenile property security offenses were se venteen or younger. As most da ta sets capable of testing GST involve surveys of adolescents (Agnew, 1992:48), the violent crime sampleâ€™s inclusion of older offenders represents an improvement over the ag e range found in most earlier research (although see, e.g., Burton et al.â€™s (1994) Me rtonian strain assessment). Measures Strain Although life stressors can number in the many hundreds (Agnew, 2001; Cohen, Kessler, and Gordon, 1995), as discussed earlier general st rain theory (GST) is organized around three scenarios in which individuals are likely to feel they are being treated badly: when others (1) prevent, or threaten to prevent, an individual from achieving positively valued goals; (2) remove, or threaten to remove, a positively valued stimuli; or (3) present, or threaten to present, noxious 6 In the pursuit of experimental homogeneity, the RISE research team excluded drunk driving offenses involving an accident causing injury or damage. Ther efore, drunk drivin g conferences stand apart as treatments that included no direct victims. Instead, a community representative attended some of the conferences and discussed the dangers of drunk driving; in addition, the police officer would provide information about blood alcohol levels and risks (Harris, 2001).
62 or negatively valued stimuli (Agnew, 1992). View ed through the lens of general strain, the Canberra reintegrative shaming experiments (RISE) survey data provide a unique opportunity to test for strain/stressors from a â€œbehavior of lawâ€ perspective. That is, to better understand offendersâ€™ negative relations with agents of social control and to assess attenuating or aggravating effects of family/friends hip networks on such stressors. As a response to tests of GST th at oftentimes fail to include a comprehensive assortment of strain measures, particularly those related to ca tegories of goal blockage and certain types of negative treatment (Agnew, 2001:323) (although see Baron, 2004; Mazerolle and Piquero, 1998), the present study measures strain as (1) disjunction between expectations and actual achievements, (2) disjunction between aspirations and expectations, (3) removal of positive stimuli, and (4) presentation of negative stimuli. The construction of each strain measure aligns with Agnewâ€™s (2001:324) recommendation that futu re tests not employ cumulative measures of strain; such measures encompass a wide range of strains, some contributing to crime and others not, thereby weakening the cumulative measureâ€™ s overall impact on crime. Complementing other studies which have examined the impact of distinct types of stra in on crime (e.g., Agnew and White, 1992; Aseltine et al ., 2000; Baron, 2004; Paternos ter and Mazerolle, 1994), the present studyâ€™s analytic plan is designed to assess how, and under what conditions, some strains may have a strong impact on cr ime while others do not. Figure 3-1 outlines all study variables. The re lationship between the studyâ€™s four measures of strain and two measures of negative affect (anger, depression) will be assessed, as will the relationship between affect and criminality. Consistent with theory (Agnew, 1992, 2001), the potential role of negative affect as mediator between strain and criminality will also be assessed, as will the relationship between th e facilitators and constraints for delinquent coping and strain,
63 affect, and criminality. Finally, a series of legal and demographic variables will be utilized as multivariate controls and for split-sample analysis of the theoryâ€™s core proposition of strain affect criminality. Disjunction between expectatio ns and actual achievements As opposed to strains that purportedly stem fr om ideal goals rooted in an economic or cultural system (e.g., Merton, 1968; Messner and Rosenfeld, 2001) , strain as a disjunction between expectations and actual achievements is rooted in research on justice in social psychology whereby a failure to achieve personal e xpectations (derived from social comparison and/or past experience) facilitate s emotions such as anger, rese ntment, rage, and disappointment (Agnew, 2001:52). This type of strain is of partic ular interest to resear chers because there is reason to believe that experienced strain is more consequential than anticipated strain (Agnew, 2002).7 Although largely unstudied within the co ntext of jurisprudence and GST, related research suggests that in a cour troom or conferencing circle a fa ilure to gain voluntary deference to the decisions of legal author ities may center on an individualâ€™s assessment that authorities are not using fair procedures and their motives are not trustworthy (Tyler and Huo, 2002). The present studyâ€™s measure taps into the argum ent that expectations are â€œexistentially basedâ€ (Agnew, 1992:52)--that they are grounded in the manner in which one would expect to achieve a valued goal. The scale for this m easure is comprised of two items: â€œYou felt pushed around in the conference/court case by people wi th more power than youâ€ and â€œDuring the conference/court you felt pushed in to things you did not agree with.â€ Higher values in this additive scale indicate a greater degree of perceived strain. Factor analysis for the scale yields 7 For the purpose of illustration, in a multilevel assessment of community, strain, and delinquency, Wareham, Cochran, Dembo, and Sellers (2005) utilized the following survey items as measures of the disjunction: â€œmy teachers don't respect my opini ons as much as I would like; people my ag e treat me like I'm still just a kid; my parents don't respect my opinions as much as I would like.â€
64 one factor confirming the scaleâ€™s unidimensiona lity; Cronbachâ€™s alpha indicates good reliability ( = .75). Item-to-scale correlati ons can be found in Appendix A. Disjunction between aspirations and expectations Harking back to traditions that argue that strain is largely a function of a disjunction between ideal goals and expected levels of goal achievement (e.g., Cloward and Ohlin, 1960; Cohen, 1955) (see Burton et al., 1994; Paternos ter and Mazerolle, 1994) , the present studyâ€™s strain measure is comprised of the single survey item: â€œDuring the conference/court case I felt that I had stuffed up at least some of my future opportunities.â€ Higher valu es indicate a greater degree of perceived strain. Presentation of negative stimuli A category of strain still la rgely understudied in tests of GST (although see Baron, 2004; Mazerolle and Piquero, 1998; Piquero a nd Sealock, 2000, 2004), Agnewâ€™s elaboration (2001:344) indicates that a partic ular form of negative stimulus --that of supervision/discipline that is strict or excessi ve given the infraction or particularly harsh--is likely to be strongly related to crime because it is seen as unjust, high in ma gnitude, and associated with low social control. Criminal justice officials who employ such beha viors are known to increase the likelihood of crime by those on the receiving end of such nega tive treatment (Sherman, 1993; Tyler, 1990). In support of this line of inqui ry, the scale for the measure is comprised of five items: â€œWere you treated in the conference/court as though you were likely to commit another offense?â€ â€œDid people during th e conference/court case make ne gative judgments about what kind of person you are?â€ â€œDuring the conference/c ourt case were you treated as though you were a criminal?â€ â€œDuring the conference/court case did any of the people who are important to you reject you because of the offence?â€ and â€œDuring the conference/court case were you treated as though you were a bad person?â€ Higher values in th is additive scale indicate a greater degree of
65 perceived strain. Factor analysis for the sc ale yields one factor confirming the scaleâ€™s unidimensionality; Cronbachâ€™s al pha indicates good reliability ( = .74). Removal of positive stimuli The removal of a positively valued stimulus is likely to create strain because it may threaten an individualâ€™s goals, values, needs, and/or identities (Agnew, 2001). Such loss is regularly featured in inventor ies of stressful life events in psychology (Thoits, 1983) and may lead to delinquency as the individual tries to prev ent the loss, retrieve the loss, or seek revenge against those responsible (Agnew, 1992:5-6). Ex tant tests of GST have operationalized this strain in several ways to include parents taking way privileges, th e death of an immediate family member or friend, the remarriage of oneâ€™s mother/f ather, or being forced to live apart from one or both parents (Capowich, Mazerolle, and Piquero, 2001). With regard to the present study, a legal contex t is especially appropr iate because of the stakes: one is vulnerable to punishment of both a formal (loss of freedom) and informal nature (e.g., loss of respect, loss of honor) (see Toby, 1957). Conceptions of justic e have been found to tie to both a personâ€™s self-esteem (Lerner, 1987) and overarching personhood (Sampson, 1983). Relatedly, concerns for respect and honor with in a social network can elicit strong negative emotions when treatment is perceived to be arbitr ary or unjust. In literature outlining the natural â€œtensionâ€ between procedural and distributive jus tice, evidence suggests th at processual threats to prestige or status weigh more heavily on an individualâ€™s mind than do unfavorable outcomes (Lind and Tyler, 1988). The RISE survey data allows for operationalizat ion of this strain th rough a scale comprised of two items: â€œDuring th e conference/court I felt like I ha d lost respect or honour among my familyâ€ and â€œDuring the conference/court I felt like I had lost respect or honour among my
66 friends.â€ Higher values in the a dditive scale indicate a greater degr ee of perceived strain. Factor analysis yields one factor confirming unidime nsionality; Cronbachâ€™s alpha indicates good reliability ( = .68). Empirical independence of strain categories Agnew (1992:50) consciously presents the major categories of strain as ideal types, or heuristics, to ensure that a full range of strainfu l events is included in empirical research. For example, one could frame negative relations betw een an individual and his/her partner as either the presentation of negative stimu li or removal of positive stimuli (i.e ., loss of healthy relations). Nonetheless, to the extent that GSTâ€™s major cat egories of strain measure distinct concepts, one would hope to find that factor analysis yields more than one factor across examined families of strain; in other words, th ere should be empirical indepe ndence between the constructs (Mazerolle and Piquero, 1998:200).8 Toward this end, the survey items utilized for the presentation of negative stimuli scale and for th e removal of positive stimuli scale were factor analyzed. Results (Table 3-1) indicate that two di stinct factors are at play: the first five survey items appear to measure the factor for presentatio n of negative stimuli, while the last two items appear to measure the factor for removal of positive stimuli. Negative Affect Distress is an unpleasant subjec tive state that takes form in anxiety, anger, and depression (Mirowsky and Ross, 1989). GST highlights negative affect (anger, mostly, but also depression) as its key mediator in the face of objectiv e and subjective strain (Agnew, 1992/2001). The mediation process may be depicted as (strain negative emotion delinquency) (Piquero and Sealock, 2004). 8 Interestingly, the theoryâ€™s original explication asserts th at there should not be an expectation that factor analysis reproduces strainâ€™s major categories (Agnew, 1992: 50, 59).
67 Of the varied affective states, anger is thought to be the most prominent mediator of the processes leading from strain to delinquency because a nger increases felt injury and the need for adaptive, delinquent, responses (A gnew, 1992:60). It disrupts cogni tive processes in ways that impede noncriminal coping and also reduces th e actual and perceived costs of crime (Agnew, 2001:327). Predictably, most empiri cal work on GST has highlighted anger as the mediator to be tested for its relationship with both strain and individualsâ€™ correc tive actions (although see depression, jealousy, and fear: Aseltine et al ., 2000; Bao et al., 2004; Sharp et al., 2000). As opposed to measures that reflect generali zed (trait-based) emoti ons (see Brezina, 1996; Mazerolle and Piquero, 1997, 1998; Piquero and Sealock, 2000, 2004), the present study utilizes reactive, situationa lly-based measures of anger and depres sion in its analyses. Although traitbased and state-based negative affect has been found to be correlated (C apowich et al., 2001) and may not (in action) be mutually exclusive (Mazerolle et al., 2003),9 the present studyâ€™s statebased measures are consistent with Agnewâ€™s ( 1992, 2001) causal chain of delinquent adaptations resulting from the negative affect caused by ne gative relations in oneâ€™s immediate social environment. Furthermore, the following vari ables are context-specif ic, supporting the call for measures that are tailored to the experi enced strain (Mazerolle et al., 2003). The studyâ€™s measures of anger and depression are comprised of the single survey items â€œThe conference/court case just made you angryâ€ and â€œDuring the confer ence/court case I felt sad or depressed.â€ Higher valu es indicate a greater degree of perceived anger or depression. There is little associat ion between anger and de pression (r = .05) (p > .05), and the two are treated as separate measures throughout the analyses. 9 Those with a greater disposition toward anger may be more likely to exhibit situational anger while exposed to strain (Mazerolle et al., 2003)
68 In sum, the present study features four cate gories of strain on two measures of negative affect. Intervening Variables As explained earlier, general strain theory sets out to not only specify sources of strain and negative emotion, but also the c onditions under which st rain and affect increase or reduce an individualâ€™s likelihood for delinque nt coping. While affect medi ates the relationship between strain and delinquency, other va riables such as social suppor t, self-efficacy, self-esteem, delinquent peers, external attribution, and emotional/cognitive coping condition strainâ€™s effect on delinquency. Each measure is presented below in one of two categories: those conditioners that are theorized to act as constrai nts to delinquent coping, and those conditioners that are theorized to act as facilitator s to delinquent coping. Constraints to delinquent coping Conventional social support. Social support is the sense of being someone important in the eyes of others, being esteemed and valu ed as a person, and of having someone who will understand and help when needed (Mirowsky an d Ross, 1989:13). General strain theory is informed by research that suggests that high degr ees of social support enable individuals to cope more successfully with stressful events, trigger behaviors that function pr oactively to eliminate stressful events before they occur, and/or may prevent a stressful event from infringing on other aspects of a personâ€™s life (Taylo r and Aspinwall, 1996). Among c linically depre ssed individuals, for example, high levels of soci al support from family members, work associates, and close friends are associated with increases in probl em-solving coping and decreases in emotional discharge coping (Fondacaro and Mo os, 1987). The present study measures social support using a scale comprised of the following five items: â€œDid people in the conference/court case say that it was not like you to do something that was wrong? ,â€ â€œDid you learn from the conference/court
69 case that there are people who care about you?,â€ â€œDuring the c onference/court case did people suggest that they loved you rega rdless of what you did?,â€ â€œDur ing the conference/court case did people talk about aspects of yourself which they like?,â€ and â€œWere you treated as a trustworthy person in the conference/court?â€ Higher values indicate a greate r degree of social support. Factor analysis yields one factor confirming unidimensionality; Cronbachâ€™s alpha indicates good reliability ( = .65). Self-efficacy. Self-efficacy refers to a belief in oneâ€™s own capacity to execute courses of action required to manage prospective situations. As such, a high degree of self-efficacy affords individuals with personal and so cial benefits; a lack may breed apprehension, apathy, or despair (Bandura, 1997:2). Relatedly, GST argues that those who are high in self-efficacy are more likely to feel that strain can be alleviat ed through nondelinquent copi ng (Agnew, 1992:71). The study measures self-efficacy by a scale comprised of two items: â€œIf I canâ€™t do a job the first time, I keep trying until I canâ€ and â€œWhen trying to le arn something new, I soon give up if I am not initially successfulâ€ (reversed). Higher values indicate a higher level of self-efficacy. Factor analysis yields one factor confirming unidimens ionality; Cronbachâ€™s alpha indicates acceptable reliability ( = .58). Self-esteem. A high degree of self-esteem relates to resilience, or the ability to achieve good outcomes as one faces stressful life circumst ances. Similar to self-efficacy as a selfreferent construct, positive self-attitude serves as a protective function be tween stressors/strains and undesirable outcomes (Kaplan, 1996). Self-esteem is measured by a scale consisting of five items: â€œI feel Iâ€™m a person of worth, at least on an equal plane with others,â€ â€œI feel that I have a number of good qualities,â€ â€œI am able to do thin gs as well as most other people,â€ â€œI take a positive attitude towards myself,â€ and â€œOn the whole, I am satisfied with myself.â€ Higher values
70 indicate higher self-esteem. Factor analysis yields one fact or confirming unidimensionality; Cronbachâ€™s alpha indicates good reliability ( = .74). Cognitive/behavioral coping strategies. Individuals cognitively reinterpret objective strains to minimize their subjec tive adversity (Agnew, 2001). The coping strategies that have received the most attention in the psychosocial stress literature are the avoidant, or minimizing, coping style and the active, confr ontive, vigilant coping style (T aylor and Aspinwall, 1996). GST outlines a typology of major coping strategies to include cognitive, emotional, and behavioral (Agnew, 1992). Of particular relevance to the present study is the cognitive coping strategy of avoidance and the behavioral c oping strategy of a desire for re venge. Complementing Mertonâ€™s (1938) adaptations of ritualis m and retreatism, avoidance ce nters on an indi vidual strongly minimizing, or altogether ignoring, the importan ce of adversity by cognitively â€œremovingâ€ him or herself from the strainful environment (Agne w, 1992). The behavioral coping strategy of a desire for revenge is, on the ot her hand, a more confrontive copi ng style and is distinguished by minimizing strain through a desire for â€œgetti ng backâ€ at those who ar e perceived to have facilitated a strain.10 In the present study, cognitive copi ng (avoidance) is measured by the item â€œDuring the conference/cour t case I felt so exposed, I wished I c ould just disappear .â€ Behavioral coping (desire for revenge) is measured by the item â€œYou wish that you could get back at the people who were accusing you in the conference/c ourt case.â€ Higher values indicate a higher degree of utilization of that coping strategy. As Agnew (1992:69) explains, vengeful attitudes can assu me either conventional or delinquent forms, although an individualsâ€™ likeli hood for delinquency is high. In order to be 10 One area that has gone largely understu died within extant tests of GST is whether the effectiveness of avoidant (versus active) coping styles are depend ent on characteristics of strain itself (e.g., short-lived v. chronic) not to mention whether environmental resources (e.g., social support, a good job) moderate the use of one type of coping strategy over another. Research indicates that this may be the case (H olahan and Moos, 1987).
