Deviant Identity and Offending: A Longitudinal Study of South Korean Youths

Abstract

This study examines deviant identity in relation to youth offending by combining items tapping both self-appraisal and reflected appraisal. In particular, using survey data from 3,446 Korean youth across five waves of the Korea Youth Panel Survey (KYPS), findings from group-based trajectory modeling (GBTM) present four distinct offending groups—a high-rate chronic group, stable non-offending group, adolescence-limited group, and declining group. Then, findings from the multinomial logit model reveal that deviant identity is a robust predictor of offending for subgroups of adolescents involved in offending at any level in comparison to stable non-offenders. Accordingly, this study supports the idea that deviant identity should be considered as a prominent predictor of a variety of types of youth offending.

Keywords

deviant identity youth delinquency labeling group-based trajectory modeling

Labeling theory argues that when individuals are labeled as deviant through the course of official intervention from the criminal justice system, they develop a deviant self-identity (Cooley, 1902; Lemert, 1951; Mead, 1934; Tannenbaum, 1938). In turn, they act in ways consistent with this wayward identity to conform to their socially applied labels, thus becoming embedded in a criminal role. Therefore, adopting one’s deviant identity through official intervention is labeling theory’s main premise and official labeling is thought to be salient in the determination of a delinquent career (Lemert, 1951; Matza, 1964; Tannenbaum, 1938). However, a great deal of prior studies on labeling did not actually measure whether an individual adopted a deviant identity and only measured the labeling event, such as formal arrest, suspension, or incarceration (Lopes et al., 2012; Mowen et al., 2020; Ward et al., 2014; Wiley et al., 2013). That is, the body of previous work tends to examine the effect of labels in the form of sanctions on offending and only assumes that a deviant identity is imparted as part of the labeling.

In fact, there is qualitative research in support of the relationship between formal intervention and transformation of one’s identity, but the findings are not always supportive of a label-identity linkage (Ageton & Elliott, 1974; Hepburn, 1977; Jensen, 1980; Ray & Downs, 1986; Wolfgang et al., 1972). For example, some findings indicate that although adopting a deviant self-identity is likely to increase deviant behaviors, this identity might not be a function of the formal labeling process (Hepburn, 1977; Ray & Downs, 1986). Moreover, other work suggests that there are other channels through which individuals embrace deviant identity—for example, individuals might adopt it because they find value in a reputation of deviant-self (Katz, 1988; Thoits, 1985; Topalli, 2005). Overall, such work implies that identity theory may be preferable over labeling theory for understanding the role of identity in deviant behavior and, thus, delinquent self-identity should be considered independently (from official labeling) in the explanation of deviant behavior (Adams et al., 2003; Asencio & Burke, 2011; Maruna et al., 2004; Paternoster & Bushway, 2009; Ray & Downs, 1986).

The present study builds upon this deviant identity theory tradition in an examination of youth offending. More specifically, the present study accounts for criminal identity in one measure that taps into both self-appraisals and reflected appraisals and considers the potentially distinct relationships between deviant identity and offending across unique subgroups defined by longitudinal offending trajectory. As will be discussed in detail below, these issues are addressed through a two-stage analysis. First, the data from 3,446 Korean youths are examined longitudinally, across five waves, in a group-based trajectory model (GBTM). The GBTM allowed us to identify groups of youths distinguished by their developmental patterns of offending over time. Upon determining the optimal number of offending groups with similar developmental trajectories in GBTM, we examined the relationship between criminal identity and offending across different offending groups by estimating a multinomial logit regression model predicting group membership. Before providing the details surrounding our analyses, we first elaborate on the theory and previous research underlying the study.

Labeling and Deviant Identity

Built on symbolic interactionism, labeling theory contends that criminal behavior is due to an individual’s reaction to labels that are given by others (Lemert, 1951; Matza, 1964; Tannenbaum, 1938). Specifically, Lemert (1951) distinguished the concepts of primary and secondary deviance and explained that, unlike primary deviance, secondary deviance occurs due to the acceptance of a deviant identity based upon labels applied by others. More fully, this perspective suggests the process of labeling deviant behavior results, first, in the adoption of a deviant identity on the part of the person labeled (Lofland, 1969). In turn, this deviant identity increases deviant behavior, as the labeled person feels compelled to take and perform the expected role of the deviant individual. The label by others becomes a master status that the person acts upon (Becker, 1963). Importantly, labeling theory proposes that the problematic label is mostly attached through formal sanctions from the criminal justice system. Specifically, a youth’s involvement in the formal justice system prompts the adoption of a deviant identity leading to more deviant behavior and, quite likely, more formal sanctioning (Bernburg, 2009).

However, as alluded to above, studies present divergent findings on the actual relevance of official labeling to deviant self-identity, suggesting that the official labeling may not always produce a deviant self-identity (Ageton & Elliott, 1974; Hepburn, 1977; Jensen, 1972; Ray & Downs, 1986). For example, Jensen (1972) examined the relationship between official labeling and delinquent self-identity using cross-sectional survey data from 11 junior and senior high schools in California. His analysis reported that official labeling has an impact on delinquent self-evaluations but that the relationship varies between Blacks and Whites, by father’s educational status among Whites, and by attachment to the law. Using a longitudinal design, Ageton and Elliott (1974) found that formal labeling through police contact is positively related to delinquent orientations. However, they also found variation in the relationship between formal labeling through police contact and deviant identity across different ethnic groups. In particular, the effects were only significant for Anglo youths; they were not significant for other racial groups, including Mexican Americans and Blacks. Subsequent studies rooted in labeling theory have not detected a significant relationship between formal labeling and delinquent self-identity (Hepburn, 1977; Ray & Downs, 1986). For example, Ray and Downs (1986) reported a null effect of formal labeling on the adoption of a deviant self-identity, though formal labeling and deviant self-identity were each positively related to increases in drug use behavior among male subjects. Similarly, Asencio and Burke (2011) suggested that similar labels do not necessarily produce similar deviant identities, though identity is related to individual differences in behavioral outcomes.

Overall, a number of studies indicate that official labeling leads to differential levels of deviant identity, and labeling theory does not account for such varied responses (Cullen & Cullen, 1978). Other work questions the tenets of labeling theory by suggesting that it should not be assumed that official labels are needed to produce a deviant or criminal self-identity, as human agents are capable of actively deriving a sense of deviant self-based on their various experiences that may or may not involve the official labeling process. For example, research suggests that individuals might find value in the adoption of a deviant identity or reputation (Katz, 1988; Thoits, 1985; Topalli, 2005). While such findings do not necessarily undermine the importance of labeling theory, they do suggest that deviant identity should be examined as a distinct component within the broader labeling theoretical framework. More specifically, the findings reported here suggest that delinquent self-identity should be viewed as a key concept independent of official labeling and that direct measurement of the extent to which individuals adopt deviant self-identities is necessary (Adams et al., 2003; Asencio & Burke, 2011; Maruna et al., 2004; Paternoster & Bushway, 2009; Ray & Downs, 1986).

