A Causal Model of Neutralization Acceptance and Delinquency

Abstract

Traditionally, neutralization theory has been conceptualized as a situational strategy employed by offenders to preemptively assuage the guilt they anticipate from contemplated offending and delinquency, and thereby promote offending. While scholars have established that neutralizing and delinquency are related, they have yet to sufficiently determine whether this relationship is causal in nature, or whether neutralizing should be thought of as an individual difference. In this study, we used trajectory analysis and structural equations modeling (SEM) techniques on GREAT (Gang Resistance Education and Training) data to find that juveniles coalesced into four stable and distinct neutralizing and delinquency groups. These trajectories were parallel across ages 12 to 16, and systematically related to each other (e.g., higher neutralizing trajectories with higher delinquency trajectories). Subsequent SEM analysis demonstrated a recursive, causal effect of neutralizations on delinquency. Our results suggest that practitioners develop measures to identify “high” versus “low” neutralizers, which may have ramifications for the offender management and counseling.

Keywords

neutralization theory delinquency individual differences offenders decision-making

In 1957, Gresham Sykes and David Matza published the now classic paper, “Techniques of Neutralization: A Theory of Delinquency” at a time when subcultural theories dominated thought on the causes of crime and delinquency. At the heart of these subcultural perspectives was the idea that working-class boys rebelled against the dominant social order by rejecting middle-class standards and replacing them with countercultural, often delinquent, values (Cloward & Ohlin, 1960; Cohen, 1955; Miller, 1958). Sykes and Matza argued that subcultural theorists overstated the extent to which delinquents rejected conventional values, believing that everyone—even the lower-class delinquent gang member—is influenced by and retains some degree of allegiance or attachment to the mainstream value system of larger society. According to Matza (1964), “That the subcultural delinquent is not significantly different from other boys is precisely the point. He is marginally different and only in process is there a cumulation sufficient to sometimes culminate in infraction” (p. 89, italics added).

However, acts that go against moral beliefs can carry with them feelings of guilt and shame. It is this potential for guilt and the corresponding negative self-image that is hypothesized to dissuade many adolescents from engaging in criminal or delinquent acts (Matza, 1964). Therefore, would-be delinquents must find ways to preemptively neutralize guilt and protect their self-image if they are to carry out the delinquent act. One way to do this is by employing “defenses to crime” (i.e., self-talk justifications and excuses) that provide episodic relief from moral constraints and allow individuals to “drift” from a conventional mindset (where offending is prohibited) to a criminogenic mindset (where offending is permissible). Drift is possible because these techniques temporarily weaken the moral force of mainstream cultural norms antithetical to the contemplated behavior, thereby neutralizing anticipated guilt and facilitating offending without serious damage to the would-be offender’s self-image (Matza, 1964). Thus, actors can remain “committed to the dominant normative system and yet so qualifies [their actions] that violations are ‘acceptable’ if not ‘right’” (Sykes & Matza, 1957, p. 667).

The original framework of neutralization theory has been useful in providing a clear understanding of how the neutralizing process works as a situationally invoked strategy (a cognitive trick of sorts) used by offenders to provide themselves with episodic relief from the deterrent effects of guilt and shame. Much to their credit, Sykes and Matza (1957) explicitly urged future researchers to develop a more “systematic approach” to understanding how this process works (p. 670). Unfortunately, the criminological community appears, for the most part, to have been largely satisfied with the theory as originally formulated and has not prioritized the systematic analysis of the subjective worlds or belief systems of the criminal actors who use them (Maruna & Copes, 2005). Granted, some scholars have elaborated on other neutralizations (e.g., Cromwell & Thurman, 2003), expanded the types of behaviors that neutralization theory can explain (e.g., Shoenberger, Heckert, & Heckert, 2012; Topalli, 2005), and merged it with other theories (see Cornish & Clarke, 1986). Despite this, the theory remains “badly underdeveloped” (Maruna & Copes, 2005, p. 221).

The most obvious issue has been the difficulty researchers have demonstrating the causal effect of neutralizing a priori on offending, a task that would seem to require prospective data. That is, while scholars have established a correlation between delinquency and neutralization acceptance, they have yet to sufficiently determine whether neutralizations come before or after delinquent acts. Another important related issue is whether the process of neutralizing represents an underlying stable belief system for offenders. If so, are some individuals more likely to employ neutralizations than others, and does this have repercussions for offending behavior over time?

While these questions have remained largely ignored in criminology, research in clinical and social psychology supports the notion that excuse-making behaviors (like neutralizing) are indicative of stable, dispositional characteristics that serve people’s desire to maintain a positive self-image and sense of control in the face of behavior they know is wrong or bad (see, for example, Snyder, 1991). Understanding the extent to which neutralizing is a stable, enduring part of one’s personality (i.e., an individual difference) rather than merely a situationally induced strategy may prove to be useful for expanding our understanding of neutralization theory and its usefulness not only as an explanation of criminal events but also as an explanation of criminality. Furthermore, a cornerstone of offender therapy is to convince individuals to accept responsibility for their actions and see themselves and their behaviors more “realistically” (see, for example, Braithwaite, 1999). Ascertaining variation in juveniles’ predilections toward employing neutralizations could prove to be useful for improving offender treatment programs, particularly if one considers that excuse-making may in fact be more adaptive than is traditionally believed (see Maruna & Mann, 2006).

Here, we employed trajectory analysis and structural equation modeling to address the stability of neutralization acceptance and the temporal order issue. Specifically, we sought to determine if there are multiple types of neutralizers, whether their patterns of neutralizing form stable trajectories, and whether these patterns were causally related to criminal outcomes. In doing so, we sought not only to endorse Sykes and Matza’s conceptualization of neutralizations as criminogenic strategies, but also to provide insights into whether neutralization acceptance is a stable, dispositional characteristic of individuals that can explain delinquent behavior over time.

Neutralization Theory, Expanded

Empirical examinations of neutralization theory are not in short supply. Scholars have tested many of its major assumptions, employing two primary approaches. First, researchers have assessed the relationship between neutralization acceptance and self-reported delinquency using a single sample. Scholars using this design have shown that neutralizations are at least moderately associated with participation in deviant acts such as digital piracy (Higgins, Wolfe, & Marcum, 2009; Morris & Higgins, 2009), white-collar crime (Stadler & Benson, 2012; Vieraitis, Piquero, Piquero, Tibbetts, & Blankenship, 2012), workplace deviance (Dabney, 1995), and shoplifting (Agnew & Peters, 1986). Second, qualitative researchers have demonstrated the theory’s usefulness and explanatory power by engaging offenders in interviews about their offending and recollections of the time preceding their crimes. Such work has shown that individuals use neutralizations to explain their participation in misconduct as diverse as drug crimes (Sandberg, 2010), sex crimes (Scully & Marolla, 1984), violent crime (Pogrebin, Stretesky, Unnithan, & Venor, 2006; Topalli, 2006), property crime (Copes, 2003; Holt & Copes, 2010; Shigihara, 2013), and various forms of white-collar crime (Benson, 1985; Copes & Vieraitis, 2009; Steinmetz & Tunnell, 2013).

