|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
The Conditional Influence of Criminological Constructs on Juvenile Delinquency: An Examination of the Moderating Effects of Self-Control by Angela Yarbrough A thesis submitted in partial fulfillment of the requirements for the degree of Master of Arts Department of Criminology College of Arts and Sciences University of South Florida Co-Major Professor: Shayne Jones, Ph.D. Co-Major Professor: Christine Sellers, Ph.D. John Cochran, Ph.D. Date of Approval: April 4, 2007 Keywords: social learning, control, deterrence, strain, general theory of cri me, interactive effects Copyright 2007, Angela Yarbrough
Acknowledgements I wish to thank several people for their invaluable assistance throughout my short time here at USF. First and foremost, I wish to thank my committee members. Dr. Shayne Jones, my major professor, I admire and respect you beyond what words could describe. As my mentor, you have helped me grow in many ways. Your expansive knowledge of criminology, psychology, and, most notably, life has forced me to see from perspectives other than my own. For that, I am forever indebted. Without your guida nce and generous patience, this thesis would not have been completed. You accept nothing but the best and while my creativity is often lacking, I hope I have provided that. On a less serious note, I will be contacting you on April 2, 2012 as this is one wager I will not lose. Additionally, Drs. Christine Sellers and John Cochran, you both have graciously provided me with an endless amount of knowledge. The confidence and intelligence that you exude can be quite intimidating, but ultimately has been a great resource for m e these past two years. Thank you both for your time and attention. Lastly, my family has always been there to push me forward. Mom, Daddy, Melanie and William, I appreciate the continuous support you all have provided me up to this point. God knows Ive had my moments and so I am very grateful for everything that you all have done for me
i Table of Contents List of Tables ii List of Figures iii Abstract iv Chapter 1 Introduction 1 Chapter 2 Literature Review 5 Social Learning Theory 5 Control Theory 8 Deterrence Theory 12 General Strain Theory 16 The General Theory of Crime 20 Summary 22 Integrated Prospective 22 Chapter 3 Methodology 30 Sample 30 Dependent Variable 31 Independent Variables 33 Social Learning Theory 33 Control Theory 35 Deterrence Theory 38 Strain Theory 39 The General Theory of Crime 40 Analytic Plan 42 Chapter 4 Results 44 Bivariate Statistics 44 Multivariate Statistics 45 Social Learning 46 Control 50 Perceptual Deterrence 53 Strain 55 Chapter 5 Discussion and Conclusion 57 References 64
iii List of Tables Table 1: Descriptive Statistics of Demographics and Theoretical Constructs 31 Table 2: Items and Factor Loadings for Measures of Self-Reported De linquency 32 Table 3: and Factor Loadings for Social Learning Measures of Definiti ons 34 Table 4: Items and Factor Loadings for Social Learning Measures of Differential Associations 34 Table 5: Items and Factor Loadings for Measures of Reinforcements 35 Table 6: Items and Factor Loadings for Control Measures of Parental Cont rol 36 Table 7: Items and Factor Loadings for Control Measures of Maternal A ttachment 37 Table 8: Items and Factor Loadings for Control Measures of Paternal A ttachment 38 Table 9: Items and Factor Loadings for Measures of Perceptual Dete rrence 39 Table 10: Items and Factor Loadings for Measures of Subjective Strains 40 Table 11: Items and Factor Loadings for Attitudinal and Behavioral Measure s of SelfControl 40 Table 12: Pearsons Zero-Order Correlations 45 Table 13: OLS Regression Results for Social Learning Variables, Self-Con trol, and Interactions 47 Table 14: OLS Regression Results for Control, Self-Control, and Interactions 51 Table 15: OLS Regression Results for Perceptual Deterrence, Self-Control and Interaction 54 Table 16: OLS Regression Results for Subjective Strains, Self-Control, and Inte raction 56
iii List of Figures Figure 1: Effects of Peers on Delinquency by Levels of Self-Control 48 Figure 2: Effects of Definitions on Delinquency by Levels of Self-Control 49 Figure 3: Effects of Reinforcements on Delinquency by Levels of Self-Cont rol 50 Figure 4: Effects of Parental Control on Delinquency by Levels of Self-Control 52 Figure 5: Effects of Maternal Attachment on Delinquency by Levels of Self -Control 53 Figure 6: Effects of Perceptual Deterrence on Delinquency by Levels of Se lf-Control 55 Figure 7: Effects of Subjective Strain on Delinquency by Levels of Self-Contr ol 56
iv The Conditional Influence of Criminological Constructs on Juvenile Delinquency: An Examination of the Moderating Effects of Self-Control Angela Yarbrough ABSTRACT Self-control and various elements comprising this construct have received much credit over the years as it has been able to account for a large amount of variance i n delinquency rates. Some research has suggested that individual difference fact ors (e.g., self-control) can overwhelm external factors (e.g., neighborhoods; see Loebe r & Wikstrm, 2000). Others have found that social influences (e.g., employment; see Wright, et al, 2001) have more pronounced effects for those most at-risk. Because of the equivocal nature of the empirical findings, this study seeks to replicate and ext end previous efforts. Specifically, the influence of constructs derived from soci al learning, control, deterrence, and strain are examined to see if any vary in their influe nce on adolescent offending as a function of self-control. Results indicate that all of these theoretical constructs (with the exception of paternal attachment) played a more important role among those who evinced the highest levels of self-control. Implica tions for criminological theory and criminal justice policy are discussed.
1 Chapter One Introduction While criminological theories have been heavily influenced by sociology, other disciplines, most notably psychology, have made significant impacts. Regardles s of whether a criminological theory uses group phenomena or individual characteristi cs to account for crime, all criminological theories are attempting to expla in why people commit criminal or deviant acts. One of the more recent criminological theorie s, the general theory of crime (GTC; Gottfredson & Hirschi, 1990), posits that the conce pt of self-control is the single best predictor of crime. Self-control refers to ones ability to consider the long-term negative consequences of antisocial behavior (Hirschi, 2004). Considerable attention to this theory has resulted in impressive empirical support for its direct effects on delinquency, yet various external factors (e.g., peers) s till appear to have significant additive effects (see Pratt and Cullen for meta-analysi s, 2000). In contrast to relying upon singular perspectives, such as that employed in the GTC, others suggest incorporating multiple theories simultaneously (Messner, Krohn, and Liska, 1989). This approach has led to more recent integrated theories that include self-control. Specifically, researchers are investigating the extent t o which self-control (or similar constructs, such as impulsivity) moderates the relationship between various external factors and delinquency. While it is only recently that criminolog ists have begun to examine how criminal behavior is the result of this interaction (Lynam, Wikst rm, Caspi, Moffitt, Loeber, & Novak, 2000; Wikstrm & Loeber, 2000; Wright, Caspi, Moffitt, & Silva, 2001; Wikstrm & Sampson, 2003), disciplines outside of criminology
2 have long recognized the interaction between the person and the environment (see Le win, 1935; Magnusson, 1988). The few studies examining the moderating effects of self-control have found divergent results. Some studies have found that external factors, such as neighborhood context, will be more influential for those with higher levels of sel fcontrol (Wikstrm & Loeber, 2000). In other words, some have suggested that external influences are simply inconsequential for those with low self-control (see als o Gottfredson & Hirschi, 1990); therefore, those with some level of self-control will be more influenced by external factors. In contrast, others have found that social infl uences (e.g., delinquent peers) have pronounced effects for those most at-risk. For instance, Wright and colleagues (2001) propose that individuals with low self-control have an elevated tendency towards deviant behavior and thus negative social influences ser ve only to exacerbate that tendency. In comparison, those with high self-control are a ble to resist the temptation of these negative social influences. Overall, classic criminological theories (e.g., social learning theory general strain theory) have attained impressive empirical support (with the exception of deterr ence theory, see Cullen, Wright, and Blevins, 2006 for review of the literature). Yet, some have argued that taking the most empirically supported components of classic theor ies and integrating them may yield theories with greater explanatory power (se e Messner et al., 1989 for overview). Some of the most recent research in this area has focused on the interactive nature of criminological constructs instead of simply viewing the theories in an additive fashion (e.g., Evans, Cullen, Burton, Dunaway, & Benson, 1997; Agnew, 2006; Pratt, Cullen, Blevins, Daigle, & Madensen, 2006). Specifically, it has been
3 suggested that certain individual characteristics could potentially moderat e the effects of factors such as those found in strain, control, and social learning (see Agnew, 2006). The proposed interactive relationships could have dramatic implications for criminological theory and just as importantly, the criminal justice sys tem. In light of the recent research that has suggested that some individuals may not be as influenc ed as others by factors outside themselves, current criminological theories ma y need to reevaluate the predictive power of their constructs. More practically, chan ges to the criminal justice system may need to be implemented to accommodate individual differences in offenders. As such, policies targeted towards those offenders who l ack selfcontrol may need to focus efforts on social factors that may not be as influenti al for those offenders exhibiting higher levels of self-control and vice-versa. Yet, as al luded to above, it remains unclear (1) how robust the interactions are and (2) in what direction t hey exert their effects. The purpose of this study is to provide further empirical guidance on both of these issues. The current analysis uses a sample of middle and high school students from Largo, Florida in an effort to replicate and extend previous findings that have invest igated the interaction between self-control and several criminological constructs. S pecifically, constructs derived from social learning, strain, deterrence, and control are examined to determine how, if at all, each varies in its influence on delinquency as a function of selfcontrol. Few studies to date have used a single sample to simultaneously examine interactions between such a full array of theoretical constructs (see Wright et al., 2001 for example). This is important in the sense that the discrepancies that exist m ay be due to methodological factors (e.g., different measurement of self-control, sample
4 idiosyncrasies). Thus, this study can supplement a relatively small literat ure that has investigated the interactive effects described above and provide direction for futur e research.
5 Chapter Two Literature Review Criminologists have long sought to find the most parsimonious way to explain a broad range of criminal behavior. This has led to the development of numerous theories with distinct and often incompatible assumptions to explain this type of behavior. These theoretical explanations range from solely sociological factors to individual characteristics, to the more recent interactive effects between the tw o. Before describing these potential interactive relationships, several theories will be discuss ed to show the important contribution each has made on its own. Social Learning Theory Akers social learning theory is one of the best known criminological theories a nd has received substantial support over the years (Akers, Krohn, Lanza-Kaduce, & Radosevich, 1979; Akers & Cochran, 1985; Akers, Greca, Cochran, & Sellers, 1989; Winfree, Backstrom, & Mays, 1994; Lee, Akers, & Borg, 2004). Burgess and Akers (1966; Akers, 1973) reformulated Sutherlands theory of differential associati on, renaming theirs social learning theory, in an attempt to explain criminal b ehavior by focusing on the concepts of differential associations, definitions, differential reinforcements, and imitation. The balance of these factors will determine w hether behavior is conforming or nonconforming (Akers, 1998). Perhaps the most important component of social learning theory, differential associations refer to interactions with different groups, focusing specifica lly on how antisocial individuals associate with one another more so than with prosocial individua ls
6 (Akers, 1985). Differential associations occur first and provide the context for t he formulation of definitions, exposure to reinforcements, and models to imitate (Akers et al., 1979). The rewards and/or punishments of established associations will influence the ability to create new associations and maintain old ones (Akers, 1998). The groups with whom one differentially associates include, but are not limited to, peers, family neighbors, church, schools, and teachers (Akers, 1985). The frequency, duration, priority, and intensity of each association determine its strength (Akers, 1998). That is, t hose associations that happen most often, last the longest, begin earlier, and involve those w ith whom one is closest will have the strongest influence on behavior. It has been sugges ted that the number of delinquent friends with whom one associates is the best predictor of delinquency (Akers et al., 1979; Warr, 2002; Akers & Jensen, 2006). Additionally, Warr (2002) concluded, based on evidence from empirical studies examining peers and delinquency, every study to date has found a significant relationship between the two. As mentioned above, differential associations provide the context for imitation. That is, the actions of those with whom one differentially associates are often m imicked. Akers (2001) posits that behavior is shaped by principles of modeling similar to those discussed by Bandura (1969). However, while imitation is potentially important for explaining the initiation or onset of delinquent behavior, its significance in maint aining or discontinuing a behavior is much less (Akers et al., 1979; Akers & Jensen, 2006). Akers and colleagues (1979) found that the imitation variables explained the least amount of variance in adolescent drinking and drug use. Additionally, Krohn and colleagues (1985) found no explanatory value of imitation variables with longitudinal data. Given the less
7 influential role of imitation, several recent studies have not even included it in tes ts of social learning theory (see Akers & Jensen, 2006 for overview). In contrast to imitation, the concept of definitions has found strong empirical support. Definitions consist of norms, attitudes, or orientations that are learned throug h imitation and reinforced through rewards and punishments by those with whom one differentially associates (Akers et al., 1979). Exposure to and reinforcemen t by others shared definitions generally lead one to accept those definitions as their own (Ake rs, 1998). Behavior is viewed as right or wrong, good or bad, and reasonable or unreasonable based on these norms, attitudes, or orientations (Akers, 1998). Definitions such as moral, religious, or conventional values can lead one to conform to the law and/or oppose delinquent acts. These negative definitions of crime decrease the likelihood of delinquency, while positive or neutral definitions toward crime increase the likeli hood of delinquent behavior (Akers & Jensen, 2006). As previously discussed, it is through rewards and punishments that definitions are fully formed and acceptable behavior is determined (Akers, 1998). In other words, t he consequences of current behavior (either by oneself or by witnessing anothers) m old future behavior. Unlike Sutherlands theory of differential association, Akers social learning theory applies Skinners (1953) idea of operant conditioning (herein refe rred to as reinforcements) to criminal behavior. These reinforcements might be posi tive, that is the consequence of the behavior is rewarding or pleasurable, or the reinforceme nts could be negative, or in other words avoids punishment or pain (Akers, 1985). Both negative and positive reinforcements help maintain or increase a behavior (Akers et al., 1979). O n the other hand, behavior can be reduced through punishments, which can be either
