Youth Gang Involvement and Long-Term Offending: An Examination into the Role of Psychopathic Traits

Abstract

Most policies to combat gang criminal behavior are rooted in deterrence and punitive strategies. This is fueled by moral panic, a get tough on crime rhetoric, and a lack of understanding for the psychological factors that may influence this behavior. Further, the extant literature has consistently observed that gang membership is associated with increased criminal behavior. In an effort to promote and shift away from punitive approaches in response to gang delinquency, the current study investigates the role psychopathic traits have in violent and property offending, longitudinally, in a sample of gang-involved youth. The study implemented count mixed effect models to investigate the topic longitudinally in waves 3, 5, 7, 8, 9, 10, and 11, while controlling for other variables with violent and property offending frequency. The current study found that some psychopathic traits are associated with offending behavior, longitudinally, in gang members and youth with a history of gang involvement. The findings suggest that gang intervention strategies should include empirically supported programs for treating psychopathic traits in gang identified youth to reduce involvement in delinquent behavior. Further, practitioners, researchers, and policymakers should collaborate to develop more empirically supported strategies to reduce and prevent gang delinquent behavior from an empathetic lens.

1. Introduction

Most strategies implemented and funded for gangs concentrate on suppression tactics (i.e., punitive gang legislation and harsh police enforcement), which are fueled by moral panic and a get tough on crime rhetoric [1,2]. McCorkle and Miethe [3] found that the gang panic facilitated by law enforcement and the media in Las Vegas resulted in the passage of punitive legislation targeted at gangs and increased law enforcement resources. Although suppression tactics have not been proven effective, many continue to implement and support these strategies because of the assumed deterrent effect and the lack of evidence-based intervention programs [2,4]. Wood et al. [5], in a review of various suppression tactics implemented against gangs, found a slight reduction in gang delinquent behavior or no effect on gang delinquent acts. Further, some of these tactics have resulted in the misidentification of community members, and promote the stereotyping of citizens as it relates to gang affiliation [5]. Unfortunately, most gang intervention strategies focus on sociological constructs and do not consider the role of psychological issues in exacerbating gang delinquency, which is also reflected in the empirical literature [2,6,7].

Most empirical investigations examining the relationship between gang involvement and criminal behavior have concentrated on sociological variables to comprehend and explain the relationship, and still fail to comprehend it [8,9,10,11,12]; however, recent investigations have begun to concentrate on psychopathic related factors. Recently, DeLisi et al. [9] found that disruptive behavior disorders (DBDs) rendered the relationship between gang involvement and criminal behavior insignificant, suggesting the gang delinquency relationship is spurious. In contrast, Wolff et al. [11] found that DBDs were associated with criminal behavior, and DBDs did not make the gang delinquency relationship spurious. Previous research has observed that psychopathic traits were associated with long term offending [13,14,15,16,17]. Research into the relationship between gang delinquency and psychopathic-adjacent psychiatric diagnosis has yielded mixed results and has not been investigated longitudinally. The inconsistent findings and lack of empirical investigation into the relationship between gang delinquency and psychopathic traits may limit the effectiveness of intervention and prevention strategies [8]. For instance, previous work has found gang membership reduced the effectiveness of multi-systemic therapy (MST) for delinquency [18]; however, gang membership is fleeting, loosely structured, and most youth members would be considered peripheral or fringe [2]. Further, Dmitrieva et al. [19] found psychopathic traits were associated with gang embeddedness and status, which could suggest that youth deeply embedded in gangs with psychopathic traits may be more resistant to intervention approaches. However, scant investigation has been conducted on psychopathic traits’ role in gang delinquency, which could potentially improve our comprehension of gang delinquency and support the improvement of intervention/prevention strategies on the topic.

1.1. Psychopathy and the Relationship to Offending

Psychopathy is a multifaceted construct comprised of interpersonal, lifestyle, affective, and antisocial characteristics [20,21]. Psychopathy, as a multi-dimensional construct with interconnected features beyond the interpersonal/affective dimensions in youth, has received consistent support in the extant literature [20,22]. The interpersonal/affective dimension of psychopathy is represented by a lack of empathy, low guilt, superficial charm, deceit, and disregard for performance; while the lifestyle/antisocial facet is represented by boredom, sensation seeking, impulsivity, criminal versatility, and serious rule violations [20,23]. Relatedly, most studies have observed that youth who scored extremely high on the interpersonal/affective dimensions are more likely to perpetrate crimes well into adulthood and are likely to mature into gang leaders [19,20]. Conversely, adolescents that score higher in the socially/deviant dimensions were more likely to be involved in short/long-term substance abuse during gang membership, and, in adult samples, were more likely to recidivate [20,24]. To identify youth that may develop into adult psychopaths and develop treatment programs, the classification of psychopathy was extended to children; and the core traits in youth are represented by grandiose–manipulative (GM) (i.e., interpersonal), callous–unemotional (CU traits) (i.e., affective), impulsive–irresponsible (i.e., lifestyle) and conduct disorder (i.e., antisocial behavior) [20,21]. Although extending the classification of psychopathy to youth is a popular approach, critics have identified that the stigmatization associated with psychopathy can result in more negative outcomes in the juvenile justice system [25]. Further, the moderate stability of personality features from childhood into adulthood, coupled with the potential for the classification of normative behavior as psychopathic during this period [21], suggest caution for the downward extension of the construct. Although, scarce research into the treatment of psychopathic traits in youth has shown it to be effective at reducing the internalized and externalized features associated with the construct [26]. The features involved in psychopathy have resulted in several modern conceptualizations that associate the construct with criminal behavior [21].

The extant literature has consistently observed a relationship between psychopathic traits and offending [20,27]. Older studies have found psychopathic traits were associated with more incidents of offending longitudinally [28,29,30]. Specifically, individuals with elevated psychopathic traits are more likely to recidivate, be versatile, and be prolific in their criminal behavior [20,28,31,32,33,34]. In a meta-analysis, Asscher et al. [31] found that youth higher in psychopathic traits were involved in more delinquent behavior and were more likely to recidivate. Although a significant amount of evidence has observed a positive relationship between both constructs, fewer empirical investigations have examined the relationship longitudinally while controlling for common risk factors associated with offending [20].