Overall, qualitative evaluations have provided strong support for the theory; however, those of a quantitative nature have found positive but weak effects of neutralizations on deviance (see Maruna & Copes, 2005). A possible limitation of these earlier examinations of neutralization theory is that they conceptualize neutralizations as a situational tool or strategy, rather than as a stable facet of the offender’s personality. This view of the theory would seem to short-change its potential value as an explanatory mechanism for persistent criminality. Certainly, criminological perspectives concerned with the development of criminality over time, such as the life-course perspective, would benefit from such a conceptualization. Furthermore, Hirschi (1969) and others (Agnew, 1994; Minor, 1981) have suggested that as offenders continue to employ neutralizations over time, they may exhibit a hardening effect whereby they become more and more dedicated to offending and less deterrable (see also, Topalli, 2005). It is therefore important to know if neutralizations are more than just “techniques” called forth in specific situations. Are they, in fact, manifestations of an underlying individual difference¹ between people, such that some are more prone to neutralize than are others? If there are habitual neutralizers, how would this change the way we think of offending over time? To conceive of neutralizing in such a manner requires that we situate it within a larger context of delinquency, criminality, and offender decision-making.

Neutralization theory’s place in the pantheon of criminological theories is sometimes difficult to establish. Certainly, it has been (and should be) linked to other theories that emphasize the individual’s need to align their behaviors with their actions. Included in this group are Mills’s (1940) notion of motive talk, Scott and Lyman’s (1968) theory of accounts, Hewitt and Stokes’s (1975) notion of disclaimers, Festinger’s (1957) theory of cognitive dissonance, and Bandura’s (1999) theory of moral disengagement. Stokes and Hewitt (1976) introduced the notion of “aligning actions” as a catch-all for the behaviors manifested in these theories. Aligning actions include “various tactics, ploys, methods, procedures, and techniques found in social interaction in those circumstances where some feature of a situation is problematic” (Stokes & Hewitt, 1976, p. 838). In their formulation, aligning actions are

largely verbal efforts to restore or assure meaningful interaction in the face of problematic situations of one kind or another, activities such as disclaiming, requesting, and giving accounts, constructing quasi-theoretical explanations of problematic situations, offering apologies, formulating the definition of a situation and talking about motives. (Stokes & Hewitt, 1976, p. 838)

In short, each of these theories, including neutralization theory, suggests that those who engage in criminal, deviant, or delinquent acts seek ways of making sense of their actions and reconciling them with their self-concept. The primary differences between them center on when the excuses, justifications, or neutralizations occur and to whom these accounts are directed. Neutralization theory contends that the excuses and justifications must come before the offense and are used for the benefit of the offender, whereas other perspectives (e.g., Scott and Lyman’s theory of accounts) are more open to the possibility that such excuses come after the offense and have a more social function (e.g., stigma management and identity construction).

Understanding how such processes persist and affect behavior is also a theme of other theories with which neutralization theory traditionally has been associated, particularly differential association (Sutherland, 1947) and social learning (Burgess & Akers, 1966). From a conceptual standpoint, neutralization theory has often been subsumed as part of the social learning process that arises from differential association, wherein delinquents learn antisocial behaviors, beliefs, motives, drives, and rationalizations from other deviant “models.” While Sutherland, and later Burgess and Akers (1966), acknowledged that these ways of thinking and behaving were adopted so as to function over time (and not merely as situationally elicited actions) they did not go so far as to view them as dispositional characteristics of the offenders themselves. In fact, Sutherland acknowledged this as a weakness of the theory in that it made the perspective vulnerable to the criticism that differential association and social learning are purely social, even symbolic, constructions (Sutherland & Cressey, 1970).

Research in psychology suggests that neutralizing may be a dispositional characteristic of individuals. Clinical and social psychologists have conceptualized excuse-making (which would include neutralizations, rationalizations, and accounts) as a stable, enduring personality characteristic with consequences for mental health. For example, Snyder and colleagues (Mehlman & Snyder, 1985; Snyder, 1989; Snyder & Higgins, 1988) have conducted research comparing the adaptiveness of accurately perceiving reality versus the usefulness of illusions about the self in helping individuals attain positive mental health outcomes. Their work was in response to the common notion that for individuals to achieve good mental health, they should have an “accurate” perception of reality, especially when it came to their own lives. Snyder’s view was that this approach ignored the potentially protective benefits of reality negotiation; altering perception of one’s self or circumstances in a manner that deemphasizes negative self-evaluation, responsibility for negative outcomes, and negative outlooks on future outcomes. The benefits that accrue from engaging in reality negotiation include the ability to see oneself as being in control of one’s actions and as being fundamentally “good.” Such favorable self-evaluations have been found to produce positive long-term mental health outcomes for individuals (Snyder, 1991).

Congruent with Sykes and Matza’s theory, and more specifically with Matza’s (1964) concept of drift, Snyder viewed the practice of reality negotiation as a key component of dynamic “self-theories” in the tradition of Adler (1927). Adler’s view was that individuals continuously construct and reconstruct a perception of the self over time through a process of self-reflection and social interaction, a “style of life” that they follow to understand themselves and the world around them. Positive illusions about oneself develop through this process, when people confront what others think of them or what they may think of themselves as they make decisions and act on those decisions. Where such decisions result in disapproval by others or by the self, individuals will often foment illusions about who they are and the outcomes of their actions to preserve a theory of the self where they are “good” people “in control” of their lives.

Snyder (1991) identifies excuse-making as a crucial tool in helping individuals achieve such adaptive illusions. His conceptualization bears obvious connections to Sykes and Matza’s theory of neutralization when he states,

the genesis of an excuse-making sequence is that the individual is linked to, or anticipates being linked to, an act that is negative . . . The act is defined as negative when it does not meet personal or societal standards. (Snyder, 1991, p. 138, italics added)

He further maintains that excuse-making is not just for others but often “the internal audience of oneself” (Snyder, 1991, p. 139). Finally, when Snyder describes the “commonness of this excuse-making process over the course of one’s life” he promotes the notion that such processes are stable and dispositional in nature. He further states, “Increasingly negative acts and increasing linkage to those acts should result in greater threats to the self theory, and in commensurate increments in the force of the reality negotiation process of excuse-making that are necessary” (Snyder, 1991, p. 139).

In the criminological literature more direct empirical evidence of the stability of neutralizing exists. Not long after Matza’s (1964) book, Delinquency and Drift, Hirschi (1969) addressed the notion that delinquent behavior led to excuse-making, which then led to further delinquency. According to Hirschi (1969),

Since a boy may commit delinquent acts episodically over an extended period of time, there is every reason to believe that neutralizations in some sense resulting from the earlier acts are causes of later acts. In fact, if we reject, as we do here, the idea that the delinquent develops a set of beliefs that positively require delinquent behavior, then the development of a series of neutralizing beliefs is exactly what we mean by the “hardening” process that presumably occurs at some point in a delinquent “career.” (p. 708)

Minor (1984) tested Hirschi’s notion of the hardening process using a two-wave panel study of college students, and found statistically weak but consistent support for the hardening process. Consequently, he suggested that future research should employ longitudinal data. Heeding this call, Agnew (1994) analyzed data from two waves of the National Youth Survey to examine the effects of neutralizations on violent behavior. Again, the evidence was weak, though supportive of an effect of neutralizations on behavior over time. Since then, there has been little research to further examine neutralizing as a stable phenomenon or as a part of an overall personality construct (but see Hamlin, 1988; Shields & Whitehall, 1994). Relatedly, there also has been mixed evidence regarding the extent to which neutralizations preemptively facilitate offending (for exceptions, see Agnew, 1994; Minor, 1984; Morris & Copes, 2012).

