The Triple Crown of Antisocial Behavior

Abstract

This study examines the effect of negative emotionality, effortful control, and community disadvantage on juvenile recidivism. Using DeLisi and Vaughn’s temperament theory as a foundation, we assess whether youth who have temperament issues and those who live in disadvantaged communities are more likely to recidivate. Findings indicate that net of a wide array of known risk factors, youth with poor temperaments, and those living in disadvantaged communities are more likely to reoffend. Additionally, those youth who face a triple threat of temperament issues and disadvantage reoffend faster post-completion. The theoretical and policy implications of these findings are discussed.

Keywords

temperament theory community context juvenile justice recidivism

Introduction

Contemporary temperament research is most prominent in the field of developmental psychology. Criminologists, however, have only periodically incorporated temperament concepts in the explanation of antisocial and deviant behavior. Most notably, negative emotionality (NE) has been implicated in Agnew’s (1992) general strain theory (1992), Gottfredson and Hirschi’s (1990) use of low self-control in the general theory of crime, Moffitt’s (1993) elaboration of the role of angry–irritableness in her dual taxonomy, and Wilson and Herrnstein’s (1985) explanation of the effects of low effortful control (EC) on offending. Recently, DeLisi and Vaughn (2014) have articulated a temperament-based theory of antisocial behavior, which employs explicit temperament concepts as the exclusive explanation of both law-breaking behavior and negative interactions with the justice system. EC and NE are the two constructs fundamental to their temperament-based theory. Furthermore, the theorists argue temperament theory to be indispensable in the advancement of understanding the effects of context on offending, stating “when low levels of behavioral self-regulation and high levels of NE are paired with disadvantaged contexts their expression is more likely to be stronger” (2014, p. 17).

The current study examines the hypothesis of an interplay between temperament theory constructs and community conditions on juvenile recidivism. We first provide a brief overview of EC and NE, followed by a review of the research on contextual effects on juvenile crime and recidivism, and the limited work examining temperament and context in tandem for understanding patterns of offending. Next we describe the data, measures, and methods, followed by the analyses and closing with a discussion regarding policy, prevention, and intervention.

Temperament Theory Constructs

EC

EC is essentially the ability to constrain a response that is perhaps more favorable (dominant) at the moment and instead perform a subdominant behavior (Kagan & Snidman, 2009; Rothbart, 1989; Sulik et al., 2010). At any given moment, there are behaviors we would prefer to engage in, yet we inhibit those emotions and behavioral impulses and instead pay attention, and act, in a more socially appropriate manner. This ability to regulate emotion and delay immediate gratification (exhibit EC) is required in every facet of life, from sitting quietly at one’s desk and raising a hand to be called on to speak in school, to arriving to work on time and interacting with coworkers and one’s superiors appropriately in later life (see also DeLisi & Vaughn, 2014; MacDonald, 2008). Juveniles with EC deficits act impulsively, seemingly unable to control their negative behavior. These youth lack techniques and skills to delay gratification, modulate reactivity, or “think before acting,” especially in emotionally charged situations (Rothbart, Evans, & Ahadi, 2000). Youth with these deficits are at increased likelihood of experiencing peer rejection of impulsive or aggressive speaking or acting out, teachers disciplining inappropriate school conduct, punitive parental responses to wandering off or inattention, and ultimately encounters with the juvenile justice system (DeLisi & Vaughn, 2014).

Individuals with lower levels of EC (often referred to as self-control in adolescents and adults) have a higher likelihood of experiencing a host of negative outcomes across many life domains, including physical health, substance dependence, personal finances, and criminal offending (see Moffitt et al., 2011). With respect to offending, EC deficits have been shown to be predictive of delinquency and reoffending (Evans-Chase, 2014; Evans-Chase & Zhou, 2014; Gordon, Diehl, & Anderson, 2012; Pratt & Cullen, 2000; Roose, Bijttebier, Van der Oord, Claes, & Lilienfeld, 2013; Stevenson & Goodman, 2001), a finding which holds across race and gender (Heatherton & Baumeister, 1996; Rothbart, 2011; Sulik et al., 2010) and across the life course (DeLisi & Vaughn, 2014, 2015; Morizot, 2015).