A recent empirical investigation has begun addressing the lack of longitudinal studies on the relationship between early childhood and adolescent samples [20]. Virtanen et al. [35] found that childhood psychopathic personality was associated with antisocial behavior later in life, while controlling for the influence of ADHD symptoms. Bergstrom and Farrington [36] measured psychopathy as a unitary construct and found it was associated with offending versatility, violent offending, and convictions throughout the life course. Lussier et al. [37] found that three core features of psychopathy (i.e., grandiose–manipulative, impulsive–irresponsible, and antisocial behavior) predicted annual convictions in a sample of adolescents. Colins et al. [38] found that elevated psychopathic traits were not associated with longitudinal offending in a sample of girls. Lee and Kim [39], using the Pathways sample, found that the relationship between psychopathic traits and offending and substance use, longitudinally, was mediated by peer delinquency. Further, Ray [40] found that the CU component increased the likelihood of being in the declining gun-carrying group, while grandiose–manipulative increased the likelihood of being in the late starter gun-carrying group. Finally, Dyck et al. [41], in a sample of adolescents, observed that all three components were significantly associated with a decrease in offending longitudinally. Most empirical investigations have found psychopathic traits are associated with offending in a theoretically expected direction. However, some studies did not observe a significant relationship between the constructs, and others found that some components (e.g., impulsivity, CU traits) were associated with antisocial behavior and offending in unexpected directions. Further, few studies have investigated the relationship longitudinally for group offending and seldom have controlled for other risk factors associated with longitudinal criminal behavior [20,42].

1.2. Gang Involvement and Offending

Consistently, research has found that gang involvement is associated with an increase in criminal offending across sex, and various explanations have been provided to make sense of the relationship [43,44,45,46,47]. One explanation for the relationship is that gang-involved youth become involved in crime due to delinquent peer saturation and are taught and reinforced to perpetrate crime [7,48,49]. Macro-level explanations postulate that socially and economically deprived neighborhoods allow for the proliferation of criminal behavior because of the lack of formal and informal social controls [12,50,51,52]. Others argue that the unstructured routine activities of gang involvement, group dynamics, culture, norms, salient events, and multi-marginality promote the push and pull factors that support criminal behavior amongst gang members [53,54,55,56,57]. Although most explanations concentrate on sociological and group constructs, recent work has begun taking a sociopsychological approach to comprehending the gang delinquency relationship.

Recently, research has investigated the relationship by examining the role of adverse childhood experiences (ACEs), mental health symptoms, and trauma on the gang delinquency relationship [58]. Chui et al. [6] observed that the moderate ACE’s group of gang members perpetrated the most criminal behavior longitudinally. Nydegger et al. [59] found that polytraumatization was associated with more mental health problems, delinquency, and drug distribution in a sample of gang-involved youth. Ross and Arsenault [60] found that trauma was associated with more violence and other delinquent acts. Further, older studies have observed that trauma was associated with the development of mental health symptoms (i.e., PTSD, post-traumatic stress, suicidal ideation) and the prescription of psychotropic medications in gang members, which facilitated more violent behavior [61,62,63,64]. A burgeoning amount of empirical investigation has found that sociopsychological constructs (e.g., trauma, ACE’s, MHS) are associated with more gang delinquency. This is compounded by the significant overlap between the variables associated with psychopathic traits, antisocial behavior, and gang involvement [65,66,67,68]. However, scant empirical investigation has examined psychopathic traits’ role in the gang delinquency relationship.

1.3. The Current Study

The current study examines psychopathic traits’ influence on the gang offending relationship longitudinally. Related studies examining psychopathy’s relationship with gang membership are mixed, and have not examined psychopathy’s impact on delinquency amongst gang members [69]. The two studies examining psychopathic traits and gang delinquency have observed a relationship between the constructs; however, the role gang status played has yielded inconsistent results [9,11]. The studies have not examined the relationship in a sample of gang members or youth with a history of gang membership; have not examined the relationship longer than a year; rely on the presence of a related psychiatric diagnosis; and have not controlled for common risk factors associated with long-term offending. Relatedly, adjacent studies have not accounted for other risk factors (e.g., group offending, victimization experiences, delinquent peer influence, moral disengagement) related to offending, and most studies are cross-sectional and retrospective. The current study attempts to address these gaps and contribute to understanding gang delinquency by investigating psychopathic traits as a multifaceted construct of violent and property offending, longitudinally, in a sample of gang-involved youth and youth with a history of gang involvement. Thus, the current study explores the following research questions:

(1)

Do CU traits significantly impact violent and property offending frequency over time in a sample of gang-involved youth or youth with a history of gang involvement?

(2)

Do impulsive–irresponsible traits significantly impact violent and property offending frequency over time in a sample of gang-involved youth or youth with a history of gang involvement?

(3)

Do grandiose–manipulative traits significantly impact violent and property offending frequency over time in a sample of gang-involved youth or youth with a history of gang involvement?

2. Method

2.1. Data

The data utilized to investigate the research question is the Pathways to Desistance dataset (a multi-site, longitudinal study of serious adolescent offenders while they mature into adulthood) [70]. Ten agencies sponsored data collection to provide policy-makers and justice officials with empirical information on various issues within juvenile justice. The Pathways to Desistance study was initially organized to investigate social and psychological variables related to desistance amongst serious delinquents. The dataset comprises 700 juvenile offenders from Philadelphia, Pennsylvania, and 654 from Phoenix, Arizona. Adolescents were identified on their adjudication charge, age, and scale of the initial population [71]. Following the signing of the appropriate consents, the preliminary meetings occurred in either the youth’s home, an agreed upon location, or the juvenile detention facility. See Mulvey [72] and Schubert et al. [71] for an in-depth discussion about the supervising researchers’ methodology and data collection procedures.

For the present study, youth that responded yes to the following two items “Have you ever been a member of a gang?” and “Gang membership six months prior” at the baseline are included in the analysis (n = 315), which is 23% of the total Pathways sample. Participants that responded yes to either of the questions at the baseline were defined as a gang member or youth with a history of gang membership and were extracted from the larger sample. The average age of the gang-involved youth in the sample was 16.02 (1.10), and most of the youth were males (91.1%). Most of the sample was Hispanic (58.4%), followed by Black (22.2%), White (13.7%), and Other (5.7%); along with a mean socio-economic score of 54.29 (12.24) (see

Table 1. Sample Description.