71 most consistent with the literature, the present study puts the coping strate gy of avoidance in the category of constraint to delinque nt coping, while the coping strate gy of a desire for revenge is included with facilitators of delinquent coping (see below). Facilitators of delinquent coping Strain as unjust. The inclusion of injustice judgments as a facilitator of delinquent coping reflects a modification to GST (Agnew, 1992). Originally, Agnew placed such judgments under the category of strain as disj unction between just/fair outcomes and actual outcomes. More recently, however, Agnew argues that the perception of unjust treatment is not a distinct type of strain but can instead be used to classify any type of strain (Agnew, 2001). In other words, a strain seen as unjust is more likely to lead to crime because it activates emotions conducive to crime (Agnew, 2001:327). With regard to the co ntext of RISE, a failure to gain voluntary deference to the decisions of le gal authorities oftentimes cente rs on an individualâ€™s assessment that authorities are not using fair procedures and their motives are not trustworthy (Tyler and Huo, 2002). Relatedly, the legal forum itself may ma ke a difference: research indicates that both plea-bargaining and mediation are rated fairer than trial (Tyler, 2003). The perceived violation of justice rules is measured by a scale consisting of ten items: â€œYou feel that people who have committed the same offense are treated the same way by conferences/courts,â€ â€œPeople were polite to you in the court/conferen ce,â€ â€œYou understood what was going on in the court/conference,â€ â€œYou understood what your rights were during the processing of the case,â€ â€œYou felt you had the opportunity to expre ss your views in the conference/court,â€ â€œAll sides got a fair chance to bring out the facts in the conference/court,â€ â€œYou felt you had enough control over the way things were run in the conference/court,â€ â€œThe conference/court took acc ount of what you said in deciding what should be done,â€ â€œYou feel you were treated with respect in the conference /court,â€ and â€œThe police were fair during the
72 court/conference.â€ In this scale, lower values indicate a greater perceived violation of justice rules. Factor analysis yields one factor confirming unidimensionality; Cronbachâ€™s alpha indicates good reliability ( = .86). Pressure/incentive to engage in criminal coping. GST draws on scholarship from social learning (Akers, 1985) primarily, but also routine activities (Felson, 1998), to argue that the type of strain that one experiences impacts the ava ilability and appeal of noncriminal and criminal coping, and that certain types of strain are a ssociated with exposure to individuals who effectively model crime, reinforce crime, and present beliefs favorable to crime (Agnew, 2001, 2006a) (see Akers, 1985). A particular form of stra in will also create pres sure or incentive when it receives strong subcultural support (Agnew, 2001). As such, and in light of research suggesting that differential associ ation with law-violating peers is among the strongest predictors of criminal behavior (e.g., Loeber and Stout hamer-Loeber, 1986; Warr, 2002), the present study measures this conditioner with the item â€œWhat do your close friends thi nk about what you did?â€ Higher values indicate a great degree of acceptan ce by oneâ€™s peers of the respondentâ€™s criminal offense.11 External attribution. Another factor affecting the dispos ition to delinquency is external attribution, or the tendency to at tribute oneâ€™s own adversity onto others. Such attribution is associated with greater degrees of anger, whic h GST then associates with crime (Agnew, 1992). Researchers in the areas of aggr ession and coercive action synthe size the process as a form of self-justification: as one shifts blame onto others, culpabilit y begets exoneration (Warner, 1986). Or put thusly, â€œThere is nothi ng wrong with me; there is some thing wrong with youâ€ (Novaco, 11 Research which taps into tolerance for deviance within a social network indicates that such tolerance may partially be a function of cynicism toward the criminal justice sy stem (Sampson and Bartusch, 1998). As such, there may be important reciprocal effects between violations of justice rules and pressure/incentive to engage in criminal coping.
73 1976:1125). The present study measures external attribution accordingly: â€œYou feel that the people who accused you in the conference/court case were more wrong than you were.â€ Higher values indicate a higher degr ee of external attribution.12 Outcome Measure Attitudinal proxies for reoffending form the basis for the outcome variable, a scale comprised of two items: â€œThe conference/court case will help prevent you from breaking the law in the futureâ€ and â€œWhat happened in the conference/court will encourage you to obey the law in the future.â€ Higher values in the scale indicate a l ack of intention to be law-abiding in the future. Factor analysis yields one factor, confirming unidimensionality; Cronbach â€™s alpha indicates good reliability ( = .75). Here, â€œcriminalityâ€ refers to a propensity for criminal behavior (Gottfredson and Hirschi, 1990). This reoffe nding proxy is consistent with Agnewâ€™s (1992:62) assertion that: Each type of strain may create a predisposition for delinquency or function as a situational event that instigates a pa rticular delinquent act (emphasis in original). In the words of Hirschi and Gottfredson (1986), then, the strain theory presented in this paper is a theory of both â€˜criminalityâ€™ and â€˜crimeâ€™ (or to use the words of Clarke and Cornish (1985), it is a theory of both â€˜criminal involvementâ€™ and â€˜criminal eventsâ€™). To Clarke and Cornish (1985), generalized criminal involvement and specific criminal events tie together in no small measure. Similar to Agnew (1992), Clarke and Cornish (1985:167) cast background influences (e.g., psychological, familial, sociodemographic) as orienting functions â€œ . . . exposing people to particular problems a nd particular opportunitie s and leading them to perceive and evaluate these in particular (criminal) ways.â€ 12 With this variable, it is important to remember that a criterion for inclusion in the RISE project is that the offender made a full admission of responsibility for the offense. Althou gh offenders may indeed feel that others are partially responsibility for their act, the admission of responsibility rein forces the variableâ€™s content validity. That is, that external attribution is in fact being measured, as oppose d to the vagaries of the criminal justice system and its questions of guilt and innocence.
74 Past tests of GST have utilized measures related to projected criminality/intentions to commit crime (e.g., Capowich et al., 2001; Hay, 2003; Mazerolle et al., 2003; Mazerolle and Piquero, 1997, 1998). Relatedly, pr ojected delinquency measures have been found to correlate with self-reported offending in longitudinal studies (Gr een, 1989; Tittle, 1980). Efforts were made to secure RISE recidivism data, but such data were not available for release from the research team (Lawrence W. Sherman, personal correspondence, 10/16/04) or from the Inter-university Consortium for Political and Social Research (ICPSR) (12/10/05). The RISE interviews do provide information on the of fenderâ€™s level of involvement in burglary, car theft, theft, vandalism, violence with and without weapon, and illic it drug use, but those survey items measure the offenderâ€™s activities for the twel ve months prior to the experiment (to include the offense for which the offender was apprehended) (Sherman et al., 2001). In other words, the survey that forms the basis for the present study does not measure criminal involvement in the period of time between court/c onference and interview. As the present study focuses on the consequences of strainful events in court or conference, only post-treatment criminal involvement is relevant. The majority of past tests of GST have uti lized cross-sectional data (Agnew, 2001), which raise issues of proper temporal ordering between strain, affect , and delinquency. On the other hand, tests that have attempted to â€œcorrectâ€ fo r the issue by utilizing longitudinal designs may have suffered from sizable lags between stra in and delinquency (e.g., Brezina, 1996; Mazerolle and Maahs, 2000). Lags can be problematic due to the typically proxim ate nature of strain, affect, and outcome (Agnew, 1992).13 13 One possible solution is for future studies to use diaries to capture daily life stressors, affect, and related conditioners (Broidy, 2001).
75 Sociodemographics/controls GST also predicts that personal characteristic s may affect oneâ€™s likelihood for criminal coping. As such, previous tests have examined the manner in which strain, negative affect, and conditioners operate across sociodemographics. Th e present study includes the social and legal variables sex (female/male), age group (non-juvenile/juve nile), minority status (nonminority/Aboriginal-Torres Strait Islander), crim e type (instrumental/expressive), and assigned treatment (court/conference) (each coded 0/1). Descriptive Statistics Descriptive statistics for the variables and scal es used in the analysis are reported in Table 3-2. Sociodemographic and legal (case) descriptives indicate that the sample is 24% female, 21% adult, and 11% Aboriginal/T orres Strait Islander. Twenty -one percent of the offenders were arrested for violence (79% combined for th e three instrumental crim es), and half of the offenders were assigned to each of the two experimental conditions.
76 Table 3-1. Factor analysis (rot ated) for presentation of negative stimuli (Factor 1) and removal of positive stimuli (Factor 2) Item Factor 1 Factor 2 1. Treated as if likely to commit another offense .649 -.057 2. Did people make negative judgments .698 -.106 3. Treated as if criminal .791 .116 4. Important people rejected you .476 .459 5. Treated as though bad person .850 .010 6. Felt I had lost respect among family -.045 .831 7. Felt I had lost respect among friends -.073 .856
77 Table 3-2. Descriptive statistics. Variable MeanSD Min Max Failure to achieve positive ly valued goals Disjunction between expect ations, achievements 5.172.06 2 10 Disjunction between aspira tions, expectations 2.561.21 1 5 Presentation of negative stimuli 9.723.25 5 20 Removal of positive stimuli 4.241.96 2 10 Negative affect: Anger 2.461.18 1 5 Negative affect: Depression 2.411.23 1 5 Social support 12.473.54 5 20 Self-efficacy 6.951.08 3 9 Self-esteem 15.042.06 6 20 Cognitive coping: avoidance 2.231.34 1 5 Behavioral coping: revenge 2.071.05 1 5 Violation of justice rules 35.666.70 10 49 Pressure/incentive for criminal coping (delinq peers) 2.251.14 1 5 External attribution 2.431.15 1 5 Offender sex .76.43 0 1 Offender age group .79.41 0 1 Offender minority status .11.31 0 1 Crime type .21.41 0 1 Assigned treatment .50.50 0 1 Outcome measure: delinquent adaptation 4.071.54 2 10
78 Figure 3-1. Study variables.
79 CHAPTER 4 ANALYSIS AND PRESENTATION OF RESULTS Inasmuch as recent tests of general strain theory (GST) have moved beyond the theoryâ€™s core tenets to examine its predictors, mediat ors, moderators, and de linquent/non-delinquent adaptations in more granular de tail (e.g., as in vicarious and anticipated strain (Agnew, 2002), differing manifestations of anger (Mazerolle a nd Piquero, 1998), or delinquency as problemsolving (Brezina, 2000)), so too have empirical tests utilized different analytic plans. Consistent with their units of analysis and sampling de signs (i.e., individual -level/cross-sectional;1 individual-level/longitudinal;2 multi-level/cross-sectional3), the over thirty tests of GST to-date have most commonly utilized ordinary least squa res (OLS) or logistic regression analysis but have also utilized contingency table analysis (Mazerolle and Maahs, 2000), structural equation methods (e.g., Aseltine et al., 2000; Brezina, 199 9; Hoffmann and Su, 1997), growth curve models (Hoffmann and Cerbone, 1999), and multil evel linear regression models (Hoffmann, 2002). The present study is the first know n test of GST to utilize data collected in a randomized, controlled setting. As such, its analytic plan op erates in concert with common statistical methods in the experimental literature (see Farri ngton and Welsh, 2005). By randomly assigning offenders at arrest to one of the two experiment al conditions, it best is olates the strains and associated emotions directly at tributable to the behavior of law. After random assignment but prior to the treatment, the aver age score of all the persons in one group (e.g., in preexisting measures of self-efficacy, self-est eem, or external attribution) s hould not have differed from the 1 The majority of tests utilize this design. 2 See e.g., Brezina, 1999; Hoffmann and Cerbone, 1999; Mazerolle, 1998. 3 See e.g., Hoffmann, 2002.