NE

Research suggests that individuals differ in how they perceive and experience interactions with others and their environment. NE involves a predominately negative outlook and perception of individuals and one’s social environment (Clark, 2005). Individuals with high NE have increased likelihood of perceiving the intentions and actions of others as hostile or directed against them. NE is composed of both “hot” (frustration, fear, anger, hostility, and irritability) and “cold” (discomfort, sadness, and soothability) variants (Rothbart, 2007; see also DeLisi & Vaughn, 2014). NE has consistently been found to be related to the externalization of behavior, internalizing problems, and adjustment difficulties, with those having higher NE evincing more problems (Clark, Watson, & Mineka, 1994; Eisenberg et al., 2001, 2005; Lengua, West, & Sandler, 1998). One vignette-based study assessing reactions to a person being singled out for punishment showed that individuals with low self-control were more likely to perceive the sanction as unfair and that both unfair sanctions and low self-control led to higher perceptions of being angry for being singled out for punishment (see Piquero, Gomez-Smith, & Langton, 2004).

Individuals with both low EC and high NE are hypothesized to evidence even greater antisocial behavior and negative interactions with and within the juvenile and criminal justice systems (DeLisi & Vaughn, 2014). There is ample evidence of the interrelatedness of EC and NE (Clark, 2005; Eisenberg, Spinrad, & Eggum, 2010; Kochanska & Knaack, 2003). The comorbidity of low EC and high NE heightens the likelihood the individual will display emotions and behaviors that will be poorly received by others as they are unable to inhibit visceral responses to emotions and negative perceptions. This coupling of poor self-regulation and the expression of negative emotions sets the stage for negative social exchanges and interactions. Those with greater propensity to perceive others and their environment as hostile and who are less able to control emotions and actions and to adhere to societal norms will be more likely rejected by friends, family, and authority, increasing the likelihood of encountering both informal and formal social control mechanisms.

Contextual Effects and Juvenile Crime

While the past two decades has seen a growth in studies examining the effects of community context on criminal behavior (Elliott et al., 1996; Sampson, Raudenbush, & Earls, 1997; Sampson, Morenoff, & Gannon-Rowley, 2002; Wikstrom & Loeber, 2000), research and theory on the recidivism of previously adjudicated juveniles has, to a large extent, neglected the potential for contextual factors to contribute to postrelease outcomes (Abrams & Snyder, 2010). We know of three studies that have examined the potential for community-level factors, such as disadvantage or immigrant concentration, to impact further involvement in criminal behavior among juveniles. Wright and Rodriguez (2014) found no relationship between returning to communities marked by concentrated immigration and recidivism among 12,000 youth in a single county in the state of Arizona. In contrast, Wolff, Baglivio, Intravia, and Piquero (2015) found youth residing in disadvantaged neighborhoods to be at an increased risk of juvenile recidivism, but those living in neighborhoods marked by a high concentration of immigrants were roughly 6% less likely to reoffend. They conclude that, consistent with the broader literature on recidivism, community context, including levels of disadvantage, is an important predictor of juvenile recidivism (Wolff et al., 2015; see also Kubrin & Stewart, 2006; Wright, Kim, Chassin, Losoya, & Piquero, 2014). A third study that examined juvenile reoffending using Pathways to Desistance data found concentrated disadvantage indirectly associated with juvenile offending, primarily through its association with exposure to deviant peers (Wright et al., 2014).

The vast majority of prior research, however, has neglected to examine the impact of temperament across varying environmental conditions and how the two may interact and contribute to recidivism among juvenile offenders. We are aware of a single study by Lynam and colleagues (2000), which examined the interaction of impulsivity and disadvantage on juvenile offending of 13-year-old males. Their findings suggest the impact of impulsivity is strongest in poorer neighborhoods and that this effect was not significant in more affluent areas. Furthermore, their results held in follow-up analyses when the youth reached late adolescence (16–17 years old), again with impulsivity contributing to delinquent behavior only in disadvantaged neighborhoods.¹

While DeLisi and Vaughn (2014) hypothesize disadvantaged contexts exacerbate the effects of the coupling of both low EC and high NE, to date, this idea has not been tested empirically. Multilevel approaches capable of accounting for community context have largely been ignored within existing temperament research, leading to relatively little information regarding the interplay between individual temperament and environmental conditions and their impact on juvenile recidivism.

Current Study

The objective of this study is to examine the interplay of temperament theory constructs and contextual measures of community disadvantage/deprivation for explaining juvenile recidivism both individually and jointly. Explicitly, we examine the effects of EC, NE, and community context on the time to rearrest from the completion of a community-based juvenile justice system service, while controlling for a host of demographic measures and key individual-level risk factors. Before presenting the results of our analysis, we discuss the data and methods used.