	%	M	SD	Min	Max
White	13.70
Black	22.20
Hispanic	58.40
Other	5.70
Male	91.10
Female	8.90
Age		16.02	1.104	14	18
SES Status		54.29	12.24	26	77

Note. All decimals rounded to the nearest hundredth. SES represents for socio-economic status.

). Attrition for the entire sample was 13.3% (n = 42).

2.2. Measures

2.2.1. Dependent Variables

Violent Offending Frequency. Violent offending frequency is represented by the sum of aggressive offenses reported across eleven items adapted from the Self-Reported Offending Inventory [34,39,73]. The items inquired about youth aggressive offending in the past six months include: (1) “destroyed/damaged property”, (2) “set fire”, (3) “forced someone to have sex”, (4) “murder”, (5) “shot someone”, (6) “shot at someone”, (7) “took by force with a weapon”, (8) took by force without a weapon”, (9) beat up someone resulting in serious injury”, (10) “participated in a fight”, and (11) “beat up someone as part of a gang”.

Property Offending Frequency. Income offending frequency is represented by the sum of income offenses reported across ten items adapted from the Self-Reported Offending Inventory [34,39,73]. The items inquired about the youth income offending in the past six months and the items include: (1) “broke in to steal”, (2) “shoplifted”, (3) “bought/received/sold stolen prop”, (4) “used check/credit card illegally”, (5) “stole care or motorcycle”, (6) “sold marijuana”, (7) “sold other drugs”, (8) “been paid by someone for sex”, (9) “took by force with a weapon”, (10) “took by force without a weapon”.

2.2.2. Predictor Variables

Psychopathic traits. Psychopathic traits are measured by the Youth Psychopathic Traits Inventory (YPI) [74], which is a self-report instrument designed to assess psychopathic traits in youth [70]. The measure taps three dimensions of psychopathy: the grandiose–manipulative dimension, the callous–unemotional dimension, and the impulsive–irresponsible dimension. The scale contains 50 items to which participants respond on a 4-point Likert scale ranging from “Does not apply at all” to “Applies very well”. Several items in the scale are reverse coded so that higher scores indicate more psychopathic characteristics [70,74,75]. The self-report nature of the YPI does not require trained interviewers to administer the instrument or official criminal files for review, and reduces socially desirable responses due to the neutral or appealing framing of items designed to measure psychopathic traits [74,75]. Further, previous work has found the YPI comparable or superior to other psychopathy assessments (i.e., PCL-YV) [75,76,77]. Since psychopathy is a multi-dimensional construct, the grandiose–manipulative dimension, callous–unemotional dimension, and impulsive–irresponsible dimension [22] are analyzed separately in the models.

2.2.3. Control Variables

Exposure to Violence. Selner-O’Hagan et al.’s [78] Exposure to Violence Inventory (ETV) was used in the sample to investigate the number of violent incidents respondents experienced [70]. The question from the ETV investigates violent incidents youth have both experienced and observed. Some items ask about the youth’s exposure to incidents of death (e.g., has anyone close to you tried to kill him/her self, has anyone close to you died, have you ever found a dead body, have you ever tried to kill yourself) [70]. The scale probes about 17 different situations; items to which the participants respond with yes have a series of follow-up questions regarding the incident. The majority of the items, except for rape, were restricted to the number of times the event occurred. Participants that responded yes to being a victim of rape or sexual assault are asked four additional questions (i.e., has it happened more than once?” “relationship of the perpetrator?” “location of the incident?” and “location if other?”) [70]. For the study, the violent victimization subscale and the violence witnessed subscale of the exposure to violence scales are used to represent violent victimization and violence witnessed.

Future Orientation. Cauffman and Woolard [79] used items from the Life Orientation Task [80], Time Perspective Scale [81], and the Future Consequences Scale [82] to develop the Future Outlook Inventory (FOI). The FOI uses Likert scale items (e.g., “I will keep working at difficult boring task if I know that will help me get ahead later”) ranging from 1 to 4 (1 = Never True to 4 = Always True) and higher scores signify more future consideration and planning by the respondent.

Perceptions of psychic rewards of crime. To operationalize the concept, the personal rewards subscale of the Indices of Personal and Social Costs and Rewards was adapted for the sample [70], which is consistent with previous studies [83,84,85]. The personal rewards subscale includes 7 items with questions like (i.e., “How much of a thrill or rush is it to break into a store or home”), inquiring about the amount of excitement or fun derived from perpetrating delinquent scenarios.

Moral Disengagement. Bandura et al.’s [86] Mechanisms of Moral Disengagement tool (MMD) was used to measure a respondent’s inclination to invoke moral disengagement strategies. The overall score of this instrument has good internal consistency and internal validity at the baseline and following timepoints (alpha = 0.88) [70]. The MMD scale has 32, three-point Likert scaled items (e.g., “Disagree” to “Agree”) with higher scores suggesting more moral detachment. The instrument contains eight dimensions: moral justification (e.g., “It is alright to beat someone who bad mouths your family.”), euphemistic language (e.g., “Slapping and shoving someone is just a way of joking.”), advantageous comparison (e.g., “It is okay to insult a classmate because beating him/her is worse.”), displacement of responsibility (e.g., “Kids cannot be blamed for using bad words when all their friends do it”), diffusion of responsibility (e.g., “A kid in a gang should not be blamed for the trouble the gang causes”), distorting consequences (e.g., “Teasing someone does not really hurt them”), attribution of blame (e.g., “If kids fight and misbehave in school it is their teacher’s fault”), and dehumanization (e.g., “Some people deserve to be treated like animals”) [70]. Following previous scholars’ advice concerning the MMD scale application, the overall score will be used.

Motivation to Succeed. The motivation to succeed items are from Eccles et al.’s [87] scale, which contains six questions inquiring about the participants’ evaluation of available opportunities in their neighborhood regarding schooling and work (e.g., “In my neighborhood it is easy for a person to get a good paying job”, “I will never have as much opportunity to succeed as kids from other neighborhoods”); and two items on adolescent perceptions regarding academic success (e.g., “How far do you think you will go in school,” “How far would you like to go in school”) [70]. This questionnaire reflects Skinner’s theoretical perspective, which suggests that achievement motivation is influenced by an interaction of mean-end beliefs (i.e., that specific causes can produce particular outcomes), agency beliefs (i.e., access to the means to accomplish tasks), and control beliefs (i.e., one’s beliefs that they can accomplish goals). Thus, assessing the role of opportunity in future success is essential. A higher score indicates more optimism regarding future success.