8 negative (e.g. removal of pleasure) or positive (e.g. presence of pain; Akers, 1998). Although the most important reinforcements are social (e.g. peers), nonsocial reinforcements such as the intrinsic pleasures derived from drugs and alcohol and can also contribute to delinquent behavior if socially reinforced as pleasurable (Bre zina & Piquero, 2003). Ultimately, it is the balance of these rewards and punishments that se rve to strengthen or extinguish behavior. Several comparative tests have supported social learning over other prominent theories (Akers & Cochran, 1985; White, Pandina, & LaGrange, 1987; Matsueda & Heimer, 1987; Akers & Lee, 1999). Social learning variables have also been found to mediate the relationships between constructs of its closest competitor, socia l control, and delinquency (see Agnew, 1993). Overall, substantial empirical evidence tends to suppor t social learning theory (see Akers & Jensen, 2003; Akers & Sellers, 2004; Akers & Jensen, 2006) and especially the effect that delinquent peer associations have on cri me and delinquency (see Simons Wu, Conger, & Lorenz, 1994; Akers, 2001; Warr, 2002). Control Theory The major rival to social learning theory is social control theory. Hirschi s (1969) Causes of Delinquency is one of the most often cited books on control theory to date. More importantly, control theory is the most empirically tested of all major criminological theories (Cohn, Farrington, & Wright, 1998). The question to be answere d for Hirschi is not why people commit crime, but rather why they do not. People are deterred from crime if they are properly socialized. Without proper social ization, people will only seek to satisfy their own desires (Hirschi, 1969). Proper socialization r esults from a youths bond to society through four important elements: belief, attachme nt,
9 involvement and commitment (Hirschi, 2002). If any of these elements are weak or broken, then delinquency is likely (Hirsch, 1969). However, the presence of strong bonds to society will create a buffer to deviance and act as a control over ones behavi or (Hirschi, 2004). One important element of the bond is belief. Belief shares considerable overlap with the previously discussed concept of definitions from social learning theory Belief can be thought of as what one can rationalize as acceptable behavior generally infl uenced by social situations and support (or lack thereof; Hirschi, 1969). However, personal beliefs can fly in the face of traditional values or what society universall y accepts as the norm. And while one can hold beliefs that preclude criminal behavior, they may, if just for a moment, be neutralized by beliefs that support criminal behavior in order to just ify it (Hirschi, 2004). What differentiates belief from definition is that an absence of preclusive beliefs rather than the presence of endorsing beliefs is what incre ases the likelihood of delinquency (Hirschi, 2002). Basic beliefs are formed from the attachments that one has with significa nt others. These attachments are perhaps the most important element of the bond and refer to feelings of fondness and positive emotion toward others of importance such as parents (Hirschi, 1969). Attachment to anything outside ones self reduces the likelihood of delinquent behavior, but most studies typically focus on family and school (Krohn, Massey, Skinner, & Laur, 1983; Krohn, Lanza-Kaduce, & Akers, 1984; Agnew, 1993; Longshore, Chang, & Messina, 2005). Hirschi (1969) posits that youth who have strong attachments to others have more to lose than those with weak or no attachments. Therefore youth will be less likely to risk jeopardizing those relationships by committing
10 delinquent acts as these acts are typically seen as incompatible with the con ventional expectations of those attachments (Hirschi, 2004). Strong attachment between child and parent creates an atmosphere of closeness through which parental values and expectations are passed (Hirschi, 2002). Magnusson (1988) suggests that the most important part of the environment is other people, especially those responsible for ones care. These significant people are re sponsible for shaping our world, creating our values, norms, and rules and how it is interpreted (Magnusson, 1988). The strength of important attachments is determined, in part, by supervision, intimacy of communication, and affectional identification (Hirschi, 2004) Direct parental supervision involves parents and youth physically spending time toge ther, while indirect parental supervision is the perception by youth that parents know of the ir whereabouts. This indirect supervision means that youth take into account how their parents would react to their behavior if they were physically present even t hough they are not (Hirschi, 2002). Hirschi (1969) suggests that since crimes require little ti me and opportunities are nearly limitless, indirect or virtual parental supervision is most important. Yet, indirect parental supervision can only be as good as the intimacy of communication that is shared between parents and youth. The intimacy of communication needs to be communication flowing from child to parent and vice versa (Hirschi, 1969). As important as supervision and communication are, if a youth does not care about ones parents, then deterrence from delinquency is unlikely. This affec tional identification is an important element of attachment as those youth who show care a nd concern towards their parents will be more likely to take that into consideration whe n presented with opportunities for delinquency (Hirschi, 1969). Therefore, those youth who
11 are not supervised (directly or indirectly), do not communicate well, and show little concern for their parents will be weakly attached to their parents and thus more likely to be delinquent. A meta-analysis of family related factors and delinquency conf irmed this, as Loeber and Stouthamer-Loeber (1986) found that both intimacy of communication and affectional identification, termed parental involvement, in addition to parental supervision, were the most robust family-related predictors of delinquency (s ee also Dornbusch, Erickson, Laird, & Wong, 2001; Wright & Cullen, 2001). Moreover, De Kemp, Scholte, Overbeek, and Engels (2006) found similar results using longitudinal data. The last two elements of the bond are commitment and involvement, which are often combined when examined in empirical studies (Krohn & Massey, 1980; Krohn et al., 1983; Akers & Lee, 1999). Involvement is typically operationalized as the tempor al dimension of commitment or in other words, as the actual amount of time spent on commitments. Hirschi (2002) states that the element of commitment suggests devot ion to a conventional activity of importance (e.g. education, work or business) such that one builds an investment or stake in conformity. Commitment to conventional lines of action should decrease the likelihood of delinquent behavior with the greater the number of commitments, the more one risks losing (Hirschi, 1969). This refers to the rational component in the decision to commit crime (Hirschi, 2002). Involvement consists of following through with ones commitments and requires time and energy (Paternost er & Bachman, 2001). When one is involved in their commitments and other conventional activities, they simply have less time or are too consumed to commit deviant act s (Hirschi, 1969).
12 Hirschi (1969) contends that the likelihood of delinquency is based considerably on ones attachments and commitments, particularly family and school, to the extent that if either is weak one believes they have little to lose. Hirschi (1969) found empir ical support for his theory as a whole, showing that the stronger the overall social bond, the less likely one is to be delinquent. Other studies have found weak to modest support for these elements, especially commitment/involvement (see Krohn et al., 1983; Ag new, 1991; Rankin & Kern, 1994). Gottfredson (2006: 80) recently concluded, however, that the foundational facts of control theory are still supported by recent resear ch. Moreover, Wright and Cullen (2001) affirmed that any model that doesnt include measures of control risk being misspecified. Deterrence Theory The rational thought process found in the elements of involvement and commitment in control theory shares overlap with rational choice theories such as deterrence. This process generally refers to weighing the costs and bene fits of crime, suggesting that all people want to maximize the benefits and minimize the costs (i.e., use a hedonistic calculus; Cullen & Agnew, 2003). Deterrence itself suggests tha t one avoids committing a criminal act in order to avoid punishment (Gibbs, 1975). This punishment could be formal, such as being apprehended by the police and becoming involved in the legal system, or informal, such as getting caught by parents (Ake rs, 1990). Deterrence theory focuses on the costs of crime, specifically increasing the costs (i.e., legal punishments) in order to decrease crime (Bentham, 1948; Beccaria, 1963). Additionally, like control and social learning theories, deterrence theory does not differentiate between offender and non-offender individual characteristics (N agin &
13 Paternoster, 1993). Rather, the theory suggests differences are due to the social c ontext and circumstances external to the individual (Pogarsky, 2007). Punishments must therefore be applied equally and the individual characteristics of the offender should not be taken into consideration (Liska & Messner, 1999). Most importantly, punishment must have certainty, severity, and celerity in order to effectively sway a potential offender from committing a crime. Certa inty of punishment is the likelihood that one will be caught and punished for a crime (Paternoster & Bachman, 2001). The deterrent effect of certainty increases when the punishm ent is thought to be quite severe (see Klepper & Nagin, 1989) and thus should effectively discourage would be offenders (Liska & Messner, 1999). The severity of punishment refers to the extent of the personal cost of the possible punishment and is typically operationalized by asking how big of a problem that punishment would be (see Grasmick & Green, 1980). Lastly, celerity implies swiftness of punishment after a cr ime occurs (Beccaria, 1963). It is derived from the psychological notion that immediate punishm ents are more effective in suppressing behavior than delayed punishments (Nagin & Pogars ky, 2001). Certainty of punishment has consistently predicted deterrence of criminal behavior better than severity or celerity of punishment (Paternoster, Saltzm an, Waldo, & Chiricos, 1985; Paternoster, 1987; Klepper & Nagin, 1989; Nagin & Pogarsky, 2001). This may be partially due to the idea that if the likelihood of being caught is very l ow, then the severity or swiftness of punishment simply matters less, if at all. Overall, previous studies have been inconsistent in predicting the contributions of severity of punishment (see Cullen & Agnew, 2003 for overview), while celerity has been given
14 relatively little importance in the literature because of its lack of empi rical support. This has lead most researchers to exclude it altogether (Gibbs, 1975; Piquero & Renger t, 1999; Nagin & Pogarsky, 2001). While examining these important aspects of punishment, research has looked at both the specific offender and the general population, with the bulk of research on the latter. Specific deterrence refers to the effect that punishment has on the one bei ng punished (Gibbs, 1975), particularly discouraging the offender from future crime (C ullen & Agnew, 2003). The direct experience with punishment and punishment avoidance serve as the deterrent effect (Stafford & Warr, 1993). In contrast, general deterrence refers to the idea that those who have been caught and punished for a crime will ser ve as examples to the general public and anyone considering crime (Paternoster & B achman, 2001). The indirect experience with punishment (Meier and Johnson, 1977) and punishment avoidance (Stafford & Warr, 1993) should be sufficient to deter the general public from crime. The general public can only be deterred from crimes when they are aware of thei r consequences. Herein lies the importance of the distinction between objective and subjective deterrence. Subjective or perceptual deterrence tends to be better supported than objective deterrence. Objective measures of severity and certainty (ge nerally ignoring celerity) were the focus when the theory was revived in the 1960s (Pat ernoster & Bachman, 2001). Objective measures of deterrence for severity include the maxim um prison sentence or the average length of prison sentence served for a particular crime, while the objective measures for certainty include the official arrest ra te (Cullen &
15 Agnew, 2003). Use of these measures has been on the decline since the 1970s as perceptual measures have gained prominence. With the works of Jensen (1969; Jensen, Erickson, & Gibbs, 1978) and Tittle (1977), perceptual or subjective measures have mostly replaced objective measur es of deterrence. Data have suggested that few people are truly aware of the a ctual likelihood of being caught and punished for a crime (Cullen & Agnew, 2003). This of course can affect ones decision to commit crime as objective threats of punishment are ir relevant if one does not perceive any risk. Subjective or perceptual measures typically foc us on asking participants if they believe or think they will be caught and punished for a cri me and how much of a problem that would be. Nagin and Paternoster (1993) found that perceptions of the certainty of informal and formal punishment and anticipated sham e effectively controlled respondents intentions to offend. Moreover, in a recent meta-analysis, non-legal sanctions were found to be overall better predictors of cri me than legal sanctions (Pratt et al., 2006). This indicates that factors such as anticipa ted shame from family, friends, and/or the community may play a bigger role than the thre at of legal punishment in the decision to commit a crime when one calculates their costs/bene fits ratio. Today deterrence theory often focuses on the perceptions that people have about the certainty and severity of punishment in their decision to conform or commit crim e, regardless of whether the punishment is legal or non-legal and also whether the ri sk is true or misperceived. Despite the influx of less than impressive support for deterr ence theory, its commonsensical approach to crime has established deterrence as the basis for
16 criminal law and policy which continues to exert much influence even to this day (see Pratt et al., 2006). General Strain Theory Agnew (1992) differentiates strain theory from social control and social learni ng by focusing specifically on negative relationships, while social control focus es on the absence of important relationships and social learning focuses on positive relati onships with deviant others. Agnew (1993) suggests that strain theory differs in explaining t he intervening mechanisms that lead to delinquency. That is, independent variables such as low social control will create the freedom to deviate for control theory and will increase deviant associations and learning of deviant definitions for social learning theor y, but strain suggests that these independent variables will lead to delinquency because of t he negative emotions, specifically the anger that it triggers (see Agnew, 1992 ). Agnews (1992) General Strain Theory is a reformulation of classic stra in theories articulated by Merton (1938), Cohen (1955), and Cloward and Ohlin (1960), proposing that it is an individuals affective response to negative events or strains that ca n foster delinquency. While these theorists proposed a social structural perspective, Agne ws General Strain Theory focuses on a social psychological perspective (Broidy, 2001). Delinquency is just one way of coping with any of several negative emotions, espe cially anger that one may feel when experiencing stressors or strains (Thaxton & Agn ew, 2004). Agnew (1992) suggests that the three major types of strain are the inabili ty to achieve positively valued goals, the loss of positively valued stimuli, and the prese ntation of negatively valued stimuli.