Finally, in determining whether a given trait or individual difference exists it is important not only to determine if it remains stable over time but if it remains differentiated across different people (see Buss & Greiling, 1999; Fiske & Taylor, 2008). Indeed, S. E. Jones, Miller, and Lynam (2011) argue strongly for a greater consideration of personality in criminological research. Their meta-analysis relating constituents of the five-factor model of personality amply demonstrated that personality dimensions like Agreeableness, Conscientiousness, and Neuroticism are causally linked to antisocial and aggressive behaviors (which would underpin or describe many, if not most of the offending behavior purported to be ascribed to neutralizing). In fact, some lower-order traits identified in their study (such as “deliberation,” which is embedded within Conscientiousness) would seem to have direct relevance to neutralizing.

In criminology, this kind of consideration of “traits” and “personality structures” has tended to take a back seat to approaches that situate stability of criminal behavior (rather than personality) within a developmental or life-course framework. Evidence for the stability of criminality, as well as the conditions under which people change their criminality, is found by differentiating groups of identified individuals along diverging, converging, or cross-cutting paths throughout the life course (see Moffitt, 1993; Nagin, Farrington, & Moffitt, 1995) to classify groups such as “life-course persistent” versus “adolescent limited” (Moffitt, 1993) or identify turning points where a trajectory of criminality changes for the better or worse (Sampson & Laub, 2005).

In this article, our analytic strategy borrows from the methodological principals of both personality psychology and developmental criminology. In personality and trait psychology research, the goal is twofold. First, determine that a personality or trait (which may have emotional and/or cognitive components) is stable over time and occurs with systematic variation in the population (i.e., people are high or low to varying degrees). Second, determine if the identified stable trait is systematically related to a particular behavioral outcome (such as violence). Criminologists have been less interested in the stability of traits as independent variables, and more interested in behavioral proclivities as dependent variables. Thus, developmental and life-course criminologists have been more focused on the stability of offending patterns, in particular, identifying distinct diverging, converging, and/or cross-cutting paths of criminal or delinquent behavior (relying on the use of multiyear longitudinal data).

Here we use longitudinal data to apply the above standards to determine the stability of neutralizing and its causal effect on delinquency. Rather than use trajectory analysis to differentiate behavioral paths related to criminality (as do life-course and developmental criminologists), we use the technique to determine the stability and differentiation assumptions that are central to the personality psychology perspective: If statistically distinct groups emerge from the data and they remain differentiated to the same degree over time this supports an individual difference or trait explanation (something that personality psychologists usually confirm through the use of factor analysis on cross-sectional data) for the effect of neutralizing on delinquent behavior. Furthermore, we sought to confirm the extent to which such a neutralizing trait would have a stable, causal effect on behavior by including it and the behavioral outcome of concern (in this case, delinquency) in a structural equation model. Unlike research in clinical psychology, which most often employs correlational analysis to link personality to behavior, this approach (more common in criminology) allows us to infer reciprocal causality over time. This was especially important for the current research, because in the criminological literature neutralizations are posited to cause delinquency rather than simply be associated (correlated) with it. As such, we tested the following hypotheses:

Hypothesis 1 (Stability Prediction): Neutralization acceptance is a stable characteristic of individuals such that the level of neutralizing is consistent from year-to-year across the data timespan (ages 12-16). That is, people who are very low neutralizers at age 12 remain very low neutralizers across ages 13, 14, 15, to 16 and likewise for other identified groups (e.g., low, moderate, and high neutralizers).

Hypothesis 2 (Individual Difference Prediction): Neutralization acceptance differentiates across two or more statistically distinct groups and such groups are stable across time (ages 12-16). That is, the difference in magnitude between very low neutralizers and other groups (low, moderate, and high) will remain the same across ages 12 to 16, as would the differences across any of these groups (i.e., the trajectories for these groups will be parallel).

Hypothesis 3 (Covariation of Neutralizing and Delinquency Prediction): Differentiated profiles of neutralization acceptance are systematically related to similarly differentiated profiles of delinquency, such that increased acceptance of neutralizations is significantly associated with increased levels of delinquency. That is, more neutralizing is associated with more delinquency, and vice versa.

Hypothesis 4 (Reciprocal Causality Prediction): Earlier neutralizing will predict later delinquency and neutralizing, which will, in turn, predict subsequent neutralizing and delinquency across time (ages 12-16). That is, there will be a recursive causal effect between these variables.

Method

Data

To determine the stability of neutralizations and their causal effect on later delinquency we use data from the Gang Resistance Education and Training (GREAT) program. Data collection for the GREAT program began with 3,500 sixth and seventh graders from 22 schools. These data came from six cities considered representative of midsized cities in the United States (Omaha, Nebraska; Lincoln, Nebraska; Philadelphia, Pennsylvania; Phoenix, Arizona; and Las Cruces, New Mexico). While these data have been used in previous studies (Esbensen & Osgood, 1999; Osgood & Schreck, 2007), our use of them differs in two ways. First, while the original study consists of six waves of data, here we used only data from Waves 2 through 6. We chose to exclude Wave 1 to minimize program effects that may have occurred after this stage. Second, this study captured a specific cohort of individuals who were 12 years old at Wave 2 and followed them until Wave 6 (age 16). This was necessary to ensure that we captured five waves of data from the same individuals. We use five waves to provide enough data to successfully estimate a cubic trajectory (Nagin, 2005).

The current analysis relies only on data from those who were aged 12 in Wave 2 and those from this group who reported to each of the next five waves of the GREAT study and who had nonmissing data on all indicators assessed herein (n = 387).² These data capture development of these youth from ages 12 through 16. The longitudinal design of the GREAT study makes it an ideal candidate for examining the trajectories of neutralizing beliefs and on the longitudinal assessment of neutralization acceptance on delinquency, because it provides data during a phase in life where delinquency is common among adolescents (Moffitt, 1993, 2003).

Measures

Neutralization Acceptance

We used the measure of neutralization acceptance from Esbensen and Osgood (1999), which consisted of items that captured the respondents’ agreement with reasons for engaging in some forms of delinquent behavior. Respondents were asked how much they agreed with nine statements: “Its’ okay to lie if it doesn’t hurt anyone”; “It’s okay to lie to keep friends out of trouble”; “It’s okay to lie to keep you out of trouble”; “It’s okay to steal from those rich enough to replace item”; “It’s okay to take little things from stores, steal if it is the only way to get it”; “It’s okay to physically fight if they hit you first”; “It’s okay to fight to protect rights”; “It’s okay to fight if friends and family are threatened”; and “A small lie is okay if no one is hurt.” The respondents indicated their level of agreement to these statements using a 5-point Likert-type scale that was anchored by strongly disagree and strongly agree. To determine a participant’s neutralization acceptance score we summed their responses to these items. The scores ranged from 9 to 45, with higher scores on this scale indicating greater neutralization acceptance. The internal consistency of these items ranged from .88 at age 12 to .86 at age 16.