Data

The current study uses data on 3 years of Florida Department of Juvenile Justice (FDJJ) community-based service completions maintained by the FDJJ.² Demographic and placement data were taken from the FDJJ’s centralized information system, and all individual-level measures were gleaned from the Community Positive Achievement Change Tool (C-PACT) risk/needs assessment administered by the FDJJ closest to the youth completing the community-based service.³ Data for this study include all youth within Florida, who completed an FDJJ community-based service between July 1, 2009, and June 31, 2012, who were administered the full C-PACT and do not include those who only received the prescreen assessment. Both the prescreen and the full assessment produce identical risk to reoffend classifications, however, the full assessment contains approximately 80 additional items (that do not calculate into overall risk level) providing a more detailed identification of the youth’s unique historical and present situation. Only the full assessment contains the requisite items to appropriately evaluate temperament theory. Additionally, 285 youth in the original sample had a home address listed, which was outside the state and were therefore dropped from the final analyses. Finally, to facilitate the multilevel design used in the current study and ensure the reliability of the Level-2 intercept, we limited inclusion to zip codes in which at least 10 youth resided, thereby removing 1,548 youth. This resulted in a final analytic sample of 26,960 unduplicated youth living in 748 zip codes (with an average of 71, σ= 46.8, youth per zip code). These youth represent 21% of all community-based completions over the 3-year study period.⁴

Measures

Official Offending Days to Failure

The outcome chosen for the current study, official offending (recidivism), is measured as subsequent delinquency referral or an adult arrest for a new law violation within 12 months from the day of service completion.⁵ The first arrest during the 365-day period postcompletion of service for a given youth was counted as recidivism, with the number of days from service completion to first subsequent arrest representing the days to failure. Youth not rearrested within their 365-day postcompletion period were “censored,” regardless of whether that youth had an arrest outside of the follow-up. There are no technical or nonlaw violations to consider, as all youth completed supervision preceding the follow-up period.

Temperament Theory Measures

EC

We include 5 items measuring EC, each of which were garnered from C-PACT items. These 5 items were standardized and combined into an EC index (α = .710) where higher values correspond to increased EC, which is expected to delay (increase time to failure) or reduce the likelihood of recidivism. The following are the 5 items included.

School conduct

School conduct was captured by combining six measures of school conduct into a single variable where graduated/general equivalency diploma ([GED] = 6), recognition of good behavior (= 5), no problems with school conduct (= 4), or problems reported by teachers (= 3), including problem calls to parents (= 2), and calls to police (= 1), with higher levels signifying less antisocial conduct in school.⁶

Behavioral impulsivity

This self-reported measure ranged from the youth reporting they are highly impulsive and usually act before thinking, impulsive, and often act without thinking, have some self-control, and use self-control and usually think before acting (coded 1–4, with higher scores representing greater levels of control).

Belief in control over antisocial behavior

The self-reported measure captures whether the youth believes antisocial behavior is out of his or her control, somewhat believes such behavior is controllable or he or she believes they can avoid or stop antisocial behavior (coded 1–3, respectively, with higher values representing higher levels of control).

Cognitive impulsivity

Impulse control ranges from the youth not knowing techniques (such as reframing, replacing antisocial thoughts with pro-social ones, relaxation) to control impulsive behavior, knowing techniques to control impulsive behavior, using techniques to control impulsive behavior, and never having a problem with impulsive behavior (coded 1–4, respectively, with higher values indicating greater impulse control).

Aggression control

Control of aggression ranged from the youth lacking alternatives to aggression, rarely using alternatives, sometimes using alternatives, often using alternatives to aggression, and youth never having a problem with aggression (coded 1–5, with higher values indicating greater control of aggression).

NE

An NE index was created by combining a number of emotionality measures into a single, additive index. Each measure was recoded (if necessary) such that higher values corresponded to higher levels of NE, and therefore higher values are expected to accelerate recidivism (α = .682).⁷ The NE index included the following items.

Depression/anxiety

Depression/anxiety ranges from no history, history of occasional feelings of depression/anxiety, history of consistent feelings of depression/anxiety, and history of impairment in everyday tasks due to depression/anxiety (coded 1–4). Higher scores indicate more depression/anxiety.