Accordingly, a number of previous studies examined the relationship between deviant identity and youth delinquency, and supported the relationship (Asencio & Burke, 2011; Bartusch & Matsueda, 1996; Burke, 1980; Felson, 1980, 1989; Heimer & Matsueda, 1994; Kinch, 1963; Matsueda, 1992; Matsueda & Heimer, 1997; Ray & Downs, 1986; Reynolds & Ceranic, 2007; Sparks & Shepherd, 1992). In contrast, some previous research using the same Korean data analyzed herein found the non-significant relationship between deviant self-identity and subsequent involvement of different types of delinquent behaviors (Kim & Lee, 2019; Na & Paternoster, 2019). Specifically, Na and Paternoster (2019) examined prosocial self-identity and its link to the involvement of eight different types of delinquent behaviors including bullying, and found the non-significant relationship net of other measures such as delinquent peer association and school attachment. Kim and Lee (2019) examined the self-deviant identity and its link to the involvement of 13 different types of delinquent behaviors including smoking, drinking, and bullying. The authors also found a non-significant relationship. While there are mixed findings, none of the previous studies focused on the direct relationship considering potential heterogeneity in offending.

Measures of Deviant Identity: Deviant Self-Appraisal and Reflected Appraisal

In the examination of the relationship between deviant self-identity and delinquency, previous studies introduced three different components of self-identity—(1) how individuals perceive themselves (self-appraisals), (2) how individuals perceive others’ view them (reflected appraisals), and (3) how others actually see individuals (others’ actual appraisals) (Kinch, 1963; Mead, 1934; Shrauger & Schoeneman, 1979). First, regarding self-appraisal, research demonstrates that how one perceives oneself (self-appraisal) is a primary factor in understanding a deviant self-identity (Burke, 1980; Matsueda, 1992; Reynolds & Ceranic, 2007; Sparks & Shepherd, 1992). Moreover, studies indicate that deviant self-appraisal is more important for understanding delinquency than broader measures of self-concept, such as self-esteem. In fact, this line of inquiry indicates that global self-esteem measures, as indicators of self-concept, are modestly or inconsistently related to delinquency (Bachman et al., 1978; Bamberg, 2011; Bynner et al., 1981; Kaplan, 1976; Restivo & Lanier, 2015; Rocque et al., 2016).

Beyond deviant self-appraisals, some studies draw heavily on the symbolic interactionist perspective and use the concept of “reflected appraisals”—the way in which the self perceives how significant others see one’s self—as an alternative conceptualization of deviant self-identity (Bartusch & Matsueda, 1996; Felson, 1980, 1989; Heimer & Matsueda, 1994; Kinch, 1963; Matsueda, 1992; Matsueda & Heimer, 1997). This perspective argues that the process of reflected appraisal is a key determinant of a person’s self-identity and hypothesizes that it is the most proximal cause of delinquent behavior. Research is largely consistent with this reflected appraisal thesis. For example, Matsueda (1992) found that reflected appraisals (as deviant) from parents were a substantial predictor of self-reported delinquency among respondents in the National Youth Survey. Similarly, Adams et al., (1998) found that perceptions of the deviant appraisals of significant others—including parents, teachers, and peers—exerted a direct effect on delinquency for non-whites. Another study by Lee et al., (2017) examined the effects of reflected parental appraisals and formal labeling during adolescence on subsequent crime. The analysis reported that only reflected parental appraisals as deviant (during adolescence) was significantly associated with criminal behavior in adulthood (Lee et al., 2017). In sum, studies suggest that reflected appraisal is another key measure of deviant identity and an important predictor of delinquent behavior (Menard & Morse, 1984; Paternoster & Iovanni, 1989).

In addition to self-appraisal and reflected appraisal, some studies examine others’ actual appraisals as a component of self-concept, finding mixed evidence regarding its link to delinquency (Kavish et al., 2016; Liu, 2000; Zhang, 1997). Furthermore, some research indicates that much of the effects on the delinquency of actual parents’ appraisals is mediated by reflected appraisals (Matsueda, 1992; Triplett & Jarjoura, 1994). In this regard, Triplett and Jarjoura (1994) explain that children might not be fully aware of their parents’ actual labeling, yet parents’ negative attitude toward them might influence their reflected appraisals and, in turn, deviant behaviors (see also Felson, 1985; Meade, 1974). Overall, studies imply that actual appraisal of others is not as key as self-appraisal and reflected appraisal in understanding deviant identity.

While the aforementioned studies delineate the importance of self-appraisal and reflected appraisal in the development of deviant self-identity, additional qualitative work has examined the reciprocal relationship between self-appraisal and reflected appraisal. In fact, a number of studies have found that reflected appraisals influence self-appraisals and vice versa (Asencio & Burke, 2011; Davis, 1961; Felson, 1985; Krueger, 1998; Paternoster & Iovanni, 1989; Rosenberg & Simmons 1972; Scimecca, 1977; Shrauger & Schoeneman, 1979). Specifically, using an incarcerated population, Asencio and Burke (2011) found that the strength of criminal identity measured by self-appraisal was influenced by the reflected appraisals of significant others. Likewise, studies suggested that self-appraisal affects the interpretation of information regarding others’ views (Davis, 1961; Felson, 1993; Krueger, 1998; Paternoster & Iovanni, 1989; Scimecca, 1977). Overall, despite the idea that self-appraisal and reflected appraisal appear as conceptually distinct components of deviant identity (Davis, 1961; Felson, 1985; Paternoster & Iovanni, 1989; Rosenberg & Simmons 1972; Scimecca, 1977; Shrauger & Schoeneman 1979), studies indicating a reciprocal relationship between self-appraisal and reflected appraisal support incorporating both into a single measure of deviant identity.

Deviant Identity and Deviant Behavior Across Trajectories

Previous literature reveals the importance of studying trajectories of offending over time due to a great deal of hidden heterogeneity in criminal behavior, and the correlates thereof (Laub & Sampson, 2001; Nagin, 2005; Nagin & Land, 1993; Piquero et al., 2007). In fact, a great deal of developmental and life-course research emphasizes the importance of distinguishing offending group membership and identifying distinct etiologies of delinquent behaviors across different offending groups (Cho & Lee, 2020; Le Blanc, 1997; Moffitt, 1993; Patterson & Yoerger 1993). Perhaps most famously, Moffitt (1993) developed two distinct offending groups: adolescence-limited and life-course persistent groups. In other seminal work on this issue, Nagin and Land (1993) assigned individuals to probabilistically appropriate groups that best represented their offending behavior, with each group exhibiting distinctive offending trajectories in terms of both level and pattern of change. They found four groups with distinctive offending trajectories over time: non-offenders, high-rate chronic offenders, adolescence-limited, and low-rate chronic offenders (see also Nagin, 1999). Studies using Korean student data classified students into three trajectories in the examination of victimization, delinquent peer association, peer delinquency, and bullying, respectively. Given the different foci of the classifications across these studies, the descriptions of three trajectory groups that emerged in each varies substantially; they are essentially non-comparable (Bax & Hlasny, 2019; Cho, 2021; Cho & Lacey, 2021; Cho, Lacey, & Kim, 2021). To our best knowledge, only one previous study using Korean student data examined trajectories of youth delinquency, with the authors separating the trajactories of male and female students for violent versus nonviolent delinquency. In this study, the description of three trajectories also varied across separate analyses. For example, regarding non-violent delinquency, the authors classified boys into moderate escalators, desistors, or stable moderates whereas girls were classified into moderate escalators, stable nondelinquents, or stable moderates (Bax & Hlasny, 2019).