Delinquency

Following Esbensen and Osgood (1999), we used 14 items to capture self-reported delinquency. These delinquent behaviors included minor deviance (e.g., lying), status offenses (e.g., skipping class), minor offenses (e.g., avoiding paying for movies or bus rides), property offenses (e.g., stole items greater than US$50, stole a motor vehicle), crimes against persons (e.g., hit someone, used a weapon or force to get money from people), and drug sales (e.g., sold marijuana or other drugs). At each age, the adolescents responded to the various delinquency items by indicating the absolute number of times that they had committed each offense in the past 6 months. Guided by Osgood and Schreck (2007), we dichotomized each item (i.e., 0 = no participation and 1 = participation one or more times) and then summed them to create a participation index (sometimes referred to as a variety index), resulting in a summed measure that ranged from 0 to 14. Higher scores indicate that the adolescent engaged in more types of delinquency. This transformation allowed us to remove the potential bias of overly skewed and kurtotic results. The internal consistency of these items at age 12 to 16 ranged from .75 to .72.

Controls

Next, we included a number of control measures pertaining to demographics and differential association. We included gender (male = 1 and female = 0) and race (1 = Black and 0 = White). Due to neutralization theory being firmly embedded in the theoretical tradition of differential association (Maruna & Copes, 2005), we included two such variables: commitment to delinquent peers and association with delinquent peers.

Commitment to Delinquent Peers

To measure commitment to delinquent peers, we summed participants’ responses to three questions. These three questions asked them the likelihood they would hang out with friends who were getting them in trouble with police, at home, and at school. Responses included not at all likely, a little likely, somewhat likely, likely, and very likely. The internal consistency for the measure was .86.

Association With Delinquent Peers

We used 16 items from Esbensen and Osgood (1999) to capture delinquent peer association. The content of these items included: friends skip school, friends lie to adults, friends destroy property, friends steal less than US$50, friends steal more than US$50, friends go into building to steal, friends steal a motor vehicle, friends hit someone, friends attack with weapon, friends armed robbery, friends sold marijuana, friends sold illegal drugs, friends use tobacco, friends use alcohol, friends use marijuana, and friends use illegal drugs. The items were anchored by (1) none of them and (5) all of them. Responses to these items were summed to create the association with delinquent peers measure. Higher scores on the scale indicate associating with more delinquent peers. The internal consistency for the measure was .95 indicating good internal consistency.

Analysis

We begin our analysis by presenting descriptive data on the general trends and the bivariate correlations. We then present the results of the trajectory models, including the selection process, posterior probability results, and graphical portrayals of neutralization and delinquency trajectories. We determined the final trajectory models using the Bayesian Information Criterion (BIC; B. Jones & Nagin, 2007). According to Nagin (2005), the maximized BIC is used to select the best model. Implementing the BIC to determine the proper model includes the inspection of additional models with multiple functional forms of the trajectories. Nagin argued that it is important to search for the proper fitting model to determine when the BIC has been maximized, and when the BIC begins to degrade. When this occurs, the proper model has to be identified. To further examine the fit, the precision of the trajectory estimates is determined through an examination of the posterior probabilities of group assignment. Nagin argued that posterior probabilities of group assignments that were .70 and above indicated a good fit of the model to the data. Third, we present multinomial regression analysis to explore the covariates that distinguish between the groups of neutralizers and delinquents. Fourth, we present cross-tabulations of the neutralization and delinquency trajectory groups to determine how they intersect.

Finally, we present a structural equation model (i.e., path analysis model) that examines the temporal ordering of neutralization and delinquency. Using Mplus 6.2, we examined the fit between the model and the data. Specifically, we examined the chi-square statistic, which should not be statistically significant (Kline, 2005). Because this is not likely to occur, given chi-square is sensitive to large samples and may give a Type I error, we examined three other measures of fit: comparative fit index (CFI; standard is below .95), root mean standard error of approximation (RMSEA; standard is below .06), and standardized root mean of the residual (SRMR; standard is below .05; see Hu & Bentler, 1999). After settling the fit between the model and the data, we provide significant paths between the variables. To make sure that we focus on the impact of the measures, we present the standardized solutions.

Results

The first step consisted of descriptive statistics and bivariate correlations.³ As shown in Table 1, the bivariate correlations for the neutralization measures from age 12 to 16 range from .51 to .77. These correlations indicate test-retest reliability of the measures. The delinquency measures from age 12 to 16 also indicate reasonable test-retest reliability (r = .34- .72). The commitment to delinquent peers is correlated with neutralization (r = .27-.51) and delinquency (r = .19-.37). The association with delinquent peers is correlated with neutralization (r = .32-.59) and delinquency (r = .32-.69).

Table 1:

Descriptive and Pearson’s Product–Moment Correlations Among Risk-Taking and Delinquency at Ages 12 to 16

Measure	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. Neutralization (12)	1.00
2. Neutralization (13)	.67**	1.00
3. Neutralization (14)	.55**	.68**	1.00
4. Neutralization (15)	.51**	.65**	.77**	1.00
5. Neutralization (16)	.52**	.60**	.65**	.74**	1.00
6. Delinquency (12)	.59**	.43**	.35**	.33**	.34**	1.00
7. Delinquency (13)	.46**	.57**	.44**	.47**	.43**	.48**	1.00
8. Delinquency (14)	.45**	.46**	.60**	.50**	.46**	.49**	.61**	1.00
9. Delinquency (15)	.39**	.43**	.51**	.53**	.51**	.44**	.56**	.72**	1.00
10. Delinquency (16)	.32**	.38**	.35**	.47**	.52**	.34**	.36**	.49**	.56**	1.00
11. Commitment to delinquent peers (12)	.51**	.35**	.29**	.27**	.28**	.34**	.37**	.35**	.28**	.19**	1.00
12. Association with delinquent peers (12)	.59**	.42**	.33**	.32	.36**	.69**	.43**	.39**	.41**	.32**	.36**	1.00
13. Sex	.25**	.22**	.19**	.23**	.27**	.16**	.23**	.09	.18**	.20**	.02	.23**	1.00
14. Race	.04	.012*	.10*	.06	.09	.04	.00	.09	.05	.05	.03	.03	−.05	1.00
M	22.69	29.68	30.74	29.36	28.40	.94	1.70	1.69	1.67	1.67	5.91	14.47	.42	.11
SD	7.24	9.14	8.74	8.15	7.91	1.49	2.38	2.26	2.30	2.05	3.00	5.40	—	—

Note. Numbers in parentheses indicate the adolescents’ age in years at the time of measurement. Scale for neutralization ranged from 15 to 55 and for delinquency the range was 0 to 14.

p < .05. **p < .01. ***p < .001.

After acknowledging these selection criteria and applying numerous iterations for the trajectory models, our second step was to determine the quadratic models that provided the best fit for the neutralization and delinquency trajectories. On the basis of the maximized BIC, a four-group model was selected for neutralization and delinquency.⁴ The posterior probabilities for neutralization and delinquency groups were above Nagin’s (2005) .70 standard implying good model fit.

Figure 1 presents the four distinct trajectories⁵ of neutralization from age 12 to 16: (a) very low level neutralization (13.21%); (b) low level neutralization (38.04%); (c) moderate level neutralization (32.86%); and (d) high level neutralization (15.90%).⁶ Very low level neutralization follows a trajectory that begins slightly above a score of 15 at age 12, increases to nearly a score of 20 at age 14, and then declines through age 16. Low level neutralization follows a similar trajectory but it begins with a score of 20, increases to and peaks at age 14 above a score of 25, and then declines to a nearly a score of 25 by age 16. Moderate level neutralization begins at a score of nearly 25 at age 12, increases and peaks at age 14 at a score nearly at 35, and then decreases to a score nearly of 33 by age 16. High level neutralization follows a trajectory that begins with a score of 31 at age 12, increases and peaks around a score of 43 at age 14, then decreases to below 40 at age 16. From these trajectories, high level of neutralization has the highest level of neutralizing belief acceptance.

