Heterogeneity in the Continuity and Change of Early and Adult Risk Factor Profiles of Incarcerated Individuals: A Latent Transition Analysis

Abstract

Considerable research has examined risk factors for offending, but far less is known on the constellations of co-occurring risk factors, such as adverse childhood experiences and low self-control, and the overall continuity in risk between childhood/adolescence and adulthood. Using data on 735 adults incarcerated in a county jail in Florida, this study examines the latent heterogeneity in risk profiles using risk factors prominent in early years and adulthood, and whether risk profile severity changes across the early and adult risk models.

Latent Class Analyses revealed three risk profiles (low, medium, high) in both the early and adulthood risk factor models. Transition probabilities indicate continuity in high and low risk in the early and adult models, while escalation was found for those in the low and medium early risk profiles. These findings demonstrate the importance of identifying and addressing risk factors at an early age to disrupt continuity and escalation in risk over the life-course.

Keywords

risk factors latent class analysis continuity profiles developmental/life-course criminology

For decades, criminologists have sought to understand the relationship between early risk factors and antisocial behavior (e.g., Farrington, 1995; Loeber, 1990; Moffit, 2015; Reid & Patterson, 1989). Some of this research has focused on the dispositional or constitutional differences between individuals, such as impulsivity or low intelligence (e.g., Farrington, 2015; Gottfredson & Hirschi, 1990; Moffitt, 1993), while other research has focused on how environmental adversity such as abuse and trauma, influences later offending (e.g., Moffitt, 1993; Widom, 1989; for a review see Jolliffe et al., 2017). However, research also suggests that individual and environmental risk factors “do not operate in isolation” (Kendziora & Osher, 2004, p. 182), and offenders often display multiple risk factors that span across several domains (Loeber & Farrington, 1998; Thornberry et al., 1995). These constellations of risk factors operate in a cumulative fashion, particularly in early years, leading to a non-linear impact on future level of risk and future offending (Herrenkohl et al., 2000; Loeber & Farrington, 1998).

For instance, adverse childhood experiences (ACEs) have received considerable attention in the criminological, psychological, and health arenas, as they represent a major environmental risk factor in childhood/adolescence and have consistently been associated with negative outcomes, which in turn increase the risk for future antisocial behavior (Anda et al., 2006, 2010; Baglivio et al., 2015; Felitti et al., 1998; Fox et al., 2015). While psychotherapy interventions (e.g. trauma therapy) may help mitigate the effects of ACEs on negative adult outcomes, ACEs still tend to co-occur with other notable early risk factors (e.g. head injury), and the biological stress response to ACEs and trauma can alter brain structures and functions important for developing self-control, emotional control, and behavioral control (Cicchetti & Toth, 2005; Hart & Rubia, 2012; Raine, 2014; Witt et al., 2010). In turn, this combination of risk factors (e.g., trauma response, low self-control), has been associated with risk factors such as delinquent peer association, poor anger control, and substance abuse later in life (Gottfredson & Hirschi, 1990). In essence, ACEs and related early risk factors can lead to the development of, and co-occur with, many additional risk factors that increase the risk of negative outcomes including offending.

Paralleling this work has been an interest in adult risk factors for crime and violence (e.g., Burton et al., 1999; Glueck & Glueck, 1968; Wolfe et al., 2016), although limited research has examined how individual differences and environmental factors interact to increase the risk of offending in adulthood. Additional work is needed to determine if there is heterogeneity (or uniformity) in cumulative risk factor profiles for the early years and adulthood, and whether there is continuity (or change) in overall risk level at these time periods. Continuity in risk from childhood/adolescence to adulthood would suggest a need for early identification and treatment to reduce later risk of offending. Escalation in risk level from early years to adulthood would indicate an accumulation of risk factors over time, and underscore the value of early prevention strategies. Therefore, understanding the hetereogeneity and continuity in risk level from childhood/adolescence through adulthood is important for criminologists, practitioners, and public health officials, as it can be used to better inform the nature, timing, and duration of intervention programming according to the co-occurrence and severity of risk factors at different points in the life-course.

To that end, this study examines the latent heterogeneity that exists within early and adult risk factors for offending, and the continuity and change in risk profiles that occur over these time periods. Drawing from developmental/life-course (DLC) theory predictions on the stability in risk based on the age reference point (i.e. time period associated with the onset and greatest impact) of risk factors (Krohn et al., 1997), and research suggesting there is heterogeneity in risk factors among individuals over time (Farrington, 2005b; Moffitt, 1993; Sampson & Laub, 2005), this study uses a Latent Transition Analysis on data obtained from a sample of incarcerated adult arrestees to address the following research questions:

Is there latent heterogeneity in the constellations of early risk factors?

Is there latent heterogeneity in the constellations of adult risk factors?

Is there continuity or change in level of risk across the resultant early and adult risk profiles?

Our overall goal is to elaborate on how distinct constellations of early-life risk factors relate to risk levels assessed in adulthood. We begin by discussing how DLC research has shown both continuity and change in risk over time, before examining early risk factors, and then moving on to adult risk factors for offending.

Continuity and Change in Risk Factors Over the Life-Course

DLC theories aim to predict and explain offending behaviors, such as onset, persistence, frequency, and desistance, across the lifespan (Farrington, 2005a). These theories tend to focus on individual and environmental characteristics that are most predictive of criminal behavior, often grouping them as risk factors occurring in early years vs adulthood (see reviews in Farrington, 2003a, 2010). As such, DLC theories are unique in that they take a holistic approach to assessing risk, and they acknowledge that a constellation of early and adult risk factors can accumulate to influence individual involvement in offending.

Research suggests that there is a compounding effect of risk factors for antisocial and criminal behavior; those who experience a plethora of risk factors in early life are at higher risk of sustaining (and developing new) risk factors for offending in adulthood (Levenson et al., 2016; Reavis et al., 2013). Two processes are theorized to underlie the relative stability in risk factor patterns over the lifespan: cumulative continuity and interactional continuity (Caspi et al., 1987, 1989). Cumulative continuity is a process where risk factors are maintained as a function of the “progressive accumulation of their own consequences” (Caspi et al., 1987, p. 308). For instance, early difficulties in emotional regulation may interfere with forming social bonds later in life (Huesmann et al., 2009). Interactional continuity refers to the perpetuation of risk where initial risk factors elicit responses from the environment that support and/or maintain risk over time through reactive, proactive, and evocative processes (Caspi et al., 1987, 1989). For example, an individual with early risk factors such as ACEs, may interpret, experience, and respond to their environment in a way that increases the perpetuation of risk, such as mistrust of others, in the future (Temcheff et al., 2008).

More transitory, versus stable, environmental risk factors are associated with offending primarily during adolescence, while continuity in offending from juvenile years to adulthood is a function of risk factors largely identifiable in childhood and stable over time (Farrington, 2003b; Moffitt, 1993). However, research also indicates that situational factors can serve as turning points, such as gainful employment or marriage, which can change the trajectory of an individual’s risk and offending patterns, introducing discontinuity in risk patterns over the life-course (e.g. Sampson & Laub, 1993). Likewise, an individual may experience or develop new risk factors in adulthood, which in turn increase their risk profile and likelihood of offending.

For instance, Jennings et al. (2016) found that the risk factors for four offender types—abstainers, adolescence-limited, life-course persistent, and recoveries—were largely stable over time, while desistance from offending was associated with more limited risk over the life-course (see also Farrington et al., 2009; Moffitt, 1993). This raises the question: to what extent do risk levels change as a function of the age reference of the risk factors, and is this continuity different for those who are members of more or less severe early risk profiles to begin with?

Despite much theoretical work, little research has specifically examined the continuity of risk profiles between early and adult risk factors, and most importantly, whether risk profile membership is constant across early and adult risk models for all levels of risk. As research indicates that offenders are a heterogeneous population with varied risk factors among them (e.g. Fox & DeLisi, 2018), we anticipate there will not be a singular risk profile, even within arrestees, most of whom have established histories of offending. Rather, we expect there to be distinct subtypes of risk profiles within the offender population, corresponding with different levels and constellations of early and adult risk factors. We also examine whether risk profiles vary across the age reference (i.e. early years vs. adulthood) of the risk factors modeled. We begin by reviewing established risk factors for offending in the early and adult years, how risk factors may differentially exist among offenders, and how profiles of risk may vary over time.

Early Risk Factors

Considerable research has examined the early and static factors influencing antisocial behavior (Akers, 1998; Anderson, 1999; Farrington, 1986, 2005b; Gottfredson & Hirschi, 1990; Laub & Sampson, 2003; Loeber, 1990; Loeber & Farrington, 1998; Moffit, 1993; Sampson & Laub, 2005; Shaw & McKay, 1942). While some theories focus on individual factors and others emphasize structural factors, most identify risk factors for offending that develop in childhood and adolescence. The early life factors consistently associated with offending include trauma/ACEs (Baglivio et al., 2014, 2015; Fox et al., 2015; Smith & Thornberry, 1995; Widom, 1989), low self-control (Farrington, 1998; 2005a; Gottfredson & Hirschi, 1990; Pratt & Cullen, 2000; Wolfe, 2015; Wolfe et al., 2016), head injuries (Leon-Carrion & Ramos, 2003; Schwartz et al., 2018; Williams et al., 2010), gender (Harris, 1977; Lanctôt & Le Blanc, 2002) and race/ethnicity (Piquero, 2008; Steffensmeier et al., 2011).

These risk factors are among the strongest early risk factors that can be measured and used to guide prevention efforts for at-risk children. ACEs, head injuries, and self-control are stable factors that will theoretically have a lasting impact on an individual’s development and behavior unless an intervention occurs (e.g. Baglivio et al., 2014; Farrington, 1998; 2005a; Gottfredson & Hirschi, 1990; Leon-Carrion & Ramos, 2003). Race and gender are stable risk factors that are correlate (not causes) of offending (e.g., Harris, 1977; Piquero, 2008), while other early risk factors, such as parental supervision and peer delinquency, while important, are temporary factors that are external and highly variant over time (Moffitt, 1993). For example, parental supervision can vary with the employment status of the parent (e.g. working multiple jobs or being a stay-at-home parent), which presents a confound when analyzing parental monitoring as a risk factor on its own. To this end, the current study examines internal, highly stable, and/or longer impact early risk factors and correlates of crime. Further, although ACEs, early head injuries, and low self-control are very consequential for antisocial behavior, the influence of these risk factors can be decreased through direct individualized interventions, such as trauma therapy or cognitive behavioral therapy (CBT) with the youth, while other early risk factors, such as structural inequalities, are more difficult to address at the individudal level. For example, poverty and other structural inequities may be underlying contributors to risk factors, such as inadequate parental supervision, but require societal-level changes to address. From an intervention perspective, focusing interventions on individual risk factors most associated with future offending and risk (such as ACEs) is an effective and efficient approach to reducing later risk of crime, and the overall level of future risk.

Adverse childhood experiences

Adverse childhood experiences (ACEs) are traumatic and stressful experiences during the early years that have been associated with numerous negative health and behavioral outcomes over the life-course (Anda et al., 2006, 2010; Felitti et al., 1998; Kalmakis & Chandler, 2015; Teicher, 2000). ACEs consist of 10 traumatic events experienced before age 18, including: emotional, physical, and sexual abuse, emotional and physical neglect, household member substance abuse and mental illness, parental separation or divorce, parental incarceration, and witnessing domestic violence (Anda et al., 2010). In their landmark ACE study, Felitti and colleagues (1998) found that higher ACE scores predicted serious negative health outcomes such as bone disease, cancer, and early death. While just 6% of the community sample experienced four or more ACEs, other research suggests that as many as 25% of juvenile offenders (Baglivio et al., 2014) and 66% of incarcerated adult offenders (Fox et al., 2019) experience at least three or more ACEs in their lives. Importantly, ACEs and trauma have consistently been linked to a host of antisocial outcomes such as delinquency, violence, and prolific offending (Baglivio et al., 2015; Fox et al., 2015; Smith & Thornberry, 1995; Widom, 1989).