Some trajectory research emphasizes trajectory-specific predictors, whereas other research reveals common risk factors across groups (Farrington & Hawkins, 1991; Fergusson et al., 2000; Moffitt, 1993; Patterson & Yoerger 1993; White et al., 2001). For example, Moffitt (1993) proposed trajectory-specific risk predictors—suggesting that early onset of delinquent behavior, neuropsychological deficits, and deprived family environments would predict life-course persistent trajectories, whereas delinquent peer association would predict adolescence-limited offending patterns. In contrast, Fergusson et al., (2000) identified common predictors associated with trajectory group membership. That is, different factors did not distinguish offending groups; instead, it was the level of exposure to these common risk or protective factors that determined offending trajectory group membership. Specifically, high exposure to family adversity and individual difficulties were associated with early onset offenders (labeled as the “chronic” group in this study), while low exposure to these same factors was related to membership in a “non-offender” group.

Accordingly, we see several possible ways that deviant identity might be associated with youth offending across groups. First, in line with Moffitt (1993) suggestion of group-specific risk factors, deviant identity might be a trajectory-specific predictor, whereby it is a key criminogenic factor for some groups but not others. Second, and by contrast, deviant identity could be a common factor for all offending groups, though the magnitude of the relationship would likely vary. This would be more consistent with research supporting the idea of uniform predictors across offending groups (Fergusson et al., 2000).

Present Study

Within an identity theory of delinquency, there are several theoretical and methodological lines of inquiry that are key to the present study. First, while self-identity as delinquent can be assessed by asking respondents to indicate whether they perceive themselves as deviant (i.e., self-appraisals), some previous studies examine other conceptualizations of deviant self-identity. In particular, key research argues that the way in which the self perceives how significant others see one’s self (i.e., reflected appraisals) is a key element of a person’s self-identity (Felson, 1980, 1989; Heimer, 1996; Matsueda, 1992; Matsueda & Heimer, 1997). Yet, research on the role of identity in youth delinquency frequently does not include measurement of both self and reflected appraisals. Second, extant work addressing the role of identity in youth delinquency does not adequately consider the potential distinct relation between deviant identity and offending across unique subgroups. Thus, to address this gap, the present study examines criminal identity by combining items tapping both self-appraisals and reflected appraisals and explores the potential distinct relationship between deviant identity and offending across subgroups using a multinomial logit model predicting group membership.

Data and Methods

Data

To test our hypotheses, this study used student data from the Korea Youth Panel Survey (KYPS) conducted by the National Youth Policy Institute (NYPI). This is a six-wave longitudinal study—spanning the second grade within middle school to the freshman year in college—designed to examine individual factors that affect deviant behavior, school dropout, participation in leisure activities, and occupational preparation across 12 regions in South Korea (Seoul Metropolitan City and 11 other metropolitan cities or provinces). The data were collected between the years of 2003 and 2008. The stratified multi-stage cluster sampling design for the KYPS first involved a probability-proportionate-to-size (stratum size) sampling of 104 middle schools across 12 regions. When a selected school refused to participate, replacement sampling was done. Next, one second year class (equivalent to the seventh graders in the U.S. schools) from each selected school was randomly sampled, with gifted and special education classes excluded. Finally, a total of 3,697 second-year students in the selected middle schools and classes were targeted for the student sample in wave 1. Among them, 3,449 provided the required parental consent forms and completed surveys were obtained from all of these students in wave 1. Students had advanced to high school after Wave 3 and were seniors at Wave 5, and many were first-year college students in Wave 6. Due to the unique contextual change that occurred among respondents who graduated from high school and entered college, this study used the survey data only from Wave 1 to Wave 5 and thus focused on changes in identity and offending among respondents through high school. Attrition rates were 7.6% at Wave 2, 2% at Wave 3, 0.1% at Wave 4, and 5% at Wave 5. The total attrition rate over the 5 years was 13.7%. After the listwise deletion of cases with missing data, there were 3,446 students who provided data across the five waves.

Measures: Dependent Variable

The dependent variable of our interest is delinquent behavior. To measure delinquent behavior, we used survey items that asked students to indicate how many times (0 = 0, 1 = 1. . . . .10 = 10+) they were involved in the following acts during the previous year: (1) severely beating other people, (2) forcing someone to give up money or property, (3) stealing something from someone when they were not around, and (4) threatening other people. Responses to the delinquent behavior items were summed per wave for each respondent. Descriptive statistics in Table 1 indicate that the average number of delinquent behavior reported by students ranged from 0.09 to 0.67 and the standard deviation from 0.81 to 2.50 across waves. These statistics show evidence of overdispersion.

Table 1.

Descriptive Statistics of Sample Data across Multiple Waves.

Variable(s)	Mean	Standard deviation	Min	Max
Delinquency
Wave 1	0.67	2.50	0	40
Wave 2	0.43	2.18	0	30
Wave 3	0.20	1.36	0	30
Wave 4	0.18	1.33	0	25
Wave 5	0.09	0.81	0	20
Deviant identity
Wave 1	1.79	0.74	1	5
Wave 2	1.75	0.73	1	5
Wave 3	1.77	0.70	1	5
Wave 4	1.81	0.76	1	5
Wave 5	1.72	0.72	1	5
Perceived informal social control
Wave 1	3.04	1.05	1	5
Wave 2	2.89	1.14	1	5
Wave 3	2.98	1.08	1	5
Wave 4	3.10	1.01	1	5
Wave 5	3.07	1.04	1	5
Low self-control
Wave 1	2.61	0.62	1	5
Wave 2	2.66	0.63	1	5
Wave 3	2.59	0.61	1	4.67
Wave 4	2.57	0.60	1	4.89
Wave 5	2.56	0.61	1	4.56
Delinquent peers
Wave 1	0.30	0.72	0	3
Wave 2	0.24	0.67	0	3
Wave 3	0.15	0.51	0	3
Wave 4	0.13	0.48	0	3
Wave 5	0.08	0.39	0	3
School attachment
Wave 1	4.08	0.76	1	5
Wave 2	1.85	0.76	1	5
Wave 3	4.15	0.76	1	5
Wave 4	4.10	0.79	1	5
Wave 5	4.32	0.74	1	5
Parent attachment
Wave 1	3.35	0.77	1	5
Wave 2	3.42	0.76	1	5
Wave 3	3.44	0.72	1	5
Wave 4	3.46	0.75	1	5
Wave 5	3.52	0.77	1	5
Gender
Wave 1	0.50	0.50	0	1
Wave 2	0.50	0.50	0	1
Wave 3	0.50	0.50	0	1
Wave 4	0.50	0.50	0	1
Wave 5	0.49	0.50	0	1
Wave	2.94	1.41	1	5

Note. N (Students) = 3,446 ( $N_{W a v e 1}$ = 3,362, $N_{W a v e 2}$ = 3,139, $N_{W a v e 3}$ = 3,083, $N_{W a v e 4}$ = 3,106, and $N_{W a v e 5}$ = 2,890).