17 The first major type of strain is the inability to achieve positively valued g oals and consists of three subtypes: (1) the traditional concept of the disjunction between aspirations and expectations, (2) the disjunction between expectations and actual achievements, and (3) the disjunction between just/fair outcomes and actual outcomes (Agnew, 1992). The first concept, the disjunction between aspirations and expectat ions, refers to the inability to attain those goals that are emphasized by society (Agnew, 1992). Originally, the focus of strain theories was on lower class individuals who were prevented from achieving purported universal goals, but Agnew (2001) has since expanded the theory to include a variety of aspirations and goals that vary according t o the individual. In contrast to the minor role that the disjunction between aspirations and expectations plays, the disjunction between expectations and actual achievements m ay be more pivotal as it is often considered more emotionally distressing (Agnew, 1992: 52). That is, expectations are more closely seeded in reality in comparison to aspi rations; therefore the disappointment may be more severe when these expectations are not met than when aspirations are not. These expectations may be formed from past personal and vicarious experiences and the inability to achieve them may lead one to various neg ative emotions such as anger, resentment, and rage (Agnew, 1992). Expectations also lead one to assume that a just or fair outcome will occur. However, this may not always be the actual outcome. The disjunction between just/fair outcomes and actual outcomes st ems from the justice literature that focuses on equitable relationships (Agnew, 1992). T hat is, a relationship is considered just or equitable when one receives what one puts in. If a relationship is not fair or just, then one may feel distress which may be allev iated through crime (Agnew, 2001). The strain is typically seen as most unjust when it is inf licted
18 intentionally by the actions of others close to the victim (Agnew, 2001). Research h as linked unjust outcomes to anger, which Agnew (1992) suggests mediates the relationship between strain and delinquency (see Averill, 1993; Berkowitz, 1993). Furthermore, other studies have found that anger increased the likelihood of crime (Aseltine, Gore, & Gordon, 2000; Mazerolle, Burton, Cullen, Evans, & Payne, 2000). While the inability to achieve positively valued goals can create strain the loss of positively valued stimuli involving serious life events (e.g., the loss of a boyfri end, girlfriend, or a family member; moving to a new town) can be detrimental to ones l ife (Akers, 1985). Losing something or someone that is positively valued causes disrupti ons to ones life which may result in criminal coping methods. Agnew (1992) suggests that delinquency could occur because one attempts to prevent, regain, substitute, avenge, and/or negatively manage (e.g. through the use of drugs or alcohol) the loss of positive ly valued stimuli. Similar to losing something/someone of importance, exposure to something negative (i.e., the presentation of negatively valued stimuli) can be equally unpleas ant (Agnew, 1992). This is typically measured by such indicators as child abuse, crim inal victimization, and negative school experiences among others (Agnew, 1992, 2003). Family and school indicators are typically used to measure this element as a predictor of delinquency because youth are often unable to avoid or escape the negatively valued stimuli that may occur within the household or school setting. Each of these strains refers to different types of negative relationships, where one is unhappy with how they are being treated (Agnew, 1992). Agnew and colleagues (2002) state that family, school, and peer groups are the most important sources of thes e
19 strains. To the extent these relationships are negative, they increase the l ikelihood that one will experience anger or frustration resulting in either criminal or noncri minal coping methods (Cullen & Agnew, 2003). The more strain, the more likely delinquency. Yet, the likelihood still depends on other factors (e.g. social supports, criminal propensitie s; Agnew, 2001; Agnew, Brezina, Wright, & Cullen, 2002). Additionally, it has been suggested that strains that cause anger are better at predicting violent cr ime than any other (see Mazerolle & Piquero, 1998; Piquero & Sealock, 2000). Recently, Agnew (2001) revised the meaning of strain by dividing it between objective and subjective measures. Objective strains are generally disl iked by the majority of members in a given group, while on the other hand, subjective strains only have to be disliked by the person experiencing them (Agnew, 2001). Like subjective deterrence, subjective strains are more important in ones decision to commit cri me (Agnew, 2006) and may be measured by asking individuals simply whether they are being treated the way they wish to be (Agnew, 2003). As stated before, there has been limited empirical support for the failure to achieve positively valued stimuli (see Agnew, 2003). Additionally, the mediating e ffects of anger have been limited in scope, and have not been especially useful in predicting drug use (see Mazerolle, Burton, Cullen, Evans, & Payne, 2000). However, many studies have found support for both the loss of positive stimuli and the presentation of negative stimuli, and their ability to explain delinquency with such predictors as child abuse a nd neglect, criminal victimization, and divorce of parents, among others (for exampl e Agnew, 1985, 1992; Agnew & Brezina, 1997; Baron & Hartnagel 1997; Piquero & Sealock, 2000; see Agnew, 2003 for complete listing).
20 The General Theory of Crime Unlike previous theories, the General Theory of Crime (GTC) seeks to explain crime as the result of a single individual characteristic (Gottfredson & H irschi, 1990). This single characteristic, namely self-control, refers to ones ability to consider the longterm, negative consequences of antisocial behavior and more recently, all of the potential consequences of ones actions (see Hirschi, 2004). Considering that pain is different ially experienced while pleasure is equally enjoyed among all people, how much one calculates these consequences will translate to ones level of self-control Thus, the less one considers future consequences, the more likely they are to commit criminal ac ts when presented with opportunities to do so (Gottfredson & Hirschi, 1990). Additionally, Hirschi and Gottfredson (1994) emphasize that their theory explains not only criminal acts, but a variety of behaviors involving immediate pleasure at the risk of long ter m pain (i.e., analogous behaviors). Those who lack self-control are often described as impulsive, risk-seeking, selfish, short-tempered, and insensitive (Gottfredson & Hirschi, 1990). Criminal a cts are committed in the pursuit of self-interest and immediate pleasure, and for those l ow in self-control these acts tend to satisfy their impulsive desires, are ris ky yet easy to accomplish, and often harm others (Gottfredson & Hirschi, 1990). Since those low in self-control tend to seek out acts that provide immediate pleasure at the risk of long term pain, it easily follows that these individuals will commit a variety of crimi nal and analogous behaviors (e.g., accidents, smoking, drinking, and drug use) with no particula r specializations (see Paternoster & Brame, 1998 for an exception). Additionally because specialization is unnecessary, those engaging in one type of crime are more li kely to
21 engage in any and all types of crime, thereby making past criminal behavior the be st predictor of future crime (see Hirschi, 2004). In addition to variety, criminal a nd analogous behaviors will be committed at relatively high frequencies by thos e with low self-control (Gottfredson & Hirschi, 1990). Gottfredson and Hirschi (1990) explicitly deny that low self-control is the r esult of a biological predisposition toward crime or the result of ineffective child re aring. Rather, low self-control becomes apparent with the absence of any child rearing (Gottfredson & Hirschi, 1990). In other words, everyone is initially prone to deviat e, but through effective child-rearing consisting of supervision, and identifying and cor recting deviant behavior, self-control is acquired (Hirschi and Gottfredson, 2001). More importantly, this characteristic will remain relatively stable throug hout the life-course (Gottfredson & Hirschi, 1990; see Mitchell & MacKenzie, 2006 for an exception). Although Gottfredson and Hirschi (1990) have stated that opportunity is a required component in addition to self-control, it has mostly been neglected in the literature (see Grasmick, Tittle, Bursik, & Arneklev, 1993; Piquero & Tibbetts, 1996; Cochran, Wood, Sellers, Wilkerson, & Chamlin, 1998 for a few exceptions). Hirschi and Gottfredson (1993: 50) later admitted, however, that opportunities to commit crime are limitless, thus minimizing the role they play. Several studies have teste d propositions set forth by Gottfredson and Hirschi and found substantial support (see Pratt & Culle n, 2000 for meta-analysis). Although Hirschi and Gottfredson (1993; see also Gottfre dson, 2006) prefer behavioral measures, these findings hold across both attitudinal and behavioral measures of self-control (see Grasmick et al., 1993; Keane, Maxim, & Teevan, 1993; Evans, Cullen, Burton, Dunaway, & Benson, 1997). Although most
22 researchers agree that the claim of this theory as a general theory of crime is overstated, Pratt and Cullen (2000) emphasized that a study would risk misspecification if meas ures of self-control were excluded. Summary As indicated in the above review, research has shown that key constructs derived from social learning (i.e. associations, definitions, and reinforcements), con trol (i.e. parental attachment and supervision), deterrence (i.e. perceptual deterrence), s train (i.e. the loss of positively valued stimuli), and the general theory of crime (i.e. sel f-control) are related to antisocial behavior. That is, constructs from each of these theor ies has demonstrated a main effect on delinquent and criminal behavior. Yet, as mentioned in Chapter 1, there is emerging evidence to indicate that there are interactive effects as well. More specifically, research has indicated that self-control moderates some (if not all) of the relationships the other theories have with delinquency. These integrated perspe ctives, however, have failed to reach consensus on the precise direction of the interaction. The following section reviews this novel approach and the equivocal findings stemming from it. Integrated Perspectives The previously discussed theories attempt to explain why an individual commits crime through direct, independent measures. These explanations can be character ized as either a social causation or social selection model. Social causation sugges ts crime is the result of deviant social relationships. More broadly stated, behavior is solely t he result of ones social context. Theories such as social learning, strain, deterrence, and c ontrol fall under this classification as each proffers that factors external to the indivi dual lead to
23 antisocial behavior. In contrast, social selection implies crime is the re sult of personal characteristics. The GTC is a good example of social selection as it sugges ts that selfcontrol predicts ones involvement with deviant peers or weak attachments to prosocial others. That is, these external influences do not cause antisocial behavior as socia l causation would predict, but rather are a result of individual difference factors. Focusing on a purely social causation or social selection model has led to an incomplete explanation of crime (see Wright et al., 1999). This has clearly bee n demonstrated by several studies that have found the effects of external factor s remained significant when individual-level measures (i.e., self-control) were includ ed (Nagin & Paternoster, 1993; Evans et al., 1997; Wright, Caspi, Moffitt, & Silva, 1999; see also Kochanska, 1993; Carlo, Roesch, & Melby, 1998 on integration). Therefore, a theoretical model that incorporates both social selection and social causation processes offe rs a more defensible perspective (Wright et al., 1999). Recently, some studies have attempted to go beyond relying exclusively on eit her a social selection or social causation model by using this proposed mixed theoretic al model. One of the most prominent perspectives in this regard is Moffitts (1993) developmental taxonomy. This taxonomy suggests that social selection and social causation models may both be accurate explanations but for two distinct types of offenders. Moffitt (1993) calls the first group of individuals life-course persi stent offenders, as attempts to change their behavior are rarely successful. Th ese offenders have relatively stable individual characteristics conducive to antisocial beh avior. Lifecourse persistent offenders suffer from cumulative continuity, which refers to the failure to learn traditional prosocial alternatives to deviant behavior and continuing to carry
24 those destructive characteristics throughout the life-course (Moffitt, 1993). For the lifecourse persistent offender, social selection is a better explanation of antis ocial behavior. On the other hand, adolescence-limited offenders are typically led to deviate bec ause they are influenced by deviant peer associations, poor school performance, and elevated strain, among others. At the same time, these offenders are also more easily apt to conf orm when presented with prosocial factors. Thus, for this type of offender, social caus ation appears to predominate as these offenders typically lack the individual charact eristics that account for deviance among life-course persistent offenders. In opposition to Moffitts (1993) dual taxonomy argument, Lahey and Waldman (2003) suggest that underlying criminal propensities exist along a continuum. La hey and Waldman (2003) propose that differential explanations are not required as distinct typologies of offenders do not exist. Additionally, several researchers have suggested that external influences may play a more important role for those either much highe r (e.g. Wright et al., 1999; Lynam et al., 2000; Wright et al., 2001) or much lower (e.g. Wikstrm & Loeber, 2000; Wikstrm & Sampson, 2003) on the continuum of criminal propensity. This suggested interactive model draws from Lewin (1935), who proposed many years ago that behavior is a function of the person and the environment. That is, one must consider the individual, their environment, and how those two forces act independently and interactively to shape behavior. Of particular interest in the pr esent study is the interaction between the two. For instance, similar environments can ha ve a differential impact on two individuals as a result of their individual characteris tics (Lewin, 1935; Magnusson, 1988). It is this interaction that underlies the relationship between an individual and the environment (Magnusson, 1988). Yet, currently most
25 criminological theories do not address how criminal behavior is the result of the interaction between the individual and the environment. The modest amount of research that has been performed on the personenvironment interaction has brought to light two opposing arguments. Both propose that ones individual characteristics will determine the level of influence by ext ernal factors in predicting deviant behavior, but in opposite directions. The first approach proposes that the effects of external factors on crime are amplified for those low in selfcontrol (see Wright et al., 1999; Lynam et al., 2000; Wright et al., 2004). That is, those lower in selfcontrol are more induced to deviate when exposed to criminogenic environments than those with higher self-control. These individuals are more easily influenced because they are at an increased likelihood of deviant behavior to begin with (Wright et al., 2001). This inclination to deviate will be exacerbated by factors such as criminal associ ations, who will expose one to criminal opportunities and define them as gratifying and worthw hile (Evans et al., 1997). Youth with self control, however, tend to be socially protected from crime throughout the life course and thus less likely to be affected by such fa ctors (see Lahey & Waldman, 2003). Using individual characteristics conceptually overlapping with self-control (i.e., high negative emotionality and low constraint), Agnew and colleagues (2002) found that these personality traits did, in fact, condition the relationship between strain and delinquency. Youth who were high in negative emotionality and low in constraint (i.e., low in self-control) were more influenced by strain, and therefore more likely t o react to strain with delinquency, than those youth who were not high in negative emotionality and low in constraint (Agnew et al., 2002). Additionally, neighborhood context appears to
26 have similar amplified effects for those who are most impulsive (see Lynam et al., 2000). With both cross-sectional and longitudinal data, Lynam and colleagues were able to show that impulsive adolescent males (an important component of low self-control) were more likely to be delinquent when exposed to criminogenic neighborhoods than nonimpulsive adolescent males. These impulsive males were more likely to take advanta ge of these criminogenic neighborhoods as they represented weak situations that generally failed to provide effective social control. Within the home, high levels of impulsivity have also been found to increase the influence of parental support in reducing antisocial behavior (Jones, Cauffman, & Piquero, 2007). That is, this study found that increased parental support is more influential in reducing antisocial behavior among impulsive youth a s opposed to nonimpulsive youth. Wright and colleagues (2004) similarly concluded that the most criminally prone (i.e. lowest in self-control) were also the most inf luenced by deterrent effects in comparison to those lacking (or with less of) a crimi nal tendency. These findings coincide with the proposed interdependence model by Wright and colleagues (2001). These researchers suggest that all social ties (prosoc ial or antisocial) are more influential for those lowest in self-control. In other words, positive socia l ties will socially protect an individual with low self-control from engaging in devi ant behavior, despite possessing an elevated propensity to do so (Wright et al., 2001), while negative social ties will amplify this inclination. Since those with self-cont rol lack the proclivity to engage in antisocial behavior, they remain unaffected by social inf luences. In opposition to the previous argument, other researchers have proposed that the effects of external factors on crime are more salient for those with more ( as opposed to less) self-control (see Wikstrm & Loeber, 2000; Piquero & Pogarsky, 2002). That i s,