Research suggests that the biological processes associated with chronic trauma can serve as the mechanisms linking ACEs and negative behavioral outcomes. Experiencing a traumatic event triggers a biological stress response, increasing levels of neurotransmitters that activate the sympathetic nervous system (SNS). This leads to a release of hormones such as cortisol and adrenaline to heighten arousal and prepare the body to deal with the threat (i.e. fight or flight) (De Billis, 2001; De Bellis & Zisk, 2014). While the stress hormones released by the SNS can help deal with these threats short-term, the repeated activation of the stress system in maltreated children can impair several regions in the brain important for response inhibition, and behavioral and emotional control (Cicchetti & Toth, 2005; Hart & Rubia, 2012; Raine, 2014). Therefore, examining the role of ACEs as an early risk factor for offending is essential.

Head injuries

The human brain is not fully developed until around age 26. Therefore, an injury to a child’s head, whether from physical abuse, a sports injury, fall, or fight, has the potential to alter the brain’s development. Head injuries are prevalent in offending populations (Fox et al., 2019; Schwartz et al., 2018; Williams et al., 2010), with childhood head injuries found in higher rates of violent versus non-violent offenders (Leon-Carrion & Ramos, 2003). This suggests that head injuries may be an important risk factor for criminal behavior.

Head injuries in children impact the developing brain in many ways. Resulting structural changes include a reduction in white and gray matter in the hippocampus and the prefrontal cortex, which helps regulate executive function skills including judgment and reasoning (Witt et al., 2010). Severe head injuries can result in changes in the amygdala, which is responsible for emotional regulation (Beauchamp et al., 2011). Functional changes to the brain after a head injury can result in impaired memory, attention, and cognitive control (Sharp et al., 2014). As such, head injuries occurring while the brain is still developing can result in physical neural changes leading to changed functioning in terms of risk perception and decision-making, increasing the risk for future offending.

Low self-control

Gottfredson and Hirschi (1990) posit that antisocial behavior is the product of low self-control, which they state is relatively stable from early adolescence through adulthood. Low self-control is measured across dimensions of impulsivity, insensitivity, preference for physical activities and simple tasks, risk-seeking, and short-sightedness (Gottfredson & Hirschi, 1990). Low self-control is a risk factor that can be identified in childhood to predict antisocial, delinquent, criminal, and violent behavior at all stages in life (Farrington, 1998, 2005a; Gottfredson & Hirschi, 1990; Pratt & Cullen, 2000; Wolfe, 2015; Wolfe et al., 2016).

A meta-analysis evaluating the literature on self-control as early risk factor for crime found the effect of low self-control on offending was strong, despite methodological and measurement variations across studies (Pratt & Cullen, 2000). Self-control is hypothesized to have a direct effect on offending, as low self-control is associated with taking less time to make decisions and more impulsivity, increasing the likelihood of criminal behavior (Pratt & Cullen, 2000). Self-control is also important for selection into turning points such as obtaining and maintaining employment, which can alter one’s life trajectory toward or away from crime (Pratt, 2016). As such, self-control is important for explaining crime across the life-course and can influence adulthood risk factors associated with turning points. Self-control warrants further examination in terms of how early risk factors translate into adulthood risk factors.

Gender

Gender is one of the strongest early risk factors for offending. Males around the globe are uniformly more likely to offend than females, spanning from adolescence through adulthood (Harris, 1977; Lanctôt & Le Blanc, 2002). Severe violent offending is almost exclusively committed by males (DeLisi, 2013; DeLisi & Vaughn, 2015), further underscoring the importance of gender as a risk factor for offending. Offenders often have clusters of risk factors, and these clusters may vary by gender. While women share some similar risk factors as men, social bonds and experiencing abuse/trauma have been found to be strong predictors of female offending (Hubbard & Pratt, 2002). This suggests that gender is not only an important early risk factor, but that it may be associated with divergent adulthood risk factors for men and women.

Race/Ethnicity

Another static demographic risk factor is individuals’ race/ethnicity. Race is a proxy measure for underlying structural and cultural risk factors such as poverty, under employment, exposure to crime and neighborhood disorganization, and adherence to the street code, as minorities are more likely to live in areas with concentrated disadvantage (Krivo & Peterson, 1996; Piquero, 2015; Sampson, 2012). Research suggests that Black and Hispanic Americans are disproportionately involved in homicide, rape, robbery, assault, and violence (Steffensmeier et al., 2011); and disproportionate rates of contact with the criminal justice system have been identified for minority children compared to White children (Piquero, 2008), and despite policy efforts to change these rates, they remain prevalent (Leiber & Rodriquez, 2011). While race does not cause crime, the concentrated disadvantage experienced by many children of color and the disproportionate rates of contact seen for minority youth warrants further examination for race/ethnicity as an early risk factor.

Adult Risk Factors

Little research has examined adult risk factors for offending. In general, this literature suggests risk factors in adulthood relate more to specific social factors and environments, as compared to the more developmental risk factors in childhood and adolescence (Eggleston & Laub, 2002). For example, social bonds, neighborhood factors, socioeconomic status, and substance abuse are known risk factors for adult offending. While there is an under-developed body of research examining adult onset offending, a rather large group of offenders begin their criminal careers in adulthood, with 40% to 50% of adult offenders showing no history of juvenile offending (Blumstein et al., 1986). Moreover, the predictors of adult offending are shared among adult offenders, regardless of past delinquency. This suggests that adult risk factors may play an important role in criminal behavior, beyond risk factors in childhood and adolescence. The current study will examine the adult risk factors of weak social bonds, living in neighborhoods with neighborhood disorganization, adherence to the street code, criminal peer association, alcohol and substance abuse, anger/irritability, low education, and poverty.

Weak social bonds

A lack of attachment to family, involvement in prosocial activities, and belief in conventional society are associated with offending among adults (Salvatore & Taniguchi, 2012). These considerations are reflected in theories such as the age-graded theory of informal social control (Sampson & Laub, 1993). Negative life events in adulthood, such as losing a job, can decrease the strength of social bonds, leading to a lower stake in conformity, and increased risk for offending (Sampson & Laub, 1993). The link between weak social bonds and crime illustrates the need to assess this risk factor in adulthood.

Similarly, marital status is a common indicator of social bonds, as individuals who are single are more likely to commit crimes compared to those who are married (Farrington & West, 1995; King et al., 2007; Warr, 1998). In fact, marriage has long been considered a turning point leading to desistance (Laub & Sampson, 2003; Sampson & Laub, 1993). Marriage offers many prosocial benefits including informal social control (Sampson & Laub, 1990), hindering deviant peer associations (Warr, 1998), and constraining offending opportunities by integrating individuals into pro-social settings (Laub & Sampson, 2003). In the absence of those benefits, individuals are at greater liberty to pursue antisocial endeavors. In fact, when men return to single status following a separation or divorce, offending increases (Theobald & Farrington, 2013). As social bonds and marriage, or lack thereof, have important effects on offending in adulthood, this status is important to study.

Neighborhood disorganization

The characteristics of the neighborhood in which an individual resides can be a risk factor for offending (Shaw & McKay, 1942), as neighborhoods with structural disadvantage tend to have higher rates of crime and violence than those with more solidarity, cohesion, and integration between residents in the community (Elliott et al., 1996; Sampson, 2012). Structural characteristics such as poverty, residential instability, and racial/ethnic heterogeneity hamper social integration and the informal social control of residents (Shaw & McKay, 1942; Wilson, 1987). These factors increase the likelihood that an individual in a socially and physically disorganized neighborhood will offend, and warrants examination as an adult risk factor.

Street code adherence

Anderson’s (1999) Code of the Street presented his qualitative research findings on the subcultural norms among a subset of Black young adults within socially isolated and economically disadvantaged urban communities. These norms and behaviors centered around earning and maintaining a status of respect, distrusting authority figures, and using violence as a means of problem-solving and sustaining a reputation for toughness (Anderson, 1999). Contemporary research has shown that individuals who strongly believe in the “street code” are at much higher risks of offending and violent behavior (McNeeley & Wilcox, 2015; Stewart et al., 2006). Adhering to street code beliefs is a strong risk factor for offending, as violence is commonly used to gain respect and handle conflicts between individuals (Anderson, 1999). Limited research has explicitly examined street code beliefs among adults (see Moule & Fox, 2020; Moule et al., 2019; Piquero et al., 2012); nonetheless, given the strong links between beliefs and behaviors generally (e.g., Akers, 1998), street code adherence is assessed as an adult risk factor for crime and violence.

Criminal peer association

Social learning theory asserts that criminal behavior is learned through differential association with offending peers, modeling, and definitions favorable to crime influenced by an individual’s associates (Akers, 2009). For example, the more exposure an individual has to peers committing crime, the more likely they are to also commit crime. Differential reinforcement poses that an individual who receives positive feedback for their behavior is likely to repeat the behavior, while negative feedback for behavior is likely to change it. Thus, individuals learn how to behave based on the feedback and imitation of those around them (Akers, 2009). As criminal peers have consistently been associated with increased risk of offending (Pratt et al., 2010), criminal peer association is examined as an adult risk factor for offending.

Alcohol and substance abuse

The use (and particularly abuse) of alcohol and drugs substantially increases the risk of engaging in other illegal behaviors. For instance, the American Public Health Association (APHA, 2010) found that “alcohol and other drugs are significant factors in all crimes, including 78% percent of violent crimes, 83% of property crimes, and 77% public order, immigration or weapons offenses as well as probation and parole violations” (para. 4). The report also indicates that inmates who have substance abuse problems have a higher chance of becoming incarcerated again, start their criminal career earlier, and come into contact with the criminal justice system more than inmates without substance abuse problems (APHA, 2010). Altogether, substance abuse and usage should be considered when examining risk factors for offending in adulthood.

Anger and irritability

Anger is a prevalent concept in strain theory, as it is indicative of negative emotionality (Agnew, 1985, 1992, 2008; Merton, 1938). Negative emotions mediate the relationship between strain/stress and offending by increasing antisocial coping and criminal propensity and reducing the perceived costs of crime (Agnew, 1992, 2008). Research supports that anger consistently predicts offending (for review see Agnew, 2008, pp. 104–105; Baglivio et al., 2015) and is used as a “triage” assessment for offenders to predict institutional misconduct (DeLisi et al., 2010). Therefore, anger/irritability is examined as a risk factor for offending in adulthood.

Education

Another important adulthood risk factor is education level. Research indicates that between 41% and 68% of incarcerated people do not have a high school education (Coley & Barton, 2006; Harlow, 2003). Incarcerated people are also considerably more likely to score in the lowest literacy levels when compared to the national population (Coley & Barton, 2006), and are generally less educated than the general population (Harlow, 2003). As reaching educational milestones (particularly college educations) are considered a significant turning point in the offending career (Arum & Beattie, 1999), having less education is considered a risk factor for offending in adulthood.

Poverty

A considerable amount of research has found a link between poverty and crime (Bjerk, 2007; Brown, 1984; Brownfield, 1986). However, less research has examined poverty as one factor in a constellation of other risk factors. Based upon previous literature, poverty is expected to exacerbate other risk factors, amplifying their effects, particularly in adulthood when responsibilities are highest (Hay & Forrest, 2009). Furthermore, poverty has been seen to persist across the life-course and spans generations within a family (Bird, 2013). As such, poverty has implications for risk of offending across the life-course.