Figure 1:

Neutralization Trajectories From Age 12 to 16

Figure 2 presents the four distinct trajectory groups for delinquency, and these were as follows: (a) abstainers (19.12%), (b) increasing delinquency (50.14%), (c) moderate level delinquency (24.65%), and (d) high level delinquency (6.05%).⁷ The abstaining delinquency group follows a trajectory that begins with 0 delinquencies at age 12 and continues to be 0 at age 16. The increasing delinquency group follows a trajectory that begins with nearly 1 delinquent act and increases to slightly above 1 delinquent act by the age of 16. The moderate level delinquency group follows a trajectory that begins with nearly 2 delinquent acts at age 12 that peaks at age 15 and then declines to below 3 delinquent acts at age 16. The high level delinquency group follows a trajectory that begins with 4 delinquent acts that peaks at age 14 and then decreases to less than 6 delinquent acts by age 16.

Figure 2:

Delinquency Trajectories From Age 12 to 16 Years

The third step is a multinomial regression analysis that explores the covariates of the neutralization trajectories. The very low level neutralization group is the reference category and membership in the other trajectories is compared with this group. Table 2 shows no covariates distinguish between membership to the low neutralization group and very low neutralization group. Individuals who have more of a commitment to delinquent peers, exp(b) = 1.27; more delinquent peer associations, exp(b) = 1.34; and are male, exp(b) = 2.34, are more likely to belong to the moderate level neutralization group than the very low level neutralization group. Individuals who have more of a commitment to delinquent peer association, exp(b) = 1.51; more delinquent peer associations, exp(b) = 1.19; and are male, exp(b) = 3.34, are more likely to belong to the high level neutralization group than the very low level neutralization group.

Table 2:

Multinomial Covariates of Neutralization Trajectory Groups

	G2 vs. G1			G3 vs. G1			G4 vs. G1
Measure	B	SE	Exp(b)	b	SE	Exp(b)	b	SE	Exp(b)
Commitment to delinquent peers	0.16	0.09	1.18	0.24**	0.09	1.27	0.41***	0.10	1.51
Delinquent peers association	0.17	0.10	1.18	0.30**	0.10	1.34	0.37***	0.10	1.19
Male	0.09	0.38	1.09	0.85*	0.39	2.34	1.21*	0.48	3.34
Black	0.08	0.11	1.09	0.09	0.12	1.09	0.19	0.14	1.20
χ² = 124.35, p = .00	−2 log likelihood = 634.59	Nagelkerke R² = .31	McFadden R² = .13	Cox and SnellR² = .29

p < .05. **p < .01. ***p < .001.

The fourth step in the analysis is a presentation of the cross-tabulation analysis of the neutralization and delinquency trajectory groups. Table 3 shows the covariates of the delinquency trajectories. The abstainer group is the reference category and membership in the other trajectories is compared with this group. Individuals with more delinquent peer associations are (Exp(b)=1.50) more likely to belong to the increasing delinquency group than abstainer group. Individuals with more commitment to delinquent peers, exp(b)=1.22; more delinquent peer associations, exp(b)=1.67; male, exp(b)=2.61; black, 2.35, are more likely to belong to the moderate level delinquency group than the abstainer group. Individuals with more commitment to delinquent peer association, exp(b)=1.45 and more delinquent peer associations, exp(b)=1.40, are more likely to belong to the high level delinquency group than the abstainer group. Table 4 shows that those abstainers from delinquency were in very low level and low level neutralization groups (89%). However, as the level of neutralization increased the level of delinquency increased: 86.3% of those in increasing delinquency were in the moderate groups of neutralization and 84% of those in the moderate level delinquency group were in the moderate to high groups of neutralization. A small subgroup of individuals (n = 21) were high on delinquency and high on neutralization.

Table 3:

Multinomial Covariates of Delinquency Trajectory Groups

	G2 vs. G1			G3 vs. G1			G4 vs. G1
Measure	b	SE	Exp(b)	b	SE	Exp(b)	b	SE	Exp(b)
Commitment to delinquent peers	0.11	0.07	1.10	0.19**	0.08	1.22	0.37***	0.10	1.45
Delinquent peers association	0.41***	0.11	1.50	0.51***	0.11	1.67	0.56***	0.11	1.40
Male	0.28	0.33	1.32	0.96**	0.39	2.61	0.81	0.60	2.25
Black	0.08	0.10	1.09	0.23*	0.11	1.25	0.16	0.17	1.17
χ² = 121.67, p = .00	−2 log likelihood = 560.26	Nagelkerke R² = .33	McFadden R² = .15	Cox and SnellR² = .30

p < .05. **p < .01. ***p < .001.

Table 4:

Cross-Tabulation of Trajectory Group Memberships

	Neutralization
Delinquency	G1	G2	G3	G4
G1	33 (40.2%)	40 (48.8%)	9 (11.0%)	0 (0.0%)
G2	16 (8.5%)	89 (47.1%)	74 (39.2%)	10 (5.3%)
G3	0 (0.0%)	14 (15.7%)	46 (51.7%)	29 (32.6%)
G4	0 (0.0%)	0 (0.0%)	2 (8.7%)	21 (91.3%)
Likelihood ratio = 227.77***
Phi = .80***
Cramer’s V = .46***

p < .05. **p < .01. ***p < .001.

We were also interested in determining whether early neutralization acceptance predicted later delinquency. Figure 3 presents the connections between neutralization and delinquency via a structural equation model. The results show that neutralization at age 12 has a link with delinquency at age 13 (standardized effect = .37). Neutralization at age 14 has a link with delinquency at age 15 (standardized effect = .40). Delinquency at age 13 has a link with neutralization at age 14 (standardized effect = .46), and delinquency at age 15 has a link with neutralization at age 16 (standardized effect = .28). This figure shows that neutralization has a prospective link with delinquency. In addition, early neutralizing beliefs are strong predictors of later ones. Specifically, neutralizations at age 12 are linked with neutralizations at age 14 (standardized effect = .67), which are linked with neutralizations at age 16 (standardized effect = .97). This suggests a stability effect for neutralizing beliefs. The data fit this model (χ² = 4.76, p = .09; CFI = 1.00; RMSEA = .06; SRMR = .02).⁸

Figure 3:

The Reciprocal Relationship of Neutralization on Delinquency

Discussion

Sykes and Matza’s (1957) insights into how juvenile delinquents preemptively cleanse their conscience to engage in norm violations has had a significant impact on how criminologists understand the etiology of delinquency and criminality. Most agree that criminal belief systems like neutralization acceptance are important for crime commission. Despite this, numerous conceptual issues about the theory exist, including whether neutralization acceptance can be represented as a stable individual difference or is simply a “technique” that is situationally applied, whether individuals differ systematically on this putative individual difference, and whether neutralizing and delinquency systematically and causally covary. By addressing these issues, we sought to expand the utility of the theory.