27 this conceptual model actually predicts that those youth scoring higher on self-control are more induced to deviate based on external factors compared to their low self-control counterparts. This perspective suggests that those low in self-control will offend regardless of external factors (Gottfredson & Hirschi, 1990). Those higher in se lf-control, on the other hand, are less likely to discount future consequences and thereby take into account external factors. Support for this position has been found with several different independent variables. For example, Wootton, Frick, Shelton, and Silverthorn (1997) examined the interaction between parenting and the individual characteristics of callousness a nd low emotionality (sharing overlap with self-control) for predicting childhood conduct problems. They concluded that ineffective parenting was influential for those wit hout significant levels of callous and unemotional characteristics in predicting c onduct problems. Meanwhile those exhibiting high levels of these characteristics ha d significant conduct problems regardless. Piquero and Pogarsky (2002) examined how deterrent effects differed in influence depending on ones level of impulsivity. They concluded that the effect of deterre nce was significantly less for impulsive youth. That is, impulsive youth are typicall y harder to deter because they do not consider future consequences (Gottfredson & Hirschi, 1990; Nagin & Pogarsky, 2001). Additional support was found by Wikstrm and Loeber (2000), who examined neighborhood context, sets of protective and risk factors, and delinquency. The factors making up this risk-protective profile run on a continuum and share some overlap with the concept of self-control (e.g. range from impulsive to nonimpulsive). The research ers
28 concluded that male juveniles exhibiting mostly protective factors, as well as a balance of risk and protective factors (or what could be considered those with high and average sel fcontrol), were most influenced by neighborhood context in predicting late onset of offending (Wikstrm & Loeber, 2000). In other words, those with low self-control offended regardless of their neighborhood context, while those youth who had average or high self-control were more affected by their environment and therefore more eas ily inclined to deviate if the neighborhood is highly disadvantaged (e.g. below the poverty level, high levels of public assistance; Wikstrm & Loeber, 2000). The models previously discussed may have prematurely articulated that person environment interactions exist in a particular direction. This is evinced by sever al studies that have used various individual difference factors resulting in divergent findings regarding these interactions. For example, parenting behaviors seem to exert the strongest influence among impulsive children. Yet this punctuated effect of parenting disappear s when considering empathy (see Jones, Cauffman, & Piquero, 2007). But as stated before Wooton et al. (1997) found that parenting behaviors were most influential for those without significant levels of callous and unemotional traits. This suggests that e xternal influences (e.g. parenting) may exert stronger effects in some instances (e .g., when the child is impulsive), or weaker effects in other instances (e.g., a child who lacks e mpathy). Thus, the specific individual difference factor being examined can affect wha t, if any, interactive effect is found. Regardless of who is more influenced, this social/psychological model (i.e. the interaction between the person and the environment) has the capability of expanding our understanding of why crime is committed. Thus far, there is reasonably strong s upport for
29 the notion that external factors of offending behavior (peers, parenting, neighborhoods, etc) vary in their influence depending on individual difference factors. However, it may be too soon to draw any conclusions about the effects that the same environments have on individuals with varying characteristics (see Bronfenbrenner, 1988). In other w ords, while most studies to date have suggested specific directions, the equivocal and contradictory findings call for continued efforts to investigate the interacti ve effects. It is important to assess this question within one sample because prior findings may be due to different operationalizations and use of individual risk factors (self-control ver sus impulsivity versus risk index). These differences may also be the result of idiosyncratic features of the sample. The current study, therefore, utilizes one sample to exa mine a wide array of criminological constructs external to the individual and a comprehensi ve scale measuring self-control similar to that described by Gottfredson a nd Hirschi (1990). Specifically, this study seeks to examine multiple external constructs r epresenting social learning, control, strain, and deterrence theories to determine how, if at all, ea ch varies as a function of self-control. Although most evidence to date agrees that the characte ristics individuals bring to their environment do in fact affect the amount of influence by that environment (see Lahey & Waldman, 2003), these influences cannot at once be greate r for both those with low and high self-control. That is, either external factors are m ore influential for those low in self-control or for those high in self-control. This study seeks to solve this dilemma.
30 Chapter Three Methodology Sample The analyses were based on information collected from students in Largo, Flor ida in 1998. This cross sectional study was designed to examine crime and delinquency in middle and high schools. Participation was voluntary and conditional upon passive parental consent. All types of students were allowed to participate including ma instream, emotionally handicapped, dropout prevention, and a 21 st Century Learning Community. In middle school, the survey was administered to all social studies classes Since all middle school students are required to take this course, this was the most logica l way to reach the greatest number of students. Two researchers remained in the room to whic h one researcher read aloud the questions in the survey, while the other assisted st udents as needed. This took approximately 50 minutes for all surveys to be completed. The response rate for the middle school sample was 81%. In high school, the survey was distributed among a random sample of third period classes. One researcher stayed in the room to give instructions and answer q uestions while students completed the survey. All surveys were completed in approximat ely 25 minutes. The response rate for this part of the sample was 79%. The total sample size was 1,674 students, with 621 from the high school (37.3%) and 1,043 from the middle school (67.7%). The sample was evenly split between males (49.9%) and females. The age distribution of Largo students was slightly positive ly skewed, reflecting the majority of students that were in middle school. The mean ag e was
31 13.8 years, while the range was 11-19. Seventy-four percent of respondents were w hite, 11.2% black, 3.9% Hispanic, 3.1% Asian, and 3.8% other. Descriptions of the remainder of the variables used in the analyses can be found in Table 1. In the following sections, the variables are described in greater detail. Table 1: Descriptive Statistics of Demographics and Theoretical Constructs N X SD Minimum Maximum Demographics : Age 1652 13.79 1.99 11 19 Sex (1=male) 1662 1.50 .50 1 2 Delinquency (before log) 1657 5.29 4.59 0 22 Social Learning Definitions 1661 .00 3.01 -4.16 8.76 Peer Associations 1649 .00 3.17 -3.27 10.99 Reinforcements 1605 -.02 3.07 -5.63 6.55 Subjective Strain 1675 .00 5.40 -6.91 25.47 Perceptual Deterrence 1643 .01 3.08 -13.10 2.42 Control Parental Control 1514 -.02 3.22 -8.21 5.06 Paternal Attach 1503 .01 4.04 -9.43 5.25 Maternal Attach 1615 .01 3.90 -10.83 4.97 Self Control 1557 .04 5.70 -21.09 12.60 Dependent Variable The dependent variable, delinquency, indicated the variety of delinquent behaviors the participant self-reported. Variety scales are preferred t o frequency scores because they are less skewed (Caspi, Moffitt, Silva, Stouthamer-Loeber, Krue ger, & Schmutte, 1994), give equal weight to all delinquent acts (Caspi et al, 1994), and it has been shown that adolescents often do not specialize in only one type of offending (Piquero et al., 1999). Delinquency was measured by asking the students how many different crimes they have ever committed. The mean number of acts committ ed was 5.29
32 with a standard deviation of 4.59 (see Table 1). However, since the variable was censored at the upper limit, the natural log (plus one) was used in the bivariate and multivar iate analyses to correct for this. Respondents could endorse up to 22 different types of delinquent behaviors. The 22 items were entered into a principal components factor analysis using promax rotation where four factors were found with eigenvalues that exceeded one. However, a Scree plot suggested a single factor solution with the biggest brea k between the first and second eigenvalues (Eigenvalue=6.70). Therefore, the 22 items wer e reentered into a principal components factor analysis, extracting a one-fact or solution. Loadings on this single factor ranged from .42 to .68. The Cronbachs alpha was .88 for this 22-item scale (see Table 2 for complete listing of items and factor loadi ngs). Table 2: Items and Factor Loadings for Measures of Self-Reported Delinquency Have you ever: Factor Loadings: Factor One 1. bought illegal drugs? .68 2. used marijuana? .65 3. sold illegal drugs such as cocaine, crack, ecsta sy, LSD, or heroin? .65 4. stolen things worth $50 or less? .63 5. stolen something worth more than $50? .61 6. skipped class without an excuse? .59 7. purposely damaged or destroyed property that did not belong to you? .58 8. used alcohol? .58 9. used tobacco products? .58 10. used other illegal drugs such as cocaine, crack ecstasy, LSD, or heroin? .57 11. stolen another students backpack, lunch money, or other personal things worth $50 or less? .56 12. gone or tried to go into a house to steal something? .55 13. attacked someone with a weapon? .54 14. lied about your age to get into some place or t o buy something? .54 15. gone or tried to go into a building to steal something? .53 16. carried a weapon for protection? .49 17. stolen or tried to steal a car or motorcycle? 48 18. hit someone with the idea of hurting them? .47
33 Table 2: (continued) Have you ever: Factor Loadings: Factor One 19. used a weapon or force to get money or things from people? .47 20. stayed out longer than youre allowed? .46 21. run away from home? .46 22. used paint, glue, or other things to get high? .42 Cronbachs Alpha: .88 Eigenvalue: 6.70 Independent Variables Social Learning Theory. Three components derived from the social learning perspective were examined: definitions, peer associations, and reinforcements. Definitions were charact erized here as attitudes one takes towards a behavior that they identify as positive, neutral, or neg ative (Akers et al., 1979). The more one considers a behavior positive, the more likely one is to engage in it. In the present analysis, definitions were operationalized as ones pos itive evaluation towards four types of delinquency 1 These items were entered into a principal components factor analysis, which yielded a one-factor solution (Eigenvalue=2.30). Loadings on this single factor ranged from .63 to .81. The four items were summed to create a scale which had a mean of 8.01 and a standard deviation of 2.84 with values ranging from 3 to 16. However, the scale was standardized (i.e. centered) ther eby changing the mean to .00 and the standard deviation to 3.01 (see Table 1). This standardized scale had a reported Cronbachs alpha of .75 (see Table 3 for complet e listing of items and factor loadings). The values for the standardized scale ra nged from 4.16 to 8.76 with higher numbers indicative of pro-criminal definitions. xxxiii 1 For the scales measuring social learning and deter rence variables, the focus is on four types of delinquency as this could be considered a conservat ive way to explain the extent of involvement in dev iant activities. That is, if some youth are involved in more serious types of criminal behavior, they would be more likely to commit, or at least approve of, thes e more minor forms of crime.
34 Table 3: Items and Factor Loadings for Social Learn ing Measures of Definitions Its okay to: Factor Loadings: Factor One 1. skip school if nothing important is going on in class. .81 2. steal little things from a store since they make so much money that it wont hurt them. .79 3. get into a physical fight with someone if they insult you or hit you first. .63 4. use marijuana since its not really harmful. .79 Cronbachs Alpha: .75 Eigenvalue: 2.30 The second component of social learning theory assessed in this analysis was pe er associations, which was measured by asking how many of the participants friends committed any of four types of delinquency. The four items were entered into a princ ipal components factor analysis. One factor was found with an eigenvalue over one (Eigenvalue=2.53). Loadings on this single factor ranged from .74 to .83. The original scale had a mean of 7.79 and a standard deviation of 3.64 with values ranging from 3 to 20. Once the scale was centered, the mean was adjusted to .00 with a standard deviation of 3.17 (see Table 1). The Cronbachs alpha was .80 for this standardized four-item scal e (see Table 4 for complete listing of items and factor loadings) with values rang ing from 3.27 to 10.99. Higher values were indicative of more deviant associations. Table 4: Items and Factor Loadings for Social Learn ing Measures of Differential Associations During the past 12 months, how many of your current friends have: Factor Loadings: Factor One 1. skipped school? .83 2. stolen something worth $50 or less? .81 3. hit someone with the idea of hurting them? .74 4. used marijuana? .80 Cronbachs Alpha: .80 Eigenvalue: 2.53 Lastly, differential reinforcements refer to the balance of rewards and punishments that strengthen or extinguish behavior. Reinforcements were measured b y
35 asking if friends respected the participant getting away with any of four types of delinquency, with responses ranging from definitely would to definitely would not. F our items were entered into a principal components factor analysis, where one factor was found with an eigenvalue over one (Eigenvalue=2.41). Loadings on this single factor ranged from .76 to .83. Originally, the mean for the scale was 9.41 and the standard deviation was 3.07 with values ranging from 3 to 16. However, the scale was centered which changed the mean to -.02 (SD= 3.07) with values for the standardized scale ranging from -5.63 to 6.55 (see Table 1). The Cronbachs alpha was .78 for this standardized four-item scale (see Table 5 for complete listing of items a nd factor loadings). Higher values were indicative of more differential reinforce ments of crime. Table 5: Items and Factor Loadings for Social Learn ing Measures of Differential Reinforcements Would your friends respect you if you got away with: Factor Loadings: Factor One 1. skipping school? .76 2. stealing something worth $50 or less? .83 3. hitting someone with the idea of hurting them? 76 4. using marijuana? .75 Cronbachs Alpha: .78 Eigenvalue: 2.41 Control Theory. One important aspect of control theory that was examined here was the effect of parenting. Hirschi (1969) stressed the value of several elements of the socia l bond, but most importantly the attachment to others, especially parents. Additionally, Loeber and Stouthamer-Loeber (1986) demonstrated in their meta-analysis of the relat ionships between family factors and delinquency that lack of parental supervision and involvement are two of the strongest predictors of delinquency among parenting
36 behaviors. Therefore, three scales of parenting were used in this study: parenta l control, maternal attachment, and paternal attachment. Parental control measured the extent to which the parents monitor the participants whereabouts. Four items assessing this construct were enter ed into a principal components factor analysis, yielding one factor with an eigenvalue ove r one (Eigenvalue=2.59). Loadings on this single factor ranged from .77 to .83. The mean for the unstandardized scale was 11.31 and the standard deviation was 2.96 with values ranging from 4 to 16. The standardized scale, however, had a mean of -.02, standard deviation of 3.22, and values ranging from -8.21 to 5.06 (see Table 1). The Cronbachs alpha was .82 for this standardized four-item scale (see Table 6 for complete li sting of items and factor loadings). Higher scores were indicative of more parent al control. Table 6: Items and Factor Loadings for Control Meas ures of Parental Control Factor Loadings: Factor One 1. My mother knows where I am when I am not at home or at school. .78 2. My father knows where I am when I am not at home or at school. .83 3. My mother knows who I am with when I am not at home. .77 4. My father knows who I am with when I am not at home. .83 Cronbachs Alpha: .82 Eigenvalue: 2.59 Maternal and paternal attachment measured the extent to which the partici pant talks, trusts, admires, and identifies with their respective mother and fathe r figure. The maternal attachment scale was based on five items which were entered into a pri ncipal components factor analysis. One factor was found with an eigenvalue over one (Eigenvalue=3.10). Loadings on this single factor ranged from .75 to .85. The mean for this scale was 21.95, having standard deviation of 6.25 and values ranging from 5 to 30.