Current Study

The purpose of this study is to expand on previous research by assessing latent profiles of early risk factors versus adult risk factors among adults incarcerated in a county jail in Florida, and examine stability of membership in risk profiles across the early and adulthood models. This is achieved by analyzing risk factors for offending that represent early and adulthood risk factors (e.g., ACEs, head injuries, self-control, neighborhood disorganization, criminal peer association, anger/irritability, poverty, and marital status) to assess heterogeneity in risk among those in the offending sample, followed by transitition analyses to assess how stable risk level is over time. This research is needed to better understand the distinct constellations of risk factors that co-occur during the early years and in adulthood, and the continuity of risk profiles across measures of early and adult risk factors.

Methodology

Data

Data for the study were collected through an interdisciplinary research effort called the Psychological Assessment of Risk and Criminality (PARC) project, in collaboration with a county detention facility in central Florida. This county has over 525,000 residents and recently experienced a 40% increase in violent crime and opioid-related offenses, leading to overcrowding in the jail. This county also has a high recidivism rate, as 42% of those released from the jail in 2017 were re-incarcerated within one year (Fox et al., 2019). Therefore, the PARC project was developed to evaluate the specific risks factors and needs among jail inmates, improve risk assessment and housing classification decisions, and identify and treat inmates with higher risks and needs while in jail, and post-release. To achieve these goals, adults incarcerated in the jail were screened, assessed, and evaluated using empirically-validated assessment tools to obtain data on their criminogenic, mental health, and structural risk factors and needs.

Sample

Data were collected on a total of 735 adults incarcerated in the jail during booking, with approximately 60 people screened monthly. Study participants were predominantly male (67.9%; n = 499), although a sizable proportion were female (32.1%, n = 236). The average age of participants was 36.6, and ranged from 18 to 73 years old. Similar to other carceral populations, most participants reported having no college education (62.1%), an annual income of less than $30,000 per year (61.0%), and were not currently married (82.9%). Unlike other jail populations, most people in this sample identified as White (71.6%; n = 526), while 28.4% identify as Black, Hispanic, or another race/ethnicity. Descriptive statistics for all measures collected for this study are presented in Table 1 ¹.

Table 1.
Risk Factor Prevalence for Study Participants.

Full Sample Under Age 25 Age 25 and Over

% M (SD) % M (SD) % M (SD) χ²/F

Early Risk Factors

Adverse Childhood Experiences 4.33 (2.75) 4.88 (2.78) 4.19 (2.72) 6.3^*

Head Injuries 0.22 (0.41) 0.30 (0.46) 0.21 (0.41) 0.8^*

Low Self-Control 2.03 (0.72) 2.12 (0.68) 2.01 (0.73) 1.2

Male 67.9 75.8 66.1 4.4^*

White 71.6 55.6 74.8 18.6^***

Adult Risk Factors

Weak Social Bonds 1.94 (0.78) 1.94 (0.82) 1.95 (0.78) 0.0

Unmarried 82.9 89.9 81.5 4.9^*

Neighborhood Disorganization 2.49 (0.58) 2.43 (0.56) 2.49 (0.58) 0.4

Street Code Adherence 1.92 (0.61) 2.07 (0.59) 1.89 (0.62) 3.1^

Criminal Peers 2.19 (0.96) 2.34 (1.00) 2.16 (0.95) 3.3

Alcohol and Drug Use 4.92 (1.70) 5.11 (1.70) 4.91 (1.69) 1.9

Anger/Irritability 3.03 (2.69) 3.69 (2.77) 2.90 (2.69) 62.4^*

No College Education 62.1 81.8 58.1 24.1^***

Low Annual Income 61.0 63.9 60.4 0.5

Note: Full sample n = 735, Under age 25 sub-sample n = 124, Age 25 and over sub-sample n = 611. M = mean, SD = standard deviation.

Early Risk Factor Measures

Adverse childhood experiences

ACEs were measured using the ACE questionnaire (Anda et al., 2010; Felitti et al., 1998), which asked participants if, during the first 18 years of life, they had been physically, emotionally or sexually abused, physically or emotionally neglected, witnessed domestic violence, lived with a family member who was ever sent to jail or prison, a family member who had a mental disorder or was suicidal, a family member who was a problem drinker, used street drugs, or misused prescription drugs, or had parents who separated or divorced. ACEs were operationalized as the total score ranging from 0 to 10, indicating the presence of total adverse childhood experiences.

Early head injuries

Participants were asked whether they had ever had an injury to the head that caused them to be knocked out (e.g. lack of consciousness) and/or dazed and confused for any period of time before age 18. Examples included a head injury resulting from a fall while sober or under the influence, road traffic accident, sports injury, fight, or other. If the participant indicated they had experienced an injury to the head, they were then asked to recall the age of the event. Head injuries that occurred before age 18 were totaled, and operationalized as a count measure.

Low self-control

Low self-control was measured using three items from the Grasmick et al. (1993) scale which asked participants whether they: (1) acted on the spur of the moment without stopping to think, (2) did things that brought them pleasure here and now, even at the cost of some future goal, and (3) were more concerned with what happened to them in the short run than in the long run. Potential responses were on a 4-point Likert scale ranging from strongly disagree “1” to strongly agree “4,” and were summed and averaged to create a total score (α = 0.708).

Race and gender

All participants completed a demographic questionnaire, which included items on two early demographic correlates of crime: gender and race/ethnicity. This study accounts for these static risk factors, with gender operationalized as male or female, and race/ethnicity operationalized as White (the primary race/ethnicity in this sample) or non-White, according to the participant’s self-reported identification.

Adult Risk Factor Measures

Weak social bonds

Social bonds were measured using two items from the Sampson and Laub (1993) social bonds scale, which asked participants whether: (1) their friends were a very important part of their life, and (2) whether they knew people who had been their close friends for years. Potential responses were on a 4-point Likert scale ranging from strongly disagree “1” to strongly agree “4.” These items were combined and averaged to obtain the mean summed scale (α = 0.668).

Marital status

Marital status was operationalized as currently married or not currently married (i.e., single, in a relationship, divorced, separated, or widowed), based on participant self-report on the demographic questionnaire.

Adherence to the street code

Adherence to the street code was measured using four items from the Stewart and Simons (2010) scale, to include: (1) “When someone disrespects you, it is important that you use physical force or aggression to teach him or her not to disrespect you.” (2) “If someone uses violence against you, it is important that you use violence against him or her to get even.” (3) “People will take advantage of you if you don’t let them know how tough you are.” and (4) “People tend to respect a person who is tough and aggressive.” Similar to the previous measures, potential responses were on a 4-point Likert scale ranging from strongly disagree “1” to strongly agree “4,” with total scores on these items used to obtain the mean summed scale (α = 0.716).

Neighborhood disorganization

Neighborhood disorganization was measured using three items from the Fox et al. (2010) scale that asked the participant whether in their neighborhood: (1) it was safe to walk outside after dark, (2) it was easy to know who belonged and who was a stranger, and (3) whether people would call the police right away if they thought a crime was being committed. Potential responses were on a 4-point Likert scale ranging from strongly disagree “1” to strongly agree “4.” The scores of these items were used to obtain the mean summed scale (α = 0.248)².

Criminal peer association

Criminal peer association was measured using two items from the Burgess and Akers (1966) scale that asked the participant whether (1) they had at least one close friend who had done something they knew to be illegal in the past six months, and (2) they had at least one close friend suggest they do something they knew to be illegal in the past 6 months. These items were also measured using a 4-point Likert scale ranging from strongly disagree “1” to strongly agree “4.” The mean summed scale was calculated and used for this measure (α = 0.718).

Alcohol and substance abuse

Alcohol and substance abuse were measured using a subscale in the MAYSI-2 (Grisso & Barnum, 2006), adapted for use in adult forensic populations (Fox et al., 2019). Eight items were used to calculate a total risk score on the alcohol/substance abuse scale, which asked whether the participant, in the past six months, (1) had done anything they wish they hadn’t while drunk or high, (2) had parents or friends who thought they drank too much, (3) had gotten in trouble while drunk or high, (4) if they were in trouble while drunk or high, if it was due to fighting, (5) had used alcohol or drugs to help them feel better, (6) had been drunk or high at work, (7) had used alcohol and drugs at the same time, and/or (8) had been so drunk or high they couldn’t remember what happened. Potential responses were “yes” or “no,” and a total summed score was created (α = 0.841).

Anger/Irritability

Anger and irritability were also measured using a subscale in the MAYSI-2 (Grisso & Barnum, 2006), adapted for adult forensic populations (Fox et al., 2019). Nine items were asked and summed to calculate a total risk score on the anger/irritability scale. Those items included whether, in the past 6 months, the participant had (1) lost their temper easily or had a “short fuse,” (2) been easily upset, (3) thought a lot about getting back at someone they were angry at, (4) been really jumpy or hyper, (5) had too many bad moods, (6) felt angry a lot, (7) got frustrated a lot, (8) stayed mad for a long time when they were mad, and (9) had hurt or broken something on purpose because they were mad. Participants responded “yes” or “no” to each item, with responses summed to create a total score on this sub-scale (α = 0.827).

Education level

The demographic questionnaire included a question about educational achievement, which was operationalized dichotomously, with participants self-reporting to have some college education (at least one class), or no college education.

Poverty

The demographic questionnaire included an item on annual household income. Poverty was operationalized as the household income self-reported by the participant, and dichotomized to indicate income less than $30,000 or $30,000 or above, in line with federal guidelines (U.S. Department of Health & Human Services, 2019).

Analytical Plan

Latent Transition Analysis

To empirically examine potential heterogeneity in the prevalence and types of risk factors present among the incarcerated sample during the early and adult time periods, and the continuity in risk profiles depending on age reference of the risk factors, a series of models were estimated using an approach called Latent Transition Analysis (LTA). LTA is a temporal extension of Latent Class Analysis (LCA; Muthén & Muthén, 2000), as it is able to estimate transition probabilities for individuals in latent classes at multiple periods (Lanza, Patrick, & Maggs, 2010).

LTA is a “person-focused” statistical classification technique designed to detect latent patterns within individuals based upon two or more covariates, using a three step process (Lanza et al., 2010). First, discrete LCAs are estimated at multiple data points, in order to estimate the fewest number of groups (i.e. latent profiles) where individuals in the group are similar to each other, but qualitatively different from individuals in other profiles (Muthén & Muthén, 2000) at each period (Lanza et al., 2010). LCA has increasingly been used to typify populations with heterogeneous features, such as offenders and juvenile delinquents (e.g., Fox & Delisi, 2018; Fox & Farrington, 2012; Reid & Loughran, 2019; Vaughn et al., 2008, 2009, 2011). LCA is also popular due to the objective nature of the analysis (Fox & Farrington, 2012), and the advantages LCA presents compared to related statistical classification techniques such as factor, k-means, and cluster analyses which require normality and linearity of the data, assumptions frequently violated in offending populations, and the fact that LCA can accommodate ordinal, nominal, and continuous levels of data, while many other classification techniques cannot.

LCA models in this study were estimated using Latent Gold v.5.1 software (Vermunt & Magidson, 2005). Potential class solutions were evaluated with three goodness-of-fit criteria: the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), and Consistent Akaike Information Criterion (CAIC) (Uebersax, 2009), which quantify differences in an estimated model and the true observations. In LCA, lower fit criteria values indicate better model fit. The solution with the majority of fit criteria in its favor is selected. This step indicates the number of sub-profiles that best fit the data at each time point in the analysis.

Second, conditional item probabilities, which reflect the correspondence between the indicators included in LCA and the latent class membership for all individuals in the sample, are examined for the early and adulthood models. These probabilities, which are similar to factor loadings in factor analysis, indicate the underlying heterogeneity and composition of the resultant classes (Lanza et al., 2010).