Neighborhood Conditions. The items from Sampson and Raudenbush’s [88] self-report measure were adapted to tap into physical and social disorder in a neighborhood [70]. The scale contains 21 items (e.g., “adults fighting or arguing loudly”, “people using needles or syringes to take drugs”), to which participants respond on a four-point Likert scale ranging from “Never” to “Often”, with higher scores indicating a greater degree of disorder within the community [70].

Association with Deviant Peers. The association with deviant peers is represented by the Peer Delinquency Antisocial behavior subscale from the Rochester Youth Study [23]. The subscale consists of 12 items (e.g., “How many of your friends have sold drugs?” “How many of your friends have suggested you sold drugs?”) on a five-point Likert-scale from “None of them to “All of them” with higher scores suggesting more association with delinquent peers [70].

Peer Delinquency Influence. Peer delinquency influence is represented by the Peer Antisocial Influence subscale from the Rochester Youth Study [89]. The subscale consists of seven items (e.g., “During the recall period how many of your friends have suggested that you should sell drugs?”) on a five-point Likert-scale from “None of them to “All of them” with higher scores suggesting more peer delinquency influence [70].

Resistant to Peer Influence. Resistance to peer influence is represented by the Resistance to Peer Influence inventory, which was developed to assess the degree adolescents act autonomously in interactions with their peer group [70]. Participants were presented with two conflicting scenarios (e.g., “Some people go along with their friends just to keep their friends happy” and “Other people refuse to go along with what their friends want to do, even though they know it will make their friends unhappy”) and are asked to choose the scenario which most closely reflects their behavior [70]. Finally, the participant is asked to rate the degree to which the statement is accurate (i.e., “sort of true” or “really true”). Ten such sequences are presented to the participant, each exploring a different dimension of potential influence: going along with friends, fitting in with friends, changing their mind, knowingly doing something wrong, hiding one’s true opinion, breaking the law, changing the way you usually act, taking risks, saying things do not really believe, and going against the crowd [70].

The overall resistance score was created by assigning each dimension a score from one to four, reflecting the particular combination of answers provided by the subject [70]. For example, choosing “some people go along with their friends just to keep their friends happy” followed by “really true of me” results in a score of one, while choosing “other people refuse to go along with what their friends want to do” followed by “really true of me” results in a score of four, and this is repeated for all ten dimensions [70].

2.3. Analytic Strategy

The research questions proposed in the study were analyzed with latent variable models (i.e., mixed effect models) due to the equivalency to latent growth models and the count nature of the dependent variables [90], and were analyzed in R v.4.2.2. The results of the Kolomogorov–Smirnov test for violent offending frequency (KS = 9.168, M/SD = 39.03 (157.52), p < 0.001) and property offending frequency (KS = 14.495, M/SD = 16.52 (32.12), p < 0.001), suggest each of the dependent variables are overdispersed, which suggests a negative binomial mixed effects model should be conducted for the dependent variables. The analysis was conducted for waves 3, 5, 7, 8, 9, 10, and 11 of the Pathway’s dataset, because each wave was collected annually [70,72]. To reduce issues related to instrumentation, the baseline was not included in the analysis because psychopathy is only measured using the Psychopathy Checklist: Youth Version (PCL-YV). For the neighborhood conditions measure, approximately (27.3–39%) was missing across the various time points within the sample, which resulted in a total of 33% missing for the variable in the total sample. Amongst the remaining independent, control, and dependent variables, approximately 8–10% of the cases were missing at various time points (see

Table 2. Independent and Dependent Variables Descriptives.