80 average scores in the other group; in other words, the biserial co rrelation of preexisting attitudes and treatment (if scored as 0/1) should not have differed from zero (Cook and Campbell, 1979). Assessment of post-treatment differences betw een randomly formed groups is less biased than with nonequivalent groups, magnitude estimates of effects can be made with greater faith, and most threats to internal validity (e.g., selection, selection-matu ration, regression) are minimized or eliminated altogether (Cook and Campbell, 1979). Further, because randomization equates conditions for all measured and unmeas ured variables, multivariate analysis that disaggregates results for subgroups (e.g., males ve rsus females) or controls for preexisting differences between conditions are genera lly unnecessary (Farri ngton and Welsh, 2005). Analytic Plan The analysis begins with a pr esentation of relationships between strain, negative affect, GST conditioners, legal/demographic information, a nd criminality. Within that discussion, the cumulative impact of strain is assessed, as is the potential for nonlinear effects between strain, affect, and criminality. In addition (the cau tion by Farrington and Welsh (2005) above duly noted), in order to facilitate comparisons of these results with related studies criminality is regressed on the independent variables to ex plore underlying relationshi ps and the mediating function of negative affect. The presentation of findings then goes on to present comparisons between the conditioners and legal/demographic ch aracteristics on mean levels of predictors (also see Eitle, 2002; Mazerolle, 1998). The sect ion ends with an exploratory analysis of outcomes of interest to restorativ e justice theory (Braithwaite, 1989). Hypothesized Relationships Table 4-1 outlines the expected relationships be tween the study variables. Each prediction is based on foundational scholarship (Agnew, 1992, 2001, 2006a). Firstly, the disjunction between expectations and actual achievements, the disjunction betw een aspirations and
81 expectations, the presentation of negative stimul i, and the removal of positive stimuli will be positively associated with anger and with depressi on. Secondly, the cumulative impact of strain will relate to higher criminality. Thirdly, there will be a nonlinear relationship between strain and criminality; that is, greater degrees of stra in will increase the likelihood of criminality. Fourthly, the strain measures and negative affect will be positively associated with criminality controlling for other predictors. Next, behavior al coping (revenge), a perceived violation of justice rules, external attribution, and the pressure/incentive to engage in criminal coping (delinquent peers) will be positivel y associated with criminality. Relatedly, social support, selfefficacy, self-esteem, and cogniti ve coping (avoidance) will be negatively associated with criminality. Additionally, mean level compar isons will support major findings from the empirical literature. Finally, st udy variables related to restorativ e justice theory (Braithwaite, 1989) will perform as expected across the experimental conditions (conference/court); specifically, offenders assigned to conference will report lesse r degrees of strain, anger, depression, and criminality and will also report a lesser degree of violations of ju stice rules--a metric that is often used to measure the efficacy of restorative justice interventions (Makkai and Braithwaite, 1994). Measure associations Appendix B reports full-sample zero-order corr elations among study variables. Each of the four measures of strain is moderately associ ated with criminality, al though two of the strain types (disjunction between aspirations and expe ctations; removal of positive stimuli) do not relate to criminality in the expected direction. First, the lack of predicted support for the rem oval of positive stimuli, measured as loss of respect or honor among family or friends, may tap into a fine nuance between a perceived loss of honor or respect as comp ared to a perceived administration of disrespectful treatment , a
82 particularly egregious form of strain that has received support in GST and related literature (Agnew, 2001, 2006a; Anderson, 1999). The former (as loss) appears to relate to inwardoriented affect (i.e., depression) ( r = .46) and a reduced likelihood for criminality ( r = -.29), while the presentation of negative stimuli (meas ured by survey items tapping into an overt administration of disrespect (i.e., offender treate d as criminal; offender treated as a bad person) shows a moderate association with anger ( r = .38) and an increased li kelihood for criminality ( r = .30). The same holds true for the two measures of strain as goal blockage: the disjunction between expectations an d actual achievements is worded as one being overtly â€œpushed aroundâ€ or â€œpushed into thingsâ€ and th is type of strain is str ongly associated with anger ( r = .55) and moderately associated with criminality ( r = .33). The other category of goal blockage, a disjunction between aspirations and expectations, is worded as a loss of future opportunity and is not positively associated with anger. Instea d, it is positively associated with depression ( r = .45) and a decreased likelihood for criminality ( r = -.31). Agnew has argued that it is unclear whether depression cause s a predisposition for crime when it is not accompanied by anger (2006a:35). Here , it is interesting to note that depression is negatively associated with criminality ( r = -.34), there is little if any association between anger and depression ( r = .05) ( p > .05), and for none of the four measures of strain does one find both anger and depression to be statistical ly significant. In short, they appear to be singularly distinct constructs. As depression may be more strongly associated with passive crimes such as drug use (see Agnew, 2006a; Jang and Johnson, 2003), th e present studyâ€™s outcome variable may inadequately capture a full range of delinquent adaptations.
83 Cumulative effect of strain Agnew (2001) argues that despite the tendency for past res earch to construct scales combining several types of strain4--an analytic strategy that he first recommended (Agnew, 1992) but that has its shortcomings due to the difficulty in disentangling empirically supported strains from those that exhibit little relationship to crime--future research should principally test strains individually to assess their empirical adequacy. Toward this end, the present study separates out the four sources of strain. As discussed, it appears that the disjuncture between expectati ons and actual achievements and the presentation of negative stimuli impact offendersâ€™ emotiona l responses and prospective criminality as predicted by theory. But what of the cumulative effect of strain on criminality (Agnew, 1992; Agnew and White, 1992)? Exploratory analysis (not shown) indicates that for those offenders who score highly on both measures of strain (i.e ., are above the median for both measures) (n = 72), the mean score for criminality is 4.65 (st.de v. = 1.72). For those offenders who score lower on both measures of strain (i.e., are below the median point for both measures) (n = 22), the mean score on criminality is 3.05 (st.dev. = .90), a percentage decrease on the scale of 34%. Relatedly, the high-strain composite is positively associated with criminality ( r = .25, p < .01) while the low-strain composite inverses the relationship ( r = -.21, p < .01). In addition, moderate effects are found between the high -strain composite and anger ( r = .482, p < .01) and the lowstrain composite and anger ( r = -.335, p < .01). 4 For example, Mazerolle and Piquero (1 997:330) constructed a composite measure of strain by summing scores of all strain variables. The researchers did, however, call fo r future research to examine the range of different strains on specific deviant outcomes (M azerolle and Piquero:339).
84 Nonlinearity There may also be a nonlinear effect between st rain and criminality; th at is, certain strains may predict delinquency more strongly after a certain threshold point is passed, and common analytical methods may underreport a relationship between strain and crime due to the relatively few respondents who surpass those high levels (Agnew, 2002:624). Whereas mild to moderate strain will not cause most people to engage in cr ime, strains at the highest levels are thought to lead to a greater degree of anger and will tax a personâ€™s legitimate coping resources (Agnew, 2005:130). As has been assessed in research on victim ization and delinquency (Agnew, 2002), Table 4-2 presents the mean criminality score across le vels of the strain most strongly related to criminality in baseline analyses: the disjunction between expectations and actual achievements. As one can see pictorially in Figure 4-1, there ap pears to be an interesting nonlinear relationship at the very highest levels of strain (i.e., an offender scoring a nine or ten on the scale). Complementary analysis (see Fi gure 4-2) mapping mean anger scores across levels of the disjunction between expectations and actual achievements yields a similar slope at the highest level of strain: mean anger of 3.14 at score of 8 on strain, 3.30 at 9, and 4.25 at 10. While the graphing of strain on anger does not depict the same degree of nonlinearity as of strain on criminality, both map as positive, concave upward curves--trend lines that have been found in research on strain (Paternoster and Mazerolle , 1994) and other areas of criminological scholarship (Tittle, 1995). Facilitators of delinquent coping As discussed earlier, GST sets out to specify not only the so urces of strain and negative emotion that tie to crime and delinquency, but also the conditions under which strain/affect increase or reduce an individualâ€™s likeli hood for delinquent coping. Therefore, those
85 facilitators/conditioners that have been found to be tied to delinquent coping in their respective literatures (e.g., stress, equity, soci al learning) should be found to tie to criminality in the present test. As predicted (see Appendix B), each of the facilitators of de linquent coping exhibit moderate to strong associations with cr iminality: (1) behavioral coping (revenge): r = .39; (2) perceived violations of justice rules: r = -.515; (3) external attribution: r = .27; (4) the pressure/incentive to engage in cr iminal coping (delinquent peers): r = .29). Conditioners are associated with changes in the direction a nd/or strength of a rela tionship (Baron and Kenny, 1986). The particularly strong relationship between th e perceived violation of justice rules and criminality ( r = -.51) may support Agnewâ€™s recent m odification to the theory where the perception of unjust treatment is no longer categorized as a distinct type of strain but can instead by used to classify any type of strain (Agnew, 2001, 2006a).6 Future research should work to better disentangle the effect. Stra ins that are viewed as unjust are more likely to lead to crime because they activate anger and related emotions, tend to reduce social control (e.g., as in unjust treatment by parents relating to strained ties), and are more likely to facilitate the social learning of crime (Agnew, 2006a:61-2). Although the cross-sec tional nature of the data limits inferences regarding the relationship between injustice judgments and the pr essure/incentive to engage in criminal coping (re: temporal ordering and delinquent peer association), it is interesting to note that perceived injustice exhibits a si milarly strong relationship with anger ( r = -.52) as it does with prospective criminality ( r = -.51). 5 The violation of justice rules measures are coded 1-5, with 1 representing the greatest degree of perceived injustice. Higher values for all other measures indicate greater degrees of that construct. 6 Within the present context, strategies of regulation based on the way in which members of the public are treated by legal authorities enhances their willingness to cooperat e with and defer to legal authorities (Tyler, 1990).
86 Also note the degree to which the facilitators of delinquent coping align with strain. Without exception, each fac ilitator associates with strain in the same direction as the strains do with criminality. For example, the pressure/incentive to engage in criminal coping (delinquent peers) is positively, and significantly, associated with the disjunction between expectations and actual achievements and with the presentation of negative stimuli, whereas it is negatively associated with the disjunction between aspirations and expectations and the removal of positive stimuli. While remaining cognizant of the limitations of cross-sectional data, it is interesting to note how the four types of strains relate to these measures of in terest in predictable ways. If two sources of strain within this sociolegal contex t associate with anger and criminality while two sources of strain associate with differentially manifested aff ect (i.e., depression), one would expect to see these fac ilitators/moderators operate exactly as they do. As depression relates to self-blame, self-isolation, wish ful thinking, and distancing (Kaplan, 1996:202), the constellation of measures discussed above may operate at some distance from those most strongly related to inwardly-oriented a ffect and adaptation. Constraints to delinquent coping As predicted, each of the constraints to de linquent coping exhibits a negative relationship with criminality, with three of the four measures as statistically significan t (see Appendix B): (1) cognitive coping (avoidance): r = -.22; (2) social support: r = -.24; (3) self-efficacy: r = -.15; (4) self-esteem: r = -.12 ( p >.05). Although related branches of research would likely predict a stronger negative association betw een social support and criminality (e.g., Cullen, 1994), its sign and significance differentiates itself among test s of GST that show mixed support (Eitle and Turner, 2003; Jang and Johnson, 2005). Social support exhibits a sm all, but significant, inverse
87 relationship with anger ( r = -.14) and depression ( r = .14), providing additional support for the measures being distinct categories of affect.7 Interestingly, one of the strongest coeffici ents in the present st udy is between cognitive coping (avoidance) and depression ( r = .55), a relationship that r eceives support in community psychology. As Billings and Moos (1984:887) expl ain, oneâ€™s â€œchoice of coping strategies has been associated with the degree of pathology in depressed patients , with active coping strategies associated with less severe dysfunction and avoidance coping linked to more serious depression.â€ Also interestingly, the relations hip between cognitive avoidance and strain is strongest for those strains least associated with criminality: th e disjunction between aspirations and expectations ( r = .35) and the removal of positive stimuli ( r = .34). Research on coping effectiveness indicates that while active behavioral coping is ge nerally adaptive (â€œhealthy,â€ if you will), avoidant coping is a nega tive strategy that may occur in re sponse to situations that are perceived as unchangeable (Mat tlin, Wethington, and Kessler, 1990:117). Similar to questions that were raised earlier, there may be a palpable difference between responses to strains that are thrust onto a person (e.g., percei ved disrespect, perceived power differentials) from those that represent the loss of something of value (e.g., future opportunities, respect and honor among family and friends). Finally, self-esteem and self-efficacy ar e found to be moderately related ( r = .42). Recent research indicates that these widely used traits in psychology are indeed strongly related and may be markers for the same underlying construct (J udge, Erez, Bono, and Thoresen, 2002). Relating back to cognitive avoidance, the strongest statis tically significant relationships for self-esteem 7 As research indicates that increased social support rela tes to coping/affect differently across sex (e.g., Fondacaro and Moos, 1987), means were compared. Males (12.41) and females (12.67) do not significantly differ on social support ( t = .470, p > .05), nor do males (2.34) and females (2.66) significantly differ on depression ( t = 1.64, p > .05). Additional split-sample analyses follow.
88 and self-efficacy are with the disjunction between aspirations and expectations ( r = -.19 / -.15) and with the removal of positive stimuli ( r = -.13 / -.16). Although so mewhat weak effects, the strains most relating to a loss of â€œselfâ€ may affect self-regard (self-esteem) and psychological control (self-efficacy) in unique ways that warrant future inquiry. Mean-level Differences between Groups of Interest Table 4-3 presents a comparison between the demographic/legal variables on mean levels of predictor variables. Sex Despite past research which has isolated important differences between men and women on measures of strain, affect, and delinquent/ nondelinquent adaptations (Beutel and Marini, 1995; Broidy, 2001; Broidy and Agnew, 1997; Ei tle and Turner, 2002; Hay, 2003; Mazerolle, 1998; Piquero and Sealock, 2004), the only significant measure in the analysis is avoidant coping ( t = 2.76) with women (24% of sample) more apt to wish to cognitively â€œremoveâ€ themselves from the strainful conference/court environment. A greater degree of avoidance coping among women has received empirical support, particularly with regard to how it interacts with low conventional social support (Cronkite and Moos, 1984). The apparent overall lack of sex differences runs contrary to evidence that males may experience more strains conducive to crime,8 engage in more crime, have lower levels of depression, exhibit higher degrees of anger,9 have lower levels of emotional social support, and are more likely to associate w ith other criminals (see Agnew, 2006a for discussion). Perhaps 8 But perhaps not the number of strains in total . Women experience specific strains in higher numbers than men, such as sexual assault/abuse, partner abuse, and gender discrimination. The key difference may be that menâ€™s strains are more conducive to criminal coping (see Agnew, 2006a:134). 9 That is, higher degrees of anger related to moral outrage. Women, on the other hand, may exhibit similar degrees of anger, but anger that is coupled with guilt, fear, depr ession, etc. (Broidy, 2001; Hoffmann and Su, 1997; Piquero and Sealock, 2004).