Finally, a matrix of transition probabilities is calculated in order to indicate the transition probabilities between latent profiles, in order to illustrate stability or change in membership of latent profiles according to the age reference of the risk factors (see Lanza & Bray, 2010; Lanza & Collins, 2008; Lanza et al., 2003, 2007). In other words, these probabilities reflect the proportion of individuals who belong in each latent profile for each model at each reference period (Lanza et al., 2010). It should be noted that while LTA is most commonly used with prospective longitudinal data, the risk factors in this analysis are modeled at distinct time periods in childhood/adolescence (t) and adulthood (t+1) using retrospective longitudinal data.

There are multiple methods available to further quantify significance and effect size of these transition probabilities. In this study we present a new strategy frequently used to measure specialization and versatility in DLC research using the Forward Specialization Coefficient (FSC; Farrington, 1986) and Adjusted Standardized Residual (ASR; Farrington et al., 1988) analyses. The FSC is one of the favored measures of specialization in behavior due to its statistical rigor and intuitive interpretation (Paternoster et al., 1998; Sullivan et al., 2009). The FSC is not influenced by sample size or distribution of cases in transition matrices, meaning that rarer profiles will not bias the analysis in any way (Farrington et al., 1988). In a transition matrix of resultant LCA profiles at t and t+1, the FSC is calculated using the formula:

$FSC = \frac{O - E}{R - E}$

where O = observed cell count, E = expected number in the cell by chance, and R = row total (Farrington, 1986; Farrington et al., 1988). By applying the FSC to each of the diagonal cells in a transition matrix, the level of consistency in risk profiles may be evaluated (Paternoster et al., 1998). The FSC ranges from 0, indicating perfect inconsistency in risk profiles (as the observed change is equal to chance), to 1, indicating perfect forward specialization in risk profiles (as membership in a certain risk profile level, such as high risk, at the early years is followed by the same risk level profile in adulthood) (Farrington et al., 1988; Stander et al., 1989). The FSC can range up to −1 if there was perfect negative specialization (i.e. a tendency for risk level profile in the early years to not be followed by the same risk level at adulthood) (Stander et al., 1989).

Next, the ASR is used to test the statistical significance of the FSC by indicating whether the observed value is significantly above or below chance expectation (Farrington et al., 1988). The ASR is computed using the formula:

$ASR = \frac{O - E}{\sqrt{E} \times \sqrt{\{[1 - (\frac{R}{T})] \times [1 - (\frac{C}{T})]\}}}$

where O = observed cell count, E = expected number in the cell by chance, R = row total, C = column total, T = grand total, * indicates multiplication, and $E = \frac{R \times C}{T}$ .

Results

To conduct the first step of the LTA, LCA models based upon the early and adult risk factors were estimated for all 735 incarcerated individuals using one to seven potential class solutions (see Table 2). Results of the early risk factor LCA indicated that a three-class solution was favored by all goodness-of-fit criteria. Results of the adult risk factor model indicate that a majority of the goodness-of-fit measures also favored a three-class solution.

Table 2.
Fit indices for all Potential Class Solutions Using LCA.

No. of classes AIC(LL) BIC(LL) CAIC(LL) LL npar Class error EntropyR2

Early Risk Factor Model

1 8,098.3 8,121.1 8,126.1 −4,044.1 5 .000 1.00

2 7,870.4 7,920.5 7,931.5 −3,924.2 11 .095 .614

3 7,825.6 7,903.1 7,920.1 −3,895.8 17 .010 .751

4 7,826.5 7,931.3 7,954.3 −3,890.2 23 .117 .738

5 7,833.5 7,965.6 7,994.6 −3,887.7 29 .183 .691

6 7,843.0 8,002.5 8,037.5 −3,886.5 35 .186 .681

7 7,854.2 8,041.0 8,082.0 −3,886.1 41 .225 .653

Adult Risk Factor Model

1 16,226.8 16,268.8 16,277.8 −8,104.4 9 .000 1.00

2 15,643.4 15,732.1 15,793.7 −7,802.7 19 .104 .666

3 15,629.4 15,764.7 15,751.1 −7,785.7 29 .142 .642

4 15,640.1 15,822.1 15,861.1 −7,781.1 39 .192 .581

5 15,654.8 15,883.4 15,932.4 −7,778.4 49 .265 .510

6 15,668.9 15,944.2 16,003.2 −7,775.4 59 .226 .620

7 15,687.0 16,008.9 16,077.9 −7,774.5 69 .280 .584

Note: Bold represents best class solution for the data. AIC = Akaike Information Criterion, BIC = Bayesian Information Criterion, CAIC = Consistent Akaike Information Criterion, LL = log likelihood, npar = number of parameters, DI = dissimilarity index.

Early Risk Profiles

Posterior probabilities for the early risk factors model present the prevalence rate of each risk factor among individuals within the three resultant profiles (see Table 3). Characteristics of the three resultant profiles differ objectively in terms of overall risk level. The first profile, which comprised 53.4% of the sample, had the highest mean score for childhood head injuries (M = 0.35), highest levels of low self-control (M = 2.12), were almost entirely male (98.9%), and had the highest prevalence of non-White members (34.9%). They experienced about five ACEs each, on average. Due to the higher scores on most risk factors, this class is labeled the Early High Risk (EHR) profile.

Table 3.
Posterior Probabilities of Early Risk Factors Across Resultant Classes.

Class 1 High Risk (%) Class 2 Medium Risk (%) Class 3 Low Risk (%)

Adverse Childhood Experiences Mean score 4.99 5.82 1.03

Head Injuries Mean score 0.35 0.13 0.11

Low Self-Control Mean score 2.12 2.01 1.83

Gender Female 1.1 98.0 33.9

Male 98.9 2.0 66.1

Race/Ethnicity Non-White 34.9 14.9 29.9

White 65.1 85.1 70.1

% of Sample 53.4 24.0 22.6

Note: n = 735. Adverse Childhood Experiences scale range: 1-10; Low self-control scale range: 1-4.

About a quarter of the sample (24.0%) were in the second latent class, which was characterized by moderate levels of early risk factors called the Early Medium Risk (EMR) profile. The EMR profile showed the highest average ACE score (M = 5.82) and was almost exclusively female (98.0%) and White (85.1%). They also showed moderate to high levels of low self-control (M = 2.01) and had an average of 0.13 childhood head injuries.

Finally, the third group exhibited generally low levels of early risk factors, and were labeled the Early Low Risk (ELR) profile. They experienced the fewest ACEs (M = 1.03) and head injuries (M = 0.11), and their low self-control scores were the lowest (M = 1.83) of any latent profile group. The ELR profile, which was about two-thirds male and 33.9% female, were predominately White (70.1%), and comprised 22.6% of the sample.

Adult Risk Profiles

Table 4 presents the posterior probabilities for the adult risk factors model, indicative of the prevalence of each risk factor for members of the three resultant profiles. Again, the nature of the three resultant profiles clearly differ in levels of overall risk. The first profile, which comprised 54.5% of the sample, had the highest mean scores for weak social bonds (M = 1.99), neighborhood disorganization (M = 2.54), street code adherence (M = 2.05), association with criminal peers (M = 2.44), alcohol and drug abuse (M = 5.37), and over five times higher mean scores on anger/irritability (M = 4.95) compared to the remaining classes. Nearly two-thirds had no college education (65.4%) and an annual income under $30,000 per year (63.4%). About 86% of the members were unmarried. Due to the overall risk associated with this subtype, this group is labeled the Adult High Risk (AHR) profile.

Table 4.
Posterior Probabilities of Adult Risk Factors Across Resultant Classes.

Class 1 High Risk (%) Class 2 Medium Risk (%) Class 3 Low Risk (%)

Weak Social Bonds Mean score 1.99 1.98 1.78

Unmarried Married 14.2 11.4 30.7

Unmarried 85.8 88.6 69.3

Neighborhood Disorganization Mean score 2.54 2.44 2.42

Street Code Adherence Mean score 2.05 1.83 1.69

Criminal Peers Mean score 2.44 1.92 1.86

Alcohol and Drug Abuse Mean score 5.37 4.11 3.98

Anger/Irritability Mean score 4.95 0.86 0.59

Education College education 34.6 28.3 57.6

No college education 65.4 71.8 42.4

Low Annual Income $30,000 or higher 36.6 3.9 86.7

Under $30,000 63.4 96.1 13.3

% of Sample 54.5 24.5 21.0

Note: n = 704. Weak social bonds, neighborhood disorganization, street code adherence and criminal peers range: 1–4. Alcohol and drug abuse range: 1–8. Anger/Irritability range: 0–9.

The second group comprised 24.5% of the sample, and was labeled the Adult Medium Risk (AMR) profile. Members of the AMR profile scored in the moderate range across adult risk factors including weak social bonds (M = 1.98), neighborhood disorganization (M = 2.44), street code adherence (M = 1.83), association with criminal peers (M = 1.92), alcohol and drug abuse (M = 4.11), and anger/irritability (M = 0.86). The vast majority of the AMR profile were low income (96.1%), unmarried (88.6%), and had no college education (71.8%).

The final adult risk factor group, characterized by the lowest prevalence of risk factors, comprised 21.0% of the sample. Specifically, members of this group had the lowest mean scores for risk factors including weak social bonds (M = 1.78), neighborhood disorganization (M = 2.42), street code adherence (M = 1.69), criminal peer association (M = 1.86), alcohol and drug abuse (M = 3.98), and anger/irritability (M = 0.59). A majority had a college education (57.6%), had a higher annual income (86.7%), and the highest proportion of married participants (30.7%). Due to the low prevalence in risk factors, this group was labeled the Adult Low Risk (ALR) profile.

Continuity and Change in Risk Profiles

In the final step of the LTA, continuity in risk profiles across the early and adult risk factors models are analyzed using transition probabilities, FSC, and ASR values in a 3 × 3 transition matrix (see Table 5). Transition probabilities reflect the probability of transitioning from a particular risk profile for the early risk model to the adult risk model, and illustrate the likelihood and associated significance of change and continuity in risk across these two models (see Jackson et al., 2006; Collins, 2002; Lanza & Collins, 2002; Lanza et al., 2010). Perfect continuity in risk profile membership is indicated by significant positive effects for the diagonal cells in the transition matrix (e.g. early high risk/adult high risk). Change in risk is indicated by significant positive effects for any non-diagonal cells (e.g. early low risk/adult medium risk).

Table 5.
Transition Probabilities from Early to Adult Risk Profiles.

Adult Risk Profiles

Early Risk Profiles High Risk Medium Risk Low Risk Total

High Risk 60.2% 20.7% 19.1%

ASR = 4.3 ASR = −4.3 ASR = −0.6

FSC = 0.16 FSC = −0.09 FSC = −0.01 100%

O = 230 O = 79 O = 73 382

E = 201.9 E = 104.2 E = 76.0

Medium Risk 59.5% 27.0% 13.5%

ASR = 1.9* ASR = −0.04 ASR = −2.3*

FSC = 0.14 FSC = 0.10 FSC = −0.08 100%

O = 97 O = 44 O = 22 163

E = 86.1 E = 44.5 E = 32.4

Low Risk 28.3% 43.4% 28.3%

ASR = −7.0* ASR = 5.2 ASR = 3.0*

FSC = −0.52 FSC = 0.22 FSC = 0.10 100%

O = 45 O = 69 O = 45 159

E = 84.0 E = 43.4 E = 31.6

Note: n = 704; ASR = adjusted standardized residual, FSC = forward specialization coefficient, O = observed, E = expected. p < .05; p < .01; p < .001.