	Wave 1					Wave 2					Wave 3					Wave 4
Variables	N/MI	Mean	S.D.	Min	Max	N/MI	Mean	S.D.	Min	Max	N/MI	Mean	S.D.	Min	Max	N/MI	Mean	S.D.	Min	Max
Grandiose/Manipulative	292/23	40.24	11.23	19	80	290/25	40.71	10.92	20	79	287/28	38.28	10.84	19	77	292/23	36.72	10.59	20	80
Callous/Unemotional	292/23	34.06	6.62	18	55	290/25	34.39	6.57	15	58	287/28	33.40	6.52	19	57	292/23	32.49	7.01	16	55
Impulsive/Irresponsible	292/23	36.62	8.68	15	60	290/25	36.99	7.75	16	57	287/28	34.82	8.41	15	55	292/23	34.58	8.69	15	56
Violence Witnessed	292/23	1.29	1.53	0	6	290/25	0.92	1.38	0	7	288/27	0.84	1.35	0	7	294/21	1.25	1.73	0	7
Violent Victimization	292/23	0.31	0.71	0	4	290/25	0.19	0.57	0	3	288/27	0.13	0.45	0	3	294/21	0.38	0.82	0	4
Future Orientation	292/23	2.43	0.58	1	4	290/25	2.54	0.51	1.38	4	288/27	2.59	0.59	1	4	292/23	2.57	0.55	1.38	4
Personal Rewards of Crime	292/23	3.06	2.51	0	10	290/25	2.70	2.43	0	10	288/27	2.20	2.31	0	10	294/21	1.99	2.29	0	10
Moral Disengagement	292/23	1.65	0.38	1	2.91	290/25	1.60	0.38	1	3	288/27	1.54	0.39	1	2.69	294/21	1.53	0.40	1.03	3
Motivation to Succeed	290/25	3.24	0.65	1.50	5	289/26	3.25	0.66	1	5	286/29	3.30	0.63	1	5	294/21	3.30	0.59	1	4.83
Neighborhood Conditions	192/123	2.31	0.78	1	4	206/109	2.37	0.79	1	4	201/114	2.41	0.76	1	4	229/86	2.34	0.84	1	4
Delinquent Peers	286/29	2.14	0.95	1	5	289/26	1.89	0.85	1	5	274/41	1.69	0.79	1	4.54	290/25	1.89	0.91	1	5
Delinquent Peer Influence	290/25	1.73	0.88	1	5	289/26	1.65	0.80	1	5	279/36	1.46	0.67	1	3.86	290/25	1.57	0.82	1	5
Resistance to Delinquent Peer Influence	290/25	2.97	0.63	1.10	4	290/25	3.07	0.60	1.40	4	287/28	3.21	0.55	1.60	4	293/22	3.26	0.55	1.50	4
Violent Offending	292/23	7.04	23.09	0	205	290/25	5.07	22.35	0	246	288/27	3.05	15.40	0	215	291/24	5.54	23.07	0	270
Property Offending	292/23	35.68	167.02	0	1607	290/25	45.66	167.22	0	1049	288/27	29.09	146.87	0	1990	291/24	40.82	143.79	0	999
	Wave 5					Wave 6					Wave 7
Grandiose/Manipulative	290/25	37.32	11.37	20	77	285/30	35.45	10.75	20	67	273/42	36.66	10.35	20	76
Callous/Unemotional	290/25	32.51	7.30	15	57	285/30	32.04	7.47	15	57	273/42	32.66	6.89	15	53
Impulsive/Irresponsible	290/25	34.55	9.04	15	60	285/30	34.47	8.88	15	60	273/42	34.05	8.39	15	60
Violence Witnessed	290/25	0.99	1.41	0	6	286/29	1.01	1.46	0	6	273/42	1.11	1.59	0	6
Violent Victimization	290/25	0.28	0.73	0	4	286/29	0.24	0.67	0	4	273/42	0.26	0.69	0	5
Future Orientation	290/25	2.59	0.55	1	2.63	285/30	2.65	0.56	1	4	273/42	2.61	0.58	1	4
Personal Rewards of Crime	291/24	1.87	2.34	0	10	286/29	1.75	2.26	0	10	274/41	1.85	2.19	0	10
Moral Disengagement	290/25	1.49	0.37	1	2.63	285/30	1.46	0.39	1	2.97	273/42	1.45	0.39	1	3
Motivation to Succeed	292/23	3.35	0.60	1	5	286/29	3.42	0.64	1	5	274/41	3.30	0.54	2	5
Neighborhood Conditions	229/86	2.28	0.81	1	4	218/97	2.13	0.83	1	3.95	197/118	2.22	0.75	1	4
Delinquent Peers	288/27	1.77	0.81	1	5	277/38	1.81	0.88	1	5	267/48	1.70	0.81	1	5
Delinquent Peer Influence	288/27	1.54	0.74	1	5	277/38	1.57	0.82	1	5	267/48	1.43	0.65	1	5
Resistance to Delinquent Peer Influence	290/25	3.33	0.56	1	4	285/30	3.34	0.55	1.2	4	273/42	3.36	0.53	1.60	4
Violent Offending	290/25	7.85	72.57	0	1190	285/30	2.68	8.56	0	103	273/42	6.49	49.49	0	736
Property Offending	290/25	44.46	169.86	0	1990	285/30	37.98	161.79	0	2221	273/42	38.30	144.62	0	1260

Note. All decimals rounded to nearest hundredth. MI = missing.

). Waves 1 and 4 were not included, to make the findings more digestible and easier to understand for practitioners and academics interested in understanding and improving intervention approaches for individuals with a history of gang membership and gang offending longitudinally. Further, waves 1 and 4 also contained more missing data points for variables because, instead of annual data, data were only collected at 6-month intervals twice, which also factored into the decision not to include the waves. Missing data were handled with the Expectation Maximization (EM) procedure using the Amelia package in R [91]. Previous work has indicated that the EM procedure is superior to traditional imputation in precision and objectivity, and is appropriate for overdispersed count data [92,93,94]. The MAR assumptions rely on the assumption that participants with higher scores on items would have scored significantly higher or lower on relevant items compared to participants without missing scores (i.e., unsystematic missingness), which is impossible to test because the necessary information is missing [95,96]. Further, Gomer and Yuan [97] assert that a mixture of missing data mechanisms is more likely at play in a dataset, even with a pattern of missingness. Secondly, previous work using this data has implemented the EM algorithm for missing data, which inspired the methodological approach of the current study [94,98]. Finally, an Adaptive Gauss–Hermite Quadrature (AGH) model was implemented because recent evidence has shown that using multiple imputations and AGH is robust against MNAR and can produce accurate data estimates [95]. Previous work using the sample found moderate agreement between self-report and official offending, which remained stable longitudinally [99].

Model Building Process. The outcome variable for the models were property and violent offending overtime. The unconditional means model simply describes the variation in initial property and violent offending scores. Next in the unconditional growth model includes time (i.e., the different time points in the data) as the only predictor to examine within-individual effects (e.g., Level 1). Next, a model was conducted with time, independent variables, and control variables. The final conditional model included time, independent variables (the grandiose–manipulative dimension, the callous–unemotional dimension, and the impulsive–irresponsible dimension), control variables, and interaction effects to evaluate within- and between-individual effects. Further, the interaction effects were analyzed to confirm if the pattern of change differed over time for the time variant variables. Time was measured ordinally, with the initial assessment deemed 0, and each subsequent year was considered a new timepoint. Considering each variable included in the model was time-varying, an AGH mixed effect model was conducted to deal with issues of autocorrelation. Recently, Nestler [100], in a simulation study, observed that the AGH approach was robust against the influence of autocorrelation in the data and can accurately provide estimates for the least number of persons and time points for a more detailed review [100]. Goodness of fit and model selection was determined by the Akaike information criteria (AIC), and the Bayesian information criterion (BIC) statistic, and −2LnLike, which were ideal for nested models. Finally, an Analysis of Variance (ANOVA) was conducted to determine the most appropriate model, which is reported in

Table 3. Mixed effects models for predictors of violent offending longitudinally.