37 Once the scale was centered, the mean changed to .01 with a standard deviation of 3.90. Values ranged from -10.83 to 4.97 (see Table 1). The Cronbachs alpha was .85 for this standardized five-item scale (see Table 7 for complete listing of items and factor loadings). Items were coded so that higher numbers were indicative of more attac hment. Table 7: Items and Factor Loadings for Control Meas ures of Maternal Attachment Think about your mother or mother-figure: Factor Lo adings: Factor One 1. I can always ask her for advice vs. I can never ask her for advice. .85 2. I can talk to her about anything vs. I cant tal k to her about anything. .83 3. I want to be the kind of person she is vs. I don t want to be the kind of person she is. .76 4. She always trusts me vs. she never trusts me. 75 5. She always praises me when I do well vs. she never praises me when I do well. .75 Cronbachs Alpha: .85 Eigenvalue: 3.10 The paternal attachment scale was based on five items which were entere d into a principal components factor analysis. One factor was found with an eigenvalue over one (Eigenvalue=3.31). Loadings on this single factor ranged from .77 to .87. The scale originally had a mean of 20.91, a standard deviation of 6.95, and values ranging from 4 to 30. Once the scale was centered, the mean for paternal attachment was .01 with a standard deviation of 4.04. These values ranged from -9.43 to 5.25 (see Table 1). The Cronbachs alpha was .87 for this standardized five-item scale (see Table 8 for complete listing of items and factor loadings). Items were coded so that higher numbers we re indicative of more attachment.
38 Table 8: Items and Factor Loadings for Control Meas ures of Paternal Attachment Think about your father or father-figure: Factor L oadings: Factor One 1. I can always ask him for advice vs. I can never ask him for advice. .87 2. I can talk to him about anything vs. I cant tal k to him about anything. .85 3. I want to be the kind of person he is vs. I don t want to be the kind of person he is. .79 4. He always trusts me vs. he never trusts me. 5. He always praises me when I do well vs. he never praises me when I do well. .78 .77 Cronbachs Alpha: .87 Eigenvalue: 3.31 Deterrence Theory. Perceptual deterrence suggests that the more one perceives the likelihood of getting caught and punished for a crime the more likely one is to be deterred from c rime (Pogarsky et al., 2004). One can only be successfully deterred from crimes when the y are aware of their consequences. This operationalization tends to be better supported tha n objective deterrence and was therefore utilized in this study. Perceptual deterrence was measured by asking how big of a problem it would be for the participant if they were caught by police for four different types of delinquency. The perceptual deterrence scale was based on four items, which were entered i nto a principal components factor analysis. One factor was found with an eigenvalue over one (Eigenvalue=2.40). Loadings on this single factor ranged from .73 to .81. The original scale had a mean of 9.93, standard deviation of 2.47, and values ranging from 0 to 12. The centered (i.e. standardized) scale, however, had a mean of .01 with a standard deviation of 3.08. Values for the centered scale ranged from -13.10 to 2.42 (see Table 1). The Cronbachs alpha was .78 for this standardized four-item scale (see Table 9 for
39 complete listing of items and factor loadings). Higher values were indicati ve of more perceptual deterrence. Table 9: Items and Factor Loadings for Measures of Perceptual Deterrence How big of a problem would it be for you if you were caught by the police for: Factor Loadings: Factor One 1. skipping school? .79 2. stealing something worth $50 or less? .81 3. hitting someone with the idea of hurting them? 73 4. using marijuana? .77 Cronbachs Alpha: .78 Eigenvalue: 2.40 General Strain Theory. Originally both objective and subjective strains were used in analyses as recommended by Agnew (2001). However upon further inspection, the objective strains scale was found to provide little explanation for the likelihood of delinquency. Therefore only subjective strains were examined for the purposes of this research. Subject ive strains have been described as situations or events that are disliked by those who have or are experiencing them (Agnew, 2001). These strains were measured by asking if any of 11 objective strains occurred to the participant and if so how big of a problem was it. The 11 items were entered into a principal components factor analysis using promax rota tion, where three factors emerged with eigenvalues over one. However, a Scree plot suggested a single factor solution (Eigenvalue=2.73). The 11 items were reentered into a principal components factor analysis, forcing a one-factor solution. Loadings on this single factor ranged from .34 to .64. The Cronbachs alpha was .68 for this standardized 11-item scale (see Table 10 for complete listing of items and factor loadings). Some items did not load as strongly as others, which explains the lower alpha reliability. However, t he alpha would have been lower had any items been deleted; therefore this standardized 11-it em
40 scale was used. The original scale had a mean of 10.14 with a standard deviation of 7.65. These values ranged from 0 to 44. Once the scale was standardized, the mean was .00 with a standard deviation of 5.40. Values ranged from -6.91 to 25.47 (see Table 1). Higher values were indicative of more subjective strain. Table 10: Items and Factor Loadings for Measures of Subjective Strains Did any of the following happen to you and if so how big of a problem was this for you? Factor Loadings: Factor One 1. Changed schools .60 2. Parents divorced .64 3. Parent moved out or away .64 4. Brother or sister moved out .34 5. Broke up with boyfriend or girlfriend .35 6. Moved to new neighborhood .63 7. Death of a relative .44 8. Lost a friendship .47 9. Pet died or disappeared .48 10. Dropped from or quit athletic team or school activities .38 11. Parent lost job for more than two months .34 Cronbachs Alpha: .68 Eigenvalue: 2.73 The General Theory of Crime. Gottfredson and Hirschi (1990) boldly suggest that low self-control is the only individual level predictor of criminal and analogous behavior. Those low in self-control are described as impulsive, insensitive, physical, risk-taking, short-sight ed, and nonverbal (Gottfredson & Hirschi, 1990: 90). These components were entered into a scale comprising a total of 11 variables measuring both behavioral and attitudi nal selfcontrol. Because the items were on different metrics, the behavioral and attitudi nal measures of self-control were standardized and summed 2 Behavioral measures focused on actual behaviors analogous to crime, while attitudinal measures of self-cont rol focused xl 2 Because the items were on different metrics and ha d to be standardized and summed, there are no unstandardized descriptive statistics to report as there were with the other scales noted above.
41 on personality traits or characteristics (Gottfredson & Hirschi, 1990). Whil e some criticism has been expressed with attitudinal measures like the ones used in the current analysis (see Hirschi, 2004), it is consistent with existing measures that have been widely used (e.g., Grasmick et al., 1993). In addition, despite Gottfredson and Hirschis (1993) preference for behavioral measures, many have argued that such an approach is tautological (Akers, 1991). In spite of this, Pratt and Cullen (2000) concluded in their meta-analysis on self-control that effect sizes were similar for both t ypes of measures and thus weakened the potentially tautological argument favoring behavioral measure s. Behavioral self-control was initially measured with five items. However, o ne item did not fare well in a reliability analysis; therefore, only four items were i ncluded. In addition to the behavioral measures, nine items that measured attitudinal selfcontrol were included. However, after running a reliability analysis, it was found that two items did not load well. A total of 11 items comprising both behavioral and attitudinal measures of self-control were entered into a principal components factor analysis using promax rotation. Three factors were found with eigenvalues that exceeded one; however, a Scree plot suggested a single-factor solution (Eigenvalue=3.03). The 11 items were reent ered into a principal components factor analysis, extracting a one-factor solution. Loa dings on this single factor ranged from .41 to .65. The Cronbachs alpha was .73 for this 11-item scale (see Table 11 for complete listing of items and factor loadings). The mean of the self-control scale was .04 with a standard deviation of 5.70. The values ranged from -21.09 to 12.60 (see Table 1). Responses were coded such that a low score was indicative of low self-control.
42 Table 11: Items and Factor Loadings for Attitudinal and Behavioral Measures of Self-Control Factor Loadings: Factor One 1. Sometimes I will take a risk just for the fun of it. .51 2. I like to test myself every now and then by doin g something a little risky. .52 3. I often act on the spur of the moment without stopping to think. .41 4. I often do whatever brings me pleasure here and now, even at the cost of some distant goal. .45 5. When things get difficult, I tend to quit. .44 6. I lose my temper pretty easily. .61 7. When Im really mad, other people better stay away from me. .62 8. More likely to hit vs. talk when mad.* .65 9. More likely to confront vs. avoid classmate who is spreading rumors about me.* .55 10. Do well on a test because I guessed vs. do well on a test because I studied hard.* .45 11. More likely to tease vs. make friends with an unpopular student.* 51 Cronbachs Alpha: .73 Eigenvalue: 3.03 behavioral measures of self-control Analytic Plan Bivariate correlations were first performed in order to analyze the ext ent to which each theoretical construct was related to delinquency. Additionally, Ordina ry Least Squares (OLS) regression techniques were used to further determine the nature of the relationships that exists between each of the theoretical constructs and del inquency as well as the interaction between these theoretical constructs and self-cont rol. As discussed previously, several studies have found divergent results in relation to person-environment interactions. Some have suggested that the relationship between external factor s and delinquency is amplified for those low in self-control (Lynam et al., 2000; Wright e t al., 2001), while others suggest this relationship to be stronger among those high in self control (Wikstrm & Loeber, 2000; Wikstrm & Sampson, 2003). The inclusion of interaction terms in the regression analyses allows for examination of this re lationship to
43 determine whether or not youth who are low in self-control are more, less, or equally susceptible to external factors (e.g., peers, parents, strains). It should be not ed that in order to avoid multicollinearity, all independent measures were centered prior to the creation of the interaction term and entered in one at a time to the regression equat ion.
44 Chapter Four Results Bivariate Findings The matrix of Pearsons zero-order correlation coefficients for all va riables examined in this research is presented in Table 12. All of the theoretical const ructs included in this study were significantly related to delinquency in the expected d irection. Self-control (r=-.495, p<.05) and perceptual deterrence (r=-.470, p<.05) were strongly and negatively correlated with delinquency. Therefore, lower levels of self-c ontrol and perception of deterrence were associated with an increased likelihood of delinque ncy. As expected, all measures of social learning were positively correlated w ith delinquency. Reinforcements were moderately related with delinquency (r=.384, p<.05), while pe ers and definitions demonstrated a much stronger correlation with delinquency (r=.634, p<.05 and r=.648, p<.05 respectively). Simply put, those who were more reinforced for their delinquent behavior, had more deviant peers, and more pro-criminal definitions were at an increased risk of delinquent behavior. Additionally, there was a modest positive correlation between subjective strains and delinquency (r=.180, p<.05), indicating that the more strain one experiences, the more likely one is to be delinque nt. The three measures of parenting were all negatively correlated with deli nquency. Parental control and delinquency demonstrated the strongest correlation (r=-.482, p<.05) among the control variables, while maternal and paternal attachment were moderate ly related to delinquency (r=-.320, p<.05 and r=-.255, p<.05 respectively). Thus, the more attached
45 one is to their mother and father and the more control parents exert on their child, the less likely the child is to be delinquent. Table 12: Pearson's Zero-Order Correlations 1 2 3 4 5 6 7 8 9 1. Delinquency 2. Self Control -.495* 3. Subj Strain .180* -.080* 4. P Attachment -.255* .203* -.102* 5. M Attachment -.320* .308* -.111* .480* 6. Pt Control -.482* .346* -.140* .378* .343* 7. P Deterrence -.470* .428* -.035 .132* .208* .37 2* 8. Reinforcements .384* -.407* .027 -.154* -.189* .237* -.266* 9. Peers .634* -.459* .160* -.234* -.283* -.421* -. 475* .388* 10. Definitions .648* -.523* .057* -.225* -.293* -. 432* -.506* .400* .623* *p<.05, two tailed test. Note: Subj Strain= Subjective Strain, P Attachment= Paternal Attachment, M Attachment= Maternal Attachment, Pt Control= Parental Control, P Deterre nce= Perceptual Deterrence Sample size ranges from 1489 to 1657 because pairwi se deletion was used. Multivariate Findings In order to test how robust the bivariate relationships were, multivariate an alyses were performed to examine the independent effects of the various constructs whi le including appropriate controls (e.g., sex, race, and age). Additionally, one of the m ain purposes of the research was to investigate the extent to which various criminologi cal constructs vary in their effects on delinquency as a function of self-control. This was accomplished through the creation of interaction terms. Attempts were made to put the composite of the interactions between each theorys constructs and self-control i nto one model, but the multicollinearity was simply too high. Therefore, the interactions of each of the constituent variables were individually examined. Delinquency was regress ed on each of the theoretical constructs, self-control, and the interactions. The tabl es listed
46 below show the results of OLS regression for each theoretical construct wit h additional models to show the moderating effects of self-control. Social Learning. The results of the regression models for self-control and social learning are presented in Table 13. In the first model, delinquency was regressed on the social learning variables and self-control to determine their main effects, controll ing for age, sex, and race. Overall, the first model was statistically significant (F=245.38, p<.05) and able to explain 54.2% of the variance in the variety of delinquent acts committed. 3 Definitions, peers, and reinforcements ( =.32, p<.05; =.29, p<.05; and =.08, p<.05, respectively) all had significant positive relationships with delinquency, a ffirming earlier bivariate findings. Thus, those with increased criminal definitions, deviant peer associations, and negative reinforcements were more likely to engage in de linquent behavior. Self-control had a negative and statistically significant effec t on delinquency ( =-.16, p<.05), meaning the more self-control one has, the less likely they are to be delinquent. xlvi 3 All variance statistics included in the analyses a re based on an adjusted R 2
47 Table 13: OLS Regression Results for Social Learnin g Variables, Self-Control, and Interactions (N=1448 ) MODEL A B se(b) MODEL B b se(b) MODEL C b se(b) MODEL D b se(b) Sex -.002 .03 -.001 .002 .03 .001 .002 .03 .001 -.002 .03 -.001 Age .06* .01 .14 .05* .01 .12 .05* 01 .12 .06* .01 .13 Race -.01 .02 -.02 -.01 .02 -.02 -.02 .02 -.02 -.02 .02 -.02 Self Control -.02* .00 -.16 -.03* .00 .19 -.03* .00 -.18 -.03* .00 -.17 Peers .08* .01 .29 .09* .01 .35 .08* .01 .32 .08* .01 .30 Definitions .09* .01 .32 .09* .01 .32 09* .01 .34 .09* .01 .32 Reinforcements .02* .01 .08 .02* .01 .0 7 .02* .01 .07 .02* .01 .08 SC x Peers .01* .00 .14 SC x Def .004* .00 .12 SC x Reinforce .003* .00 .07 Adjusted R 2 .542 .555 .554 .547 *p<.05 two tailed test Note: SC= Self-Control, Def= Definitions, Reinforce = Reinforcements In Model B, all variables from Model A were included with the addition of the interaction between self-control and delinquent peers (see Table 13). This was the strongest interaction with self-control among the social learning variabl es ( =.14, p<.05). The positive coefficient for the interaction term revealed that the positive eff ect of delinquent peers was stronger among those higher in self-control (see Figure 1) In other words, delinquent peers have a stronger effect on delinquency for those with higher self control versus those lower in self-control. The interaction term demonstrated incr emental validity in model fit (R 2 Change=.01, p<.05). Simply put, the addition of the interaction resulted in significant increased explanation of variance.