Anger/irritability

History of anger or irritability was assessed along a continuum ranging from no history, history of occasional feelings of anger/irritability, history of consistent feelings of anger/irritability, and history of aggressive reactions to feelings of anger/irritability (coded 1–4, respectively). Higher scores indicate more anger/aggression.

Frustration tolerance

Ranging from rarely getting upset over small things, sometimes gets upset over small things or has temper tantrums, and often gets upset over small things or has temper tantrums (coded 1–3, respectively). Higher scores indicate a lower level of tolerance.

Hostile interpretation

Hostile interpretation ranges from primarily positive, primarily negative, and primarily hostile view of the intentions of others in a nonconfrontational setting (coded 1–3). Higher scores indicating greater levels of hostility attributed to others’ actions and intent.

Community Measures

In the current study, we use three distinct measures of community socioeconomic conditions drawn from past research devoted to contextual effects (Baumer, Messner, & Felson, 1998; Kubrin & Stewart, 2006; Morenoff, Sampson, & Raudenbush, 2001). Data used to construct the community-level measures were drawn from the 2009–2013 American Community Survey (ACS) 5-year estimates for zip codes in the State of Florida (U.S. Census Bureau, 2014), which corresponds closely with the 2009–2012 study time period. It is believed that these measures represent the best available estimate of the conditions present during the time youth completed their service with FDJJ. From this data, the three measures described below were created.

Resource deprivation index

The first measure of community disadvantage is resource deprivation (Land, McCall, & Cohen, 1990). Six zip code–level variables were used to form the measure including the proportion of individuals living below the poverty line, median family income (logged and reverse coded), the proportion of female-headed households, the unemployment rate, the percentage of the population who are non-Hispanic Black, and the Gini index of income inequality. Consistent with previous research, these variables are strongly correlated to one another at the zip code level, and factor analyses indicated that these variables loaded on a single factor in our sample. The items were standardized and combined to form an additive index of neighborhood disadvantage (α = .855).

Concentrated disadvantage index

A second measure of disadvantage, concentrated disadvantage, is composed of six zip code–level variables drawn from the ACS: the proportion of families living below the poverty line, median family income (logged and reverse coded), the proportion of female-headed households, the unemployment rate, the proportion of the population with a high school degree (reverse coded), and the proportion of households receiving public assistance. Similar to the resource deprivation index, these items were standardized and combined to form an additive index of neighborhood disadvantage (α = .893).

Concentrated affluence index

Importantly, focusing exclusively on disadvantage masks the potential protective effects afforded to those live in affluent areas (Brooks-Gunn, Duncan, Kato, & Sealand, 1993; Massey, 2001; Morenoff et al., 2001). As such, we also employ a measure of concentrated affluence drawn from Massey’s (2001) work. This measure, known as index of concentration at the extremes (ICE), captures the degree to which affluence is concentrated, relative to the concentration of poverty in a given area (in our case zip code). Consistent with prior research, the ICE was calculated using the following formula: ([number of affluent families − number of poor families]/total number of families; Kubrin & Stewart, 2006). Families defined as “poor” are those below the poverty line, while “affluent” is demarcated as families with incomes 2 SDs above the mean (for the current study M = US$58,547; σ = US$19,987, which equates to US$98,522). The final measure ranges from +1 to −1, with +1 indicating that all families in a given neighborhood are affluent, a value of −1 indicates all families are poor, and a value of 0 indicates an equal balance between affluent and poor families (Kubrin & Stewart, 2006; Massey, 2001; Morenoff et al., 2001).

Independent Variable Controls

Demographics

We include gender, race and ethnicity, and age at completion of the community-based service as demographic controls. Gender was measured as female (= 0) and male (= 1), while race–ethnicity is measured using a set of dichotomous variables, with 1 = Black, 1 = Hispanic, 1 = “other” (reference group = White). Age at completion was measured continuously.

Individual-level risk factors

Substance use

Substance use is categorized into no past use (= 0), past use (= 1), and abuse (= 2) where such use caused problems in family conflict, health, pro-social peer associations, withdrawal, increased tolerance to drugs/alcohol, or contributed to criminal behavior.

Antisocial peer association

Antisocial peer association was assessed using a self-report measure of the youth’s friendship network (= 1 if youth reported having antisocial peers or associating with gang members, else = 0).

Age at first offense

Age at first offense includes 12 and under, 13 to 14, 15, 16, and over 16 (coded 1–5, respectively). These categories are those captured by the C-PACT, with higher values signifying the youth was older when first arrested.