89 there is something unique about th is legal context, and it may st em from the fact that general strain theory centers on one key point: the initial perception of one or more significant strains in oneâ€™s life. Without that perception, the theoryâ€™s intermediate measures do not come into play. Whereas males may be more likely to experience a whole host of high magnitude strains in their daily lives (e.g., abusive peer relations, threats to masculine status, monetary problems, criminal victimization, homelessness) (Agnew, 2006a), th e present studyâ€™s legal context may preclude males from perceiving a measurably greater level of high magnitude , unjust strains as compared to females. Put another way, an offenderâ€™s full life history is not really bei ng tapped into because the study measures focus solely on strains experi enced in court or conference--not in oneâ€™s school or neighborhood or in a patrol car. Particularly in light of the preponderance of in strumental crimes in the sample (79%), the apparent lack of gender differences may also tap into a closing gender gap in crime and of womenâ€™s increased economic marginalization (H eimer, 2000). Further, GSTâ€™s emphasis on violations of justice rules (Agnew, 2001, 2006a) may relate to empirical evidence that like people in general, even high-risk males are willing to defer to the decisions of legal authorities when those decisions are viewed as procedurally just and trustworthy (Tyler and Huo, 2002). In short, it may be possible that sex differe nces are â€œwashed outâ€ at their inception. Age Table 4-3 indicates statistica lly significant age-based differences for two measures, the violation of justice rule s and the pressure/incenti ve to engage in criminal coping (delinquent peers), with adults (21% of samp le) expressing a greater degree of perceived violations of justice rules ( t = 2.08) and a greater perceived accep tance of their crime among peers ( t = 2.12). In essence, this analysis by age is that of all j uvenile offenders (in all three offense categories) crossed with adult violent offenders, so age may act more as a proxy for offense type
90 (instrumental v. expressive) than for oneâ€™s deve lopmental stage.10 Mean-level differences by offense type are outlined later. Nonetheless, more detailed procedural justic e measures would provide information into whether these adult offenders are more sensitiz ed to perceived injust ice than their younger counterparts, perhaps due to perceived chronic thr eats to social identity (see Tyler, 2003). And although much of the research on social learning relates to the experiences of juveniles, an inclusion of multi-level measures may explain wh ether an adult offenderâ€™s structural conditions (e.g., economic inequality, social disorganization) impact the breadth a nd depth of his or her delinquent peer network (see Akers, 1992). Minority/majority Aboriginal/Torres Strait Islander offenders (11% of sample) express a lesser degree of perceived violations of justice rules ( t = 2.66). This unanticipated effect is particularly interesting given the relationship between the minority experience and procedural injustice (Tyler, 2003), and of Australian Aboriginals some times rejecting the â€œwhite manâ€™sâ€ criminal justice system as discriminatory against bl acks, adopting different standards regarding appropriate punishment (e.g., corpor al punishment), and viewing thei r traditional justice system as superior (Braithwaite, 1989:95) . As such, it would not have been unreasonable to assume differences in affect (anger or depression), external attribution, and/or behavioral and cognitive coping. The overall lack of si gnificant differences across the study measures (save one) also runs contrary to evidence that disadvantaged se gments within a society experience unique forms 10 Due to the relatively few cases in the juvenile violen t crime category, the RISE research team amended its selection criteria mid-project to include violent offenders up to the age of twen ty-nine. Property offenders are age seventeen or younger (Sherman et al., 2001).
91 of strain, particularly with re gard to discrimination grounded in ascribed characteristics (Agnew, 2001). Crime type Mean-level differences (Table 4-3) indicate a good deal of differentiation by crime type (instrumental v. expressive), with offenders arrest ed for violence expressing a greater degree of (1) strain as disjunction between ex pectations and actual achievements ( t = 1.99), (2) selfefficacy ( t = 2.13), (3) violations of justice rules ( t = 2.10), (4) extern al attribution ( t = 3.82), (5) pressure/incentive to engage in cr iminal coping (delinquent peers) ( t = 4.19), and (6) prospective criminality ( t = 2.77). Instrumental offenders, on the ot her hand, express a grea ter degree of (1) strain as removal of positive stimuli ( t = 3.74). Despite evidence that many offenders exhibit diverse and versatile offending patterns (Mazer olle, Brame, Paternoster, Piquero, and Dean, 2000), past research has split by offense type beca use some offenses (e.g., theft) are thought to call on risks/rewards calc ulations as compared to expres sive crimes, which call for more immediate action (Schoe pfer and Piquero, 2006). As discussed within the sect ion on age, it is important to question whether offenders arrested for violence may already be more sens itized to perceived injustice because of the context of their arrests and the fact that many e xpressive crimes are tied to threats to social identity (Anderson, 1999; Tyler, 2003). Victim ization, for example, a phenomenon already tied to high rates of perceived injust ice, cannot be understood apart fr om offending (Lauritsen et al., 1991). It is interesting to note, however, that violent offenders do not appear to differ in a likely concomitant emotion: anger. As was raised ear lier, the studyâ€™s focus on offendersâ€™ reactions to their experimental condition may â€œwash outâ€ preexperimental cognitive and emotional states. Furthermore, it is interesting to note that violent offenders score higher on self-efficacy ( t = 2.13), a construct measuring percei ved psychological control that GST argues is a facilitator of
92 nondelinquent coping (Agnew, 1992). Perhaps the observed effect centers on oneâ€™s use of psychological control as research indicates that individuals who have a history of successful social influence through coercion may develop high self-efficacy/self-confidence in coercive means to solve problems (Tedeschi and Felson, 1994:210). Finally, the relationship between expressive crime in particular and the pressure/incentive to engage in criminal co ping (delinquent peers) ( t = 4.19) is interesting c onsidering that social learning theory has been applied to nearly all crime types and problem behaviors (Akers, 1985). Future research should assess whether in a legal, partly shaming-centered context such as this, other-directed (violent) offenders differenti ally draw on purported â€œpeer acceptanceâ€ as a neutralization tec hnique (Sykes and Matza, 1957). Regression Estimations In order to facilitate comparisons to past tests of general strain theory, and particularly to those of active offenders (e.g., Baron, 2004; Ba ron and Hartnagel, 2002; Baron et al., 2001; Piquero and Sealock, 2000, 2004), Table 4-4 presents five stepwise OLS regression estimations for the full sample. Each model presents both standardized and unstandardized coefficients.11 Model 1 includes each of the strain variables (disjunction between expectations and actual achievements, disjunction between as pirations and expecta tions, presentation of negative stimuli, and removal of positive stimuli) and four contro l variables (sex, age group, minority status, and crime type). As expected, the re sults are substantly similar to the earlier zero-order correlations (see that discussion) with all four st rain variables exhibiting significant ( p < .01) effects on criminality. Offender sex, age group, and minority group fail to exhibit significant effects as 11 Correlations among the independent variables do not indicate problems with multicollinearity.
93 anticipated, while crime type (i.e., expressive crime) fails to achieve significance as it had in the correlation matrix (Appendix B). Model 2 presents the same equation as Model 1, but includes anger as measure of negative affect. As expected, ange r exhibits a significant ( p < .01) positive effect on criminality. Of particular importance is that th e introduction of anger into the m odel does not substantively alter the statistical significance of the strain variables. In other words, anger does not largely mediate the effects of strain on criminality (i.e., strain has a direct, but not indirect, effect on criminality). The inclusion of anger into the model does, however, lower the coe fficients for three of the four strain variables (disjunction between expectati ons and actual achievements, presentation of negative stimuli, and remova l of positive stimuli). Model 3 replaces anger with the measure of de pression. Consistent with earlier analysis, depression exhibit a significant ( p < .01) negative effect on cr iminality. Interestingly, unlike with anger (Model 2), the inclusi on of depression into the model renders strain as disjunction between aspirations and expecta tions and strain as removal of positive stimuli statistically insignificant, lending support to GSTâ€™s prediction th at affect (in this cas e, depression) largely mediates the relationship between strain and crim inality (Agnew, 1992). It is important to note, once again, that the baseline relationship of these two strains on criminality is inversed from theoretical expectation (i.e., nega tive), raising questions about whether strain as the loss of something of value (e.g., opportunity, honor among fa mily and friends) is qualitatively different from strain as presentation of stressors. Model 4 estimates the effects on criminality with both measures of affect in the model. Anger and depression remain statistically significant ( p < .05, p < .01, respectively), and the personal/legal variables still fail to exhibit signi ficant effects. As compared to the previous
94 model, strain as disjunction between aspirations and expect ations becomes statistically significant ( p < .05), which is likely a byproduc t of its high coefficient ( = -.23) in the angeronly model. Model 5 adds the remaining study measures to the equation. As opposed to earlier analysis that indicates that adults mo re strongly perceive violati ons of justice rules and the pressure/incentive to engage in criminal coping (d elinquent peers) (Table 4-3), Model 5 indicates that adults report higher prospect ive criminality than juveniles ( = .14). Crime type (i.e., violent offenders) ( = .13) exerts a significant and positive effect on criminality, as does a greater degree of perceived viola tions of justice rules ( = -.41). High social support ( = -.12), high self-efficacy ( = -.13), and cognitive coping (avoidance) ( = -.15) all exert negative and significant effects on criminality. All four strain variables fail to exert significant effects on criminality in the full model. The relative in fluence of the standardized coefficient for the violation of ju stice rules ( = -.41) should be noted. A word of caution should be made regarding these regression estimates, however. Just as Farrington and Welsh (2005) argue that controlled, randomized st udies generally do not require multivariate analysis that controls for preexis ting differences, simulations have shown that stepwise regression can indicate stat istically significant results for pr edictors that are unrelated to the outcome variable (Bushway, Sweeten, and Wilson, 2006). As such, the regression of criminality on the study variables should be viewed as exploratory, with all due emphasis placed on the complementary analyses outlined earlier. Study Measures and Restorative Justice The analysis to this point has concentrated on manifestations of strain, affect, and delinquent coping across the full offender group, rega rdless of experimental treatment. In other
95 words, as a direct test of genera l strain theory and not of restorative justi ce theory (Braithwaite, 1989),12 to this point little attention has been pa id to whether emotions important to general strain theory differ across an offenderâ€™s random assignment to court or conference. Table 4-5 presents the statistically significant mean-level differences. Consistent with RISE assessments outlined in the review of literature, offenders a ssigned to court are considerably more likely to perceive violations of justice rules ( t = 5.22). Taken on balance, this finding is not only consistent with the RISE research teamâ€™s assessm ent, but is also consistent with procedural justice scholarship that emphasizes the power of participation (T hibaut and Walker, 1975). As Tyler (2003) explains, (P)eople are more satisfied with a procedure that allows them to participate by explaining their situation and communicati ng their views to the authorit ies about that situation and how should be handled. This participation e ffect makes clear why procedures such as mediation are more popular than the courts. Of primary importance is the ability to state oneâ€™s views to an authority and to feel that those views are being considered (p. 300). Offenders assigned to court are al so more likely to feel angry ( t = 2.02), perceive a level of support for their delinquen cy among their peers ( t = 2.35), exhibit a desire to behaviorally cope through revenge ( t = 2.95), and to perceive strain s related to a disjunction between expectations/actual achievement s and aspirations/expectations. Overall, it is interesting to note that all the st atistically significant facilitators of delinquent coping associate with court assignm ent. It is perhaps equally im portant to note, however, that no theorized inhibitors to delinquent copi ng associate with conference assignment. In addition, and perhaps more centrally, th e present studyâ€™s measure of prospective criminality does not differ across experimental treat ment with regard to statistical significance.13 12 For example, see Tosouni (2004). 13 Means for criminality do differ (descriptively) for court (x-bar = 4.18) and conference (x-bar = 3.96).
96 Although restorative justi ce initiatives are often deemed successful with criteria other than reoffending (Makkai and Braithwaite, 1994), this is still noteworthy. Taking into account the RISE recidivism report (Sherman et al., 2002) that found that vi olent offenders assigned to conference reoffended at substantially lower levels than those assigned to court, an additional comparison-of-means was conducted for only that offense. Criminality was found to not significantly differ across court or conference for violent offenders ( t = 1.21) ( p > .05).14 Agnew (1992) asserts that GST is a theory of both â€œcrimeâ€ and â€œcriminalityâ€ (p. 62); the present analysis may cast these outcome measures as potentially dis tinct from one another. Nonetheless, as was expressed in the discussion of OL S estimates, caution should be us ed in interpreting a lack of statistical significance. In both offender samples, prospective criminality is found to reduce in the expected direction: that of those assigned to conference. In this randomized, controlled environment, it is altogether possi ble for effects to be analytically interesting but statistically non-significant (Bushway et al., 2006). 14 Means for criminality among violent offe nders do differ (d escriptively) for co urt (x-bar = 4.88) and conference (xbar = 4.32).