Results of the LTA indicate both continuity and change in the risk profile memberships across the early and adult risk factor models. Strong and significant levels of continuity were found for those in the high and low risk profiles across the early and adult models. This was especially the case for the high risk profile, as a majority of the members of the EHR profile were also members of the AHR profile (60.2%, FSC = 0.16). This continuity in high risk was statistically significant (ASR = 4.3, p < .01), as more participants were in the early and adult high risk profiles than expected by chance. Similarly, significant risk continuity was observed for the ELR to ALR profiles (ASR = 3.0, p < .01), so that more individuals were members of same risk level profiles than statistically expected. However, the percentage of participants (28.3%, FSC = 0.10) who remained in the low risk profile across early to adult models was fairly low.

There was also notable discontinuity observed in the transition of risk profiles across the early and adult models, with risk levels increasing from the early to the adult risk factors models. Substantial change in risk profile membership was observed for the ELR profile, with 43.4% of the group transitioning to the AMR profile (FSC = 0.22). This transition from early low risk to adult medium risk profiles occurred significantly more often than statistically expected (ASR = 5.2, p < .01). A sizable and significantly larger proportion of individuals in the EMR profile were later members of the AHR profile (59.5%, FSC = 0.14) than expected by chance (ASR = 1.9, p < .05).

Results of the LTA also indicate that there are certain transitions in risk profiles that are significantly less likely to occur than expected. For instance, members of the EHR profile are unlikely to later be members of the AMR profile (ASR = −4.3, p < .01), EMR members are significantly less likely to be in the ALR profile (ASR = −2.3, p < .05), and the ELR members are significantly unlikely to later be members of the AHR profile (ASR = −7.0, p < .001).

Risk Profile Invariance by Age

To further assess the risk profiles, additional LTA models were estimated using participant age as a grouping variable (under age 25 vs. age 25 and over) so that measurement invariance for risk profiles by younger and older inmates could be assessed. This allows us to control for issues of retrospective reporting of childhood risk factors, a problem that would presumably increase with age. To do this, a model with item-response probabilities freely estimated within each age group was compared to a model where these probabilities were constrained to be equal across ages. A likelihood-ratio (G²) test was conducted to test the hypothesis that measurement invariance holds across age groups (see Magidson & Vermunt, 2004). Results suggest that the resultant latent classes do not significantly vary by age group. Although fewer classes were found for the younger inmates (low and high risk), the patterns of risk and continuity were consistent across age groups in the sample (results available in online supplemental material).

Discussion

Criminologists have long expressed an interest in understanding the influence that early life experiences have on antisocial behavior and related negative life outcomes. Indeed, DLC theories focus on within-individual changes in risk and behavior over time, spanning from early childhood into later life (Farrington, 2005b; Moffitt, 1993; Sampson & Laub, 2005). However, far less research has examined the heterogeneity in risk profiles as assessed by measures of early and adult risk factors, and the continuity or change in risk factor profiles across models of early versus adult risk levels. These considerations were the basis for the current study. Using a large sample of incarcerated individuals in Florida, the current study sought to understand the diversity of risk profiles associated with offending, and the relationship between risk profile membership when assessing early versus adult risk factors. Based on results from a Latent Transition Analysis, three findings warrant broader discussion.

First, notable heterogeneity in risk profiles was found across early and adult life risk factors models, even within the offender sample. Specifically, risk factors appear to co-occur in one of three patterns: low risk (i.e. low severity or few risk factors), medium risk (i.e. moderate severity or several risk factors), and high risk (i.e. high severity and nearly all risk factors present). This indicates that there is substantial latent variation in overall risk for individuals within this population, despite the fact that this sample was made up of arrestees who experienced substantial disadvantage and came from high risk contexts. While the high risk profiles were most common in both early and adult models, with over half the sample endorsing risk factors that qualified as high risk across both models, about a quarter of the sample qualified as medium risk in early years and adulthood, and just about 20% are classified as low risk for offending in either model.

These findings have notable implications for theory and practice. Given the varied level of risk and risk factors found within this offender sample, we expect that different mechanisms are likely at play when it comes to the underlying causes of the offending behavior since all the participants have offended at one time or another. It is unlikely that one criminological theory fully explains heterogeneity identified for early and adulthood models. Instead, different theories can best explain each risk level. For instance, Moffitt’s (1993) dual taxonomy of offenders that are adolescence-limited (relatively low risk in early years) and life-course persistent (high levels of risk, particularly neuropsychological deficits, over the lifespan) may be used to explain two of the risk profiles (EHR and LHR, respectively). However, strain (Agnew, 1985) and age-graded informal social control (Sampson & Laub, 1993) theories can better explain the patterns of risk factors relating to anger/irritability and alcohol and drug abuse, lack of education and marriage, poverty, and weak social bonds observed in various adult risk profiles (AHR and AMR).

Second, these results support assertions that risk factors “do not operate in isolation” (Kendziora & Osher, 2004, p. 182), and that risk is apparent and co-occurs across early and adult assessments of risk factors, particularly among the highest risk profiles. Related to the previous point, most offenders displayed risk factors that spanned multiple theoretical domains, illustrating the benefits of utilizing cross-disciplinary and multi-theoretical models, such as those proposed in the DLC literature (see Loeber & Farrington, 1998). Moreover, it also suggests that any intervention or prevention strategy aimed at reducing risk within these risk groups must be multi-faceted, as there is no single set of risk factors exhibited by all offenders. Instead, it appears that more tailored rehabilitation and prevention strategies aimed at neutralizing the varied constellations of risk factors present in each risk group would be most effective, in line with the risk-needs-responsivity model (Bonta & Andrews, 2007).

Third, the results indicate that the transition from early to adult risk profiles is marked by both continuity and change. The biggest continuity was present for those who were high risk according to measures referencing childhood, as the majority of these were also categorized into the high risk profile according to the adult measures. The findings for the highest risk profiles have theoretical implications, as the relative stability at this level of risk is supportive of both cumulative and interactional continuity (Caspi et al., 1987, 1989). It also suggests that the highest risk may be largely static and neuropsychological in nature, and likely identifiable in early years (Farrington, 2003b; Moffitt, 1993). The fact that a measure of head injury was part of this group of early risk factors is consistent with this explanation.

A similar, although much more modest, pattern of continuity occurred for those in the early low risk profile, who were significantly more likely to stay low risk according to the adult models. However, the early low risk profile also experienced substantial discontinuity, as a large number (over 40%) qualified as medium risk according to measures of adult risk factors. Importantly, there was no circumstance where individuals were more likely to fall in a lower risk profile between childhood/adolescence and adulthood models. Unfortunately, there was little evidence to suggests that turning points, such as gainful employment or marriage, helped to disrupt the level of risk for a substantial proportion of the sample, as the discontinuity in risk patterns over the life-course were always upward in nature. This finding suggests that many low and medium risk offenders experienced new external risk factors in adulthood, increasing their risk profile from the early to the adult models. Taken together, these points have multiple implications for policy and prevention, which are discussed below.

Policy Implications

A clear policy goal is to implement prevention and intervention strategies aimed at reducing risk factors from childhood and adolescence to adulthood, as only increases in risk were observed in our sample. By disrupting the continuity or escalation in risk over time, a decrease in severity and frequency of offending is likely to follow. Use of interventions shown to successfully address early risk factors such as cognitive behavioral therapy (CBT), functional family therapy (FFT), multi-systemic therapy (MST), dialectical behavior therapy (DBT), and trauma therapy may help to mitigate issues that contribute to deleterious risk in adulthood.

However, the use of more specialized interventions designed for specific risk groups may be most effective in reducing risk and disrupting the transition of risk over the lifespan. For instance, the early moderate risk (EMR) profile consisted largely of females scoring high on ACEs (with an average of over 5 ACEs per person), suggesting that the impact of trauma and adversity may be particularly impactful for women’s offending patterns. This is consistent with gendered pathways perspectives (Javdani et al., 2011) and continuing recognition for the need for gender-specific risk assessments (Daly, 1992, 1994; Wattanaporn & Holtfreter, 2014). Similarly, the adult medium risk (AMR) profile, which had lower rates of college education, lower income, and lowest rates of marriage compared to the other risk groups, appear to have more limited stakes in conformity and could benefit from approaches that re-enfranchises them into society, such as occupational rehabilitation and stable living circumstances.

Conversely, the early low risk (ELR) profile is likely associated with low level opportunistic offending, which is more challenging to prevent, but also more limited in severity and duration than other offending types (Moffitt, 1993). This behavior is more normative in adolescence and early adulthood, and not dispositional in nature, as these results largely suggest. However, given the relatively low base rate in stability for low risk profile membership across the childhood/adolescence and adulthood risk factors models (28% of ELR offenders fit this pathway), it appears that events such as snares (Moffitt, 1993) may be responsible for entrapping individuals who are otherwise low to moderate risk, which may have led to the their incarceration, despite their otherwise low risk profiles. This suggests that structural factors and criminal justice policies that remove obstacles to reentry may benefit those who appear low risk, to prevent the escalation of risk (43% of ELR transitioned from low to medium risk over time).

Finally, the high risk offenders have both the most identifiable and severe risk factors, making interventions for this group much more critical. For instance, these individuals reported substantial trauma and head injuries during childhood, and low self-control, and they also reported high anger/irritability, alcohol and drug abuse, criminal peer association, and street code adherence in their adult years. This illustrates the complex presentations of risk and needs, and the multifaceted nature of the interventions needed to address problems in these high risk groups. Moreover, these interventions should begin early to prevent the continuity of risk across time. Implementing such interventions is particularly important considering the overlap in risk from early to adult years among these offenders. In other words, without intervention, there is a high likelihood that these high risk youth will transition to become high risk adults, committing crime, and exhibiting other negative outcomes that impact themselves and society at large.

Of course, the current study is not without limitations. First, the biggest limitation was that our study did not draw upon prospective longitudinal data. This presents various problems. The study involved asking respondents about things that happened many years in the past, and respondents may not always recall these events clearly (see Hänninen & Soininen, 2012). Second, although we asked respondents about events that occurred in the past and link these events with contemporary factors, we cannot formally establish causal ordering between all of these constructs. Relatedly, much of the DLC literature emphasizes within-individual change in a number of constructs, including many of the experiences and traits examined in the current study. Owing to these considerations, exploring the dynamics of change in these constructs using prospective longitudinal data should be a priority of future research. However, part of the goal of this study was to examine whether risk profiles would differ cross-sectionally when using measures meant to capture early risk versus adult risk factors. These limitations notwithstanding, the current study also brought to bear a sizable sample of jail inmates and a rich set of variables to better understand risk profiles.

In the end, the consistency in risk factors across measures of early risks/dispositions and measures of adulthood risks/dispositions found among this sample of incarcerated individuals indicate the need for early interventions, particularly among high-risk populations. Clearly, criminal justice interventions, for many at-risk people, come too late. Identifying and addressing risk factors at an early age, through comprehensive and multi-faceted efforts across social institutions and agencies, is necessary and worthwhile. As our findings show, there is notable continuity in individuals at high risk for offending in childhood/adolescence and adulthood. Disrupting this continuity therefore requires early intervention, which should produce long-term benefits for individuals and public safety. Furthermore, the movement of low and medium early risk individuals into medium and high adult risk categories similarly highlights the possibility that interventions and treatments in adolescence may prevent the accumulation of additional risk factors and negative behaviors in adulthood. Overall, the findings from the current study provide insight into an underdeveloped area of risk factor literature and we encourage future research to continue examining how risk factors change over the life-course and how interventions may work to negate or inhibit the deleterious consequences associated with these risk factors.