Recall our initial three hypotheses. First, we predicted that neutralizing would be stable and differentiated across time (Hypotheses 1 and 2). Consistent with Piquero (2008), our trajectory analyses of the neutralization items in the GREAT data clearly demonstrated that (a) juveniles coalesced into four differentiated groups (very low neutralizers to high neutralizers) and (b) the parallel nature of their trajectories demonstrates that these categories remain stable and differentiated across time, a key requirement for neutralizing to be considered an individual difference (see Mischel & Shoda, 1998). Second, the trajectory analysis of the delinquency measure also produced four distinct groups (very low delinquency to high) that remained differentiated (parallel) and stable from ages 12 to 16, consistent with findings from previous trajectory analyses (Jennings & Reingle, 2012). Third, these factors (neutralization acceptance and delinquency) were systematically related to each other. The cross-tabular analysis juxtaposing the delinquency and neutralization trajectory groups demonstrated that neutralizing and delinquency were linked to one another (Hypothesis 3), such that low neutralizer trajectories were more strongly associated with low delinquency trajectories, and higher neutralizing trajectories with higher delinquency trajectories, demonstrating a systematic relationship between neutralizing and delinquency.

What of Hypothesis 4 (predicting reciprocal causality between neutralizing and delinquency)? Based solely on the above-described analyses, we would caution that the direction of causation between neutralizing and delinquency is difficult to establish, leaving open the question of how or whether neutralizing is driving delinquent behavior or vice versa. While the causal, temporal relationship between neutralizing and offending has received mixed support in previous research (see Agnew, 1994; Minor, 1981; Morris & Copes, 2012) we thought it important to determine whether it existed within the current data as this is key not only to conceptualizing neutralization usage as an individual difference, but also as having causal implications for delinquency.

To address this temporal issue, we employed structural equation modeling to determine effects between neutralization and delinquency across time, a technique that allows us to observe serial and simultaneous effects of these two variables. We found that earlier neutralizing had an effect on later neutralizing and later delinquency, and that this effect was reciprocal moving forward (i.e., resultant neutralizing and delinquency predicted subsequent neutralizing and delinquency through to age 16). This confirmed previous research suggesting a causal role of neutralizing in offending (see Agnew, 1994; Minor, 1981; Morris & Copes, 2012). More importantly, the positive change in the value of obtained coefficients between the 12-14 interval and the 14-16 intervals implies a stabilizing influence, suggesting that the relationship strengthens over time.

This would seem at least partially supportive of Hirschi’s (1969) notion of the “hardening” effect (see also Minor, 1981). Recall that hardening is assumed to occur over time as a consequence of offenders continuously employing neutralizations to engage in lawbreaking and other negative behaviors. The paradoxical outcome of this process is that, over time, the behaviors and beliefs associated with such continued neutralizing (i.e., participation in offending and beliefs about the appropriateness of offending) become more and more ingrained within the offender until neutralizing becomes an unnecessary step in the process. When that happens, the offender, having internalized offending as a positive or necessary activity, loses the need to neutralize at all. Previous research has identified such offenders as “hardcore” and notes that when they do neutralize, it is not when they contemplate engaging in behaviors that violate conventional norms, but rather when they contemplate engaging in positive ones (see Copes, Brookman, & Brown, 2013; Topalli, 2005, 2006).

Having said that, it is important to note that these results can only be suggestive of a hardening effect. Their weakness lies in the limitations of the data, which only measure neutralization acceptance and delinquency up to age 16. Although Hirschi did not specify the amount of time required for hardening to occur or the age at which a hardening inflection point would happen, we could at least speculate that it is most likely after this age based on long-standing findings confirming the age-crime curve (Hirschi & Gottfredson, 1983). Typically, research has found there to be a sharp incline in offending behavior during early adolescence that then peaks during the mid/late teenage years and then declines steeply as young people age into their mid-20s (Farrington, 1986). With regard to the current study then, more complete longitudinal data covering the years following adolescence into adulthood would be necessary to fully confirm the existence of a neutralizing-precipitated hardening effect. That said, when combined with our previously mentioned analyses, the path analysis supports the notion that neutralizing happens over time, is causally related to delinquency, and can be conceived of as a stable individual difference.

Still, the age-limited nature of GREAT presents other challenges. To begin, the narrow age range (12-16) limits our ability to make stronger assertions regarding stability over longer periods. This is problematic not only with respect to confirming a hardening effect but also for identifying our conceptualization of neutralizing as predictive of later adult offending behavior, particularly during the age range most strongly associated with offending by life-course theorists (the mid-20s). In addition, other weaknesses in the data highlight the need for further research and improved data to more strongly confirm our findings. First, our analysis strategies required us to employ a limited subset of the overall data. Furthermore, to follow the same group of individuals across all the remaining waves we limited participants in the data set to those who were 12 years of age at Wave 2 and followed only them until age 16 (five waves of data). This left us with a smaller sample size, which would limit power. This may explain the lack of effects of the control variables on neutralization trajectories. Second, our analysis implemented the publicly available GREAT data, which has a lower response rate than the more recent restricted access version implemented in work by Esbensen and colleagues (see, for example, Esbensen, Peterson, Taylor, & Osgood, 2012; Melde & Esbensen, 2012) again presenting challenges in relation to power. Finally, the age range of 12 to 16 presents limitations on the younger end of the continuum in that we cannot tell when neutralization as a trait may first develop and take hold. Nor do we know if the proposed hardening process begins at age 12 or some time earlier. Thus, future research should measure a wider age range than was available to us to more fully determine the development and effects of neutralization as a personality construct.

With such limitations having been acknowledged, our results suggest intriguing questions regarding the role of neutralizations in offending, both conceptual and applied. What does viewing neutralizing as indicative of an individual difference say about long-standing conceptualizations of both differential association and social learning perspectives? If neutralizing is indeed a stable individual difference, then what is its genesis and how does it interact with other individual differences commonly studied by criminologists and psychologists, such as impulsivity, attachment, or self-control? For example, those who are risk-takers and who lack impulse control may be more likely to accept neutralizations than would others. Such an interpretation may be seen as validation of other theorists who have subsumed neutralization in their theories (e.g., Akers, 1985; Hirschi, 1969). Alternatively, Van Gelder and De Vries (2012) recently argued for the development of trait-state models of criminal decision-making, pointing out that traits (individual differences) and states interact in interesting and important ways with implications for understanding and predicting offending. Finally, it is important to remember that our conceptualization of neutralization is related specifically to youth offenders. But, we know that adults employ neutralizations as well, often in support of committing white-collar crimes. Thus, neutralizing as a personality construct may develop early (probably in childhood) but manifest itself differently across the life course based on the age of the actor, suggesting a trait by time interaction, with neutralizing manifest in earlier years with regard to delinquent acts and in later years with regard to more “adult” crimes (like embezzlement or tax evasion).

This may have ramifications for theory integration. For example, how can neutralization theory as a trait or individual difference model be integrated with situationally oriented rational choice theories of crime? A related case has been put forth by Piquero, Paternoster, Pogarsky, and Loughran (2011) who argue that to comprehensively apply the tenets of deterrence theory it is necessary to fully understand how individual differences (what they refer to as the “kinds-of-people” dimension) in social bonding, morality, impulsivity, and decision-making competence interact with situational differences in emotions and alcohol or drug use. Finally, if it is indeed an individual difference, can neutralizing be measured in a systematic, standardized, and psychometrically valid way? Would such a measure permit us to identify individuals at various ages as very low, low, medium, or high neutralizers and how would this be implemented in the prediction or prevention of delinquency and offending?