Measures: Independent Variables

To measure our key independent variable, deviant identity, we averaged responses for each student per wave from four survey items tapping both self-appraisal and reflected appraisal as deviant. Specifically, students were asked to indicate the extent to which they agreed (1 = strongly disagree. . . 5 = strongly agree) with four survey items: (1) “I regard myself as a problem youth,” and (2) “I regard myself as a juvenile delinquent,” (3) “People around me regard me as a problem youth,” and (4) “People around me regard me as a juvenile delinquent.” We thus take the approach that deviant identity is multidimensional, having both a personal dimension (self-perception) and a public dimension (perception of how viewed by others). However, we have theoretical and empirical reasons to support combining these multiple dimensions into a single measure of deviant identity. For example, studies (reviewed above) suggest a reciprocal relationship between self-appraisal and reflected appraisal (Asencio & Burke, 2011; Bartusch & Matsueda, 1996; Burke, 1980; Davis, 1961; Felson, 1980, 1985, 1989; Heimer & Matsueda, 1994; Kinch, 1963; Matsueda, 1992; Matsueda & Heimer, 1997; Paternoster & Iovanni, 1989; Reynolds & Ceranic, 2007; Rosenberg & Simmons, 1972; Scimecca, 1977; Shrauger & Schoeneman, 1979; Sparks & Shepherd, 1992;). Consistent with the overlap implied with such mutual influence, the items tapping self-appraisals and reflected appraisals in the current study are highly intercorrelated (Cronbach’s alpha = .91). Finally, a principal-components analysis with Varimax with Kaiser Normalization rotation was used for assessing the number of latent factors underlying the four measured items tapping self-appraisals and reflected appraisals. This analysis yielded one factor, with the eigenvalue for the component larger than one, and factor loadings for all four items were above 0.5. These statistics in factor and reliability analyses were based on pooling all values for students across waves.

Given that previous studies suggest criminal persistence can be attributable to the process of “cumulative continuity” in terms of limited support and resources from significant others (Braithwaite, 1989; Horney et al., 1995; Maruna, 2001; Sampson & Laub, 1997), this study included perceived informal social control as a correlate of youth delinquent behavior. Perceived informal social control was measured as the average score per wave across two survey items asking students to indicate the extent to which they agreed (1 = strongly disagree. . . 5 = strongly agree) with the two following statements: (1) “If I do something wrong, I will be shamed by others around me,” and (2) “If I do something wrong, I will be blamed by others around me.” The correlation between these two items was 0.88. It is important to note that shame in the item used for this variable connotes something different from shame as self-imposed costs—as in “I feel ashamed when I do something wrong” (Grasmick & Bursik, 1990). Specifically, the item addressing shame in this study taps a perceived cost, but it is a cost imposed by others. Thus, the items composing perceived informal social control assess the perception of how others would sanction respondents through shame and blame as forms of direct informal social control.

We also control for other key correlates of youth delinquent behavior including low self-control, delinquent peers, social bonds measured by school attachment and parent attachment, and gender. The control variables were measured in a manner similar to other studies examining youth violence, especially among school-based samples (e.g., Johnson et al., 2019; O & Wilcox, 2018). Low self-control was measured by taking the mean of a total of nine survey items (1 = very untrue/strongly disagree. . .5 = very true/strongly agree) which tap the student’s attention span, temper, and impulsivity (Cronbach’s alpha = .74). The specific items were as follows: (1) “I jump into exciting things even if I have to take an examination tomorrow,” (2) “I abandon a task once it becomes hard and laborious to do,” (3) “I don’t do my homework habitually,” (4) “I am apt to enjoy risky activities,” (5) “I enjoy teasing and harassing other people,” (6) “I lose my temper whenever I get angry,” (7) “I am often seized by an impulse to throw an object when I get angry,” (8) “Sometimes I can’t suppress an impulse to hit other people,” and (9) “I sometimes feel like a bomb ready to explode.” The scores across these nine items were averaged for each student per each wave in order to create the low self-control variable, with higher values reflecting high levels of low self-control.

Delinquent peers was calculated in each wave by taking the sum of three items that tapped whether the student’s closest friends (1 = yes, 0 = no) were involved in the following delinquent behaviors during the last year: (1) severely beating other people, (2) forcing someone to give up money or property, and (3) stealing something from someone when they were not around. The resulting sums ranged from 0 to 3. Social bonds in the form of school attachment and parent attachment were measured by calculating the mean of three items and six items, respectively for each student per each wave. First, school attachment was measured by calculating the mean of response to items that asked students to indicate how much (1 = very untrue. . . 5 = very true) they agreed with the following statement about their feelings toward their school, teachers, and education (Cronbach’s alpha = .89): (1) “I find it difficult to follow school rules and regulations,” (2) “I am not in good terms with school teachers,” and (3) “I am not interested in school work, and find it difficult to keep up.” Items were reverse coded before averaging, with higher score on the resulting variable thus representing greater levels of school attachment. Parent attachment was measured by calculating the mean of six items that asked respondents to indicate how much (1 = very untrue. . . 5 = very true) they agreed with the following statements (Cronbach’s alpha = .88): (1) “My parents and I try to spend much time together,” (2) “My parents always treat me with love and affection,” (3) “My parents and I understand each other well,” (4) “My parents and I candidly talk about everything,” (5)” I frequently talk about my thoughts and what I experience away from home with my parents,” and (6) “My parents and I have frequent conversations.” Student gender (0 = male, 1 = female) was also used as a control variable.

Analytic Procedure

In the current study, a group-based trajectory model (GBTM) and a multinominal logistic regression model were carried out in sequence. The first step was to estimate distinct developmental trajectories of offending among the 3,446 sampled adolescents across five waves of data collection. Our goal here was to classify changes in offending counts through a longitudinal latent class model with a random-intercept, accounting for the possibility that each student group (i.e., each latent class) may have a different starting point in terms of their offending propensity.

Following Nagin and Odgers (2010), we determine the appropriate number of classes using unconditional trajectory models. That is, we use the basic model without covariates except the time variables to control for the possible fluctuations in offending over time per each group. Specifically, the trajectories were estimated with up to a third-order polynomial function of wave (or time), as we assumed at least one group would have multiple up and down fluctuations over time (e.g., high-rate chronic, adolescence-limited, or declining). This possible fluctuation can be controlled using a quadratic or cubic function for time (wave) in our regression model. Accordingly, our Poisson-based trajectory model for analyzing youth offending as count data is described in equation (1) as follows:

l n λ_{t}^{j} = β_{0}^{j} + β_{1}^{j} W a v e_{t} + β_{2}^{j} W a v e_{t}^{2} + β_{3}^{j} W a v e_{t}^{3}

(1)

Where $λ_{t}^{j}$ is the Poisson rate parameter with j indexing trajectory group and t indexing wave. The two most important outputs of this basic model are estimates of the curve of each group’s trajectory and the size of the group as measured by the proportion of the population following that trajectory (Nagin & Odgers, 2010). This proportion can be interpreted as the probability of trajectory group membership. We will later discuss our decision to retain four groups for further analysis.

Upon determining the optimal number of latent groups with similar developmental trajectories in offending, we consider analyses examining how the probability of trajectory group membership varies with five conventional risk factors of delinquent behavior and deviant identity. To do so, we use the pre-determined trajectory group membership as our dependent variable in the multinomial logistic regression and examine the determinants of group membership across different trajectory groups. Since the determinants (independent variables) are repeatedly measured for each student over multiple waves, we cluster independent variables by student identifiers using the clustering command (i.e., Cluster (StudentID)) in Stata 14 with robust standard errors. We also use the time variables (Wave, Wave², and Wave³) in the regression model to detrend the pre-determined group trajectories over five waves. We report our findings in the following section.