97 Table 4-1. Predicted relationships , study measures and criminality. Independent Variable Variable criminality Disjunction between Expectations a nd Actual Achievements Positive Disjunction between Aspirations and Expectations Positive Presentation of Negative Stimuli Positive Removal of Positive Stimuli Positive Anger Positive Depression Positive Behavioral Coping (Revenge) Positive Violation of Justice Rules Positive External Attribution Positive Delinquent Peers Positive Social Support Negative Self Efficacy Negative Self Esteem Negative Cognitive Coping (Avoidance) Negative Sex No Prediction Age Group No Prediction Minority No Prediction Crime Type No Prediction Treatment (1=conference) Negative
98 Table 4-2. Nonlinear effect , strain on criminality. Level of strain: Disjunction between expectations and actual achievements Mean criminality score N 2 (low) 3.39 23 3 3.33 12 4 3.88 85 5 3.90 21 6 4.03 38 7 4.71 17 8 4.64 22 9 4.90 10 10 6.00 8
99 Table 4-3. Mean-level differen ces by social/legal measures. Males Female T -Value Mean SD Mean SD Cognitive coping: avoidance 2.09 1.28 2.67 1.43 2.76 Juveniles Adults T -Value Mean SD Mean SD Violation of justice rules 36.09 6.49 33.80 7.29 2.08 Pressure/incentive (delinq peers) 2.15 1.03 2.60 1.42 2.12 Aboriginal/Torres Strait Non-Aborig/Torres T -Value Mean SD Mean SD Violation of justice rules 37.96 4.44 35.35 6.91 2.66 Instrumental Crime Expressive Crime T -Value Mean SD Mean SD Disjunction betwn expect/achieve 5.03 2.04 5.68 2.10 1.99 Removal of positive stimuli 4.47 1.96 3.40 1.74 3.74 Self-efficacy 6.88 1.08 7.24 1.01 2.13 Violation of justice rules 36.22 6.06 33.56 8.40 2.10 External attribution 2.27 1.04 3.04 1.32 3.82 Pressure/incentive (delinq peers) 2.05 .93 2.98 1.50 4.19 Criminality 3.93 1.48 4.60 1.64 2.77 Note: Only measures w/ significant differences are included ( p < .05). T -values are presented as absolute values. For violation of justice rules, lower valu es indicate greater perceived injustice. For all other measures, higher values indicate a greater degree of that construct.
100Table 4-4. Criminality regressed on study variables. Measures Model 1 Model 2 Model 3 Model 4 Model 5 Disjunction between expectations, achievements .25 (.18)* .17 (.12)* .23 (.17)* .16 (.12)** -.07 (-.05) Disjunction between aspirations, exp ectations -.20 (-.26)* -.23 (-.29)* -.13 (-.17) -.15 (-.20)** -.12 (-.16) Presentation of negative stimuli .18 ( .08)* .16 (.07)* .24 (.12)* .21 (.11)* .14 (.07) Removal of positive stimuli -.16 (-.12)* .15 (-.12)* -.05 (-.04) -.05 (-.04) .02 (.01) Offender sex -.05 (-.16) -.04 (-.14) -.06 (-.20) -.05 (-.17) -.03 (-.13) Offender age group .04 (.17) .04 ( .16) .09 (.33) .08 (.33) .14 (.55)** Offender minority group .01 (.05) .01 (.05) .01 (.06) .01 (.04) .07 (.33) Crime type .09 (.37) .09 (.32) .11 (.42) .11 (.42) .13 (.51)** Anger .16 (.20)* .15 (.19)** -.02 (-.03) Depression -.27 (-.35)* -.27 (-.34)* -.09 (-.11) Social support -.12 (-.05)** Self efficacy -.13 (-.19)** Cognitive coping: avoidance -.15 (-.18)** Behavioral coping: revenge .11 (.15) Violation of justice rules -.41 (-.09)* Pressure/incentive crim coping (del inquent peers) .05 (.07) External attribution .06 (.09) Constant 3.40 3.34 3.31 3.24 9.33 Adjusted R2 .24 .25 .31 .32 .43 Note: Standardized effects shown, with unstandardized effects in parentheses. * p < .01; ** p < .05
101 Table 4-5. Mean-level differe nces by RISE treatment. Court Conference T Value MeanSDMeanSD Disjunction betwn expect/achieve 5.412.034.922.07 1.82+ Disjunction betwn aspir/expect 2.701.152.431.25 1.72+ Anger 2.621.202.311.16 2.02* Behavioral coping: revenge 2.2126.96.36.199 2.95* Violation of justice rules 33.506.4137.836.30 5.22* Pressure/incentive (delinq peers) 2.421.192.081.05 2.35* Note: Only measures w/ signifi cant differences are included. T -values are presented as absolute values. For violation of justice rules, lower values indicate greater perceived injustice. For all other measures, higher values indicate a greater degree of that construct. + p < .10 * p < .05
102 3 3.5 4 4.5 5 5.5 6 12345678910 DISJUNCTION BETWEEN EXPECTATIONS, ACTUAL ACHIEVEMENTS Figure 4-1. Strain on criminality: Nonlinearity.
103 1 1.5 2 2.5 3 3.5 4 12345678910 DISJUNCTION BETWEEN EXPECTATIONS, ACTUAL ACHIEVEMENT S Figure 4-2. Strain on an ger: Nonlinearity.
104 CHAPTER 5 DISCUSSION AND CONCLUSION Key Contributions The present study builds on general strain theoryâ€™s empi rical record in several ways. Most notably, no known tests have examined general st rain from a sociolegal perspective and no known tests have utilized experimental data. Additionally, few tests have explored GST internationally (although see Bao et al., 2004; Me sch et al., 2003; Pinhey and Okinaka, 2004), tests have not traditionally utilized active offenders (although see Baron, 2004; Baron and Hartnagel, 2002; Baron et al., 2001; Piquero and Sealock, 2000, 2004) , most tests have failed to utilize subjective measures of strain (i.e., t hose types of strains st emming from events or conditions that are disliked by the people e xperiencing them) (although see Agnew and White, 1992), and most tests have not feat ured as rich a set of GST-spec ific strains, conditioners, and measures of affect. While the study finds major support for many of the theoryâ€™s key tenets, it also raises important questions about how certain categorie s of strain, negative affect, and conditioners operate sociolegally. More spec ifically, the disjunction between aspirations and expectations, the removal of positive stimuli, and the expression of depression separated themselves as measures that will likely require more detailed specifications in future tests. In short, the perceived loss of future opportunities and/or loss of respect or honor among family or friends operated very differently from an administration of disrespectful treatment. The former (as loss) related to inward-oriented a ffect (i.e., depression) a nd to a lower degree of criminality, while the latter associated with outward-oriented affect (i.e., anger) and a higher degree of criminality. Considering that depression may be more strongly a ssociated with passive crimes (see Agnew, 2006a), it is possible that the present stud yâ€™s criminality measure does not
105 capture a full range of delinquent adaptations. In light of the fact that none of the offenders was arrested for a drug offense, and questions rega rding future criminality were queried in an environment centering on explicit expressive and instrumental crimes, it is likely that outwardoriented criminality was at the perceptual fore. As recent critiques indicate that different strains can impact people differently (Agnew, 2001, 2006a; also see Briere, 1992), the findings raise interesting possibilities. Consistent with theoretical prediction, an ex amination of the cumulative effect of the strains most associated with criminality in th is study (disjunction be tween expectations and actual achievements; the presentation of negative stimuli) indicated a sizable difference between high-strain and low-strain offenders and criminal ity; a similar effect was also found for anger. Also consistent with prediction regarding non linearity in GST, offenders experiencing the highest levels of the studyâ€™s most empirically supported strain (disjunction between expectations and actual achievements) were found to exhi bit a disproportionally high degree of both criminality and anger. Similar to the case of cu mulative effects, future tests may benefit from understanding whether common anal ytical methods underreport a re lationship between strain and crime due to the relatively few respondents who surpass very high levels (Agnew, 2002:624).1 The present analysis also assessed the relative influence of facilitato rs and inhibitors to delinquent coping. As predicted, each of the facil itators of delinquent coping exhibited moderate to strong associations with crim inality. Without exception, each facilitator of delinquent coping was found to associate with strain in the same di rection as the strains did with criminality. As explained earlier, if two sources of strain within this sociolegal context associate with anger and 1 Related research has found that composite measures ba sed on individuals at the â€œright tailâ€ can elucidate interesting trends in perceptual data. In one such study, blacks were found to be over three times more likely than whites to surpass the highest levels of perceived violations of justice rules (Rice and Piquero, 2005).
106 criminality while two sources of strain associat e with differentially manifested affect (i.e., depression), one would have expect ed to see the facilitators oper ate exactly as they did. The violation of justice rules separated itself in particular. It was strongly associated with the types of strains most associated with criminality, with anger, with the facilita tors of delinquent coping, with criminality itself, and then also (as expe cted) exhibited an inverse relationship with the inhibitors of delinquent c oping (e.g., social support). With regard to the inhibitors to delinquent coping, as predicted each associated with a reduced likelihood for criminality, an d all but one associated with a lower degree of anger. The one outlier is the cognitive copi ng mechanism of avoidance, a condi tioner that raised particularly interesting questions. As a measure with a strong association with depression, cognitive avoidance also associated most strongly with the strains negatively associated with criminality. Future tests of GST may wish to work to better understand whether avoida nce coping is actually an inhibitor or a facil itator of delinquency, depending on the sp ecific strain and form of negative affect. Unlike past tests of GST that have found im portant differences across sex, age, and minority status, the present study found few differen ces. Violations of ju stice rules once again played a prominent role, with adult (expressi ve) offenders and Aust ralian non-Aboriginals perceiving greater injustice. Across crime type (instrumental v. expressive), more differences were noted. Offenders arrested for violence expressed a greater degree of strain as disjunction between expectations and actual achievements, self -efficacy, violations of justice rules, external attribution, pressure/incentive to engage in criminal coping (de linquent peers), and prospective criminality. Instrumental offenders, on the other hand, expressed a greater degree of strain as removal of positive stimuli. Future research should examine whether such trends reflect
107 differing cost/benefit calculations across offense type or are mispecified in light of diverse offending patterns among many criminals. The multivariate analysis was somewhat simila r to what was found at the bivariate level, although it is important to note that contrary to theoretical prediction, anger was found not to mediate the effects of strain on criminality. Interestingly, the replacement of depression for anger in the multivariate models indicated that depression mediated the (inverse) relationship between strain as disjunction be tween aspirations/expectations and strain as the removal of positive stimuli onto criminality, on ce again supporting a call for future research to better understand how depression operates w ithin this context. As opposed to earlier analysis, a multivariate model including all study variables indi cated that juveniles were higher than adults in prospective criminality, and the four strain variables failed to exert significant effects on criminality in the full model. As a randomize d, controlled experiment, due attention should be paid to the relative utility of multivariate findings, on balance. Finally, differences were also assessed from th e perspective of restor ative justice theory. Largely consistent with both th eory and RISE program and outcomes assessments (e.g., Sherman et al., 2000, 2004), offenders assigned to court were mo re likely to perceive violations of justice rules, to feel angry, to percei ve a level of support for their de linquency among their peers, and to exhibit a desire to behaviorally cope through re venge. These offenders were also found to more likely perceive strains related to a disjuncti on between expectations/actual achievements and aspirations/expectations. In othe r words, offenders assigned to c ourt were more likely to feel that they felt pushed around by people with more power, pushed into things s/he did not agree with, and had â€œstuffed upâ€ future opportunities.
108 Interestingly, all the significan t facilitators of de linquent coping were associated with court assignment. As discussed earlier, however, no theorized inhibitors to delinquent coping (e.g., perceived social support, high self esteem) associated with conference assignment. In addition, and perhaps most centrally, the studyâ€™s measure of prospective criminality did not significantly differ (statistically) across experimental treatment fo r the full sample or violent subsample. As an experimental context, however, it should be no ted that descriptive (mean-level) measures of criminality were higher for court assign ees in both full and violent samples. Limitations Several important limitations to the present study are worth mentioning. Despite the past tests of GST that have utilized projected crimin ality/intentions to commit crime as dependent variable, the present stud y would have most certainly benefited from RISE recidivism data. At a minimum, it would have been interesting to assess to what degree projected criminality correlated with self-reported and/ or officially reported offendi ng (see Green, 1989; Tittle, 1980). Next, it is critical to note that the RISE su rveys were based upon completion rates of 76 percent of juvenile personal property offenders, 73 percent of juvenile property security offenders, and 72 percent of violent crime offe nders (Sherman et al., 2001). Ongoing outcomes assessments by the RISE research teams may help us to understand whether, and to what degree, these completion rates reflect random, or non-ra ndom, missing data. Relatedly, in a working paper Sherman and Strang (2004:7) drew attention to the fact that in the property experiment, one-third of the offenders assigned to c onferencing did not meet their victims. Additionally, the measures for the disjunction be tween aspirations and expectations, anger, depression, cognitive coping, emoti onal coping, pressure/incentive to engage in criminal coping, and external attribution were based on individual survey items , not scales. As events and
109 conditions related to strain are by all mean s multifaceted, future research may find that instruments made up of several indicato rs are best suited to the task. Policy Implications As might be expected in light of the legal context, the studyâ€™s findings have several important policy implications. Most notably, it is quite clear that despite the criminal justice systemâ€™s best intentions, many offenders are being treated in a manner that most predisposes them to future criminality (see Sherman, 1993). Mo re specifically, the tr aditional processing of offenders appears to be associated with greate r degrees of several forms of strain, a greater degree of anger, a greater desire for revenge, and an overarching sense of injustice. An imposing demonstration of state power may counter a just ice systemâ€™s purported purpose and may bespeak intractable problems for that state in years to come. As such, justice systems may be well served to consider the relati ve shortcomings of a retributive paradigm. Coupled with restorativ e justice and reintegrative shaming principles (Braithwaite, 1992; Sherman and Strang, 1997), genera l strain theory would appear to be an excellent way to frame inputs to an effective legal system--a syst em that administers sanctions that are seen as just, a ppropriate in magnitude, do not create pressure or incentive to engage in crime, and do not reduce key elements of so cial control (also s ee Agnew, 2001:352). Future Directions for Research The studyâ€™s findings also suggest several direct ions for future research. For example, despite past efforts that have been undertak en to better understand subcategories of negative affect (e.g., Mazerolle and Piquer o, 1998; Mazerolle et al., 2003), fu ture research may wish to test the social functions of affect. That is, to assess under what circumstances anger is mistakenly entangled with reactive aggression . Literature on violence and aggression suggests that there are multiple functions of anger in social situations to include defensive, energizing,
110 expressive, disruptive, and that of self-presentation (Novaco, 1976; Tedeschi and Felson, 1994). Relatedly, and as has been the case with effort s to better specify process-based judgments onto the social regulatory activities of agents of so cial control (Tyler, 2003), future research should better specify perceived violations of relevant justice norms onto the varied social functions of affect. For example, whethe r the reactive expression of se lf-presentation maps onto the prevention of achieving positively valued goals , whether expressive anger maps onto the removal of positively valued stimuli, and so forth. In addition, as has been argued in Thornberryâ€™s (1987) interac tional theory and in Agnewâ€™s (2005) general theory, future research may wish to better understand whet her the characteristics of strain have reciprocal e ffects on one another; for exampl e, whether injustice judgments influence perceptions of high magnitude, while high magnitude also influences injustice judgments. Further, research may wish to gauge the impact of crime on these strain characteristics, such as in the possible case of cr ime leading to lower degrees of social control, which then leads to ever-higher involvement in crime. As such, there may be any number of effects underway that warrant addi tional inquiry. One such effect may be the â€œstoryline,â€ or the temporal level between background and situational f actors in offending; in other words, an event that is out of the ordinary that alters an indivi dualâ€™s characteristics, inte ractions, or settings for interactions and increases oneâ€™s likelihood for offending (e.g., in an unresolved dispute or a brief, tempting opportunity for crime) (Agnew, 2006b:129-134). In closing, it would appear that a legal context can become a particular ly fruitful area for general strain theory scholarship. Just as resear ch suggests that in judgi ng the courts people are less concerned with performance issues than they are with quality of treatment (Tyler and Huo, 2002:188), so too does general strain theory focus on the implications of negative relations with
111 others. A propositional alignment is clearly eviden t, and the theory will likely benefit from the undertaking.