Supplemental Material

YVJJ939648_supplemental_material - Heterogeneity in the Continuity and Change of Early and Adult Risk Factor Profiles of Incarcerated Individuals: A Latent Transition Analysis

YVJJ939648_supplemental_material for Heterogeneity in the Continuity and Change of Early and Adult Risk Factor Profiles of Incarcerated Individuals: A Latent Transition Analysis by Bryanna Fox, Kelly Kortright, Lexi Gill, Daniela Oramas Mora, Richard K. Moule and Edelyn Verona in Youth Violence and Juvenile Justice

	Full Sample	Under Age 25	Age 25 and Over
Early Risk Factors
Adverse Childhood Experiences		4.33 (2.75)		4.88 (2.78)		4.19 (2.72)	6.3^*
Head Injuries		0.22 (0.41)		0.30 (0.46)		0.21 (0.41)	0.8^*
Low Self-Control		2.03 (0.72)		2.12 (0.68)		2.01 (0.73)	1.2
Male	67.9		75.8		66.1		4.4^*
White	71.6		55.6		74.8		18.6^***
Adult Risk Factors
Weak Social Bonds		1.94 (0.78)		1.94 (0.82)		1.95 (0.78)	0.0
Unmarried	82.9		89.9		81.5		4.9^*
Neighborhood Disorganization		2.49 (0.58)		2.43 (0.56)		2.49 (0.58)	0.4
Street Code Adherence		1.92 (0.61)		2.07 (0.59)		1.89 (0.62)	3.1^**
Criminal Peers		2.19 (0.96)		2.34 (1.00)		2.16 (0.95)	3.3
Alcohol and Drug Use		4.92 (1.70)		5.11 (1.70)		4.91 (1.69)	1.9
Anger/Irritability		3.03 (2.69)		3.69 (2.77)		2.90 (2.69)	62.4^***
No College Education	62.1		81.8		58.1		24.1^***
Low Annual Income	61.0		63.9		60.4		0.5

No. of classes	AIC(LL)	BIC(LL)	CAIC(LL)	LL	npar	Class error	EntropyR2
Early Risk Factor Model
1	8,098.3	8,121.1	8,126.1	−4,044.1	5	.000	1.00
2	7,870.4	7,920.5	7,931.5	−3,924.2	11	.095	.614
3	7,825.6	7,903.1	7,920.1	−3,895.8	17	.010	.751
4	7,826.5	7,931.3	7,954.3	−3,890.2	23	.117	.738
5	7,833.5	7,965.6	7,994.6	−3,887.7	29	.183	.691
6	7,843.0	8,002.5	8,037.5	−3,886.5	35	.186	.681
7	7,854.2	8,041.0	8,082.0	−3,886.1	41	.225	.653
Adult Risk Factor Model
1	16,226.8	16,268.8	16,277.8	−8,104.4	9	.000	1.00
2	15,643.4	15,732.1	15,793.7	−7,802.7	19	.104	.666
3	15,629.4	15,764.7	15,751.1	−7,785.7	29	.142	.642
4	15,640.1	15,822.1	15,861.1	−7,781.1	39	.192	.581
5	15,654.8	15,883.4	15,932.4	−7,778.4	49	.265	.510
6	15,668.9	15,944.2	16,003.2	−7,775.4	59	.226	.620
7	15,687.0	16,008.9	16,077.9	−7,774.5	69	.280	.584

		Class 1 High Risk (%)	Class 2 Medium Risk (%)	Class 3 Low Risk (%)
Adverse Childhood Experiences	Mean score	4.99	5.82	1.03
Head Injuries	Mean score	0.35	0.13	0.11
Low Self-Control	Mean score	2.12	2.01	1.83
Gender	Female	1.1	98.0	33.9
Male	98.9	2.0	66.1
Race/Ethnicity	Non-White	34.9	14.9	29.9
White	65.1	85.1	70.1
% of Sample		53.4	24.0	22.6

		Class 1 High Risk (%)	Class 2 Medium Risk (%)	Class 3 Low Risk (%)
Weak Social Bonds	Mean score	1.99	1.98	1.78
Unmarried	Married	14.2	11.4	30.7
	Unmarried	85.8	88.6	69.3
Neighborhood Disorganization	Mean score	2.54	2.44	2.42
Street Code Adherence	Mean score	2.05	1.83	1.69
Criminal Peers	Mean score	2.44	1.92	1.86
Alcohol and Drug Abuse	Mean score	5.37	4.11	3.98
Anger/Irritability	Mean score	4.95	0.86	0.59
Education	College education	34.6	28.3	57.6
No college education	65.4	71.8	42.4
Low Annual Income	$30,000 or higher	36.6	3.9	86.7
Under $30,000	63.4	96.1	13.3
% of Sample		54.5	24.5	21.0

	Adult Risk Profiles
High Risk	60.2%	20.7%	19.1%
	ASR = 4.3**	ASR = −4.3**	ASR = −0.6
FSC = 0.16	FSC = −0.09	FSC = −0.01	100%
O = 230	O = 79	O = 73	382
E = 201.9	E = 104.2	E = 76.0
Medium Risk	59.5%	27.0%	13.5%
ASR = 1.9*	ASR = −0.04	ASR = −2.3*
	FSC = 0.14	FSC = 0.10	FSC = −0.08	100%
	O = 97	O = 44	O = 22	163
	E = 86.1	E = 44.5	E = 32.4
Low Risk	28.3%	43.4%	28.3%
	ASR = −7.0***	ASR = 5.2**	ASR = 3.0*
	FSC = −0.52	FSC = 0.22	FSC = 0.10	100%
	O = 45	O = 69	O = 45	159
	E = 84.0	E = 43.4	E = 31.6

Footnotes

Acknowledgments

The authors would like to thank the Sheriff’s Office and Detention Facility that worked with us on this project for their collaboration and partnership, and thanks to the dozens of research assistants who helped collect data on the Psychological Assessment of Risk and Criminality (PARC) study. We would also like to thank the anonymous Reviewers, Guest Editor Jessica Caudill, and Editor Chad Trulson for their helpful feedback and suggestions, which greatly improved this study.

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 iD

Bryanna Fox

Supplemental Material

Supplemental material for this article is available online.

Notes

References

Agnew

(1985). A revised strain theory of delinquency. Social Forces, 64(1), 151–167.

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30(1), 47–88.

Agnew

(2008). General strain theory: Current status and directions for further research. In Cullen

F. T.

Wright

J. P.

Belvins

(Eds.), Taking stock: The status of criminological theory. Advances in criminological theory (pp. 101–123). Transaction.

Akers

R. L.

(1998). Social learning and social structure: A general theory of crime and deviance. Northeastern University Press.

Akers

R. L.

(2009). Social learning and social structure: A general theory of crime and deviance. Routledge.

American Public Health Association. (2010). Online only: Report finds most U.S. inmates suffer from substance abuse or addiction. The Nation’s Health, 40(3), E11.

Anda

R. F.

Butchart

Felitti

V. J.

Brown

D. W.

(2010). Building a framework for global surveillance of the public health implications of adverse childhood experiences. American Journal of Preventive Medicine, 39(1), 93–98.

Anda

R. F.

Felitti

V. J.

Bremner

J. D.

Walker

J. D.

Whitfield

C. H.

Perry

B. D.

Dube

S. R.

Giles

W. H.

(2006). The enduring effects of abuse and related adverse experiences in childhood. European Archives of Psychiatry and Clinical Neuroscience, 256(3), 174–186.

Anderson

(1999). Code of the street: Decency, violence, and the moral life of the inner city. W. W. Norton.

10.

Arum

Beattie

I. R.

(1999). High school experience and the risk of adult incarceration. Criminology, 37(3), 515–540.

11.

Baglivio

M. T.

Epps

Swartz

Huq

M. S.

Sheer

Hardt

N. S.

(2014). The prevalence of adverse childhood experiences (ACE) in the lives of juvenile offenders. Journal of Juvenile Justice, 3(2), 1–23.

12.

Baglivio

M. T.

Wolff

K. T.

Piquero

A. R.

Epps

(2015). The relationship between adverse childhood experiences (ACE) and juvenile offending trajectories in a juvenile offender sample. Journal of Criminal Justice, 43(3), 229–241.

13.

Beauchamp

M. H.

Ditchfield

Maller

J. J.

Catroppa

Godfrey

Rosenfeld

J. V.

Kean

M. J.

Anderson

V. A.

(2011). Hippocampus, amygdala and global brain changes 10 years after childhood traumatic brain injury. International Journal of Developmental Neuroscience, 29(2), 137–143.

14.

Bird

(2013). The intergenerational transmission of poverty: An overview. In Shepherd

Brunt

(Eds.), Chronic poverty (pp. 60–84). Palgrave Macmillan.

15.

Bjerk

(2007). Measuring the relationship between youth criminal participation and household economic resources. Journal of Quantitative Criminology, 23(1), 23–39.

16.

Blumstein

Cohen

Roth

J. A.

Visher

C. A.

(1986). Criminal careers and career criminals. National Academy Press.

17.

Bonta

Andrews

D. A.

(2007). Risk-need-responsivity model for offender assessment and rehabilitation. Rehabilitation, 6(1), 1–22.

18.

Brown

S. E.

(1984). Social class, child maltreatment, and delinquent behavior. Criminology, 22(2), 259–278.

19.

Brownfield

(1986). Social class and violent behavior. Criminology, 24(3), 421–438.

20.

Burgess

R. L.

Akers

R. L.

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

21.

Burton

V. S.

Jr Evans

T. D.

Cullen

F. T.

Olivares

K. M.

Dunaway

R. G.

(1999). Age, self-control, and adults’ offending behaviors: A research note assessing a general theory of crime. Journal of Criminal Justice, 27(1), 45–54.

22.

Caspi

Bem

D. J.

Elder

G. H.

Jr . (1989). Continuities and consequences of interactional styles across the life course. Journal of Personality, 57(2), 375–406.

23.

Caspi

Elder

G. H.

Bem

D. J.

(1987). Moving against the world: Life-course patterns of explosive children. Developmental Psychology, 23(2), 308–313.

24.

Cicchetti

Toth

S. L.

(2005). Child maltreatment. Annual Review Clinical Psychology, 1, 409–438.

25.

Coley

R. J.

Barton

P. E.

(2006). Locked up and locked out: An educational perspective on the US prison population. Policy information report. Educational Testing Service.

26.

Collins

L. M.

(2002). Using latent transition analysis to examine the gateway hypothesis. In Kandel

(Ed.), Stages and pathways of drug involvement: Examining the gateway hypothesis (pp. 254–269). Cambridge University Press.

27.

Daly

(1992). Women’s pathways to felony court: Feminist theories of lawbreaking and problems of representation. Southern California Review of Law and Women’s Studies, 2, 11–52.

28.

Daly

(1994). Gender, crime, and punishment. Yale University Press.

29.

De Bellis

M. D.

(2001). Developmental traumatology: The psychobiological development of maltreated children and its implications for research, treatment, and policy. Development and Psychopathology, 13(3), 539–564.

30.

De Bellis

M. D.

Zisk

(2014). The biological effects of childhood trauma. Child and Adolescent Psychiatric Clinics, 23(2), 185–222.

31.

DeLisi

(2013). Criminal psychology. Bridgepoint Education.

32.

DeLisi

Caudill

J. W.

Trulson

C. R.

Marquart

J. W.

Vaughn

M. G.

Beaver

K. M.

(2010). Angry inmates are violent inmates: A Poisson regression approach to youthful offenders. Journal of Forensic Psychology Practice, 10(5), 419–439.

33.

DeLisi

Vaughn

M. G.

(2015). Ingredients for criminality require genes, temperament, and psychopathic personality. Journal of Criminal Justice, 43(4), 290–294.

34.

Eggleston

E. P.

Laub

J. H.

(2002). The onset of adult offending: A neglected dimension of the criminal career. Journal of Criminal Justice, 30(6), 603–622.

35.