Results

Group-Based Trajectory Analysis

The goal of our Poisson-based GTBM was to accurately capture the overall heterogeneity in patterns of youth offending by determining the number of trajectory groups observed in the data. To determine the most optimal number of latent classes to retain, we assessed multiple statistics including Bayesian Information Criterion (BIC), Akaike Information Criterion (AIC), log-likelihood values, posterior probabilities, odds of correct classification, and group proportion. The BIC and AIC are closely related model selection criteria representing the model’s goodness-of-fit along with Log-likelihood values. Generally, the BIC penalizes free parameters more strongly than does the AIC, so the BIC favors more parsimonious models. Typically, higher (or less negative) values of BIC, AIC, and log-likelihood, are preferable, though these statistics need to be considered in conjunction with a priori theory as well as posterior probabilities. Posterior probabilities reflect the accuracy of the classification of cases in groups. The higher the posterior probability, the greater the probability that the cases in the groups are classified accurately. Model fit statistics for two-, three-, four-, and five-class solutions are presented in Table 2. Table 2 also presents posterior probabilities, associated odds of correct classification, and estimated group membership for models with 2, 3, 4, and 5 classes.

Table 2.

Summary of Latent Class Analysis Process with Delinquency.

	2 Group	3 Group	4 Group	5 Group
BIC	−9696.71	−8888.99	−8093.28	−8187.00
AIC	−9669.06	−8845.98	−8034.90	−8113.26
Log-likelihood	−9660.06	−8831.98	−8015.9	−8089.26
Posterior probabilities and class proportions for 2 Group model
	Posterior prob.	Odds of correct classification	Estimated group membership (%)	t-statistics for H₀: Parameter=0
Class 1	0.995	23.896	89.304	160.542
Class 2	0.978	378.359	10.696	19.229
Posterior probabilities and class proportions for 3 Group model
Class 1	0.982	14.487	78.838	104.523
Class 2	0.974	184.884	16.822	24.069
Class 3	0.969	691.862	4.340	11.750
Posterior probabilities and class proportions for 4 Group model
Class 1	0.976	1396.782	2.791	8.769
Class 2	0.986	14.902	82.268	109.504
Class 3	0.934	99.903	12.418	17.825
Class 4	0.959	894.400	2.523	8.864
Posterior probabilities and class proportions for 5 Group model
Class 1	0.917	376.518	3.136	7.845
Class 2	0.977	14.012	74.091	94.232
Class 3	0.905	51.708	15.078	21.061
Class 4	0.906	191.293	5.739	11.804
Class 5	0.968	1855.678	1.956	8.300

The top panel in Table 2 provides evidence that the four (4) group model has the best fit to our data, with the 4-group model associated with the highest BIC, AIC, and log-likelihood values. The bottom four panels of Table 2, reporting the posterior probabilities (and associated odds of correct classification and estimated group membership) for models with 2, 3, 4, and 5 classes, also support a 4-group solution. Though all posterior probabilities are high in their values, the 4-group model shows the highest values relative to 2, 3, or 5 group models.

Given the collective statistical evidence in conjunction with a-priori theoretical expectations regarding offending trajectories (Nagin, 1999, 2005), we assess the 4-class solution as the most appropriate. The estimated trajectories are graphically illustrated in Figure 1. It shows four distinct trajectory profiles with their confidence interval bandwidths surrounding each trajectory. Borrowing from prior researchers’ descriptions of offending trajectories (Moffitt, 1993; Nagin et al., 2010;), we termed the profiles of each group based on the level of offending as: (1) a high-rate chronic offending group (2.8%), (2) a stable non-offending group (82.3%), (3) a declining group (12.4%), and (4) adolescence-limited group (2.5%).

Figure 1.

Youth offending trajectory profiles*.

Figure 1 shows that the adolescence-limited class displays high levels of youth offending in waves 1 and 2, but then a dramatic decline in offending levels after wave 2. On the other hand, the high-rate chronic class exhibits moderately high levels of youth offending in waves 1 and 2 but then an increase in the level of offending that begins in wave 3 and remains steady through wave 5. By wave 3 the high-rate chronic offending group has levels far greater than all three other groups. The declining group, similar to the high-rate chronic-offending group, shows moderately high levels of offending in wave 1. However, the average number of offenses among students in the declining group immediately dropped by wave 2, and it decreased to almost zero in the subsequent waves. Finally, students in the stable non-offending group persisted in reporting the lowest number of offenses at each wave (virtually no offending). The Appendix A depicts individual student trajectories per class.

Predicting Group Membership

The second component of the analysis in Table 3 involved analyses examining how the probability of trajectory group membership varies with deviant identity and other covariates. This involved using a nominal indicator of class membership as an outcome variable (1 = High-rate chronic offending, 2 = Stable non-offending, 3 = Declining, and 4 = Adolescence-limited) for a multinomial logistic regression model in which the low-offending class was the reference category.

In Table 3, the level of deviant identity is positively related to (1) being a high-rate chronic offending relative to a persistently stable non-offending, (2) having a declining relative to a stable non-offending trajectory, and (3) exhibiting an adolescence-limited trajectory relative to a stable non-offending trajectory. Therefore, the three panels of Table 3 collectively indicate that the level of deviant identity is positively associated with being in any of groups that exhibit offending (high-rate chronic offending, adolescence-limited offending, or declining offending) versus stable non-offending. Accordingly, deviant identity distinguishes high-rate chronic, declining, and adolescence-limited groups from the stable non-offending group. Moreover, the effect of deviant identity does not differ substantially across the three panels in Table 3. It appears that deviant identity is a common, robust predictor of any sort of trajectory that involves youth delinquency versus a non-delinquent trajectory.

Table 3.

Multinomial Logistic Regression Predicting Class Membership.

Variable	Coefficient estimate	Standard error	z-Score	p-Value
High-rate Chronic versus Stable Non-offending
Intercept	−6.28	0.87	−7.22	.000
Deviant identity**	0.66	0.12	5.64	.000
Perceived informal social control	−0.06	0.08	−0.80	.423
School attachment**	−0.15	0.05	−3.03	.002
Parent attachment	−0.01	0.11	−0.07	.945
Low self-control**	0.96	0.15	6.23	.000
Delinquent peers**	0.95	0.06	15.08	.000
Female**	−0.89	0.24	–3.75	.000
Wave*	−1.14	0.47	−2.40	.016
Wave²**	0.50	0.17	2.91	.004
Wave³**	−0.06	0.02	−3.13	.002
Declining versus Stable Non-offending
Intercept	−3.77	0.42	−9.00	.000
Deviant identity**	0.31	0.05	5.99	.000
Perceived informal social control	0.01	0.04	0.31	.759
School attachment*	−0.05	0.03	−2.01	.045
Parent attachment	−0.04	0.06	−0.77	.441
Low self-control**	0.65	0.07	9.37	.000
Delinquent peers**	0.62	0.04	14.08	.000
Female**	−0.32	0.11	−2.94	.003
Wave*	−0.49	0.24	−2.02	.043
Wave²*	0.22	0.09	2.52	.012
Wave³**	−0.03	0.01	−2.73	.006
Adolescence-limited versus stable non-offending
Intercept	−7.27	0.70	−10.40	.000
Deviant identity**	0.54	0.10	5.69	.000
Perceived informal social control	−0.05	0.07	−0.80	.426
School attachment	0.01	0.06	0.10	.919
Parent attachment*	−0.22	0.11	−1.99	.046
Low self-control**	1.00	0.13	7.67	.000
Delinquent peers**	1.07	0.06	17.33	.000
Female	−0.27	0.23	−1.15	.249
Wave	0.11	0.48	0.22	.822
Wave²	0.02	0.18	0.13	.895
Wave³	0.00	0.02	−0.21	.834

*p ≤ 0.05. **p ≤ 0.01.