112 APPENDIX A INDIVIDUAL SURVEY ITEM S AND SCALE CONSTRUCTION Strain 1: Disjunction between expect ations and actual achievements. ( = .75) 1. You felt pushed around in the conference/court case by people with more power than you. ( r = .91) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 2. During the conference/court you felt pushed into things you did not agree with ( r = .88) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree Strain 2: Disjunction between aspirations and expectations. 1. During the conference/court case I felt that I had stuffed up at least some of my future opportunities. (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree Strain 3: Presentation of negative stimuli. ( = .74) 1. Were you treated in the conference/cour t as though you were likely to commit another offense? ( r = .66) (1) Not at all, (2) A little, (3) Somewhat, (4) A lot 2. Did people during the conference/court case make negative judgments about what kind of person you are? ( r = .71) (1) Not at all, (2) A little, (3) Somewhat, (4) A lot 3. During the conference/cour t case were you treated as t hough you were a criminal. ( r = .78) (1) Not at all, (2) A little, (3) Somewhat, (4) A lot 4. During the conference/court case did any of the people who are important to you reject you because of the offence? ( r = .53) (1) Not at all, (2) A little, (3) Somewhat, (4) A lot 5. During the conference/court case were you treated as thou gh you were a bad person? ( r = .82) (1) Not at all, (2) A little, (3) Somewhat, (4) A lot Strain 4: Removal of positive stimuli. ( = .68) 1. During the conference/court cas e I felt like I had lost respec t or honour among my family. ( r = .90) (1) Not at all, (2) A little, (3) Quite a bit, (4) A lot, (5) Felt overwhelmed by it 2. During the conference/court cas e I felt like I had lost respec t or honour among my friends. ( r = .85) (1) Not at all, (2) A little, (3) Quite a bit, (4) A lot, (5) Felt overwhelmed by it Negative affect 1: Anger 1. The conference/court case just made you angry. (1) Strongly Disagree, (2) Disagree, (3 ) Neither, (4) Agree, (5) Strongly Agree
113 Negative affect 2: Depression 1. During the conference/court case I felt sad or depressed. (1) Not at all, (2) A little, (3) Quite a bit, (4) A lot, (5) Felt overwhelmed by it Conventional social support. ( = .65) 1. Did people in the conference/court say that it was not like you to do something that was wrong? ( r = .59) (1) Not at all, (2) A little, (3) Somewhat, (4) A lot 2. Did you learn from the conference/court case that there are people w ho care about you? ( r = .69) (1) Not at all, (2) A little, (3) Somewhat, (4) A lot 3. During the conference/court case did people suggest that they loved you regardless of what you did? ( r = .72) (1) Not at all, (2) A little, (3) Somewhat, (4) A lot 4. During the conference/court case did people talk about aspects of yourself which they like? ( r = .73) (1) Not at all, (2) A little, (3) Somewhat, (4) A lot 5. Were you treated as a trustworthy person in the conference/court? ( r = .51) (1) Not at all, (2) A little, (3) Somewhat, (4) A lot Self-efficacy. ( = .58) 1. If I canâ€™t do a job the first time, I keep trying until I can. ( r = .82) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 2. When trying to learn something new, I soon give up if I am not initially successful (reversed). ( r = .85) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree Self-esteem. ( = .74) 1. I feel that Iâ€™m a person of worth, at least on an equal plan e with others. ( r = .75) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 2. I feel that I have a nu mber of good qualities. ( r = .72) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 3. I am able to do things as well as most other people. ( r = .61) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 4. I take a positive attitude towards myself. ( r = .72) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 5. On the whole, I am satisfied with myself. ( r = .69) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree
114 Cognitive coping strategy: avoidance. 1. During the conference/court case I felt so exposed, I wished I could just disappear. (1) Not at all, (2) A little, (3) Quite a bit, (4) A lot, (5) Felt overwhelmed by it Emotional coping strategy: revenge. 1. You wish that you could get back at the peop le who were accusing you in the conference/court case. (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree Violation of relevant justice rules. ( = .86) 1. You feel that people who have committed the same offense are treated the same way by conferences/courts. ( r = .47) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 2. People were polite to you in the court/conference. ( r = .62) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 3. You understood what was going on in the court/conference. ( r = .57) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 4. You understood what your ri ghts were during the proc essing of the case. ( r = .67) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 5. You felt you had the opportunity to express your views in the conference/court. ( r = .76) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 6. All sides got a fair chance to bring out the facts in the conference/court. ( r = .76) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 7. You felt you had enough control over the way thi ngs were run in the conference/court. ( r = .77) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 8. The conference/court took account of what you said in deciding what should be done. ( r = .71) (1) Strongly Disagree, (2) Disagree , (3) Neither, (4) Agree, (5) Strongly Agree 9. You feel you were treated with re spect in the conference/court. ( r = .73) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree 10. The police were fair duri ng the court/conference. ( r = .65) (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree Pressure/incentive to engage in criminal coping. 1. What do your close friends think about what you did? Do they think that the offence you committed was: (1) Totally wrong, (2) In some way wrong, (3) Neither right nor wrong, (4) In some way right, (5) Totally right
115 External attribution. 1. You feel that the people who accused you in the conference/court case were more wrong than you were. (1) Strongly Disagree, (2) Disagree, (3) Neither, (4) Agree, (5) Strongly Agree Outcome measure: delinquent adaptation ( = .73) 1. The conference/court case will help prevent you from breaking the law in the future. ( r = .91) (1) Strongly Agree, (2) Agree, (3) Neith er, (4) Disagree, (5 ) Strongly Disagree 2. What happened in the conferen ce/court will encourage you to obe y the law in the future. ( r = .87) (1) Strongly Agree, (2) Agree, (3 ) Neither, (4) Disagree, (5) Strongly Disagree Background/Control variables 1. Offender sex. Female (0), Male (1) 2. Offender age group. Non-juvenile (0), Juvenile (1) 3. Offender minority status (Aborig inal/Torres Strait Islander). No (0), Yes (1) 4. Crime type. Instrumental (0), Expressive (1) 5. Assigned treatment. Court (0), Conference (1) Note: A computation of the sum of individual su rvey items was used to construct the scales. Tests for unidimensionality indicate the loading on one factor for each scale. Item-to-scale correlations are presented in pa rentheses for each scale item.
116APPENDIX B CORRELATION MATRIX (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) Expect/achv (1) 1.00 Aspir/expct (2) -.05 1.00 Negat stim (3) .49 .03 1.00 Posit stim (4) .02 .56 .02 1.00 Anger (5) .55 .12 .38 .07 1.00 Depress (6) .01 . 45 .02 .46 .05 1.00 Justice (7) .65 .12 -.48 .19 -.52 .18 1.00 Soc supp (8) -.16 .08 .02 .11 -.14 .14 .19 1.00 Self effic (9) -.08 -.15 .05 -.16 -.10 .00 .05 .24 1.00 Self est (10) -.06 -.19 -.09 -.13 -.08 -.02 .04 .33 .42 1.00 Ext attr (11) .31 -.12 .20 -.25 .29 -.10 -.32 -.08 .01 .01 1.00 Del peer (12) .31 -.15 .25 -.29 .18 -.16 -.37 -.04 .18 . 22 .38 1.00 Cope:revenge(13) .40 -.16 .24 -.11 .41 -.24 -.37 -.19 -.15 -.04 .45 .24 1.00 Cope:avoid (14) .18 .35 .15 .34 .14 .55 -.04 .13 -.07 -.07 .04 -.03 -.02 1.00 Criminalty (15) .33 -.31 .30 -.29 .27 -.34 -.51 -.24 -.15 -.12 .27 .29 .39 -.22 1.00 Sex (16) .01 -.02 -.04 -.00 -.04 -.11 .11 -.03 .00 -.01 -.04 -.14 .00 -.19 -.08 1.00 Age grp (17) -.01 .07 .06 .01 .03 .11 .14 .11 -.10 -.06 -.10 -.16 .01 .08 .00 -.11 1.00 Minority (18) -.12 -.12 -.04 -.04 -.08 .01 .12 .07 .10 .01 -.08 -.11 -.06 -.07 .04 .01 .05 1.00 Crime type (19) .13 -.11 .10 -.22 .08 -.03 -.16 .04 .14 .12 .28 .33 .09 .03 .18 -.01 -.30 -.02 1.00 Treatment (20) -.12 -.11 -.05 .10 -.13 .17 .32 .23 .00 -.02 -.09 -.15 -.19 .03 -.07 .11 -.04 .07 .02 1.00 Note: ( p < .05) in bold (2-tailed)
117 LIST OF REFERENCES Agnew, R. (1980). A revised strain theory of delinquency (Doctoral diss ertation, University of North Carolina, 1980). Dissertation Abstracts International (UMI No. 8114781). Agnew, R. (1985). A revised st rain theory of delinquency. Social Forces, 64, 151-167. Agnew, R. (1992). Foundation for a general st rain theory of crime and delinquency. Criminology, 30, 47-87. Agnew, R. (1995). The contribution of social-psyc hological strain theory to the explanation of crime and delinquency. In F.Adler & W.S. Laufer (Series Eds.), The legacy of anomie theory : Vol. 6. Advances in criminological theory (pp. 113-137). New Brunswick, NJ: Transaction. Agnew, R. (1997). Stability and change in crime over the life course: A strain theory explanation. In T. P. Thornberry (Ed.), Advances in criminological theory : Vol 7. Developmental theories of crime and delinquency . New Brunswick, NJ: Transaction. Agnew, R. (1999). A general strain theory of community differences in crime rates. Journal of Research in Crime and Delinquency, 36, 123-155. Agnew, R. (2001). Building on the foundation of ge neral strain theory: Sp ecifying the types of strain most likely to lead to crime and delinquency. Journal of Research in Crime and Delinquency, 38, 319-361. Agnew, R. (2002). Experienced, vicarious, and anticipated strain: An exploratory study on physical victimization and delinquency. Justice Quarterly, 19, 603-632. Agnew, R. (2005). Why do criminals offend?: A general theory of crime and delinquency. Los Angeles: Roxbury Publishing Company. Agnew, R. (2006a). Pressured into crime: An ove rview of general strain theory. Los Angeles: Roxbury Publishing Company. Agnew, R. (2006b). Storylines as a neglected cause of crime. Journal of Research in Crime and Delinquency, 43, 119-147. Agnew, R., & Brezina, T. (1997) . Relational problems with peers, gender, and delinquency. Youth and Society, 29, 84-111. Agnew, R., Cullen, F. T., Burton, V. S., Evans, T. D., & Dunaway, R. G. (1996). A new test of classic strain theory. Justice Quarterly, 13, 681-704. Agnew, R., & White, H. R. (1992). An empi rical test of general strain theory. Criminology, 30, 475-499.
118 Agnew, R., Brezina, T., Wright, J. P., & Cullen, F. T. (2002). Strain, personality traits, and delinquency: Extending general strain theory. Criminology, 40, 43-71. Akers, R. L. (1973). Deviant behavior: A soci al learning approach. Belmont, CA: Wadsworth. Akers, R. L. (1985). Deviant behavior: A so cial learning approach (3rd ed) . Belmont, CA: Wadsworth. Akers, R. L. (1992). Linking sociology and its specialties. Social Forces, 71, 1-16. Akers, R. L., & Sellers, C. S. (2004). Criminological theories: Introduction, evaluation, and application (4th ed.). Los Angeles: Roxbury Publishing Company. Anderson, E. (1999). Code of the street. New York: Norton. Aseltine, R. H., Gore, S., & Gordon, J. (2000). Life stress, anger and anxiety, and delinquency: An empirical test of general strain theory. Journal of Health and Social Behavior, 41, 256-275. Averill, J. R. (1982). Anger and aggression. New York: Springer-Verlag. Avison, W. R., & Turner, R. J. (1988). Stre ssful life events and depressive symptoms: Disaggregating the effects of acute stressors and chronic strains. Journal of Health and Social Behavior , 29, 253-264 Bandura, A. (1983). Psychological mechanisms of aggression. In R. G. Green & E. I. Donnerstein (Eds.), Aggression: Theoretica l and empirical reviews : Vol. 1 (pp. 1-40). San Diego, CA: Academic Press. Bandura, A. (1997). Self-efficacy in changing societies. Cambridge, UK: Cambridge University Press. Bao, W., Haas, A., & Pi, Y. (2004). Life stra in, negative emotions, and delinquency: An empirical test of general strain theory in the Peopleâ€™s Republic of China. International Journal of Offender Therapy and Comparative Criminology, 48, 281-297. Baron, R. M., & Kenny, D. A. (1986). The moderato r-mediator variable di stinction in social psychological research: Conceptual, stra tegic, and statisti cal considerations. Journal of Personality and Social Psychology, 51, 1173-1182. Baron, S. W. (2004). General strai n, street youth and crime: A test of Agnewâ€™s revised theory. Criminology, 42, 457-483. Baron, S. W., & Hartnagel, T. F. (2002). Street youth and labor market strain. Journal of Criminal Justice, 30, 519-533.