Predictors	IRR	CI	p	IRR	CI	p	IRR	CI	p
(Intercept)	1.43	1.18–1.74	<0.001	2.17	1.72–2.73	<0.001	0.02	0.01–0.07	<0.001
Time				0.81	0.76–0.86	<0.001	1.50	1.09–2.16	0.014
Grandiose/Manipulative							1.00	0.99–1.02	0.752
Callous/Unemotional							1.02	1.00–1.05	0.072
Impulsive/Irresponsible							1.03	1.00–1.05	0.024
Violence Witnessed							1.45	1.32–1.59	<0.001
Violent Victimization							1.30	1.07–1.57	0.008
Future Orientation							1.11	0.87–1.40	0.406
Personal Rewards of Crime							1.08	1.03–1.14	0.003
Moral Disengagement							1.13	0.80–1.59	0.495
Motivation to Succeed							1.02	0.89–1.17	0.789
Neighborhood Conditions							1.02	0.88–1.18	0.788
Delinquent Peers							1.45	1.17–1.80	0.001
Delinquent Peer Influence							1.02	0.81–1.27	0.873
Resistance to Delinquent Peer Influence							1.17	0.94–1.45	0.156
Time × Grandiose/Manipulative							1.00	0.99–1.00	0.754
Time × Callous/Unemotional							0.99	0.99–1.00	0.050
Time × Impulsive/Irresponsible							1.00	0.99–1.00	0.258
Time × Violence Witnessed							0.97	0.94–1.00	0.023
Time × Violent Victimization							1.05	1.00–1.11	0.058
Time × Future Orientation							0.97	0.91–1.04	0.382
Time × Personal Rewards of Crime							1.00	0.98–1.01	0.812
Time × Moral Disengagement							1.07	0.97–1.17	0.199
Time × Motivation to Succeed							1.01	0.95–1.08	0.790
Time × Neighborhood Conditions							1.02	0.98–1.06	0.345
Time × Delinquent Peers							1.05	0.98–1.12	0.183
Time × Delinquent Peer Influence							1.00	0.94–1.08	0.938
Time × Resistant Peer Delinquent Influence							0.92	0.86–0.99	0.018
Random Effects
Within person (σ2)	1.46		1.39				1.15
Initial Status (τ00)	2.38		1.87				0.10
							0.66
Rate of Change (τ11)			0.06				0.04
Covariance (ρ01)			0.03				−0.09
AIC	8632.6		8558.2				7914.7
BIC	8649.7		8592.4				8097.0
−2 loglikelihood	−4313.3		−4273.1				−3925.3
N	315		315				315
							799
Marginal R2/Conditional R2	0.000/0.620		0.041/0.677				0.381/0.699

a. Predictors: Grandiose/Manipulative, Callous/Unemotional, Impulsive/Irresponsible, Violence Witnessed, Violent Victimization, Future Orientation, Deviant Beliefs, Moral Disengagement, Motivation to Succeed, Neighborhood Conditions, Delinquent Peers, Delinquent Peer Influence, Resistance to Delinquent Peer Influence. b. Dependent Variable: Violent Offending. c. bold text are significant findings.

and

Table 4. Mixed effects models for predictors of income offending longitudinally.

Predictors	IRR	CI	p	IRR	CI	p	IRR	CI	p
(Intercept)	3.07	2.16–4.37	<0.001	39.12	3.53–433.66	0.003	0.07	0.00–1.34	0.078
Time				1.00	0.51–1.96	0.996	0.63	0.29–1.41	0.263
Grandiose/Manipulative							1.02	0.99–1.06	0.240
Callous/Unemotional							0.94	0.88–0.99	0.025
Impulsive/Irresponsible							1.08	1.02–1.13	0.004
Violence Witnessed							1.65	1.32–2.06	<0.001
Violent Victimization							2.12	1.32–3.41	0.002
Future Orientation							0.89	0.51–1.57	0.692
Personal Rewards of Crime							1.14	1.01–1.29	0.040
Moral Disengagement							2.43	1.14–5.20	0.022
Motivation to Succeed							0.67	0.43–1.06	0.086
Neighborhood Conditions							0.73	0.53–1.03	0.070
Delinquent Peers							1.45	0.86–2.43	0.165
Delinquent Peer Influence							1.46	0.85–2.50	0.170
Resistance to Delinquent Peer Influence							1.20	0.74–1.94	0.470
Time × Grandiose/Manipulative							1.00	0.99–1.01	0.885
Time × Callous/Unemotional							1.01	0.99–1.02	0.411
Time × Impulsive/Irresponsible							0.99	0.98–1.00	0.227
Time × Violence Witnessed							0.98	0.92–1.04	0.467
Time × Violent Victimization							0.95	0.83–1.08	0.414
Time × Future Orientation							1.01	0.86–1.18	0.916
Time × Personal Rewards of Crime							0.98	0.94–1.01	0.216
Time × Moral Disengagement							0.95	0.78–1.17	0.650
Time × Motivation to Succeed							1.03	0.91–1.16	0.668
Time × Neighborhood Conditions							1.09	0.99–1.19	0.071
Time × Delinquent Peers							1.17	1.00–1.36	0.043
Time × Delinquent Peer Influence							1.05	0.90–1.23	0.537
Time × Resistant Peer Delinquent Influence							1.00	0.87–1.14	0.978
Random Effects
Within person (σ2)	0.28			6.93			1.97
Initial Status (τ00)	10.07			0.00			0.04
							4.24
Rate of Change (τ11)				0.00			0.00
Covariance (ρ01)				−0.99			1.00
AIC	183,013			16,073			10,421
BIC	183,030			16,107			10,609
−2 loglikelihood	−91,503			−8030			−5177.5
N	315			315			315
							799
Observations	2205			2205			2205
Marginal R2/Conditional R2	0.000/0.973			0.000/0.000			0.319/0.625

a. Predictors: Grandiose/Manipulative, Callous/Unemotional, Impulsive/Irresponsible, Violence Witnessed, Violent Victimization, Future Orientation, Deviant Beliefs, Moral Disengagement, Motivation to Succeed, Neighborhood Conditions, Delinquent Peers, Delinquent Peer Influence, Resistance to Delinquent Peer Influence. b. Dependent Variable: Property Offending. c. bold text are significant findings.

.