48 0 0.5 1 1.5 2 2.5 -1SD+1SD PeersDelinquency Low SC High SC Figure 1: Effects of Peers on Delinquency by Levels of Self-Control Note: Model included only the peer association scal e, self-control, and the interaction. Low SC=individuals scoring one standard deviation b elow the mean on self-control High SC= individuals scoring one standard deviation above the mean on self-control In Model C, all variables from Model A were included with the addition of the interaction between self-control and definitions (see Table 13). This interacti on was positive and statistically significant ( =.12, p<.05), indicating that as one increases in self-control, the importance of criminal definitions on delinquency also increases (see Figure 2). In other words, criminal definitions were more influential in increa sing delinquency among those higher in self-control. The R 2 change was statistically significant, suggesting that the introduction of the interaction term signif icantly increased the model fit (R 2 Change=.01, p<.05).
49 0 0.5 1 1.5 2 2.5 -1SD+1SD DefinitionsDelinquency Low SC High SCFigure 2: Effects of Definitions on Delinquency by Levels of Self-Control Note: Model included only the definitions scale, se lf-control, and the interaction. Low SC= individuals scoring one standard deviation below the mean on self-control High SC= individuals scoring one standard deviation above the mean on self-control Lastly, in Model D, all variables from Model A were included with the addition of the interaction between self-control and reinforcements (see Table 13). The i nteraction was positive and statistically significant ( =.07, p<.05). Thus, the relationship between reinforcements and delinquency was stronger for those higher in self-control (s ee Figure 3). Stated differently, reinforcements increase the likelihood of delinquency m ore strongly for those higher in self-control than those lower in self control. Additi onally, the R 2 change was statistically significant, indicating that the introduction of the interaction term significantly increased the model fit (R 2 Change=.01, p<.05).
50 0 0.5 1 1.5 2 2.5 -1SD+1SD ReinforcementsDelinquency Low SC High SC Figure 3: Effects of Reinforcements on Delinquency by Levels of Self-Control Note: Model included only the reinforcements scale, self-control, and the interaction. Low SC= individuals scoring one standard deviation below the mean on self-control High SC= individuals scoring one standard deviation above the mean on self-control Control. Table 14 shows the results of the OLS regression analyses used to assess the relationships between delinquency, three measures of parenting, and self-control controlling for age, sex, and race. Overall, the first model was statistica lly significant and able to explain 41.5% of the variance in the variety of delinquent acts committed (F=132.77, p<.05). In this model, delinquency was regressed on parental control, maternal attachment, paternal attachment and self-control to determine their main effects. Parental control was the strongest parenting predictor of delinquency ( =-.31, p<.05) followed by maternal attachment ( =-.09, p<.05). Those with less parental control and less attachment to their mothers were more likely to be delinquent. Self-c ontrol also proved to be a strong predictor of delinquency ( =-.35, p<.05), meaning those with less
51 self-control were more likely to be delinquent. The significant bivariate re lationship between paternal attachment and delinquency was no longer significant in multi variate analyses. Table 14: OLS Regression Results for Control, SelfControl, and Interactions (N=1302) MODEL A B se(b) MODEL B b se(b) MODEL C B se(b) MODEL D b se(b) Sex -.04 .04 -.03 -.04 .04 -.03 -.04 .04 -.02 -.04 .04 -.03 Age .09* .01 .22 .09* .01 .21 .09* .01 .21 .09* .01 .21 Race .02 .02 .02 .02 .02 .02 .02 .02 .02 .02 .02 .02 Self Control -.05* .00 -.35 -.05* .00 .36 -.05* .00 -.35 -.05* .00 -.35 Parental Control -.08* .01 -.31 -.08* .01 -.31 -.08* .01 -.31 -.08* .01 -.30 M Attachment -.02* .01 -.09 -.02* .01 .08 -.02* .01 -.09 -.02* .01 -.08 P Attachment -.004 .01 -.02 -.004 .01 .02 -.004 .01 -.02 -.01 .01 -.03 SC x Pt Control -.002* .00 -.06 SC x M Attach -.002* .001 -.06 SC x P Attach -.001 .001 -.04 Adjusted R 2 .415 .418 .418 .416 *p<.05 two tailed test Note: SC= self-control, Pt Control= Parental Contro l, M Attach/ment= Maternal Attachment, P Attach/ment= Paternal Attachment Model B shows the results of the variables from Model A with the addition of the interaction between parental control and self-control (see Table 14). The interac tion was negative and significant at ( =-.06, p<.05), indicating that parental control had a greater negative effect on delinquency for those with higher self-control (see Figure 4). Tha t is, parental control was stronger inhibitor of delinquency among those higher in self-co ntrol. The R 2 change was statistically significant, indicating that the introduction of t he interaction term significantly increased the model fit (R 2 Change=.004, p<.05).
52 0 0.5 1 1.5 2 2.5 -1SD+1SD Parental ControlDelinquency Low SC High SC Figure 4: Effects of Parental Control on Delinquenc y by Levels of Self-Control Note: Model included only the parental control scal e, self-control, and the interaction. Low SC= individuals scoring one standard deviation below the mean on self-control High SC= individuals scoring one standard deviation above the mean on self-control In Model C, all variables from Model A were included with the addition of the interaction between maternal attachment and self-control (see Table 14). The i nteraction term was significant and negative at the p<.05 level ( =-.06). This showed that maternal attachment had a greater negative effect on delinquency for those higher in self -control (see Figure 5). In other words, maternal attachment had a greater influen ce in reducing delinquency for those youth who are higher in self-control. Additionally, the inclusion of the interaction term resulted in a significant increase in explained variance (R 2 change=.004, p<.05).
53 0 0.5 1 1.5 2 2.5 -1SD+1SD Maternal AttachmentDelinquency Low SC High SCFigure 5: Effects of Maternal Attachment on Delinqu ency by Levels of Self-Control Note: Model included only the maternal attachment s cale, self-control, and the interaction. Low SC= individuals scoring one standard deviation below the mean on self-control High SC= individuals scoring one standard deviation above the mean on self-control Lastly, Model D includes the results of all variables from Model A and the interaction between paternal attachment and self-control (see Table 14). Once a gain, as in the main effect OLS regression model, paternal attachment proved insignificant a nd the interaction followed suit. Perceptual Deterrence. Model A of Table 15 shows the results of delinquency regressed on perceptual deterrence and self-control, controlling for age, sex, and race. This overall model was statistically significant and able to explain 39.2% of the variance of variet y of delinquent acts committed (F=192.17, p<.05). Perceptual deterrence and self-control were bot h significant, negative predictors of delinquency ( =-.29, p<.05 and =-.38, p<.05 respectively). Therefore, the more one perceives getting caught as a problem and the higher one is in self-control, the less likely one is to be delinquent.
54 Table 15: OLS Regression Results for Perceptual Det errence, Self-Control and Interaction (N=1485) MODEL A b se(b) MODEL B b se(b) Sex .01 .04 .01 .03 .03 .02 Age .10* .01 .24 .09* .01 .22 Race .01 .02 .01 .01 .02 .01 Self Control -.06* .00 -.38 -.06* .00 .39 P Deterrence -.08* .01 -.29 -.09* .01 .34 SC x P Deterrence -.004* .00 -.12 Adjusted R 2 .392 .403 *p<.05 two tailed test Note: SC= self-control, P Deterrence= Perceptual De terrence Model B shows the results of OLS regression analyses for all variables from Model A with the addition of the interaction between perceptual deterrence and self control (see Table 15). This interaction was statistically significant and ne gative ( =-.12, p<.05), indicating that perceptual deterrence had a stronger negative effect on delinquency among those higher in self-control (see Figure 6). In simple terms perceptual deterrence is more influential in reducing delinquency among those hi gher in self-control. Additionally, the R 2 change was statistically significant, indicating that the introduction of the interaction term significantly increased the model fit (R 2 Change=.01, p<.05).
55 0 0.5 1 1.5 2 2.5 -1SD+1SD Perceptual DeterrenceDelinquency Low SC High SCFigure 6: Effects of Perceputal Deterrence on Delin quency by Levels of Self-Control Note: Model included only the perceptual deterrence scale, self-control, and the interaction. Low SC= individuals scoring one standard deviation below the mean on self-control High SC= individuals scoring one standard deviation above the mean on self-control Strain. Table 16 presents the results of OLS regression with delinquency and subjective strains. In Model A, delinquency was regressed on the subjective strains scal e and selfcontrol to assess the main effects controlling for age, sex, and race. Overall, t he first model was statistically significant and able to explain 34.1% of the variance in t he variety of delinquent acts committed (F=155.77, p<.05). Subjective strains had a positive and statistically significant relationship with delinquency ( =.13, p<.05), while self-control has a negative and statistically significant relationship with delinquency ( =-.48, p<.05). The more those who experience strains perceive it as a problem and the less self -control they have, the more likely they are to participate in delinquent activity.