119 Baron, S. W., Kennedy, L. W., & Forde, D. R. ( 2001). Male street youthsâ€™ conflict: The role of background, subcultural, a nd situational factors. Justice Quarterly, 18, 759-789. Bazemore, G., & Umbreit, M. (2001). A comparis on of four restorative conferencing models. OJJDP Juvenile Justice Bulletin, February. Beirne, P. (1997). Inventing criminology: Th e â€œscience of manâ€ in Cesare Beccariaâ€™s Dei delitti e delle pene (1764). In M. McShane & F.P. Williams III (Eds.), Criminological theory: Vol. 3. Contemporary literature in theory and practice (pp. 21-65). New York: Garland Publishing. Bernard, T. J. (1987). Testing structural strain theories. Journal of Research in Crime and Delinquency 24, 262-280. Bernard, T. J. (1990). Angry aggressi on among the â€œtruly disadvantaged.â€ Criminology, 28, 7396. Bernard, T. J. (1995). Merton vers us Hirschi: Who is faithful to Durkheimâ€™s heritage? In F.Adler & W.S. Laufer (Series Eds.), The legacy of anomie theory : Vol. 6. Advances in criminological theory (pp. 81-90). New Brunswick, NJ: Transaction. Beutel, A. M., & Marini, M. M. (1995). Gender and values. American Sociological Review, 60, 436-448. Billings, A. G., & Moos, R. H. (1984). Coping, stress, and social resources among adults with unipolar depression. Journal of Personality and Social Psychology, 46, 877-891. Bonta, J., Wallace-Capretta , S., & Rooney, J. (1998). Restorative justice: An evaluation of the restorative resolutions project. Ottawa, Canada: Solicitor General Canada. Braithwaite, J. (1989). Crime, shame, and reintegration. Cambridge, UK: Cambridge University Press. Braithwaite, J., & Braithwaite, V. (2001). Sham e, shame management and regulation. In E. Ahmed, N. Harris, J. Braithwa ite, & V. Braithwaite (Eds.), Shame management through reintegration (pp. 3-72). Cambridge, UK: Cambridge University Press. Braithwaite, J., & Mugford, S. (1994). Cond itions for a successful reintegration ceremony. British Journal of Criminology, 34, 139-171. Brezina, T. (1996). Adapting to strain: An examination of delinquent coping responses. Criminology, 34, 39-60. Brezina, T. (1998). Adolescent maltreatment a nd delinquency: The question of intervening processes. Journal of Research in Crime and Delinquency, 35, 71-99.
120 Brezina, T. (1999). Teenage violence toward pa rents as an adaptation to family strain. Youth and Society, 30, 416-444. Brezina, T. (2000). Delinquent problem-solving: An interpretive framework for criminological theory and research. Journal of Research in Crime and Delinquency, 37, 3-30. Briere, J. N. (1992). Child abuse trauma: Theory and treatm ent of the las ting effects. Newbury Park, CA: Sage Publications. Broidy, L. M. (2001). A test of general strain theory. Criminology, 39, 9-35. Broidy, L. M., & Agnew, R. (1997). Gender and crim e: A general strain th eory perspective. Journal of Research in Crime and Delinquency, 34, 275-306. Burton, V. S., Cullen, F. T., Evans, T. D., & Dunaway, R. G. (1994). Reconsidering strain theory: Operationalization, rival theories, and adult criminality. Journal of Quantitative Criminology, 10, 213-239. Bushway, S. D., Sweeten, G., & Wilson, D. B. ( 2006). Size matters: Standard errors in the application of null hypothesis significance testing in criminology and criminal justice. Journal of Experime ntal Criminology, 2, 1-22. Capowich, G. E., Mazerolle, P., & Piquero, A. (2001) . General strain theory, situational anger, and social networks: An assessment of conditioning influences. Journal of Criminal Justice, 29, 445-461. Cernkovich, S. A., & Giordano, P. C. (1992) . Delinquency, opportunity, and gender. Journal of Criminal Law and Criminology, 70 , 145-151. Cernkovich, S. A., Giordano, P. G., & Rudolph, J. L. (2000). Race, crime, and the American dream. Journal of Research in Crime and Delinquency, 37, 131-170. Christie, N. (1977). Conflicts as property. British Journal of Criminology, 17, 1-5. Clarke, R. V., & Cornish, D. B. (1985). M odeling offendersâ€™ decisions: A framework for research and policy. In M. Tonry & N. Morris (Eds.), Crime and justice: An annual review of research: Vol. 6. Chicago: University of Chicago Press. Clinard, M. B. (1964). Anomie and deviant behavior. New York: Free Press. Cloward, R. A., & Ohlin, L. E. (1960). Delinquency and opportunity: A theory of delinquent gangs. New York: Free Press. Cohen, A. K. (1955). Delinquent boys: The culture of the gang. Glencoe, IL: Free Press.
121 Cohen, S., & Hoberman, H. M. (1983). Positive ev ents and social supports as buffers of life change stress. Journal of Applied Social Psychology, 13, 99-125. Cohen, S., Kessler, R. C., & Gordon, L. U. (1995). Measuring stress. New York: Oxford University Press. Colvin, M. (2000). Crime and coercion: An integrated theory of chronic criminality. New York: St. Martinâ€™s Press. Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: De sign & analysis issues for field settings. Boston: Houghton Mifflin Company. Cornish, D., & Clarke, R. (1986). The reasoning criminal: Rati onal choice perspectives on offending. New York: Springer-Verlag. Cronkite, R., & Moos, R. H. (1984). The role of predisposing and moderating factors in the stress-illness relationship. Journal of Health and Social Behavior, 25, 372-393. Cullen, F. T. (1994). Social support as an organizing concept for criminology. Justice Quarterly, 11, 527-559. Cullen, F. T., Fisher, B. S., & Applegate, B. K. (2000). Public opinion about punishment and corrections. In M. Tonry (Ed.), Crime and justice: A review of research: Vol. 27 (pp. 179). Chicago: University of Chicago Press. Cunneen, C. (2002). Restorative justice: Emerging views. The drawing board: An Australian review of public affairs. Sydney, Australia: University of Sydney School of Economics and Political Science. Daly, K. (2001). Revisiting the relationship between retributive and restorative justice. In H. Strang & J. Braithwaite (Eds.) Restorative justice: Philosophy to practice (pp. 33-54). Aldershot, England: Dartmouth/Ashgate. Durkheim, E. (1951). Suicide. Glencoe, IL: The Free Press. Eitle, D. J. (2002). Exploring a source of devi ance-producing strain for females: Perceived discrimination and general strain theory. Journal of Criminal Justice, 30, 429-442. Eitle, D. J., & Turner, R. J. (2002). Exposure to community violence and young adult crime: The effects of witnessing violence, traumatic vict imization, and other stressful life events. Journal of Research in Crime and Delinquency, 39, 214-237. Eitle, D. J., & Turner, R. J. (2003). Stress exposure, race, and young adult male crime. The Sociological Quarterly, 44, 243-269.
122 Fagan, J. A., & Wexler, S. (1987). Fa mily origins of violent delinquents. Criminology, 25, 643669. Farnworth, M., & Leiber, M. J. (1989). Strain theory revisited: Ec onomic goals, educational means, and delinquency. American Sociological Review, 54, 263-274. Farrington, D. P., & Welsh, B. C. (2005). Rando mized experiments in criminology: What have we learned in the last two decades? Journal of Experime ntal Criminology, 1, 9-38. Felson, M. (1998). Crime and everyday life (2nd ed). Thousand Oaks, CA: Pine Forge Press. Fondacaro, M. R., & Moos, R. H. (1987). Social support and coping: A longitudinal analysis. American Journal of Community Psychology, 15, 653-673. Gottfredson, M. R., & Hirschi, T. (1990). A general theory of crime. Stanford, CA: Stanford University Press. Green, D. (1989). Measures of illegal behavior in individual level deterrence research. Journal of Research in Crime and Delinquency, 26, 253-75. Hagan, J., & McCarthy, B. (1992). Streetlife and delinquency. British Journal of Sociology, 43, 533-561. Hagan, J., & McCarthy, B. (1997). Mean streets: Youth crime and homelessness. Cambridge, UK: Cambridge University Press. Harris, N. (2001). Shaming and shame: Regulatin g drink-driving. In E. Ahmed, N. Harris, J. Braithwaite, & V. Braithwaite (Eds.), Shame management through reintegration (pp. 73207). Cambridge, UK: Cambridge University Press. Hay, C. (2003). Family strain, gender, and delinquency. Sociological Perspectives, 46, 107-135. Heimer, K. (2000). Changes in the gender gap in crime and womenâ€™s economic marginalization. In E. Jefferis & R. Titus (Eds.), The nature of crime: Contin uity and change, criminal justice 2000 (Vol. 1). Washington, DC: Na tional Institute of Justice. Henry, B., Moffitt, T. E., Caspi, A., Langley, J ., et al. (1994). On the â€˜remembrance of things pastâ€™: A longitudinal evaluati on of the retrospective method. Psychological Assessment, 6, 92-101. Hirschi, T. (1969). Causes of delinquency. Berkeley, CA: University of California Press. Hirschi, T., & Gottfredson, M. (1986). The distin ction between crime and criminality. In T. F. Hartnagel & R. A. Silverman (Eds), Critique and explanation. New Brunswick, NJ: Transaction Books.
123 Hoffmann, J. P. (2002). A contextual analysis of differential associati on, social control, and strain theories of delinquency. Social Forces, 81, 753-785. Hoffmann, J. P., & Cerbone, F. G. (1999). Stress ful life events and deli nquency escalation in early adolescence. Criminology, 37, 343-374. Hoffmann, J. P., & Ireland, T. O. (2004). Strain and opportunity structures. Journal of Quantitative Criminology, 20, 263-292. Hoffmann, J. P., & Su, S. S. (1997). The conditi onal effects of stress on delinquency and drug use: A strain theory assessment of sex differences. Journal of Research in Crime and Delinquency, 34, 46-78. Holahan, C. J., & Moos, R. H. (1991). Life st ressors, personal and social resources, and depression: A four-year structural model. Journal of Abnormal Psychology, 100, 31-38. Hudson, J., Galaway, A., Morris, X., & Maxwell, G. (1996). Family group conferences: Perspectives on policy and practice. Leichhardt, PA: The Federation Press. Jang, S. J., & Johnson, B. R. (2003). Strain, negative emotions, and deviant coping among African Americans: A test of general strain theory. Journal of Quantitative Criminology, 19, 79-105. Jensen, G. F. (1995). Salvaging structure through strain: A theoretical and empirical critique. In F. Adler & W. S. Laufer (Series Eds.), The legacy of anomie theory : Vol. 6. Advances in criminological theory (pp. 139-158). New Brunswick, NJ: Transaction. Judge, T. A., Erez, A., Bono, J. E., & Thorese n, C. J. (2002). Are measures of self-esteem, neuroticism, locus of control, and genera lized self-efficacy indicators of a common core construct? Journal of Personality and Social Psychology, 83, 693. Kaplan, H. B. (1996). Psychosocial stress from th e perspective of self th eory. In H. Kaplan (Ed.), Psychosocial stress: Perspectives on stru cture, theory, life-course, and methods (pp. 175-244). San Diego, CA: Academic Press. Katz, J. (1988). Seductions of crime. New York: Basic Books. Kaufman, J. M., Rebellon, C., Thaxton, S., & Agnew, R. (2005). A general strain theory of the race-crime relationship. In P. Mazerolle & R. Agnew (Eds.), General strain theory: Essential readings. Los Angeles: Wadsworth. Kornhauser, R. R. (1978). Social sources of delinquency. Chicago: University of Chicago Press. Kurki, L. (2000). Restorative and comm unity Justice. In M. Tonry (Ed.), Crime and justice: A review of research (pp. 235-303). Chicago: Univer sity of Chicago Press.
124 Landau, S. F. (1997). Crime patterns and their re lation to subjective soci al stress and support indicators: The role of gender. Journal of Quantitative Criminology, 13, 29-56. Landau, S. F. (1998). Crime, subjective social stress, support indicators , and ethnic origin: The Israeli experience. Justice Quarterly, 15, 243-272. Latimer, J., Dowden, C., & Muise, D. (2001). The effectiveness of restor ative justice practices: A meta-analysis. Research and Statistics Divisi on Methodological Series. Canada: Department of Justice. Lauritsen, J. L., Sampson, R. J., & Laub, J. H. (1991). The link between offending and victimization among adolescents. Criminology, 29, 265-292. Lemert, E. M. (1967). Human deviance, social probl ems, and social control. Englewood Cliffs, NJ: Prentice Hall. Lerner M. J. (1987). Integrating societal and psychological rules of entitlement: The basic task of each social actor and a fundament al task for the social sciences . Social Justice Research, 1 , 107. Lind E. A., & Tyler T. R. (1988). The social psychology of procedural justice. New York: Plenum. Liska, A. (1971). Aspirations, expectations, an d delinquency: Stress an d additive models. Sociological Quarterly, 12, 99-107. Loeber, R., & Stouthamer-Loeber, M. (1986). Family factors as correlates and predictors of juvenile conduct problems and delinque ncy. In M. Tonry & N. Morris (Eds.), Crime and justice: An annual review of research : Vol. 7 (pp 29-149). Ch icago: University of Chicago Press. Makkai, T., & Braithwaite, J. (1994). Reintegr ative shaming and compliance with regulatory standards. Criminology, 32, 361-83. Massey, D. S. (2002). A brief histor y of human society: The origin and role of emotion in social life. American Sociological Review, 67, 1-29. Matsueda, R. L., Drakulich, K., & Kubrin, C. E. (2006). Race and neighborhood codes of violence. In R. D. Peterson, L. J. Krivo, & J. Hagan (Eds.), The many colors of crime: Inequalities of race, ethn icity, and crime in America (pp. 334-356) . New York: New York University Press. Matt, G. E., & Dean, A. (1993). Social support from friends and psychological distress among elderly persons: Moderator effects of age. Journal of Health and Social Behavior, 34, 187-200.