Worst prior offense

We measure the most serious offense a youth has ever been adjudicated for as separate dichotomous measures of felony other (= 1), property felony (= 1), and violent felony (= 1), with misdemeanor serving as the reference category.

History of residential placement

This measure categorizes whether the youth has had a history of residential commitment placement with the FDJJ (no history = 1, one residential placement = 2, and two or more residential placements = 3).⁸

Service type

The type of community-based service the juvenile completed preceding the 365-day follow-up was grouped into five types: diversion services, probation supervision, day treatment, redirections, and aftercare services. Each service type was dichotomous (e.g., probation = 1), with the least restrictive service, diversion, serving as the reference group.⁹

Table 1 presents the descriptive statistics for the variables included in our analysis. The 1-year recidivism rate is 41%, with an average number of days to failure of 276. The diverse sample had an average age at service completion of 17, was 23% female, 47% Black, and 15% Hispanic. Probation supervision (37%) was the most commonly completed community-based service completed prior to tracking recidivism, followed by aftercare, diversion, day treatment, and finally redirection. Seventy-nine percent of the youth had a prior felony (with 47% being a violent felony). Before examining the multivariate relationships present in our analysis, it was necessary to assess the potential for collinearity to impact our results. Bivariate analyses, available upon request, indicate that the vast majority of the correlations between variables included in the models shown below were under .75, with only a single correlation approaching .80, suggesting that collinearity does not pose a large risk to the results obtained.

Table 1.

Descriptive Statistics for the Analysis of Temperament and Community Context.

Dependent Variables	Mean	SD
Time to failure	276.18	124.73
Rearrest	0.410	0.492
Temperament measures
Effortful control index	0.000	0.680
Negative emotionality index	0.000	0.684
Community measures (n = 746)
Resource deprivation index	0.000	0.732
Concentrated disadvantage index	0.000	0.770
Concentrated affluence index	0.109	0.194
Demographics
Age	17.004	1.675
Gender	0.769	0.421
Black	0.467	0.499
Hispanic	0.152	0.359
Other race	0.005	0.071
Individual-level risk factors
Substance abuse	0.892	0.748
Antisocial peers	0.472	0.499
Age at first offense	2.189	1.119
Prior misdemeanor	0.208	0.406
Prior other felony	0.041	0.198
Prior property felony	0.281	0.450
Prior violent felony	0.470	0.499
Residential placement history	1.367	0.590
Diversion	0.227	0.419
Probation	0.371	0.483
Redirection	0.048	0.214
Day treatment	0.094	0.292
Aftercare	0.260	0.438

Note. N = 26,960.

Analytic Strategy

We utilize a three-step strategy to examine the interplay between community context and youth temperament. First we employ multilevel logistic regression to examine the effects of EC, NE, and community disadvantage on juvenile recidivism. These models are an extension of traditional regression models that account for the nested nature of the data across multiple levels of analysis (i.e., youth nested in communities from across the state; Raudenbush & Bryk, 2002).

Second, we assess whether youth with the lowest levels of EC and highest levels of NE are more likely live in communities characterized by high levels of economic disadvantage. This was done by taking youth who score in the top 10% of the sample of NE and the bottom 10% of EC and comparing the communities in which they live using simple descriptive and bivariate statistics.

Finally, we examine whether the concurrence of these three adverse factors (high emotionality, low EC, and living in a disadvantaged community) remains predictive of juvenile recidivism once a wide range of individual-level risk factors are considered. This was done by creating a final indicator where those highest NE/lowest EC youth who also lived in a neighborhood characterized by a high degree of disadvantage (i.e., 1 SD above the mean, = 1) were compared to all other youth (= 0). For this portion of the analysis, we use a series of Cox proportional hazard models. This final specification assesses the potential repercussions of a confluence of temperament issues and community disadvantage on the probability of reoffending and time-to-failure.

Results

The Effect of Temperament and Community Context on Juvenile Recidivism

Our first objective was to assess the effect of EC, NE, and community context on juvenile recidivism while controlling for relevant demographic characteristics and several well-known risk factors. The multilevel logistic regression results of these analyses are shown in Table 2. Model 1 of Table 2 displays the main effects of NE and EC net of a wide range of demographic and individual characteristics. Models 2–4 examine the effect of each community-level construct considered in the analysis independently (resource deprivation, concentrated disadvantage, and concentrated affluence), while also including the temperament constructs and all demographic and individual-level risk factors.¹⁰ Note: As we present odds ratios (ORs), values less than 1.0 indicate a negative or preventative effect on recidivism, while values greater than 1.0 indicate a positive or adverse effect of the variable on recidivism.