56 Table 16: OLS Regression Results for Subjective Str ains, Self-Control, and Interaction (N=1494) MODEL A B se(b) MODEL B b se(b) Sex -.07 .04 -.04 -.07 .04 -.04 Age .12* .01 .28 .12* .01 .28 Race .02 .02 .02 .02 .02 .02 Self Control -.07* .00 -.48 -.07* .00 .48 Subjective Strains .02* .00 .13 .02* .00 .14 SC x Subjective Strains .001* .00 .05 Adjusted R 2 .341 .343 *p<.05 two tailed test Note: SC= self-control In Model B, all variables from Model A were included with the addition of an interaction between subjective strains and self-control. The interaction proved t o be significant at p<.05 ( =.05). The significant interaction suggests that strains have a greater effect on delinquency for those higher in self-control. In other words, t hose higher in self-control are more likely to react to subjective strains with delinquency (see Figure 7). Additionally, the inclusion of the interaction term resulted in a significant incr ease in variance explained (R 2 change=.002, p<.05). 0 0.5 1 1.5 2 2.5 -1SD+1SD Subjective StrainsDelinquency Low SC High SCFigure 7: The Effect of Subjective Strain on Delinq uency by Levels of Self-Control Note: Model included only the subjective strain sca le, self-control, and the interaction. Low SC= individuals scoring one standard deviation below the mean on self-control High SC= individuals scoring one standard deviation above the mean on self-control
57 Chapter 5 Discussion and Conclusions Most criminological theory testing has focused on the main effects of theore tical constructs, where rival theories are often pitted against one another in an attem pt to examine which theory is a better predictor of criminal behavior. Others prefer t o integrate different criminological theories in an effort to provide more comprehensive ex planatory models (see Messner et al., 1989). More recently, researchers have examined the interactive nature of various criminological constructs (e.g. Wright et a l., 2001). Some have investigated the extent to which self-control (or similar constructs, such a s impulsivity) moderates the relationships between various criminological cons tructs and antisocial behavior (see Wooton et al., 1997, Carlo et al., 1998; Agnew et al., 2002; Jones et al., 2007). However, consistent support is lacking, leaving little to be reliably concluded. Specifically, two major hypotheses exist that suggest very different findings. The first suggests that those most at-risk will be more strongly influence d by external factors. The other proffers that the least at-risk will be more easily sw ayed by external influences. Both of these hypotheses have received empirical support resulting in much ambiguity. In order to test these two competing arguments about the potential moderating role of self-control, this study sought to remedy the equivocal findings by focusi ng on a broad array of criminological constructs more so than any single study to date. Thr ee reasons could be cited for the divergent findings. First, the divergent findings pr eviously noted could be the result of idiosyncratic sample characteristics. That is, ther e may be
58 something peculiar to the samples used in previous research that is influencing the findings. This raises the possibility that the interactive effects are not r obust. A second possibility is that different operationalizations of at-risk individuals have been used. For example, Wright et al. (2001) have relied on a measure of self-control that incl udes symptoms of ADHD and antisocial behavior generated from various sources (e.g., parents, teachers, etc.). Others have focused on particular personality construc ts such as negative emotionality and constraint (Agnew et al., 2002) or impulsivity (Lynam et al., 2000) as operationalizations of self-control. Lastly, several studies have ex amined only one or a few important criminological constructs and how self-control (or a relat ed concept) moderated their relationship with delinquency (e.g. Wooton et al., 1997; Piquero & Pogarsky, 2002). That is, important criminological constructs have been neglect ed that may in fact have important interactive effects with self-control. The curr ent study sought to partially remedy these issues by relying on a single sample, using a comprehensive measure of self-control, and examining multiple theoretical constructs. Whil e this strategy cannot offer a definitive conclusion to these issues, they may help to res olve some of the ambiguity and thus offer fruitful guidance for future research. One theory examined was social learning theory. Specifically, peers, defini tions, and reinforcements were explored, and the extent to which their effects were modi fied by self-control. No study to date has included definitions and reinforcements, two centr al components to social learning theory, in the context of the interactive hypothesis. As expected, peers and definitions were very strong direct predictors of delinquenc y, while reinforcements were slightly weaker. Self-control also exerted signifi cant influence on delinquency, but demonstrated a smaller effect size than peers and definitions. All three
59 social learning constructs significantly interacted with self-contr ol in the same direction. As such, delinquent peers, criminal definitions, and criminal reinforcements all ex erted a significantly stronger effect among those higher in self-control. Constructs representing control theory (i.e. parental control, maternal attach ment, and paternal attachment) were also examined. While paternal attachment was not significantly related to self-reported delinquency, parental control and mater nal attachment were both significant correlates of lower delinquency. The lack of significance for paternal attachment could be due to the fact that matern al attachment may overwhelm this factor. That is, mothers might play a more central role in t he adolescents life, and while fathers may be important by themselves, they do not of fer any additional control beyond that of mothers. Additionally, self-control had the strongest direct effect on delinquency in this model. Interactions were found for both parental control and maternal attachment, which, once again, indicated that these fa ctors were more influential among those higher in self-control. Deterrence theory was the third criminological theory examined. Both perceptua l deterrence and self-control exerted significant main effects on delinquency, with selfcontrol being the strongest predictor in the model. The interaction of the two further added to the model, and indicated that perceptual deterrence is more influential for thos e higher in self-control. Lastly, strain theory was tested. Subjective strains were found to have sig nificant main effects on delinquency. Once again, self-control was the strongest dire ct predictor in the model. As in the previous models that incorporated other theoretical constructs ( see above), the interaction between subjective strains and self-control was signifi cant and
60 indicated that subjective strains played a more important role for those higher in se lfcontrol. Collectively, the results consistently indicated that external factors we re more influential among those high in self-control. That is, with the exception of paternal attachment (which failed to demonstrate a significant main effect as we ll), all external factors measured in this study induced those with high self-control to conform or devia te more so than low self-control counterparts. Stated alternatively, those with l ower selfcontrol were influenced less by peers, parents, and strains, among other factors than those evincing higher self-control. Delinquent peers, pro-criminal definitions, dif ferential reinforcements of crime, and subjective strain increased the likelihood of delin quency while perceptual deterrence, parental control, and maternal attachment decr eased the likelihood of delinquency to a greater extent among those higher in self-control com pared to those low in self-control. This is in contrast to some previous arguments, and suggests that the interdependence model may, in fact, work in the opposite direction of that proposed by Wright and his colleagues (2001). With several controversial and competing arguments concerning the role of selfcontrol, these findings bear important implications that warrant further discuss ion. In particular, the explanatory power of constructs derived from the theories exa mined here (and potentially all criminological theories) may be partially conting ent upon ones level of self-control. This suggests that Lewin (1935) and Magnusson (1988) were both correct in suggesting that similar environments can have a differential impact on two indi viduals as a result of their individual characteristics. That is, this study has found that behavior is better explained and therefore better predicted when the interaction between the person
61 and environment is taken into account. Furthermore, this suggests that previous explanations of crime may be partly misspecified if they do not take into consi deration that the impact of some social processes may vary as a function of ones level of se lfcontrol. In order to more accurately predict crime, theories may need to be somew hat modified to account for these individual differences. Specifically, integrating s ocial selection and social causation models via examination of their interactive e ffects will improve the predictive power of the overall model. However, more research is needed to determine what additional social influences vary as a function of self-control. Practically speaking, policies should address all risk factors, both individual and social. The findings from the current study suggest two specific approaches depending on the type of individual being targeted. First, prevention policies should be aimed towar d the acquisition of self-control early in life to avoid later delinquent behavior. Spec ifically, parenting classes aimed at proper supervision, discipline, control, conventional prosoci al attitudes, and strong bonds, among others, will be most cost-effective and influential a s effective parenting will increase the likelihood of acquiring self-control (see Gottfredson & Hirschi, 1990). Individuals low in self-control will also be harder to deter once they reach adolescence and will offend at higher rates throughout the life course. Thus, traditional interventions that target family issues, peer relationships, or ot her life stressors might prove to be ineffective among those low in self-control. Therefore focusing on t his period of early development will be most effective and should have the greatest impac t on overall offending rates (see Gottfredson & Hirschi, 1995). Even if strategies to develop self-control early in life are implemented, thi s does not eliminate the possibility of delinquent behavior among youth with higher levels of
62 self-control. Results of this study indicate that these youth are at an incr eased risk of delinquency if they are exposed to negative external factors (e.g. delinquent pee rs, strain) compared to low self-control counterparts. Therefore, intervention policies that focus on deterring those offenders high in self-control from recidivating should also prove ver y effective. This could be accomplished first by making a concentrated effort to reduce the presence of negative factors, especially delinquent peers, among these youth. These high self-control offenders could also be deterred from recidivating by increasin g exposure to positive external factors (e.g. increased parental control, better matern al attachment). Getting youth involved in prosocial activities (e.g. Boy Scouts, after school clubs, s ports) should also prove successful as this will positively influence youth (e.g. via pro-s ocial peers; forming definitions unfavorable to law violations). This study found that high se lfcontrol youth were more strongly persuaded to conform when they experience these positive factors and thus emphasis placed on positive social influences should decrease delinquent activity. Therefore, youth with higher self-control will make respons ible decisions when they are not exposed to negative social influences, but introducing these influences will induce these youth to deviate when they would not have otherwise. From this it can easily be concluded that any study that neglects either psychologic al or sociological factors will risk misspecification. This study is not without its limitations, however. Since only cross-sectional dat a were analyzed, it cannot be concluded that the findings are a result of developmenta l processes or that these relationships will change as a process of development. A lthough much research attests to this, it is not the focus here. Additionally, not all variabl es of each theoretical tradition were examined and fully measured in this study. T herefore, this
63 study cannot be conceptualized as a comprehensive and definitive test of the inter actional hypothesis. That being said, this study did investigate more theoretical construc ts than any other to date. Lastly, the results of the current study might not be genera lizable. The sample used in this analysis was drawn from middle and high school students from Largo, Florida. It is unknown to what degree this sample differs from others that w ould be of interest to criminologists. Despite these limitations, this study has provided a useful test of competing arguments regarding the moderating role of self-control by examining how rel ationships between constructs derived from several criminological theories varied in the ir effects on delinquency as a function of ones self-control. Ultimately, results from this study suggest that further replication is needed to confirm if, in fact, it is those indivi duals with higher levels of self-control who are more strongly influenced by social fa ctors. If this is the case, the theoretical and practical implications previously listed should be taken into consideration as they will likely contribute to reducing delinquency among adoles cents.
65 References Agnew, R. (1985). Social control theory and delinquency: A longitudinal test. Criminology 23, 1, 47-61. Agnew, R., (1992). Foundation for a general strain theory of crime and delinquency. Criminology 30, 1, 47-88. Agnew, R. (1993). Why do they do it? An examination of the intervening mechanisms between social control variables and delinquency. Journal of Research in Crime and Delinquency, 30, 3, 245-266. Agnew, R. (2006). General strain theory: Current status and directions for further research. In F. Cullen, J. Wright, and K. Blevins (Eds), Taking Stock: The Status of Criminological Theory (pp. 101-123). New Brunswick, NJ: Transaction Publishers. Agnew, R., & Brezina, T. (1997). Relational problems with peers, gender, and delinquency. Youth & Society 29, 1, 84-111. Agnew, R., Brezina, T., Wright, J., & Cullen, F. (2002). Strain, personality traits, and delinquency: Extending general strain theory. Criminology 40, 1, 43-71. Akers, R. (1973). Deviant behavior: A social learning approach. Belmont, CA: Wadsworth. Akers, R. (1985). Deviant behavior: A social learning approach 3 rd ed. Belmont, CA: Wadsworth. Akers, R. (1998). Social learning and social structure: A general theory of crime and deviance. Boston: Northeastern University Press.
66 Akers, R., & Cochran, J. (1985). Adolescent marijuana use: A test of three theories of deviant behavior. Deviant Behavior 6, 323-346. Akers, R., Greca, A., Cochran, J., & Sellers, C. (1989). Social learning theory and alcohol behavior among the elderly. The Sociological Quarterly 30, 4, 625-638. Akers, R., & Jensen, G. (Eds) (2003). Social learning theory and the explanation of crime. New Brunswick, NJ: Transaction Publishers. Akers, R. & Jensen, G. (2006). The empirical status of social learning theory of cri me and deviance: The past, present, and future. In F. Cullen, J. Wright, and K. Blevins (Eds), Taking Stock: The Status of Criminological Theory (pp. 37-76). New Brunswick, NJ: Transaction Publishers. Akers, R., Krohn, M., Lanza-Kaduce, L., & Radosevich, M. (1979). Social learning and deviant behavior: A specific test of a general theory American Sociological Review 44, 4, 636-655. Akers, R., & Lee, G. (1999). Age, social learning, and social bonding in adolescent substance use. Deviant Behavior 19, 1-25. Akers, R., & Sellers, C. (2004). Criminological theories: Introduction, evaluation, and application 4 th Ed. Los Angeles: Roxbury Publishing Company. Aseltine, R., 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, 3, 256-275. Averill, J. (1993). Illusions of anger. In R. Felson and J. Tedeschi (Eds), Aggression and Violence. New York: Springer-Verlag.
67 Bandura, A. (1969). Principles of behavior modification New York: Holt, Rinehart, and Winston, Inc. Baron, S., & Hartnagel, T. (1997). Attributions, affect, and crime: Street youths reactions to unemployment. Criminology 35, 3, 409-434. Beccaria, C. (1963). On crimes and punishments Indianapolis, IN: The Bobbs-Merrill Company, Inc. Berkowitz, L. (1993). Aggression: Its causes, consequences, and control Philadelphia: Temple University Press. Bentham, J. (1948). The Principles of Morals and Legislation. New York: Hafner Press. Bohm, R. (2001). A primer on crime and delinquency theory 2 nd Ed. Belmont, CA: Wadsworth/Thomson Learning. Brezina, T., & Piqeruo, A. (2003). Exploring the relationship between social and nonsocial reinforcement in the context of social learning theory. In R. Akers and G. Jensen (Eds), Social Learning Theory and the Explanation of Crime (pp. 265288). New Jersey: Transaction Publishers. Broidy, L. (2001). A test of general strain theory. Criminology 39, 1, 9-35. Bronfenbrenner, U. (1988). Interacting systems in human development: Research paradigms: Present and future. In N. Bolger; A. Caspi, G. Downey, and M. Moorehouse (Eds.), Persons in Context: Developmental Processes ( pp. 25-49) New York: Cambridge University Press. Burgess, R., & Akers, R. (1966). A differential association-reinforcement theor y of criminal behavior Social Problems 14, 2, 128-147.
68 Carlo, G., Roesch, S., & Melby, J. (1998). The multiplicative relations of parenting and temperament to prosocial and antisocial behaviors in adolescence. Journal of Early Adolescence 18, 3, 266-290. Caspi, A., Moffitt, T., Silva, P., Stouthamer-Loeber, M., Krueger, R., & Schmutte, P. (1994). Are some people crime-prone? Replications of the personality-crime relationship across countries, genders, races, and methods. Criminology 32, 2, 163-196. Cloward, R., & Ohlin, L. (1960). Delinquency and opportunity: A theory of delinquent gangs New York: Free Press. Cochran, J., Wood, P., Sellers, C., Wilkerson, W., & Chamlin, M. (1998). Academic dishonesty and low self-control: An empirical test of a general theory of crim e. Deviant Behavior, 19, 227-255. Cohen, Albert K. (1955). Delinquent boys: The culture of the gang New York: The Free Press. Cohn, E., Farrington, D., & Wright, R. (1998). Evaluating criminology and criminal justice Westport, CT: Greenwood Press. Cullen, F., & Agnew, R. (Eds) (2003). Criminology theory: Past to present 2 nd Ed. Los Angeles: Roxbury Publishing Company. Cullen, F., Wright, J., & Blevins, K. (Eds). (2006). Taking Stock: The Status of Criminological Theory New Brunswick, NJ: Transaction Publishers. DeKemp, R., Scholte, R., Overbeek, G., & Engels, R. (2006). Early adolescent delinquency: The role of parents and best friends. Criminal Justice and Behavior 33, 4, 488-510.