Elliott

D. S.

Wilson

W. J.

Huizinga

Sampson

R. J.

Elliott

Rankin

(1996). The effects of neighborhood disadvantage on adolescent development. Journal of Research in Crime and Delinquency, 33(4), 389–426.

36.

Farrington

D. P.

(1986). Age and crime. Crime and Justice, 7, 189–250.

37.

Farrington

D. P.

(1995). The development of offending and antisocial behaviour from childhood: Key findings from the Cambridge study in delinquent development. Journal of Child Psychology and Psychiatry, 6(36), 929–964.

38.

Farrington

D. P.

(1998). Predictors, causes, and correlates of male youth violence. Crime and Justice, 24, 421–475.

39.

Farrington

D. P.

(2003a). Developmental and life-course criminology: Key theoretical and empirical issues. Criminology, 41, 221–255.

40.

Farrington

D. P.

(2003b). Key results from the first forty years of the Cambridge study in delinquent development. In Thornberry

T. P.

Krohn

M. D.

(Eds.), Taking stock of delinquency: An overview of findings from contemporary longitudinal studies (pp. 137–183). Springer.

41.

Farrington

D. P.

(2005a). Childhood origins of antisocial behavior. Clinical Psychology & Psychotherapy: An International Journal of Theory & Practice, 12(3), 177–190.

42.

Farrington

D. P.

(2005b). Introduction to integrated developmental and life-course theories of offending. In Integrated developmental and life-course theories of offending (pp. 1–14). New Brunswick, NJ: Transaction.

43.

Farrington

D. P.

(2010). Life-course and developmental theories in criminology. In McLaughlin

Newburn

(Eds.), The sage handbook of criminological theory (pp. 249–269). SAGE.

44.

Farrington

D. P.

(2015). Prospective longitudinal research on the development of offending. Australian & New Zealand Journal of Criminology, 48(3), 314–335.

45.

Farrington

D. P.

Snyder

H. N.

Finnegan

T. A.

(1988). Specialization in juvenile court careers. Criminology, 26(3), 461–487.

46.

Farrington

D. P.

Ttofi

M. M.

Coid

J. W.

(2009). Development of adolescence-limited, late-onset, and persistent offenders from age 8 to age 48. Aggressive Behavior: Official Journal of the International Society for Research on Aggression, 35(2), 150–163.

47.

Farrington

D. P.

West

D. J.

(1995). Effects of marriage, separation, and children on offending by adult males. In Blau

Z. S.

Hagan

(Eds.), Current perspectives on aging and the life cycle (Vol. 14, pp. 249–281). JAI Press.

48.

Felitti

V. J.

Anda

R. F.

Nordenberg

Williamson

D. F.

Spitz

A. M.

Edwards

Koss

M. P.

Marks

J. S.

(1998). Relationship of childhood abuse and household dysfunction to many of death in adults. The adverse childhood experiences (ACE) study. American Journal of Preventive Medicine, 14(4), 245–258.

49.

Fox

Farrington

D. P.

(2012). Creating burglary profiles using latent class analysis: A new approach to offender profiling. Criminal Justice and Behavior, 39(12), 1582–1611.

50.

Fox

Perez

Cass

Baglivio

M. T.

Epps

(2015). Trauma changes everything: Examining the relationship between adverse childhood experiences and serious, violent and chronic juvenile offenders. Child Abuse & Neglect, 46, 163–173.

51.

Fox

DeLisi

(2018). From criminological heterogeneity to coherent classes: Developing a typology of juvenile sex offenders. Youth Violence and Juvenile Justice, 16(3), 299–318.

52.

Fox

Verona

Fournier

(2019). Psychological assessment of risk in a county jail: Implications for reentry, recidivism and detention practices in the USA. Journal of Criminal Psychology, 9(4), 173–186.

53.

Fox

K. A.

Lane

Akers

R. L.

(2010). Do perceptions of neighborhood disorganization predict crime or victimization? An examination of gang member versus non-gang member jail inmates. Journal of Criminal Justice, 38(4), 720–729.

54.

Glueck

E. T.

(1968). Delinquents and nondelinquents in perspective. Harvard University Press.

55.

Gottfredson

M. R.

Hirschi

(1990). A general theory of crime. Stanford University Press.

56.

Grasmick

H. G.

Tittle

C. R.

Bursik

R. J.

Arneklev

B. J.

(1993). Testing the core empirical implications of Gottfredson and Hirschi’s general theory of crime. Journal of Research on Crime and Delinquency, 30, 5–29.

57.

Grisso

Barnum

(2006). Massachusetts youth screening instrument, version 2: MAYSI-2: User's manual and technical report. Sarasota, FL: Professional Resource Press.

58.

Hänninen

D. T.

Soininen

(2012). Age-associated memory impairment. Drugs and Aging, 11, 480–489.

59.

Harlow

C. W.

(2003). Education and correctional populations. US Department of Justice, Office of Justice Programs.

60.

Harris

A. R.

(1977). Sex and theories of deviance: Toward a functional theory of deviant type-scripts. American Sociological Review, 42(1), 3–16.

61.

Hart

Rubia

(2012). Neuroimaging of child abuse: A critical review. Frontiers in Human Neuroscience, 6, 1–24.

62.

Hay

Forrest

(2009). The implications of family poverty for a pattern of persisting offending. In Savage

(Ed.), The development of persistent criminality (pp. 54–70). Oxford University Press.

63.

Herrenkohl

T. I.

Maguin

Hill

K. G.

Hawkins

J. D.

Abbott

R. D.

Catalano

R. F.

(2000). Developmental risk factors for youth violence. Journal of Adolescent Health, 26(3), 176–186.

64.

Hubbard

D. J.

Pratt

T. C.

(2002). A meta-analysis of the predictors of delinquency among girls. Journal of Offender Rehabilitation, 34(3), 1–13.

65.

Huesmann

L. R.

Dubow

E. F.

Boxer

(2009). Continuity of aggression from childhood to early adulthood as a predictor of life outcomes: Implications for the adolescent-limited and life-course-persistent models. Aggressive Behavior, 35(2), 136–149.

66.

Jackson

K. M.

O’Neill

S. E.

Sher

K. J.

(2006). Characterizing alcohol dependence: Transitions during young and middle adulthood. Experimental and Clinical Psychopharmacology, 14(2), 228–244.

67.

Javdani

Sadeh

Verona

(2011). Expanding our lens: Female pathways to antisocial behavior in adolescence and adulthood. Clinical Psychology Review, 31(8), 1324–1348.

68.

Jennings

W. G.

Rocque

Fox

B. H.

Piquero

A. R.

Farrington

D. P.

(2016). Can they recover? An assessment of adult adjustment problems among males in the abstainer, recovery, life-course persistent, and adolescence-limited pathways followed up to age 56 in the Cambridge study in delinquent development. Development and Psychopathology, 28(2), 537–549.

69.

Jolliffe

Farrington

D. P.

Piquero

A. R.

Loeber

Hill

(2017). Systematic review of early risk factors for life-course-persistent, adolescence-limited, and late-onset offenders in prospective longitudinal studies. Aggression and Violent Behavior, 33, 15–23.

70.

Kalmakis

K. A.

Chandler

G. E.

(2015). Health consequences of adverse childhood experiences: a systematic review. Journal of the American Association of Nurse Practitioners, 27(8), 457–465.

71.

Kendziora

K. T.

Osher

D. M.

(2004). Fostering resilience among youth in the juvenile justice system. In Clauss-Ehlers

C. S.

Weist

M. D.

(Eds.), Community planning to foster resilience in children (pp. 177–195). Springer.

72.

King

R. D.

Massoglia

MacMillan

(2007). The context of marriage and crime: Gender, the propensity to marry, and offending in early adulthood. Criminology, 45(1), 33–65.

73.

Krivo

L. J.

Peterson

R. D.

(1996). Extremely disadvantaged neighborhoods and urban crime. Social Forces, 75(2), 619–648.

74.

Krohn

Lizotte

Perez

(1997). The interrelationship between substance use and precocious transitions to adult statuses. Journal of Health and Social Behavior, 38(1), 87–103.

75.

Lanctôt

Le Blanc

(2002). Explaining deviance by adolescent females. Crime and Justice, 29, 113–202.

76.

Lanza

S. T

Collins

L. M.

(2002). Pubertal timing and the onset of substance use in females during early adolescence. Prevention Science, 3(1), 69–82.

77.

Lanza

S. T.

Bray

B. C.

(2010). Transitions in drug use among high-risk women: An application of latent class and latent transition analysis. Advances and Applications in Statistical Sciences, 3(2), 203–235.

78.

Lanza

S. T.

Collins

L. M.

(2008). A new SAS procedure for latent transition analysis: Transitions in dating and sexual risk behaviors. Developmental Psychology, 44(2), 446–456.

79.

Lanza

S. T.

Collins

L. M.

Lemmon

D. R.

Schafer

J. L.

(2007). PROC LCA: A SAS procedure for latent class analysis. Structural Equation Modeling: A Multidisciplinary Journal, 14(4), 671–694.

80.

Lanza

S. T.

Flaherty

B. P.

Collins

L. M.

(2003). Latent class and latent transition analysis. In Schinka

J. A.

Velicer

W. F.

(Eds.), Handbook of psychology: Vol. 2. Research methods in psychology (pp. 663–685). Wiley.

81.

Lanza

S. T.

Patrick

M. E.

Maggs

J. L.

(2010). Latent transition analysis: Benefits of a latent variable approach to modeling transitions in substance use. Journal of Drug Issues, 40, 93–120.

82.

Lanza

S. T.

Rhoades

B. L.

Nix

R. L.

Greenberg

M. T.

(2010). Modeling the interplay of multilevel risk factors for future academic and behavior problems: A person-centered approach. Development and Psychopathology, 22, 313–335.

83.

Lanza

S. T.

Savage

J. S.

Birch

(2010). Identification and prediction of latent classes of weight-loss strategies among women. Obesity, 18, 833–840.

84.

Laub

J. H.

Sampson

R. J.

(2003). Shared beginnings, divergent lives: Delinquent boys to age 70. Harvard University Press.

85.

Leiber

Rodriguez

(2011). The implementation of the disproportionate minority confinement/contact (DMC) mandate: A failure or success? Race and Justice, 1(1), 103–124.

86.

Leon-Carrion

Ramos

F. J. C.

(2003). Blows to the head during development can predispose to violent criminal behaviour: Rehabilitation of consequences of head injury is a measure for crime prevention. Brain Injury, 17(3), 207–216.

87.

Levenson

J. S.

Willis

G. M.

Prescott

D. S.

(2016). Adverse childhood experiences in the lives of male sex offenders: Implications for trauma-informed care. Sexual Abuse, 28(4), 340–359.

88.

Loeber

(1990). Development and risk factors of juvenile antisocial behavior and delinquency. Clinical Psychology Review, 10(1), 1–41.

89.

Loeber

Farrington

D. P.

(Eds.). (1998). Serious and violent juvenile offenders: Risk factors and successful interventions. SAGE.

90.

Magidson

Vermunt

J. K.

(2004). Latent class models. In Kaplan

(Ed.), Sage handbook of quantitative methodology for the social sciences (pp. 175–198). SAGE.

91.

McNeeley

Wilcox

(2015). Street codes, routine activities, neighbourhood context and victimization. British Journal of Criminology, 55(5), 921–943.

92.

Merton

R. K.

(1938). Social structure and anomie. American Sociological Review, 3(5), 672–682.

93.

Moffitt

T. E.

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

94.

Moffitt

T. E.

(2015). Life-course-persistent versus adolescence-limited antisocial behavior. In Cicchetti

Cohen

D. J.

(Eds.), Developmental psychopathology (2nd ed., pp. 570–598). John Wiley & Sons.