Table 2.

Hierarchical Logistic Regression of Temperament and Community Context.

	(1)	(2)	(3)	(4)
	OR (SE)	OR (SE)	OR (SE)	OR (SE)
Temperament measures
Effortful control index	0.879** (.021)	0.879** (.021)	0.879** (.021)	0.879** (.021)
Negative emotionality index	1.109** (.027)	1.109** (.027)	1.109** (.027)	1.109** (.027)
Community context
Resource deprivation index	—	1.166** (.027)	—	—
Concentrated disadvantage index	—	—	1.161** (.027)	—
Concentrated affluence index	—	—	—	0.559** (.057)
Demographics
Age	0.978* (.011)	0.978* (.011)	0.978* (.011)	0.978* (.011)
Gender	1.952** (.068)	1.951** (.068)	1.951** (.068)	1.951** (.068)
Black	1.500** (.055)	1.500** (.055)	1.500** (.055)	1.500** (.055)
Hispanic	1.124* (.052)	1.125* (.052)	1.125* (.052)	1.125* (.052)
Other race	0.955 (.190)	0.953 (.190)	0.953 (.190)	0.954 (.190)
Individual-level risk factors
Substance abuse	1.261** (.026)	1.261** (.026)	1.261** (.026)	1.261** (.026)
Antisocial peers	1.058* (.030)	1.058* (.030)	1.058* (.030)	1.058* (.030)
Age at first offense	0.871** (.013)	0.871** (.013)	0.871** (.013)	.871** (.013)
Prior other felony	1.092 (.081)	1.091 (.081)	1.091 (.081)	1.091 (.081)
Prior property felony	1.162** (.049)	1.161** (.049)	1.161** (.049)	1.162** (.049)
Prior violent felony	1.078 (.044)	1.077 (.044)	1.077 (.044)	1.077 (.044)
Residential placement history	1.300** (.051)	1.299** (.051)	1.299** (.051)	1.299** (.051)
Probation	1.486** (.068)	1.487** (.068)	1.487** (.068)	1.487** (.068)
Redirection	2.064** (.143)	2.064** (.143)	2.064** (.143)	2.064** (.143)
Day treatment	1.983** (.114)	1.982** (.114)	1.982** (.114)	1.983** (.114)
Aftercare	1.721** (.110)	1.720** (.110)	1.721** (.110)	1.721** (.110)

Note. N = 26,960. Multilevel logistic regression model showing ORs with SDs in parentheses. All Level-1 variables have been group mean centered. OR = odds ratio; SE = standard errors.

*p < .05; **p < .01.

Model 1 suggests that youth with lower levels of EC and higher levels of NE are more likely to reoffend, net of all other characteristics considered. In addition to the temperament constructs, males as well as Black and Hispanic youth were more likely to recidivate. Additionally, youth reporting greater substance use issues, those who were younger at the time of their first arrest, those who have prior property felony offenses, residential placement history, or antisocial peers, and anyone not completing diversion programs are also more likely to recidivate.

Models 2–4 assess the impact of disadvantage on the likelihood of recidivism using three distinct measures of the socioeconomic conditions present in the communities in which youth reside. ORs greater than 1.0 shown in Models 2 and 3 indicate that youth living in communities characterized by high levels of disadvantage, measured by either the resource deprivation or concentrated disadvantage, are significantly more likely to recidivate, even after controlling for the multitude of well-known individual-level risk factors. The results shown in Model 4 are also consistent with theory and previous research on concentrated affluence. Youth who reside in areas with higher concentrations of affluence, as measured by the ICE index, are significantly less likely to reoffend following their completion of FDJJ services. Across Models 2–4 it is important to note as well that controlling for community context does little to alter the significance or strength associated with the EC and NE indexes. Results shown in Table 2 underscore the importance of considering both temperament and community context in the examination juvenile recidivism.

The Combination of Difficult Temperament and Adverse Contexts on Juvenile Recidivism

Next, we assessed for the potential for all three of these adverse risk factors to coalesce and have a significant effect on the time to failure following release from FDJJ programming. First we examined the extent to which youth with difficult temperaments (high NE and low EC) reside in communities characterized by high levels of disadvantage. Out of the 1,016 youth who fell in the bottom 10% of EC and the top 10% on NE, 31% lived in highly disadvantaged neighborhoods (defined as 1 SD above the mean in resource deprivation).¹¹ This suggests a moderate degree of overlap between temperament problems and levels of disadvantage and raises the question, how do youth with difficult temperaments who reside in bad neighborhoods fare with respect to recidivism once released from FDJJ programming? To answer that question, we turn to our final set of analyses presented in Table 3.