This potential for neutralizing to be thought of as a measurable trait or personality concept brings us to the policy consequences of this research. Shifting our view of neutralization theory from the prediction of when people will implement aligning actions (Stokes & Hewitt, 1976) in response to situational demands to conceptualizing the offender as someone who inherently seeks out such strategies has important implications for offender treatment programs. This is particularly true for those where offenders are under consideration for or in the process of reentry. Based partly on Snyder’s finding that reality negotiation can have positive mental health benefits, Maruna and Mann (2006) cogently point out that most such programs require that offenders “take responsibility” for their actions and “face the truth of what they have done” and “who they are” as almost ritualistic preconditions for release into society (see, for example, Beech & Mann, 2002; Salter, 1988). If, as they contend, such “parochial” exercises exacerbate offending rather than limit it (and lead to a lack of success in reentry) then understanding neutralizing personalities becomes more than simply a theoretical exercise. To this end, establishing a standardized and psychometrically valid measure capable of differentiating different kinds of neutralizers early on may tell us which young people are at greater risk for offending and suggest specified prevention plans. This would allow service-providers to differentiate offenders based on this dimension has potential repercussions for individualized delivery of services to offenders. High neutralizers may require a very different cognitive-behavioral intervention approach from low neutralizers for example. In fact, such individuals would seem to require a treatment regimen that simultaneously capitalizes on the positive effects of neutralizing (to retain high self-efficacy for example) and discourages antisocial behavior.

Footnotes

Acknowledgements

We wish to thank the following scholars for their extremely insightful and helpful suggestions: Robert Agnew, Robert Morris, Alex Piquero, T. J. Taylor, and Brent Teasdale. We also wish to thank Kathleen Tuner for editing the manuscript.

Notes

Volkan Topalli, PhD, is an associate professor in the Department of Criminal Justice & Criminology at Georgia State University in the Andrew Young School of Policy Studies. His research and teaching interests are in the areas of offender decision-making and street violence. His recent publications have appeared in such journals as Criminology, Justice Quarterly, Criminal Justice & Behavior, and the British Journal of Criminology, with funding from such agencies as the National Science Foundation, the National Institute of Justice, and the National Institutes of Health.

George E. Higgins is a professor in the Department of Justice Administration at the University of Louisville. He received his PhD in criminology from Indiana University of Pennsylvania in 2001. His most recent publications appear in Journal of Criminal Justice, Deviant Behavior, Criminal Justice and Behavior, Youth and Society, and American Journal of Criminal Justice.

Heith Copes is an associate professor in the Department of Justice Sciences at the University of Alabama at Birmingham. His primary interest is in understanding the decision-making process and identity construction of offenders. His most recent book, with Lynne M. Vieraitis, is Identity Thieves: Motives and Methods.

References

Adler

(1927). The practice and theory of individual psychology. New York, NY: Harcourt.

Agnew

(1994). The techniques of neutralization and violence. Criminology, 32, 555-580.

Agnew

Peters

(1986). The techniques of neutralization: An analysis of predisposing and situational factors. Criminal Justice and Behavior, 13, 81-97.

Akers

R. L.

(1985). Deviant behavior: A social learning approach. Belmont, CA: Wadsworth.

Bandura

(1999). Moral disengagement in the perpetration of inhumanities. Personality and Social Psychology Review, 3, 193-209.

Beech

Mann

(2002). Recent developments in the assessment and treatment of sexual offenders. In McGuire

(Ed.), Offender rehabilitation and treatment: Effective programmes and policies to reduce re-offending (pp. 259-288). Chichester, UK: Wiley.

Benson

M. L.

(1985). Denying the guilty mind: Accounting for involvement in white-collar crime. Criminology, 23, 583-608.

Braithwaite

(1999). Restorative justice: Assessing optimistic and pessimistic accounts. Crime and Justice: A Review of Research, 25, 1-127.

Burgess

R. L.

Akers

R. L.

(1966). A differential association-reinforcement theory of criminal behavior. Social Problems, 14, 128-147.

10.

Buss

D. M.

Greiling

(1999). Adaptive individual differences. Journal of Personality, 67, 209-243.

11.

Cloward

R. A.

Ohlin

L. E.

(1960). Delinquency and opportunity: A theory of delinquent gangs. New York, NY: Free Press.

12.

Cohen

(1955). Delinquent boys: The culture of the gang. Glencoe, IL: Free Press.

13.

Copes

(2003). Societal attachments, offending frequency, and techniques of neutralization. Deviant Behavior, 24, 101-127.

14.

Copes

Brookman

Brown

(2013). Accounting for violations of the convict code. Deviant Behavior, 34, 841-858.

15.

Copes

Vieraitis

(2009). Bounded rationality of identity thieves: Using offender-based research to inform policy. Criminology & Public Policy, 8, 237-262.

16.

Cornish

Clarke

R. V.

(1986). The reasoning criminal. New York, NY: Springer.

17.

Cromwell

Thurman

(2003). The devil made me do it: Use of neutralizations by shoplifters. Deviant Behavior, 24, 535-550.

18.

Dabney

D. A.

(1995). Neutralization and deviance in the workplace: Theft of supplies and medicines by hospital nurses. Deviant Behavior, 16, 313-331.

19.

Delisi

Piquero

(2011). New frontiers in criminal careers research, 2000-2011: A state of the art review. Journal of Criminal Justice, 39, 289-301.

20.

Esbensen

F. A.

Osgood

D. W.

(1999). Gang Resistance Education and Training (G.R.E.A.T.): Results from the national evaluation. Journal of Research in Crime and Delinquency, 36, 194-225.

21.

Esbensen

F. A.

Peterson

Taylor

T. J.

Osgood

D. W.

(2012). Results from a multi-site evaluation of the GREAT program. Justice Quarterly, 29, 125-151.

22.

Farrington

D. P.

(1986). Age and crime. In Tonry

Morris

(Eds.), Crime and justice: An annual review of research (Vol. 7, pp. 189-250). Chicago, IL: University of Chicago Press.

23.

Festinger

(1957). A theory of cognitive dissonance. Evanston, IL: Row, Peterson.

24.

Fiske

S. T.

Taylor

S. E.

(2008). Social cognition: From brains to culture. New York, NY: McGraw-Hill.

25.

Hamlin

J. E.

(1988). The misplaced role of rational choice in neutralization theory. Criminology, 26, 425-438.

26.

Hewitt

J. P.

Stokes

(1975). Disclaimers. American Sociological Review, 40, 1-11.

27.

Higgins

G. E.

Wolfe

S. E.

Marcum

C. E.

(2009). Digital piracy and neutralization: A trajectory analysis from short-term longitudinal data. International Journal of Cyber Criminology, 2, 324-336.

28.

Hirschi

(1969). Causes of delinquency. Berkeley: University of California Press.

29.

Hirschi

Gottfredson

(1983). Age and the explanation of crime. American Journal of Sociology, 89, 552-584.

30.

Holt

T. J.

Copes

(2010). Transferring subcultural knowledge online: Practices and beliefs of persistent digital pirates. Deviant Behavior, 31, 625-654.

31.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equations Modeling: A Multidisciplinary Journal, 6, 1-55.

32.