The pattern of the relationships between other factors and offending trajectories was consistent with those just reported for deviant identity. Specifically, both low self-control and delinquent peers association were positively associated with membership in the high-rate chronic, adolescence-limited, and declining trajectories relative to the stable non-offending group. Thus, both low self-control and delinquent peers contrasted any of the offending groups from the non-offending group; they were common, robust factors for membership in any offending trajectory. School and parental attachment variables demonstrated a different pattern of effects. Specifically, a low level of school attachment was negatively associated with membership in both high-rate chronic and declining groups relative to the stable non-offending group, whereas it was not associated with membership in the adolescence-limited group (versus the non-offending group). In contrast, parental attachment was negatively associated with adolescence-limited group membership relative to stable non-offending group membership, but parental attachment did not distinguish high-rate chronic or declining offending membership versus non-offending group membership. Similar to school attachment, being male was associated with high-rate chronic and declining trajectories relative to low-offending trajectories. Perceived informal social control (tapping shame and blame) was not related to membership in any of the three offending groups relative to the non-offending group.

Conclusion and Discussion

We examined how deviant identity independently from official labeling is associated with deviant behaviors among youth. In fact, though some studies using the same data analyzed herein found a non-significant relationship between identity and deviance, much other previous work supports the relationship (Adams et al., 2003; Asencio & Burke, 2011; Bartusch & Matsueda, 1996; Burke, 1980; Felson, 1980, 1989; Heimer & Matsueda, 1994; Kim & Lee, 2019; Kinch, 1963; Maruna et al., 2004; Matsueda, 1992; Matsueda & Heimer, 1997; Na & Paternoster, 2019; Paternoster & Bushway, 2009; Ray & Downs, 1986; Reynolds & Ceranic, 2007; Sparks & Shepherd, 1992). That said, none of the previous studies yielding such mixed findings focused on the role of identity in distinguishing heterogeneous offending trajectories.

First, findings from GBTM suggest that four distinct groups are appropriate to capture the heterogeneity in youth offending: (1) high-rate chronic group, (2) stable non-offending group, (3) declining group, and (4) adolescence-limited group. Then, the main analysis in the multinomial logistic regression model reveals that deviant identity distinguishes each high-rate chronic, declining, and adolescence-limited group from the stable non-offending group. At the same time, deviant identity is a common predictor for those offending groups. That is, a higher level of deviant identity is more positively associated with being a high-rate chronic, declining, and adolescence-limited group relative to being a stable non-offending group. We think it is reasonable to see overlapping effects of deviant identity across those offending groups despite a difference in continuity in offending. In particular, in terms of the level of offending, those in high-rate chronic, declining, adolescence-limited groups engaged in a higher level of offending than those in the stable non-offending group. In terms of the offending pattern, those groups display a peak point of offending across trajectories (see Figure 1). Beyond deviant identity, those in high-rate chronic, declining, and adolescence-limited groups (relative to those in the stable non-offending group) also tend to display lower self-control and greater association with delinquent peers. Thus, there is a good deal of overlap in risk factors for membership in any of the offending groups (versus non-offenders). However, some distinct risk and protective factors were also noted. For example, social bonding (i.e., school attachment and parent attachment) and gender were distinguished with membership in high-rate chronic and declining groups versus the adolescence-limited group. Such results suggest that, among groups reporting any offending, high-rate chronic and declining groups share more commonality (than adolescence-limited group). We encourage further exploration and debate regarding this issue.¹

Regarding the policy implication, the findings in terms of deviant identity and its positive association with high-rate chronic, declining, and adolescence-limited groups (in relative to the stable non-offending group) suggest that we should provide continuous efforts to help youth across almost all offending groups acquire a new prosocial identity—such as the narrative identity of a redemption script (Maruna, 2001). Such implications, of course, are preliminary, and we encourage further research in order to better understand deviant identity and offending across subgroups and to provide multiple avenues of interpretation based on different theoretical perspectives.

In thinking about future research, several limitations of the present study deserve attention. First, the data we used come from students in South Korea. Thus, our findings may not necessarily be generalizable to students in the U.S. context or other countries. Therefore, further research is still needed to examine whether there is cross-national generalizability (or, alternatively, context specificity) in the role deviant identity plays in longitudinal trajectories of offending. Second, our analysis was unable to examine various characteristics of schools such as school efficacy or school socioeconomic status in which students in the analysis are nested. Since different school characteristics could influence students’ behavior, future research should carefully consider this possibility and include such characteristics in a contextual, multilevel analysis. Finally, since the offense data were limited to self-reported crimes, our data set may not account for offenses not reported or under-reported during the survey.

Despite these limitations, we believe results from this study point to the importance of deviant identity as an explanation of delinquent involvement across different offending trajectory groups. Thus, we suggest that insights into the deviant identity should continue to be considered in future research, as doing so holds important potential practical implications for how we might alter trajectories, bringing about more timely desistance.

Supplemental Material

sj-docx-1-cad-10.1177_00111287221102061 – Supplemental material for Deviant Identity and Offending: A Longitudinal Study of South Korean Youths

Supplemental material, sj-docx-1-cad-10.1177_00111287221102061 for Deviant Identity and Offending: A Longitudinal Study of South Korean Youths by SooHyun O, YongJei Lee, Pamela Wilcox and Francis T. Cullen in Crime & Delinquency

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

SooHyun O

YongJei Lee

Notes

Author Biographies

SooHyun O is an assistant professor in the Department of Criminology at the University of South Florida. Her current research examines juvenile delinquency, juvenile victimization at school, crime hot spot forecasting, and the relationship between land use and crime.

YongJei Lee is an assistant professor in the Department of Criminology at the University of South Florida. His current research examines police effectiveness, spatio-temporal patterns of crime hot spots, crime hot spot forecasting algorithm, measures of crime concentration phenomenon, and machine learning models for recidivism forecasting.

Pamela Wilcox is a professor in the Department of Sociology and Criminology at Pennsylvania State University. Her research examines crime and victimization from a multilevel perspective that highlights how individual-level and place-level factors interact with environmental contexts.

Francis T. Cullen is Distinguished Research Professor Emeritus and a Senior Research Associate in the School of Criminal Justice at the University of Cincinnati. He is a past president of the American Society of Criminology and the Academy of Criminal Justice Sciences. His current research focuses on racial beliefs and public policy, building a rehabilitation-redemption model of offender reform, and the criminology of Donald Trump.