95.

Moule

R. K.

Jr Burruss

G. W.

Gifford

F. E.

Parry

M. M.

Fox

(2019). Legal socialization and subcultural norms: Examining linkages between perceptions of procedural justice, legal cynicism, and the code of the street. Journal of Criminal Justice, 61, 26–39.

96.

Moule

R. K.

Fox

(2020). Belief in the code of the street and individual involvement in offending: A meta-analysis. Youth Violence and Juvenile Justice. Advance online publication. https://doi.org/10.1177/1541204020927737

97.

Muthén

L. K.

(2000). Integrating person-centered and variable-centered analyses: Growth mixture modeling with latent trajectory classes. Alcoholism: Clinical and Experimental Research, 24(6), 882–891.

98.

Paternoster

Brame

Piquero

Mazerolle

Dean

C. W.

(1998). The forward specialization coefficient: Distributional properties and subgroup differences. Journal of Quantitative Criminology, 14(2), 133–154.

99.

Piquero

A. R.

(2008). Disproportionate minority contact. The future of children, 18(2), 59–79.

100.

Piquero

A. R.

(2015). Understanding race/ethnicity differences in offending across the life course: Gaps and opportunities. Journal of Developmental and Life-Course Criminology, 1(1), 21–32.

101.

Piquero

A. R.

Intravia

Stewart

Piquero

N. L.

Gertz

Bratton

(2012). Investigating the determinants of the street code and its relation to offending among adults. American Journal of Criminal Justice, 37(1), 19–32.

102.

Pratt

T. C.

(2016). A self-control/life-course theory of criminal behavior. European Journal of Criminology, 13(1), 129–146.

103.

Pratt

T. C.

Cullen

F. T.

(2000). The empirical status of Gottfredson and Hirschi’s general theory of crime: A meta-analysis. Criminology, 38(3), 931–964.

104.

Pratt

T. C.

Cullen

F. T.

Sellers

C. S.

Winfree

T. L.

Jr Madensen

T. D.

Daigle

L. E.

Fearn

N. E.

Gau

J. M.

, (2010). The empirical status of social learning theory: A meta-analysis. Justice Quarterly, 27(6), 765–802.

105.

Raine

(2014). The anatomy of violence: The biological roots of crime. Vintage.

106.

Reavis

J. A.

Looman

Franco

K. A.

Rojas

(2013). Adverse childhood experiences and adult criminality: How long must we live before we possess our own lives? The Permanente Journal, 17(2), 44–48.

107.

Reid

J. A.

Loughran

T. A.

(2019). A latent class typology of justice-involved youth victims and exploration of trauma-related psychological symptoms. Justice Quarterly. Advance online publication. https://doi.org/10.1080/07418825.2019.1595700

108.

Reid

J. B.

Patterson

G. R.

(1989). The development of antisocial behaviour patterns in childhood and adolescence. European Journal of Personality, 3(2), 107–119.

109.

Salvatore

Taniguchi

T. A.

(2012). Do social bonds matter for emerging adults? Deviant Behavior, 33(9), 738–756.

110.

Sampson

R. J.

(2012). Great American city: Chicago and the enduring neighborhood effect. University of Chicago Press.

111.

Sampson

R. J.

Laub

J. H.

(1990). Crime and deviance over the life course: The salience of adult social bonds. American Sociological Review, 55, 609–627.

112.

Sampson

R. J.

Laub

J. H.

(1993). Crime in the making: Pathways and turning points through life. Harvard University Press.

113.

Sampson

R. J.

Laub

J. H.

(2005). When prediction fails: From crime-prone boys to heterogeneity in adulthood. The Annals of the American Academy of Political and Social Science, 602(1), 73–79.

114.

Schwartz

J. A.

Connolly

E. J.

Valgardson

B. A.

(2018). An evaluation of the directional relationship between head injuries and subsequent changes in impulse control and delinquency in a sample of previously adjudicated males. Journal of Criminal Justice, 56, 70–80.

115.

Sharp

D. J.

Scott

Leech

(2014). Network dysfunction after traumatic brain injury. Nature Reviews Neurology, 10(3), 156–166.

116.

Shaw

C. R.

McKay

H. D.

(1942). Juvenile delinquency and urban areas. University of Chicago Press.

117.

Smith

Thornberry

T. P.

(1995). The relationship between childhood maltreatment and adolescent involvement in delinquency. Criminology, 33(4), 451–481.

118.

Stander

Farrington

D. P.

Hill

Altham

P. M.

(1989). Markov chain analysis and specialization in criminal careers. The British Journal of Criminology, 29(4), 317–335.

119.

Steffensmeier

Feldmeyer

Harris

C. T.

Ulmer

J. T.

(2011). Reassessing trends in black violent crime, 1980–2008: Sorting out the “Hispanic effect” in uniform crime reports arrests, national crime victimization survey offender estimates, and US prisoner counts. Criminology, 49(1), 197–251.

120.

Stewart

E. A.

Schreck

C. J.

Simons

R. L.

(2006). “I ain’t gonna let no one disrespect me” Does the code of the street reduce or increase violent victimization among African American adolescents? Journal of Research in Crime and Delinquency, 43(4), 427–458.

121.

Stewart

E. A.

Simmons

R. L.

(2010). Race, code of the street, and violent delinquency: A multilevel investigation of neighborhood street culture and individual norms of violence. Criminology, 42(2), 569–605.

122.

Sullivan

C. J.

McGloin

J. M.

Ray

J. V.

Caudy

M. S.

(2009). Detecting specialization in offending: Comparing analytic approaches. Journal of Quantitative Criminology, 25, 419–441.

123.

Teicher

M. H.

(2000). Wounds that time wouldn’t heal: The neurobiology of childhood abuse. Cerebrum, 2(4), 50–67.

124.

Temcheff

C. E.

Serbin

L. A.

Martin-Storey

Stack

D. M.

Hodgins

Ledingham

Schwartzman

A. E.

(2008). Continuity and pathways from aggression in childhood to family violence in adulthood: A 30-year longitudinal study. Journal of Family Violence, 23(4), 231–242.

125.

Theobald

Farrington

D. P.

(2013). The effects of marital breakdown on offending: results from a prospective longitudinal survey of males. Psychology, Crime & Law, 19, 391–408.

126.

Thornberry

T. P.

Huizinga

Loeber

(1995). The prevention of serious delinquency and violence: Implications from the Program of Research on the Causes and Correlates of Delinquency. In Howell

Krisberg

Hawkins

Wilson

J. D.

(Eds.), Sourcebook on serious, violent and chronic juvenile offenders (pp. 213–237). Thousand Oaks, CA: Sage.

127.

U.S. Department of Health & Human Services. (2019). 2019 poverty guidelines. Retrieved March 17, 2020, from https://aspe.hhs.gov/2019-poverty-guidelines

128.

Uebersax

(2009). Latent class analysis: Frequently asked questions. Latent Structure Analysis. Retrieved February 25, 2020, from www.johnuebersax.com/stat/index.htm

129.

Vaughn

M. G.

DeLisi

Beaver

K. M.

Howard

M. O.

(2008). Toward a quantitative typology of burglars: A latent profile analysis of career offenders. Journal of Forensic Science, 53(6), 1387–1392.

130.

Vaughn

M. G.

DeLisi

Beaver

K. M.

Howard

M. O.

(2009). Multiple murder and criminal careers: A latent class analysis of multiple homicide offenders. Forensic Science International, 183, 67–73.

131.

Vaughn

M. G.

DeList

Gunter

Beaver

K. M.

Perron

B. E.

Howard

M. O.

(2011). The severe 5%: A latent class analysis of the externalizing behavior spectrum in the United States. Journal of Criminal Justice, 39(1), 75–80.

132.

Vermunt

J. K.

Magidson

(2005). Technical guide for latent GOLD 4.0: Basic and advanced. Statistical Innovations.

133.

Warr

(1998). Life-course transitions and desistance from crime. Criminology 36, 183–216.

134.

Wattanaporn

K. A.

Holtfreter

(2014). The impact of feminist pathways research on gender-responsive policy and practice. Feminist Criminology, 9(3), 191–207.

135.

Widom

C. S.

(1989). Child abuse, neglect, and violent criminal behavior. Criminology, 27(2), 251–271.

136.

Williams

W. H.

Cordan

Mewse

A. J.

Tonks

Burgess

C. N.

(2010). Self-reported traumatic brain injury in male young offenders: A risk factor for re-offending, poor mental health and violence? Neuropsychological Rehabilitation, 20(6), 801–812.

137.

Wilson

W. J.

(1987). The truly disadvantaged: The inner city, the underclass, and public policy. University of Chicago.

138.

Witt

S. T.

Lovejoy

D. W.

Pearlson

G. D.

Stevens

M. C.

(2010). Decreased prefrontal cortex activity in mild traumatic brain injury during performance of an auditory oddball task. Brain Imaging and Behavior, 4(3–4), 232–247.

139.

Wolfe

S. E.

(2015). Low self-control, gender, race, and offending in late life. Psychology, Crime & Law, 21(5), 426–451.

140.

Wolfe

S. E.

Reisig

M. D.

Holtfreter

(2016). Low self-control and crime in late adulthood. Research on Aging, 38(7), 767–790.

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

0.07 MB

	Full Sample		Under Age 25		Age 25 and Over
	%	M (SD)	%	M (SD)	%	M (SD)	χ²/F
Early Risk Factors
Adverse Childhood Experiences		4.33 (2.75)		4.88 (2.78)		4.19 (2.72)	6.3^*
Head Injuries		0.22 (0.41)		0.30 (0.46)		0.21 (0.41)	0.8^*
Low Self-Control		2.03 (0.72)		2.12 (0.68)		2.01 (0.73)	1.2
Male	67.9		75.8		66.1		4.4^*
White	71.6		55.6		74.8		18.6^***
Adult Risk Factors
Weak Social Bonds		1.94 (0.78)		1.94 (0.82)		1.95 (0.78)	0.0
Unmarried	82.9		89.9		81.5		4.9^*
Neighborhood Disorganization		2.49 (0.58)		2.43 (0.56)		2.49 (0.58)	0.4
Street Code Adherence		1.92 (0.61)		2.07 (0.59)		1.89 (0.62)	3.1^**
Criminal Peers		2.19 (0.96)		2.34 (1.00)		2.16 (0.95)	3.3
Alcohol and Drug Use		4.92 (1.70)		5.11 (1.70)		4.91 (1.69)	1.9
Anger/Irritability		3.03 (2.69)		3.69 (2.77)		2.90 (2.69)	62.4^***
No College Education	62.1		81.8		58.1		24.1^***
Low Annual Income	61.0		63.9		60.4		0.5

	Adult Risk Profiles
Early Risk Profiles	High Risk	Medium Risk	Low Risk	Total
High Risk	60.2%	20.7%	19.1%
	ASR = 4.3**	ASR = −4.3**	ASR = −0.6
	FSC = 0.16	FSC = −0.09	FSC = −0.01	100%
	O = 230	O = 79	O = 73	382
	E = 201.9	E = 104.2	E = 76.0
Medium Risk	59.5%	27.0%	13.5%
Medium Risk	ASR = 1.9*	ASR = −0.04	ASR = −2.3*
	FSC = 0.14	FSC = 0.10	FSC = −0.08	100%
	O = 97	O = 44	O = 22	163
	E = 86.1	E = 44.5	E = 32.4
Low Risk	28.3%	43.4%	28.3%
	ASR = −7.0***	ASR = 5.2**	ASR = 3.0*
	FSC = −0.52	FSC = 0.22	FSC = 0.10	100%
	O = 45	O = 69	O = 45	159
	E = 84.0	E = 43.4	E = 31.6