3. Results

A mixed effects models was conducted to examine the relationship between psychopathic traits, (i.e., the grandiose–manipulative dimension, the callous–unemotional dimension, and the impulsive–irresponsible dimension), victimization witnessed, violent victimization, future orientation, perceptions of psychic rewards and crime, moral disengagement, motivation to succeed, neighborhood conditions, association with deviant peers, peer delinquency influence, and resistant to peer influence on violent offending frequency over time. Gang-involved youth and youth with a history of gang involvement that perpetrated a higher amount of violent offenses at the start were 50% more likely to perpetrate violent offenses over time (γ₀₁ = 1.50, p < 0.05). Gang-involved youth that scored higher on the impulsive–irresponsible dimension at the start were 3% more likely to perpetrate violent offenses over time (γ₀₄ = 1.03, p < 0.05). Gang-involved youth and youth with a history of gang involvement that witnessed more violent victimization at the start were 45% more likely to perpetrate violent offenses over time (γ₀₅ = 1.45, p < 0.001). Gang-involved youth and youth with a history of gang involvement that experienced more violent victimization at the start were 30% more likely to perpetrate violent offenses over time (γ₀₆ =1.30, p < 0.01). Gang-involved youth and youth with a history of gang involvement with higher perceptions of personal rewards for crime at the start were 8% more likely to perpetrate violent offenses over time (γ₀₈ = 1.08, p < 0.01). Gang-involved youth and youth with a history of gang involvement that associated with more delinquent peers at the start were 45% more likely to perpetrate violent offenses over time (γ₁₂ = 1.45, p = 0.001). Gang-involved youth and youth with a history of gang involvement that witnessed more violent victimization annually had a 3% decrease in violent offending annually (γ₁₈ = 0.97, p < 0.05). Gang-involved youth and youth with a history of gang involvement that were more resistant to delinquent peers annually had an 8% decrease in violent offending annually (γ₂₆ = 0.92, p < 0.05).

A mixed effects model was conducted to examine the relationship between psychopathic traits, (i.e., the grandiose–manipulative dimension, the callous–unemotional dimension, and the impulsive–irresponsible dimension), victimization witnessed, violent victimization, future orientation, deviant beliefs, moral disengagement, motivation to succeed, neighborhood conditions, association with perceptions of psychic rewards and crime, peer delinquency influence, and resistant to peer influence on property offending frequency over time. Gang-involved youth and youth with a history of gang involvement that scored higher on the callous unemotional dimension at the start were 6% less likely to perpetrate property offenses annually over time (γ₀₃ = 0.94, p < 0.05). Gang-involved youth and youth with a history of gang involvement that scored higher on the impulsive–irresponsible dimension at the start were 8% more likely to perpetrate property offenses annually over time (γ₀₄ = 1.08, p = 0.01). Gang-involved youth and youth with a history of gang involvement that witnessed more violent victimization at the start were 65% more likely to perpetrate property offenses annually over time (γ₀₅ = 1.65, p < 0.001). Gang-involved youth and youth with a history of gang involvement that experienced more violent victimization at the start were 112% more likely to perpetrate property offenses annually over time (γ₀₆ = 2.12, p < 0.01). Gang-involved youth and youth with a history of gang involvement with higher perceptions of psychic rewards and crime at the start were 14% more likely to perpetrate property offenses annually over time (γ₀₈ = 1.14, p < 0.05). Gang-involved youth and youth with a history of gang involvement that were more morally disengaged at the start were 143% more likely to perpetrate property offenses annually over time (γ₀₉ = 2.43, p < 0.05). Gang-involved youth and youth with a history of gang involvement that associated with more delinquent peers annually had a 17% increase in property offending annually (γ₂₄ = 1.17, p < 0.05).

4. Discussion

The current study’s goal was to contribute to understanding gang involvement and psychopathic traits, violent and property offending longitudinally. Before this study, scant empirical investigation had examined the relationship between psychopathic traits, violence, and property offending longitudinally in a sample of gang-involved youth and youth with a history of gang involvement. Most empirical investigations have been cross-sectional and retrospective, and concentrate on comparing non-gang members to gang members. The current study found that the initial perceptions of psychic rewards and crime, violence witnessed and experienced, and impulsivity irresponsible dimension were associated with more violent and property offenses over time. Previous research has identified that psychopathy, in general, is consistently associated positively associated with offending [20,27,31,33,34,101]. Previous work using the Pathways sample found higher GM scores in young adults and gang leaders [19]; considering this, it is likely low-level gang members are involved in violent and property crime at the behest of leaders or older members. In other words, the direct involvement of individuals high in GM is limited, which can contribute to insignificant findings. Ray [66] found that direct victimization experiences and delinquent peer association were associated with higher scores on the impulsive–irresponsible dimension of psychopathy. Previous work has identified that potential gang members are evaluated on delinquent versatility and potential and their relationship with current gang members, coupled with the fact that youth gang joining is motivated by the desire to prevent vicarious and direct victimization [7,45,102,103]. Thus, in the context of the study’s findings, it is probable that subsequent gang involvement increased exposure to intra/inter-gang violence and income crime, and youth higher in the impulsive–irresponsible dimension are more likely to perpetrate these acts because they are more likely to become embedded in the gang subculture, due to their propensity for risk-taking and thrill-seeking [7,56,104,105]. Previous work has found that youth higher in psychopathic traits may perceive crime as more rewarding and have a reward dominant response style [14,26,104,106,107,108]; in the context of the current findings, it is probable that gang-involved youth and youth with a history of gang involvement that scored high in the impulsive–irresponsible domain, and are more morally disengaged, are less likely to consider the risk of offending and perpetrating violent and property offenses with their peers. Another probable explanation is that gang-involved youth and youth with a history of gang involvement high in psychopathic traits that perceived crime as more rewarding may be more likely to become more embedded in the gang subculture, subsequently adopting more components of the “Street Code”, increasing their violent and property offenses [56,105]. Turanovic and Young [109] suggest violent youth are more likely to create social networks with other violent youth; it is probable that gang-involved youth and youth with a history of gang involvement are higher in the impulsive–irresponsible domain, more morally disengaged, and those that perceive crime as rewarding are more likely to cooperate in perpetrating violent and income crime over time.