References

Adams

M. S.

Johnson

J. D.

Evans

T. D.

(1998). Racial differences in informal labeling effects. Deviant Behavior, 19(2), 157–171.

Adams

M. S.

Robertson

C. T.

Gray Ray

Ray

M. C.

(2003). Labeling and delinquency. Adolescence, 38(149), 171–187.

Ageton

S. S.

Elliott

D. S.

(1974). The effects of legal processing on delinquent orientations. Social Problems, 22(1), 87–100.

Asencio

E. K.

Burke

P. J.

(2011). Does incarceration change the criminal identity? A synthesis of labeling and identity theory perspectives on identity change. Sociological Perspectives, 54(2), 163–182.

Bachman

J. G.

O’Malley

P. M.

Johnston

(1978). Adolescence to adulthood—change and stability in the lives of young men. Institute for Social Research.

Bamberg

(2011). Who am I? Narration and its contribution to self and identity. Theory & Psychology, 21(1), 3–24.

Bartusch

D. J.

Matsueda

R. L.

(1996). Gender, reflected appraisals, and labeling: A cross-group test of an interactionist theory of delinquency. Social Forces, 75(1), 145–176.

Bax

Hlasny

(2019). The causes and courses of nonviolent and violent delinquency among South Korean adolescents. Deviant Behavior, 40(7), 816–834.

Becker

H. S.

(1963). The outsiders: Studies in the sociology of deviance. The Free Press.

10.

Bernburg

J. G.

(2009). Labeling theory. In Krohn

M. D.

Lizotte Hall

A. J.

Penly

(Eds.), Handbook on crime and deviance (pp. 187–207). Springer Science and Business Media, LLC.

11.

Braithwaite

(1989). Crime, shame and reintegration. Cambridge University Press.

12.

Burke

P. J.

(1980). The self: Measurement requirements from an interactionist perspective. Social Psychology Quarterly, 43(1), 18–29.

13.

Bynner

J. M.

O’Malley

P. M.

Bachman

J. G.

(1981). Self-esteem and delinquency revisited. Journal of Youth and Adolescence, 10(6), 407–441.

14.

Cho

(2021). Modeling the reciprocal relationships between group-based developmental trajectories of peer delinquency and self-reported delinquency during adolescence: An application of interactional theory. Criminal Justice and Behavior, 48(2), 234–255.

15.

Cho

Lacey

(2021). The impact of life domains on developmental trajectories of peer delinquency among Korean adolescents: A partial test of Agnew’s general theory of crime and delinquency with a group-based trajectory modeling approach. International Journal of Offender Therapy and Comparative Criminology. Advance online publication. https://doi.org/10.1177/0306624X211016326

16.

Cho

Lacey

Kim

(2021). Developmental trajectories of delinquent peer association among Korean adolescents: A latent class growth analysis approach to assessing peer selection and socialization effects on online and offline crimes. Journal of Contemporary Criminal Justice, 37(3), 379–405.

17.

Cho

Lee

J. R.

(2020). Joint growth trajectories of bullying perpetration and victimization among Korean adolescents: Estimating a second-order growth mixture model–factor-of-curves with low self-control and opportunity correlates. Crime & Delinquency, 66(9), 1296–1337.

18.

Cooley

C. H.

(1902). Human nature and the social order. Scribners and Sons.

19.

Cullen

F. T.

Cullen

J. B.

(1978). Toward a paradigm of labeling theory (No. 58). University of Nebraska at Lincoln.

20.

Davis

(1961). Deviance disavowal: The management of strained interaction by the visibly handicapped. Social Problems, 9(2), 120–132.

21.

Farrington

D. P.

Hawkins

J. D.

(1991). Predicting participation, early onset and later persistence in officially recorded offending. Criminal Behaviour and Mental Health, 1(1), 1–33.

22.

Felson

R. B.

(1980). Communication barriers and the reflected appraisal process. Social Psychology Quarterly, 43(2), 223–233.

23.

Felson

R. B.

(1985). Reflected appraisal and the development of self. Social Psychology Quarterly, 48(1), 71–78.

24.

Felson

R. B.

(1989). Parents and the reflected appraisal process: A longitudinal analysis. Journal of personality and social psychology, 56(6), 965–971.

25.

Felson

R. B.

(1993). The (somewhat) social self: How others affect self-appraisals. In Suls

(Ed.), Psychological Perspectives on the Self (Vol. 4, pp. 13–38). Erlbaum.

26.

Fergusson

D. M.

Horwood

L. J.

Nagin

D. S.

(2000). Offending trajectories in a New Zealand birth cohort. Criminology, 38(2), 525–552.

27.

Grasmick

H. G.

Bursik

R. J.

Jr (1990). Conscience, Significant Others, and Rational Choice: Extending the Deterrence Model. Law and Society Review, 24(3), 837–861.

28.

Heimer

(1996). Gender, interaction, and delinquency: Testing a theory of differential social control. Social Psychology Quarterly, 59(1), 39–61.

29.

Heimer

Matsueda

R. L.

(1994). Role-taking, role commitment, and delinquency: A theory of differential social control. American Sociological Review, 59(3), 365–390.

30.

Hepburn

J. R.

(1977). The impact of police intervention upon juvenile delinquents. Criminology, 15(2), 235–262.

31.

Horney

Osgood

D. W.

Marshall

I. H.

(1995). Criminal careers in the short-term: intra-individual variability in crime and its relation to local life circumstances. American Sociological Review, 60(5), 655–673.

32.

Jensen

G. F.

(1972). Delinquency and adolescent self-conceptions: A study of the personal relevance of infraction. Social Problems, 20(1), 84–103.

33.

Jensen

G. F.

(1980). Labeling and identity: Toward a reconciliation of divergent findings. Criminology, 18(1), 121–129.

34.

Johnson

C. L.

Wilcox

Peterson

(2019). Stressed out and strapped: Examining the link between psychological difficulties and student weapon carrying and use. Criminal Justice and Behavior, 46(7), 980–998.

35.

Kaplan

H. B.

(1976). Self-attitudes and deviant response. Social Forces, 54(4), 788–801.

36.

Katz

(1988). Seductions of Crime: A Chilling Exploration of the Criminal Mind–From Juvenile Delinquency to Cold-blooded Murder. Basic Books

37.

Kavish

D. R.

Mullins

C. W.

Soto

D. A.

(2016). Interactionist labeling: Formal and informal labeling’s effects on juvenile delinquency. Crime & Delinquency, 62(10), 1313–1336.

38.

Kim

Lee

(2019). Does it take a school? Revisiting the influence of first arrest on subsequent delinquency and educational attainment in a tolerant educational background. Journal of Research in Crime and Delinquency, 56(2), 254–302.

39.

Kinch

J. W.

(1963). A formalized theory of the self-concept. American Journal of Sociology, 68(4), 481–486.

40.

Krueger

(1998). On the perception of social consensus. In Zanna

M. P.

(Ed.). Advances in experimental social psychology (Vol 30, pp. 163–240). Academic Press.

41.

Laub

J. H.

Sampson

R. J.

(2001). Understanding desistance from crime. Crime and Justice, 28, 1–69.

42.