Table 3.

Multivariate Analysis of DeLisi and Vaughn’s Temperament-Based Theory of Antisocial Behavior and Juvenile Recidivism, Examination of the Confluence of Low Control, High Emotionality, and Community Disadvantage.

	(1)	(2)	(3)
	OR (SE)	OR (SE)	OR (SE)
	Resource Deprivation	Concentrated Disadvantage	Limited Affluence
Low control/high emotionality and disadvantaged community context	1.155* (.075)	1.185* (.097)	1.278** (.111)
Demographics
Age	0.977** (.007)	0.977** (.007)	0.978** (.007)
Gender	1.667** (.045)	1.668** (.045)	1.670** (.045)
Black	1.374** (.030)	1.374** (.030)	1.374** (.030)
Hispanic	1.138** (.034)	1.137** (.034)	1.137** (.034)
Other race	0.990 (.153)	0.990 (.154)	0.991 (.154)
Individual-level risk factors
Substance abuse	1.208** (.017)	1.208** (.017)	1.208** (.017)
Antisocial peers	1.089** (.021)	1.089** (.021)	1.089** (.021)
Age at first offense	0.892** (.010)	0.892** (.010)	0.892** (.010)
Prior other felony	1.064 (.059)	1.064 (.059)	1.064 (.059)
Prior property felony	1.140** (.036)	1.140** (.036)	1.140** (.036)
Prior violent felony	1.116** (.034)	1.116** (.034)	1.116** (.034)
Residential placement history	1.151** (.029)	1.151** (.029)	1.151** (.029)
Probation	1.426** (.050)	1.425** (.049)	1.425** (.049)
Redirection	1.886** (.093)	1.885** (.093)	1.884** (.093)
Day treatment	1.838** (.075)	1.838** (.075)	1.838** (.074)
Aftercare	1.622** (.074)	1.621** (.074)	1.621** (.074)

Note. N = 29,690. Cox survival regression models showing ORs with clustered SEs in parentheses.ORs = odds ratio; SE = standard errors.

*p < .05; **p < .01.

Table 3 presents the results of a series of Cox proportional hazard models in which the effect of high NE, low EC, and residing in a community characterized by a high degree of disadvantage is examined.¹² Here, youth who fell in the tails of the distribution in EC and NE and live in areas with high levels of disadvantage (= 1) are compared to all other youth (= 0). Results shown in Models 1–3 confirm that youth who face this combination of temperament problems and live in a disadvantaged area are significantly more likely to be rearrested sooner than other youth, even after controlling for a wide range of prominent individual-level risk factors.¹³ Although this sample of youth is relatively small (between 273 and 317) in comparison to the full sample, the finding that this relationship between temperament and community context holds is substantively meaningful and has important theoretical and policy implications, which are discussed below.¹⁴

Discussion

DeLisi and Vaughn (2014) articulated a temperament-based theory of antisocial behavior and interactions with the justice system predicted on EC and NE. Within that theoretical framework they hypothesized that individuals with low EC and high NE would be most at risk for deviant and delinquent behavior, and furthermore that these deficits would be more deleterious in disadvantaged contexts. The current study examined these suppositions, finding support for faster time to rearrest for juveniles with low EC, high NE, and for those living in disadvantaged (and less affluent) communities. Furthermore, youth evidencing all three risks (low EC, high NE, and disadvantaged context) recidivate sooner when compared to youth without all three risks. These results held across three separate measures of context (to examine sensitivity), and even when controlling for demographic measures, many prominent risk factors, and the community-based service the juvenile completed.