69 Dornbusch, S., Erickson, K., Laird, J., & Wong, C. (2001). The relation of family and school attachment to adolescent deviance in diverse groups and communities. Journal of Adolescent Research 16, 396-422. Evans, T., Cullen, F., Burton, V., Dunaway, G., & Benson, M. (1997). The social consequences of self-control: Testing the general theory of crime. Criminology 35, 3, 475-504. Gibbs, J. (1975). Crime, punishment and deterrence New York: Elsevier. Gottfredson, M. (2006). The empirical status of control theory in criminology. In F. Cullen, J. Wright, and K. Blevins (Eds), Taking Stock: The Status of Criminological Theory (pp. 77-100). New Brunswick, NJ: Transaction Publishers. Gottfredson, M., & Hirschi, T. (1990). A general theory of crime Stanford, CA: Stanford University Press. Gottfredson, M. & Hirschi, T. (1995). National crime control policies. Society 32, 2, pp. 22-28. Grasmick, H., & Green, D. (1980). Legal punishment, social disapproval and internalization as inhibitors of illegal behavior. Journal of Criminal Law and Criminology, 71, 3, 325-335. Grasmick, H., Tittle, C., Bursik, R., & Arneklev, B. (1993). Testing the core empirical implications of Gottfredson and Hirschis general theory of crime. Journal of Research in Crime and Delinquency 30, 1, 5-29. Hirschi, T. (1969). Causes of delinquency Berkeley: University of California Press. Hirschi, T. (2002). Causes of delinquency New Brunswick, NJ: Transaction Publishers.
70 Hirschi, T. (2004). Self-Control and Crime. In R. Baumeister and K.Vohs (Eds.), Handbook of Self-Regulation: Research, Theory, and Applications (pp. 537-552). New York: The Guilford Press. Hirschi, T., & Gottfredson, M. (1993). Commentary: Testing the general theory of cri me. Journal of Research in Crime and Delinquency 30, 1, 47-54. Hirschi, T., & Gottfredson, M. (1994). The generality of deviance. In T. Hirschi and M. Gottfredson (Eds.), The Generality of Deviance (pp. 1-22). New Brunswick, NJ: Transaction. Jeglum-Bartusch, D., Lynam, D., Moffitt, T., & Silva, P. (1997). Is age important? Testing a general versus a development theory of antisocial behavior. Criminology 35, 13-48. Jensen, G. (1969). Crime doesnt pay: Correlates of a shared misunderstanding. Social Problems 17, 189-201. Jensen, G., Erickson, M., & Gibbs, J. (1978 ). Perceived risk of punishment and selfreported delinquency Social Forces 57, 1, 57-78. Jones, S., Cauffman, E., & Piquero, A. (2007). The influence of parental support among incarcerated adolescent offenders: The moderating effects of self-control. Criminal Justice and Behavior 34, 2, 229-245. Keane, C., Maxim, P. & Teevan, James. (1993). Drinking and driving, self-control, and gender: Testing a general theory of crime. Journal of Research in Crime and Delinquency 30, 1, 30-46. Klepper, S. & Nagin, D. (1989). The deterrent effect of perceived certainty and se verity of punishment revisited. Criminology 27, 721-746.
71 Kochanska, G. (1993). Toward a synthesis of parental socialization and child temperament in early development of conscience. Child Development 64, 325347. Krohn, M., & Massey, J. (1980). Social control and delinquent behavior: An examination of the elements of the social bond. The Sociological Quarterly 21, 529-543. Krohn, M., Lanza-Kaduce, L., & Akers, R. (1984). Community context and theories of deviant behavior: An examination of social learning and social bonding theories. The Sociological Quarterly 25, 353-371. Krohn, M., Massey, J., Skinner, W., & Lauer, R. (1983). Social bonding theory and adolescent cigarette smoking: A longitudinal analysis. Journal of Health and Social Behavior 24, 4, 337-349. Krohn, M., Skinner, W., Massey, J., & Akers, R. (1985). Social learning theory and adolescent cigarette smoking: A longitudinal study. Social Problems 32, 5, 455473. Lahey, B., Moffitt, T., & Caspi, A. (Eds) (2003). Causes of conduct disorder and juvenile delinquency. New York: Guildford Press. Lahey, B., & Waldman, I. (2003). A developmental propensity model of the origins of conduct problems during childhood and adolescence. In B. Lahey, T. Moffitt, & A. Caspi (Eds), Causes of Conduct Disorder and Juvenile Delinquency (pp. 76117). New York: Guildford Press. Lanier, M., & Henry, S. (2004). Essential criminology 2 nd Ed. Boulder, CO: Westview Press.
72 Lee, G., Akers, R., & Borg, M. (2004). Social learning and structural factors in adolescent substance use. Western Criminological Review 5, 1, 17-34. Lewin, K. (1935). A dynamic theory of personality New York: McGraw-Hill Book Company, Inc. Liska, A., & Messner, S. (1999). Perspectives on crime and delinquency 3 rd Ed. Upper Saddle River, NJ: Prentice-Hall, Inc. Loeber, R., & Stouthamer-Loeber, M. (1986). Family factors as correlates and pr edictors of juvenile conduct problems and delinquency Crime and Justice 7, 29-149. Longshore, D., Chang, E., & Messina, N. (2004). Self-control and social bonds: A combined control perspective on juvenile offending. Journal of Quantitative Criminology 21, 4, 542-564. Lynam, D., Wikstrm, P., Caspi, A., Moffitt, T., Loeber, R., & Novak, S. (2000). The interaction between impulsivity and neighborhood context on offending: The effects of impulsivity are stronger in poorer neighborhoods. Journal of Abnormal Psychology 109, 4, 563-574 Magnusson, D. (1988). Individual development from an interactional perspective: A longitudinal study New Jersey: Lawrence Erlbaum Associates, Inc. Matsueda, R., & Heimer, K. (1987). Race, family structure, and delinquency: A test of differential association and social control theories. American Sociological Review 52, 6, 826-840. Mazerolle, P., Burton, V., Cullen, F., Evans, T., & Payne, G. (2000). Strain, anger, and delinquent adaptations: Specifying general strain theory. Journal of Criminal Justice 28, 89-101.
73 Mazerolle, P. & Piquero, A. (1998). Linking exposure to strain with anger: An investigation of deviant adaptations. Journal of Criminal Justice 26, 195-211. Meier, R., & Johnson, W. (1977). Deterrence as social control: The legal and extrale gal production of conformity. American Sociological Review 42, 2, 292-304. Merton, R. (1938). Social structure and anomie. American Sociological Review, 3, 672682. Messner, S., Krohn, M., & Liska, A. (Eds). (1989). Theoretical integration in the study of deviance and crime: Problems and prospects Albany: State University of New York Press. Mitchell, O. & MacKenzie, D. (2006). The stability and resiliency of self-cont rol in a sample of incarcerated offenders. Crime and Delinquency 52, 3, 432-449. Moffitt, T. (1993). Adolescence-limited and life-course-persistent antisocial behavior: a developmental taxonomy. Psychological Review 100, 4, 674-701. Moffitt, T. (2003). Life-course persistent and adolescence-limited antisocia l behavior. In B. Lahey, T. Moffitt, & A. Caspi (Eds), Causes of Conduct Disorder and Juvenile Delinquency (pp. 49-75). New York: Guildford Press. Moffitt, T., & Caspi, A. (2001). Childhood predictors differentiate life-course persist ent and adolescence-limited antisocial pathways among males and females. Development and Psychopathology 13, 2, 355-375. Nagin, D., & Paternoster, R. (1993). Enduring individual differences and rational choice theories of crime. Law and Society Review 27, 3, 467-496.
74 Nagin, D., & Pogarsky, G. (2001). Integrating celerity, impulsivity, and extra legal sanction threats into a model of general deterrence: Theory and evidence. Criminology 39, 4, 865-891. Ousey, G. & Wilcox, P. (in press). The interaction of antisocial propensity and lif e-course varying predictors of delinquent behavior: Differences by method of estimation and implications for theory. Paternoster, R. (1987). The deterrent effect of the perceived certainty and seve rity of punishment: A review of the evidence and issues. Justice Quarterly 4, 2, 173217. Paternoster, R., & Bachman, R. (Eds) (2001). Explaining criminals and crime: Essays in contemporary criminological theory Los Angeles: Roxbury Publishing Company. Paternoster, R., & Brame, R. (1998). The structural similarity of processes g enerating criminal and analogous behaviors. Criminology 36, 3, 633-669. Paternoster, R., Saltzman, L., Waldo, G., & Chiricos, T. (1985). Assessments of risk and behavioral experience: An exploratory study of change. Criminology 23, 417433. Piquero, A., & Pogarsky, G. (2002). Beyond Stafford and Warrs reconceptualization of deterrence: Personal and vicarious experiences, impulsivity, and offending behavior. Journal of Research in Crime and Delinquency 39, 2, 153-186. Piquero, A., & Rengert, G. (1999). Studying deterrence with active residential burg lars. Justice Quarterly 16, 2, 451-471.
75 Piquero, N., & Sealock, M. (2000). Generalizing general strain theory: An exami nation of an offending population. Justice Quarterly 17, 3, 449-484. Piquero, A., & Tibbetts, S. (1996). Specifying the direct and indirect effects of low se lfcontrol and situational factors in offenders decision making: Toward a more complete model of rational offending. Justice Quarterly 13, 3, 481-504. Pogarsky, G. (2007). Deterrence and individual differences among convicted offend ers. Journal of Quantitative Criminology 23, 5974. Pratt, T. & Cullen, F. (2000). The Empirical Status of Gottfredson and Hirschis Ge neral Theory of Crime: A Meta-Analysis. Criminology, 38, 3, 931-964. Pratt, T., Cullen, F., Blevins, K., Daigle, L., & Madensen, T. (2006). The empirical status of deterrence theory: A meta-analysis. In F. Cullen, J. Wright, and K. Blevins (Eds), Taking Stock: The Status of Criminological Theory (pp. 367-395). New Brunswick, NJ: Transaction Publishers. Rankin, J., & Kern, R. (1994). Parental attachments and delinquency. Criminology 32, 4, 495-515. Simons, R., Wu, C., Conger, R., & Lorenz, F. (1994). Two routes to delinquency: Differences between early and late starters in the impact of parenting and deviant peers. Criminology 32, 2, 247-276. Skinner, B. (1953). Science and human behavior New York: Macmillan. Stafford, M. & Warr, M. (1993). A reconceptualization of general and specific deterrence. Journal of Research in Crime and Delinquency 30, 2, 123-135. Thaxton, S., & Agnew, R. (2004). The nonlinear effects of parental and teacher attachment on delinquency. Justice Quarterly 21, 4, 763-791.
76 Tittle, C. (1977). Sanction fear and the maintenance of social order. Social Forces 55, 579-596. Tittle, C., Ward, D., & Gramick, H. (2003). Self-control and crime/deviance: Cognitive vs. behavioral measures. Journal of Quantitative Criminology 19, 4, 333-365. Warr, M. (2002). Companions in crime: The social aspects of criminal conduct Cambridge, UK: Cambridge University Press. White, H., Pandina, R., & LaGrange, R. (1987). Longitudinal Predictors of serious substance use and delinquency. Criminology 25, 3, 715-740. Wikstrm, P., & Loeber, R. (2000). Do disadvantaged neighborhoods cause well-adjusted children to become adolescent delinquents? A study of male juvenile serious offending, individual risk and protective factors, and neighborhood context. Criminology 38, 4, 1109-1142. Wikstrm. P., & Sampson, R. (2003). Social mechanisms of community influences on crime and pathways in criminality. In B. Lahey, T. Moffitt, and A. Caspi (Eds), Causes of Conduct Disorder and Juvenile Delinquency (p. 118-1148). New York: Guilford Press. Winfree, L., Backstrom, T., & Mays, G. (1994). Social learning theory, self-repor ted delinquency, and youth gangs: A new twist on a general theory of crime and delinquency. Youth & Society 26, 2, 147-177. Wood, P., Gove, W., Wilson, J., & Cochran, J. (1997). Nonsocial reinforcement and habitual criminal conduct: An extension of learning theory. Criminology 35, 2, 335-366.
77 Wooton, J., Frick, P., Shelton, K., & Silverthorn, P. (1997). Ineffective parenting and childhood conduct problems: The moderating role of callous-unemotional traits. Journal of Consulting and Clinical Psychology 65, 2, 301-308. Wright, B., Caspi, A., Moffitt, T., & Paternoster, R. (2004). Does the perceived risk of punishment deter criminally prone individuals? Rational choice, self-control, and crime. Journal of Research in Crime and Delinquency 41, 2, 180-213. Wright, B., Caspi, A., Moffitt, T., & Silva, P. (1999). Low self-control, social bonds, and crime: Social causation, social selection, or both? Criminology 37, 3, 479-514. Wright, B., Caspi, A., Moffitt, T., & Silva, P. (2001). The effects of social ties on crime vary by criminal propensity: A Life-course model of interdependence. Criminology 39, 2, 321-351. Wright, J., & Cullen, F. (2001). Parental efficacy and delinquent behavior: Do control and support matter? Criminology 39, 3, 677-705.
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001928535
007 cr mnu|||uuuuu
008 080225s2007 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001942
The conditional influence of criminological constructs on juvenile delinquency :
b an examination of the moderating effects of self-control
h [electronic resource] /
by Angela Yarbrough.
[Tampa, Fla] :
University of South Florida,
ABSTRACT: Self-control and various elements comprising this construct have received much credit over the years as it has been able to account for a large amount of variance in delinquency rates. Some research has suggested that individual difference factors (e.g., self-control) can overwhelm external factors (e.g., neighborhoods; see Loeber & Wikstrm, 2000). Others have found that social influences (e.g., employment; see Wright, et al, 2001) have more pronounced effects for those most at-risk. Because of the equivocal nature of the empirical findings, this study seeks to replicate and extend previous efforts. Specifically, the influence of constructs derived from social learning, control, deterrence, and strain are examined to see if any vary in their influence on adolescent offending as a function of self-control. Results indicate that all of these theoretical constructs (with the exception of paternal attachment) played a more important role among those who evinced the highest levels of self-control. Implications for criminological theory and criminal justice policy are discussed.
Thesis (M.A.)--University of South Florida, 2007.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 77 pages.
Co-adviser: Shayne Jones, Ph.D.
Co-adviser: Christine Sellers, Ph.D.
General theory of crime.
t USF Electronic Theses and Dissertations.