125 Mattin, J. A., Wethington, E., & Kessler, R. C. (1990). Situational determinants of coping and coping effectiveness. Journal of Health and Social Behavior, 31, 103-122. Mazerolle, P. (1998). Gender, general strain and delinquency: An empirical examination. Justice Quarterly, 15, 65-92. Mazerolle, P., Brame, R., Paternoster, R., Piquero, A., & Dean, C. (2000). Onset age, persistence, and offending versat ility: Comparisons across gender. Criminology, 38, 1143-1172. Mazerolle, P., & Maahs, J. (2000). General strain and delinquency: An alternative examination of conditioning influences. Justice Quarterly, 17, 753-778. Mazerolle, P., & Piquero, A. (1997). Violent respons es to strain: An examination of conditioning influences. Violence and Victims, 12, 323-343. Mazerolle, P., & Piquero, A. (1998). Linking exposure to strain with anger: An investigation of deviant adaptations. Journal of Criminal Justice, 26, 195-211. Mazerolle, P., Piquero, A., & Capowich, G. E. (2003). Examining the links between strain, situational and dispositiona l anger, and crime: Further specifying and testing general strain theory. Youth and Society, 35, 131-157. Melville, H. (1853). Bartleby, the sc rivener: A story of wall-street. Putnamâ€™s Monthly , 546557; 609-615. Merton, R. K. (1938). Social structure and anomie. American Sociological Review, 3, 672-682. Merton, R. K. (1968). Social theory and social structure. New York: Free Press. Merton, R. K. (1995). Opportunity structure: The emergence, di ffusion and differentiation of a sociological concept. In F. Ad ler & W. S. Laufer (Series Eds.), The legacy of anomie theory : Vol. 6. Advances in criminological theory (pp. 3-78). New Brunswick, NJ: Transaction. Mesch, G. S., Fishman, G., & Eisikovitz, Z. (200 3). Attitudes supporting vi olence and aggressive behavior among adolescents in Israel : The role of family and peers. Journal of Interpersonal Violence, 18, 1132-1148. Messner, S. F. (1988). Mertonâ€™s social st ructure and anomie: The road not taken. Deviant Behavior, 9, 33-53. Messner, S. F., & Rosenfeld, R. (2001). Crime and the American dream. Belmont, CA: Wadsworth.
126 Mikula, G. (1993). On the experience of injustice. European Journal of Social Psychology, 4, 223-244. Mirowsky, J., & Ross, C. E. (1989). Social causes of psychological distress. New York: Aldine de Gruyter. Moffitt, T. E. (1993). Adolescence limited and li fe course persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-701. Moffitt, T. E. (2003). Life-course persistent and adolescence-limited antisocial behavior: A 10year research review and a research agenda. In B. Lahey, T. Moffitt & A.Caspi (Eds.). The causes of conduct disorder and serious juvenile delinquency. NY: Guilford. Moore, D. B., & O'Connell, T. (1994). Family conferencing in Wagga Wagga: A communitarian model of justice. In C. Alder & J. Wundersitz (Eds.), Family conferencing and juvenile justice: The way forward or misplaced optimism? (pp. 15-44). Canberra: Australian Institute of Criminology. Morris, A., & Maxwell, G. (1998). Restora tive justice in New Z ealand: Family group conferences as a case study. Western Criminology Review, 1. Available at http://wcr.sonoma.edu/v1n1/morris.html (accessed August, 2006). Nicholas, K. B., & Bieber, S. L. (1996). Parent al abusive versus supporti ve behaviors and their relation to hostility and aggression in young adults. Child Abuse and Neglect 20, 11951211. Novaco, R. W. (1976). The functions and regulation of the arousal of anger. American Journal of Psychiatry, 133, 1124-1128. Office of Juvenile Justice and Delinquency Prevention (1998). Guide for implementing the balanced and restorative justice model. Washington, DC: U.S. Department of Justice. Passas, N. (1995). Continuities in the anomie traditi on. In F. Adler & W. S. Laufer (Series Eds.), The legacy of anomie theory : Vol. 6. Advances in criminological theory (pp. 91-112). New Brunswick, NJ: Transaction. Paternoster, R., Brame, R., & Bachman, R. (1997) . Do fair procedures matter? The effect of procedural justice on spouse assault. Law & Society Review, 31, 163-204. Paternoster, R., & Mazerolle, P. (1994). General strain theory and delinquency: A replication and extension. Journal of Research in Crime and Delinquency, 31, 235-263. Paternoster, R., & Simpson, S. S. (1996). Sanction threats and appeals to morality: Testing a rational choice model of corporate crime. Law & Society Review, 30, 549-583.
127 Patterson, G. R. (1975). Families: Applications of soci al learning to family life. Champaign, IL: Research Press. Pinhey, T. K., & Okinaka, A. M. (2004). Expl oring the purging behavi or of Asian-Pacific adolescents in Guam: Does heavy te levision viewing make a difference? Deviant Behavior, 25, 27-41. Piquero, N. L., & Sealock, M. D. (2000). Generali zing general strain theory: An examination of an offending population. Justice Quarterly, 17, 449-484. Piquero, N. L., & Sealock, M. D. (2004). Gender and general strain theory: A preliminary test of Broidy and Agnewâ€™s gender/GST hypotheses. Justice Quarterly, 21, 125-158. Quinney, R. (1970). The social reality of crime. Boston: Little, Brown. Rice, S. K., & Piquero, A. R. (2005). Percepti ons of discrimination and justice in New York City. Policing: An International Journal of Police Strategies & Management, 28, 98117. Robertson, J. (1996). Research on family group conf erences in child welfare in New Zealand. In J. Hudson, B. Galaway, A. Morris, & G. Maxwell (Eds.), Family group conferences: Perspectives on policy and practice (pp. 49-64). Leichhardt, PA : The Federation Press. Rossi, P. H., Lipsey, M. W., & Freeman, H. E. (2004). Evaluation: A systematic approach. Thousand Oaks, CA: Sage Publications. Sampson, E. E. (1983). Justice and the critique of pure psychology. New York: Plenum. Sampson, R. J., & Bartusch, D. J. (1998). Le gal cynicism and (subcul tural?) tolerance of deviance: The neighborhood context of racial differences. Law and Society Review, 32, 777-804. Sampson, R. J., & Laub, J. H. (1993). Crime in the making: Pathways and turning points through life. Cambridge, MA: Harvard University Press. Sampson, R. J., & Lauritsen, J. (1993). Viol ent victimization and offending: Individual-, situational-, and community-level risk factors. In A.J. Reiss Jr. (Ed.), Understanding and preventing violence: Vol. 3. Washington, DC: National Research Council. Sarat, A. (2001). When the state kills: C apital punishment and th e American condition. Princeton, NJ: Princeton University Press. Schoepfer, A., & Piquero, A. R. (2006). Self-con trol, moral beliefs, and criminal activity. Deviant Behavior, 27, 51-71.
128 Sharp, S. F., Terling-Watt, T. L., Atkins, L. A ., & Gilliam, J. T. (2001). Purging behavior in a sample of college females: A research note on general strain theory and female deviance. Deviant Behavior: An Inte rdisciplinary Journal, 22, 171-188. Sherman, L. W. (1993). Defiance, deterrence, and i rrelevance: A theory of the criminal sanction. Journal of Research in Crime and Delinquency 30 , 445-473. Sherman, L. W. (2003). Reason for emotion: Reinve nting justice with theo ries, innovations and research. 2002 ASC Pr esidential address. Criminology, 41, 1-38. Sherman, L. W., & Barnes, G. C. (1997). Restora tive justice and offenders â€™ respect for the law. Canberra RISE Working Papers. Canberra, Australia: Australian National University (April). Sherman, L. W., Braithwaite, J., St rang, H., & Barnes, G. C. (2001). Reintegrative shaming experiments (RISE) in Australia, 1995-1999. Ann Arbor, MI: Inter-university Consortium for Political and Social Research (ICPSR 2993). Sherman, L. W., & Strang, H., (1997). The righ t kind of shame for crime prevention. Canberra RISE Working Papers. Canberra, Australia: Australian National University (April). Sherman, L. W., & Strang, H., ( 2004). Restorative justice: What we know and how we know it. Jerry Lee Program on Randomized Controlled Trials in Restorative Justice, Working Paper #1. Philadelphia, PA: University of Pennsylvania. Sherman, L. W., Strang, H., Ange l, C., Woods, D., Barnes, G. C., Bennett, S., & Inkpen, N. (2005). Effects of face-to-face restorative justice on victims of crime in four randomized, controlled trials. Journal of Experimental Criminology, 1, 367-395. Sherman, L. W., Strang, H., & Woods D. J. (2000). Recidivism patterns in the Canberra reintegrative shaming experiments (RISE). Centre for Restorative Justice, Research School of Social Sciences. Canberra, Au stralia: Australian National University. Simons, R. L., Chen, Y. F., Stew art, E. A., & Brody, G. H. (2003) . Incidents of discrimination and risk for delinquency: A longitudinal test of strain theory with an African American sample. Justice Quarterly, 20, 827-854. Slocum, L. A., Simpson, S. S., & Smith, D. A. (2005). Strained lives and crime: Examining intra-individual variation in strain and offending in a sample of incarcerated women. Criminology, 43, 1067-1110. Smith, C. A., Wallston, K. A. (1992). Adaptation in patients with chronic rheumatoid arthritis: Application of a general model. Health Psychology, 11, 151-162. Strang, H. (2004). Repair or revenge: Victim s and restorative justice. New York: Oxford University Press.
129 Sykes, G. M., & Matza, D. (1957). Techniques of neutralization: A theory of delinquency. American Sociological Review, 22, 664-670. Taylor, S. E., & Aspinwall, L. G. (1996). Me diating and moderating processes in psychosocial stress: Appraisal, coping, resistance, a nd vulnerability. In H. Kaplan (Ed.), Psychosocial stress: Perspectives on structur e, theory, life-course, and methods (pp. 71-110). San Diego, CA: Academic Press. Tedeschi, J. T., & Felson, R. B. (1994). Violence, aggression, and coercive actions. Washington, DC: American Ps ychological Association. Thibaut, J. W., & Walker, L. (1975). Procedural justice: A psychological analysis. Hillsdale, NJ: Erlbaum. Thoits, P. (1983). Dimensions of life events th at influence psychological distress: An evaluation and synthesis of the literature. In H. B. Kaplan (Ed.), Psychosocial stress: Trends in theory and research. New York: Academic Press. Thornberry, T. P. (1987). Toward an interactional theory of delinquency. Criminology, 25, 863891. Thornberry, T. P., Ireland, T. O., & Smith, C. A. (2001). The importance of timing: The varying impact of childhood and adolescent maltr eatment on multiple problem outcomes. Development and Psychopathology, 13, 957-979. Tittle, C. (1980). Sanctions and social deviance: The question of deterrence. New York: Praeger. Tittle, C. (1995). Control balance: Towards a gene ral theory of deviance. Boulder, CO: Westview Press. Toby, J. (1957). Social disorganizat ion and stake in conformity. Journal of Criminal Law and Criminology, 48 , 12-17. Tosouni, A. (2004). Reintegrativ e shaming among youthful offende rs: Testing the theory through a secondary data analysis. Long Beac h, CA: California State University. Tyler, T. (1990). Why people obey the law. New Haven, CT: Yale University Press. Tyler, T. (2003). Procedural just ice, legitimacy, and the effective rule of law. In M. Tonry (Ed.), Crime and justice: A review of research (pp. 283-358). Chicago: University of Chicago Press. Tyler, T., & Huo, Y. J. (2002). Trust in the law: Encouraging public cooperation with the police and courts. New York: Russell Sage Foundation.
130 Umbreit, M. (1994). Victim meets offender: The impact of restorative justice and mediation. Monsey, NY: Criminal Justice Press. Umbreit, M., & Bradshaw, W. (1997). Victim expe rience of meeting adult vs. juvenile offenders: A cross-national comparison. Federal Probation, 61, 33-39. Vold, G. B., Bernard, T. J., & Snipes, J. B. (2002). Theoretical criminology (5th ed.). New York: Oxford University Press. Wareham, J., Cochran, J. K., Dembo, R., & Se llers, C. S. (2005). Community, strain, and delinquency: A test of a multi-level model of general strain theory. Western Criminology Review, 6, http://wcr.sonoma.edu/v6n1/wareham.htm (accessed August, 2006). Warner, C. T. (1986). Anger and sim ilar delusions. In R. Harre (Ed.), Social construction of emotions (pp. 135-165). Cambridge, MA: Basil Blackwell. Warr, M. (2002). Companions in crime: The social aspects of criminal conduct. Cambridge: Cambridge University Press. Weatherburn, D., & Lind, B. (1997). Social and economic stress, child neglect, and juvenile delinquency. New South Wales Bureau of Crime Statistics and Research. Sydney, Australia. Weisburd, D., Lum, C. M., & Petrosino, A. (2001) . Does research design affect study outcomes in criminal justice? The Annals of the American Acade my of Political and Social Science, 578, 50-70. Wheaton, B., Roszell, P., & Hall, K. (1997). The impact of twenty childhood and adult traumatic stressors on the risk of psychi atric disorder. In I. H. Gottlib (Ed.), Stress and adversity over the lif e course: Trajectories and turning points. New York: Cambridge University Press. Wright, J. P., Cullen, F. T., Agnew, R. S., & Br ezina, T. (2001). â€œThe root of all evil?â€ An exploratory study of money a nd delinquent involvement. Justice Quarterly, 18, 239-268. Zehr, H. (1990). Changing lenses: A new focus for crime and justice. Scottdale, PA: Herald Press.
131 BIOGRAPHICAL SKETCH Stephen K. Rice was born on July 2, 1967, in New York City. Growing up in New York City and Daytona Beach, Florida, he graduate d from Spruce Creek High School in 1985. He earned a B.A. in English from the University of Florida in 1989, an M.S. in criminology and criminal justice from Florida State Univers ity in 1993, and an M.A. in sociology from the University of Florida in 2003. Stephen is an Assistant Profe ssor at John Jay College of Criminal Justice in New York City. His research interests in clude criminological theory and th e sociology/social psychology of punishment. His scholarship ha s appeared in the journals Criminology, Journal of Criminal Justice, Journal of Criminal Justice Education, Policing, Journal of Ethnicity in Criminal Justice, Australian and New Zealand Journal of Criminology, and the Brill series Race and Ethnicity: Across Time, Space and Discipline. Stephen has been married to Karen Collins Rice for 16 years. They have two sons: Connor, age 9, and Brogan, age 2.