Le Blanc

. (1997). A generic control theory of the criminal phenomenon: The structural and dynamic statements of an integrative multilayered control theory. Developmental Theories of Crime and Delinquency, 7, 215–285.

43.

Lee

J. S.

Tajima

E. A.

Herrenkohl

T. I.

Hong

(2017). Effects of formal and informal deviant labels in adolescence on crime in adulthood. Social Work Research, 41(2), 97–110.

44.

Lemert

(1951). Social pathology. McGraw-Hill

45.

Liu

(2000). The conditional effect of peer groups on the relationship between parental labeling and youth delinquency. Sociological Perspectives, 43(3), 499–514.

46.

Lofland

(1969). Deviance and identity. Prentice Hall.

47.

Lopes

Krohn

M. D.

Lizotte

A. J.

Schmidt

N. M.

Vasquez

B. E.

Bernburg

J. G.

(2012). Labeling and cumulative disadvantage: The impact of formal police intervention on life chances and crime during emerging adulthood. Crime & Delinquency, 58(3), 456–488.

48.

Maruna

(2001). Making good: How ex-convicts reform and rebuild their lives. American Psychological Association.

49.

Maruna

Lebel

T. P.

Mitchell

Naples

(2004). Pygmalion in the reintegration process: Desistance from crime through the looking glass. Psychology, Crime & Law, 10(3), 271–281.

50.

Matsueda

R. L.

(1992). Reflected appraisals, parental labeling, and delinquency: Specifying a symbolic interactionist theory. American journal of sociology, 97(6), 1577–1611.

51.

Matsueda

R. L.

Heimer

(1997). A symbolic interactionist theory of role-transitions, role-commitments, and delinquency. Developmental Theories of Crime and Delinquency, 44(3), 163–213.

52.

Matza

(1964). Delinquency and Drift. Wiley.

53.

Mead

G. H.

(1934). Mind, self and society. University of Chicago Press.

54.

Meade

A. C.

(1974). The labeling approach to delinquency: State of the theory as a function of method. Social Forces, 53(1), 83–91.

55.

Menard

Morse

B. J.

(1984). A structuralist critique of the IQ-delinquency hypothesis: Theory and evidence. American Journal of Sociology, 89(6), 1347–1378.

56.

Moffitt

T. E.

(1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100(4), 674–701.

57.

Mowen

T. J.

Brent

J. J.

Boman

J. H.

IV (2020). The effect of school discipline on offending across time. Justice Quarterly, 37(4), 739–760.

58.

Paternoster

(2019). Prosocial identities and youth violence. Journal of Research in Crime and Delinquency, 56(1), 84–128.

59.

Nagin

D. S.

(1999). Analyzing developmental trajectories: A semiparametric, group-based approach. Psychological Methods, 4(2), 139–157.

60.

Nagin

D. S.

(2005). Group-based modeling of development. Harvard University Press.

61.

Nagin

D. S.

Land

K. C.

(1993). Age, criminal careers, and population heterogeneity: Specification and estimation of a nonparametric, mixed Poisson model. Criminology, 31(3), 327–362.

62.

Nagin

D. S.

Odgers

C. L.

(2010). Group-based trajectory modeling in clinical research. Annual Review of Clinical Psychology, 6, 109–138.

63.

Wilcox

. (2018). Routine activity theory, target congruence, and school context: A multilevel analysis of teacher victimization. Victims and Offenders, 13(3), 349–372.

64.

Paternoster

Bushway

(2009). Desistance and the “feared self”: Toward an identity theory of criminal desistance. Journal of Criminal Law and Criminology, 99(4), 1103–1157.

65.

Paternoster

Iovanni

(1989). The labeling perspective and delinquency: An elaboration of the theory and an assessment of the evidence. Justice Quarterly, 6(3), 359–394.

66.

Patterson

G. R.

Yoerger

(1993). Developmental models for delinquent behavior. In Hodgins

(Ed.), Mental disorder and crime (pp. 140–172). Sage Publications, Inc.

67.

Piquero

A. R.

Farrington

D. P.

Blumstein

(2007). Key issues in criminal career research: New analyses of the Cambridge study in delinquent development. Cambridge University Press.

68.

Ray

M. C.

Downs

W. R.

(1986). An empirical test of labeling theory using longitudinal data. Journal of Research in Crime and Delinquency, 23(2), 169–194.

69.

Restivo

Lanier

M. M.

(2015). Measuring the contextual effects and mitigating factors of labeling theory. Justice Quarterly, 32(1), 116–141.

70.

Reynolds

S. J.

Ceranic

T. L.

(2007). The effects of moral judgment and moral identity on moral behavior: An empirical examination of the moral individual. The Journal of Applied Psychology, 92(6), 1610–1624.

71.

Rocque

Posick

Paternoster

(2016). Identities through time: An exploration of identity change as a cause of desistance. Justice Quarterly, 33(1), 45–72.

72.

Rosenberg

Simmons

R. G.

(1972). Black and white self-esteem: The urban school child. rose monograph series. American Sociological Association.

73.

Sampson

R. J.

Laub

J. H.

(1997). A life-course theory of cumulative disadvantage and stability of delinquency. In Thornberry

T. P.

(Ed.), Developmental theories of crime and delinquency (pp. 133–161).Transaction.

74.

Scimecca

J. A.

(1977). Labeling theory and personal construct theory: Toward the measurement of individual variation. The Journal of Criminal Law & Criminology, 68(4), 652–659.

75.

Shrauger

J. S.

Schoeneman

T. J.

(1979). Symbolic interactionist view of self-concept: Through the looking glass darkly. Psychological Bulletin, 86(3), 549–573.

76.

Sparks

Shepherd

(1992). Self-Identity and the theory of planned behavior: Assesing the role of identification with “Green Consumerism.” Social Psychology Quarterly, 55(4), 388–399.

77.

Tannenbaum

(1938). Crime and the community. Ginn and Company.

78.

Thoits

P. A.

(1985). Self-labeling processes in mental illness: The role of emotional deviance. American Journal of Sociology, 91(2), 221–249.

79.

Topalli

(2005). When being good is bad: An expansion of neutralization theory. Criminology, 43(3), 797–836.

80.

Triplett

R. A.

Jarjoura

G. R.

(1994). Theoretical and empirical specification of a model of informal labeling. Journal of Quantitative Criminology, 10(3), 241–276.

81.

Ward

J. T.

Krohn

M. D.

Gibson

C. L.

(2014). The effects of police contact on trajectories of violence: A group-based, propensity score matching analysis. Journal of Interpersonal Violence, 29(3), 440–475.

82.

White

H. R.

Bates

M. E.

Buyske

(2001). Adolescence-limited versus persistent delinquency: Extending Moffitt’s hypothesis into adulthood. Journal of Abnormal Psychology, 110(4), 600–609.

83.

Wiley

S. A.

Slocum

L. A.

Esbensen

F. A.

(2013). The unintended consequences of being stopped or arrested: An exploration of the labeling mechanisms through which police contact leads to subsequent delinquency. Criminology, 51(4), 927–966.

84.

Wolfgang

M. E.

Figlio

R. M.

Sellin

(1972). Delinquency in a birth cohort. University of Chicago Press.

85.

Zhang

(1997). Informal reactions and delinquency. Criminal Justice and Behavior, 24(1), 129–150.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

3.52 MB