Although the extant literature regarding CU traits suggests that youth with elevated CU traits identifies a subgroup of youth more likely to display a stable, chronic pattern of offending longitudinally [13,14,15,16,17], the current study found evidence to the contrary. Specifically, elevated CU traits at the start were associated with decreased property offending over time, consistent with studies examining antisocial behavior and recidivism [40,41,42]. The current findings are consistent with research that identified the impulsive–irresponsible dimension of psychopathy as the most important dimension for predicting future offending and distinguishing a unique subgroup [28,29]. Considering adolescents high in CU traits display higher pre-planning and preparation prior to antisocial acts [110,111], it is likely gang-involved youth and youth with a history of gang involvement with elevated CU traits may be disinterested in perpetrating property offenses that are spontaneous compared to more impulsive gang members [42]. Another probable explanation is that the relationship between CU traits and offending has been observed for severe and violent offenses, which does not apply to the items that make up the income offense scale (e.g., sold drugs, been paid for sex, shoplifting). Finally, although previous studies using the Pathways data have observed a relationship between CU traits and offending, the current study examines gang-involved youth and youth with a history of gang involvement longitudinally, while the previous studies have examined the entire sample or male-only samples.

4.1. Desistance from Violent and Property Offending

Finally, the current study also found that gang-involved youth and youth with a history of gang involvement who witnessed more violence and were more resistant to delinquent peer influences annually were less likely to perpetrate long-term violent crime. Previous work has found that repeated exposure to violence (e.g., loss of friends, loss of family members, and peers) can cause members to become disillusioned and burned out with the gang lifestyle, and these events can serve as push factors toward desistance [112,113,114,115,116], subsequently resulting in reduced violent offending annually. Gang members who reach the contemplation and exploration stages will be more resistant to their delinquent peers and will want to explore alternative lifestyles that do not involve engaging in aggressive behavior throughout their life course [112]. Finally, it is also likely that pull factors (e.g., associating with prosocial peers, encouragement from non-guardians or parents, involvement in prosocial programs) make gang members more resistant to their delinquent associates, eventually leading to gang desistance and reduced involvement with violent offending over time [112,113,114,116,117].

4.2. Treatment and Policy Implications

The findings suggest adopting intervention strategies that target specific psychopathic traits (i.e., impulsivity/irresponsibility) to reduce long-term violent behavior [21,66]. Although previous research has identified gang membership as a potential barrier for intervention strategies targeted at reducing delinquency [18]; research examining gang desistance suggests developing intervention programs for victimized and traumatized gang members in collaboration with former gang members [112]. The programs should concentrate on post-traumatic growth, encourage developing positive connections with prosocial peers and groups, and mitigate the risk of psychopathic traits through empirically based treatment programs [32,107,112,118,119,120,121]. For instance, the Mendota Juvenile Treatment Center (MJTC) is an intervention program that is quick, transparent, and rewards participants for engaging in prosocial behavior while withholding rewards for antisocial behavior [107,118]. This program is designed to take advantage of the reward–dominant response styles associated with adolescents high in psychopathic traits. Caldwell et al. [118] found that 18–22% of the improvement in psychopathic traits and a 5–11% reduction in violent and non-violent reoffending was directly attributed to the MJTC program. Another potentially effective intervention program may be the wrap-around method, which includes developing a personalized plan for the child and using mental health services available within the child’s network area [122]. The wrap-around method is a multi-pronged approach for treating disruptive behavior disorders symptoms and delinquency. In the context of the current findings, potentially getting gang-involved youth and youth with a history of gang involvement involved in intervention programs designed to reduce psychopathic traits, delinquency, and encourage reintegration into mainstream society may be effective in preventing future and long-term criminal involvement [112]. These suggestions would be effective treatment interventions in the criminal justice system; however, broader policy practices can also assist with prevention.

Most gang legislation is rooted in specific deterrence and usually implements punitive approaches for combating gangs [2]. For example, a disproportionate amount of funding is shifted toward gang enhancement laws, civil injunctions, and police gang units, while diversion/intervention programs specific to gang involvement are inadequately funded throughout the juvenile justice system [2,123]. The first policy suggestion is the reallocation of funds for combating gang related crime into empirically supported intervention programs designed to reduce, remedy, and prevent gang involvement and psychopathic traits. Secondly, more resources should be allocated to evaluate the effectiveness of the current gang intervention programs and consider restructuring these programs to consider the role of psychopathic traits and develop a comprehensive model for gang desistance. Finally, collaboration should be carried out amongst policymakers, community leaders, gang researchers, social service workers, and criminal justice practitioners to address gang crime and the influence psychopathic traits have on their behaviors, in order to develop effective policies and interventions [123,124].

4.3. Limitations and Future Research

The current study provides insight into the role psychopathic traits have in influencing criminal behavior longitudinally in a sample of gang-involved youth and youth with a history of gang involvement; however, several weaknesses are apparent in the study. The Pathways sample is comprised of youth involved in serious offenses, which is not representative of most youth involved in the justice system; thus, the findings cannot be generalized to all justice-involved youth who join gangs. The Pathways sample does not allow for the consideration of gang demographics (e.g., mixed-sex, racially diverse) and is not a comprehensive representation of all gangs (e.g., skinheads, bikers, international). It does not follow participants during early childhood, which makes accurately identifying temporal ordering difficult, limiting the scope of the study’s findings. Although the self-report nature of gang membership is vulnerable to deceit and memory issues, previous research has verified this as an effective approach for determining gang status [125,126]. Self-report psychopathy measures have been criticized with regard to validity due to the manipulative, deceitful nature of psychopaths; however, the instruments used in the current study have been validated in previous studies [74,75,76].

Despite the limitations, the study relies on empirically validated measures to operationalize the variables tested. The study carefully considers the contributions of previous studies on similar topics using the Pathways Data [19,39,40,66,117,119,127,128]. The current study uniquely contributes to increasing the understanding between psychopathic traits and criminal behavior longitudinally amongst gang-involved youth and youth with a history of gang involvement. Future research would benefit from considering the psychopathic traits’ influence on offending longitudinally in gang samples to better comprehend the factors involved in offending. Previous work has suggested the relationship between gang involvement and antisocial behavior may be spurious when controlling for the influence of disruptive behavior disorders [9]; more research must be conducted in the area to understand the relationship better and develop effective intervention strategies for gang-involved youth. Further, more gang research would benefit from examining the role psychological factors have in gang involvement and related antisocial behavior. Finally, future work would benefit from examining the potential moderating and mediating of psychopathic traits and moral disengagement longitudinally in sexual crimes, considering that we found that both constructs are associated with predatory sexual behavior [129].