Jennings

Reingle

J. M.

(2012). On the number and shape of developmental/life-course violence, aggression, and developmental trajectories: A state-of-the-art review. Journal of Criminal Justice, 40, 472-489.

33.

Jones

Nagin

(2007). Advances in group-based trajectory modeling and an SAS procedure for estimating them. Sociological Methods & Research, 35, 542-571.

34.

Jones

S. E.

Miller

J. D.

Lynam

D. R.

(2011). Personality, antisocial behavior, and aggression: A meta-analytic review. Journal of Criminal Justice, 39, 329-337.

35.

Kaufman

Rousseeuw

P. J.

(2009). Finding groups in data: An introduction to cluster analysis (Vol. 344). Hoboken, NJ: John Wiley.

36.

Kline

R. B.

(2005). Principles and practice of structural equation modeling (2nd ed.). New York, NY: Guilford.

37.

Maruna

Copes

(2005). What have we learned from fifty years of neutralization research? Crime and Justice: A Review of Research, 32, 221-320.

38.

Maruna

Mann

R. E.

(2006). A fundamental attribution error? Rethinking cognitive distortions. Legal and Criminological Psychology, 11, 155-177.

39.

Matza

(1964). Delinquency and drift. New York, NY: Wiley.

40.

Mehlman

R. C.

Snyder

C. R.

(1985). Excuse theory: A test of the self-protective role of attributions. Journal of Personality and Social Psychology, 49, 994-1001.

41.

Melde

Esbensen

F. A.

(2013). Gangs and violence: Disentangling the impact of gang membership on the level and nature of offending. Journal of Quantitative Criminology, 29, 143-166.

42.

Miller

(1958). Lower class culture as a generating milieu of gang delinquency. Journal of Social Issues, 14, 5-19.

43.

Mills

C. W.

(1940). Situated actions and vocabularies of motive. American Sociological Review, 5, 904-913.

44.

Minor

W. W.

(1981). Techniques of neutralization: A reconceptualization and empirical examination. Journal of Research in Crime & Delinquency, 18, 295-318.

45.

Minor

W. W.

(1984). Neutralization as a hardening process: Considerations in the modeling of change. Social Forces, 62, 995-1019.

46.

Mischel

Shoda

(1998). Reconciling processing dynamics and personality dispositions. Annual Review of Psychology, 49, 229-258.

47.

Moffitt

T. E.

(1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-701.

48.

Moffitt

T. E.

(2003). Life-course-persistent and adolescence-limited antisocial behavior: A 10-year research review and a research agenda. In Lahey

Moffitt

T. E.

Caspi

(Eds.), Causes of conduct disorder and juvenile delinquency (pp. 49-75). New York, NY: Guilford.

49.

Morris

R. G.

Copes

(2012). Exploring the temporal dynamics of the neutralization/delinquency relationship. Criminal Justice Review. Advance online publication. doi:0734016812456548

50.

Morris

R. G.

Higgins

G. E.

(2009). Neutralizing potential and self-reported digital piracy: A multi-theoretical exploration among college undergraduates. Criminal Justice Review, 34, 173-195.

51.

Muthen

L. K.

Muthen

B. O.

(2002). How to use a Monte Carlo study to decide on sample size and determine power. Structural Equation Modeling, 9, 599-620.

52.

Nagin

(2005). Group-based modeling of development. Cambridge, MA: Harvard University Press.

53.

Nagin

Farrington

D. P.

Moffitt

T. E.

(1995). Life-course trajectories of different types of offenders. Criminology, 33, 111-139.

54.

Osgood

W. D.

Schreck

C. J.

(2007). A new method for studying the extent, stability, and predictors of individual specialization in violence. Criminology, 45, 273-312.

55.

Piquero

(2008). Taking stock of developmental trajectories of criminal activity over the life course. In Liberman

(Ed.), The long view of crime: A synthesis of longitudinal research (pp. 23-78). New York, NY: Springer.

56.

Piquero

Paternoster

Pogarsky

Loughran

(2011). Elaborating the individual difference component in deterrence theory. Annual Review of Law and Social Science, 7, 335-360.

57.

Pogrebin

Stretesky

P. B.

Unnithan

N. P.

Venor

(2006). Retrospective accounts of violent events by gun offenders. Deviant Behavior, 27, 479-501.

58.

Salter

A. C.

(1988). Treating child sex offenders and their victims: A practical guide. Newbury Park, CA: SAGE.

59.

Sampson

R. J.

Laub

J. H.

(2005). Crime in the making: Pathways and turning points through life. Cambridge, MA: Harvard University Press.

60.

Sandberg

(2010). The sweet taste of sin: A Muslim drug dealer in a Nordic welfare state. Journal of Scandinavian Studies in Criminology and Crime Prevention, 11, 103-118.

61.

Scott

M. B.

Lyman

S. N.

(1968). Accounts. American Sociological Review, 33, 46-61.

62.

Scully

Marolla

(1984). Convicted rapists’ vocabulary of motive: Excuses and justifications. Social Problems, 31, 530-544.

63.

Shields

I. W.

Whitehall

G. C.

(1994). Neutralization and delinquency among teenagers. Criminal Justice and Behavior, 21, 223-235.

64.

Shigihara

(2013). It’s only stealing a little a lot: Techniques of neutralization for theft among restaurant workers. Deviant Behavior, 34, 494-512.

65.

Shoenberger

Heckert

(2012). Techniques of neutralization and positive deviance. Deviant Behavior, 33, 774-791.

66.

Snyder

C. R.

(1989). Reality negotiation: From excuses to hope and beyond. Journal of Social and Clinical Psychology, 8, 130-157.

67.

Snyder

C. R.

(1991). Explanatory style as reality negotiation with one’s self-theory, or vice versa. Psychological Inquiry, 2, 36-39.

68.

Snyder

C. R.

Higgins

R. L.

(1988). Excuses: Their effective role in the negotiation of reality. Psychological Bulletin, 104, 23-25.

69.

Stadler

Benson

(2012). Revisiting the guilty mind: Neutralization of white-collar crime. Criminal Justice Review. Advance online publication. doi:10.1177/0734016812465618

70.

Steinmetz

Tunnell

(2013). Under the pixelated Jolly Roger: A study of on-line pirates. Deviant Behavior, 34, 53-67.

71.

Stokes

Hewitt

J. P.

(1976). Aligning actions. American Sociological Review, 41, 838-849.

72.

Sutherland

E. H.

(1947). Principles of criminology (4th ed.). Philadelphia, PA: J. B. Lippincott.

73.

Sutherland

E. H.

Cressey

(1970). Criminology (8th ed.). Philadelphia, PA: J. B. Lippincott.

74.

Sykes

Matza

(1957). Techniques of neutralization: A theory of delinquency. American Sociological Review, 22, 664-670.

75.

Topalli

(2005). When being good is bad: An expansion of neutralization theory. Criminology, 43, 797-835.

76.

Topalli

(2006). The seductive nature of autotelic crime: How neutralization theory serves as a boundary condition for understanding hardcore street offending. Sociological Inquiry, 76, 475-501.

77.

Van Gelder

J. L.

De Vries

R. E.

(2012). Traits and states: Integrating personality and affect into a model of criminal decision-making. Criminology, 50, 637-671.

78.

Vieraitis

Piquero

Tibbetts

Blankenship

(2012). Do women and men differ in neutralizations of corporate crime? Criminal Justice Review, 37, 478-493.