The key finding that youth who possess poor impulse control, negatively experience others and their environment, and live in disadvantaged communities reoffend faster not only provides support for a central hypothesis of temperament theory but also has implications for correctional policy. Prominent juvenile justice paradigms (such as risk, needs, and responsivity) advocate targeting limited resources to the highest risk youth. Assessment tools, such as the C-PACT used in the current study, capture items relevant to EC and NE, and the home address of justice-involved youth is known to the system and its actors. Using this information, juvenile justice systems can focus increased services on interventions targeted to youth assessed as having all three risk factors. Furthermore, selective prevention efforts can be targeted to youth not involved in the justice system, but at increased risk of future involvement based on low EC, high NE, and community context. Crime concentration and criminology of place research has repeatedly demonstrated a small segment of populations, even in toxic contexts, commit the majority of offenses (Braga, Papachristos, & Hureau, 2014; Sherman, Gartin, & Buerger, 1989; Weisburd, 2015). Assessing temperament may be important in targeting prevention/intervention to those individuals. There is evidence that interventions can improve self-regulation and self-control (Evans-Chase, 2014; Piquero, Jennings, & Farrington, 2010) and address mismanagement of anger and aggression (Goldstein et al., 1986; Holmqvist, Hill, & Lang, 2009; Reddy & Goldstein, 2001) to reduce delinquency. Additionally, prevention and intervention models that are holistic and involve standardized skills and socialization training for the at-risk youth’s parents, as well as the youth (such as multisystemic therapy, functional family therapy, and nurse–family partnerships), have been among the most successful and/or prolific, generally (Baglivio, Jackowski, Greenwald, & Wolff, 2014; Crimesolutions .gov , 2014; Olds, 2007; Sawyer & Borduin, 2011; Sexton & Turner, 2010). Thus, not only should policy responses to juvenile offending be informed by “the malleable aspects of psychosocial functioning in a developmentally informed manner” (Monahan, Steinberg, & Piquero, 2015, p. 1), but future empirical research should examine the efficacy of these, and other, models with youth exposed to the triple threat of low EC, high NE, and disadvantaged contexts.

The current study is not without limitation. The items from the C-PACT used to construct the EC and NE measures were not specifically designed for that purpose, though we believe they map nicely into the EC and NE concepts as laid out by DeLisi and Vaughn (2014). Additionally, we lack a measure of self-reported delinquency, relying instead on official arrests. However, our key outcome measure (rearrest) is a central outcome for juvenile justice systems as well as the funders of services, namely the Florida legislature, that comprise the data we employed herein. Furthermore, prior work has indicated congruence in that the probability of a self-reported offense leading to a conviction is highest at age 15–18 (Farrington, Ttofi, Crago, & Coid, 2014), providing some assurance that our findings would be comparable to those using self-report. We were limited to the use of youth assessed with the Full C-PACT, removing those only assessed with the C-PACT prescreen. While this increased the proportion of youth that were assessed as moderate–high or high risk, it limits generalizability of our findings to all juvenile offenders in Florida. As well, although the effects we observe are in line with temperament theory, it is important to recognize that large sample sizes, such as the sample used here, tend to bias results in favor of finding significant effects. Accordingly, we urge others to replicate our work using different samples and different measures in order to assess the robustness of the results described here. Finally, we limited inclusion to zip codes where 10 or more juvenile offenders completed a community-based service during the 3-year study period. We were fortunate however in that the current study did not suffer from problems associated with limited variation across homogeneous contexts found in prior work (Tillyer & Vose, 2011). Future research should extend our initial analyses to examine recidivism of low EC and high NE youth across contexts, using different contextual measures (e.g., collective efficacy, informal social control), and across different levels of measurement, such as census tracts or street segments.

This study found strong support for the explicit EC and NE constructs of DeLisi and Vaughn’s (2014) temperament-based theory. Furthermore, we provided additional evidence that context matters in the recidivism of already delinquent youth, adding to the scant literature in that arena. Youth having the tripartite confluence of risk factors of low EC, high NE, and residence in a disadvantaged context reoffend faster than other juvenile offenders. It is our hope that more criminologists will take notice of temperament as an explanation of antisociality and crime and continue to examine the ways in which context affects offending.

Footnotes

Declaration of Conflicting Interests

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

Funding

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

Notes

References

Abrams

L. S.

Snyder

S. M.

(2010). Youth offender reentry: Models for intervention and directions for future inquiry. Children and Youth Services Review, 32, 1787–1795.

Agnew

(1992). Foundation for a general strain theory of crime and delinquency. Criminology, 30, 47–87.

Andrews

D. A.

Bonta

(2003). The psychology of criminal conduct (3rd ed.). Cincinnati, OH: Anderson.

Baglivio

M. T.

Jackowski

Greenwald

M. A.

Wolff

K. T.

(2014). Comparison of multisystemic therapy and functional family therapy effectiveness: A multiyear statewide propensity score matching analysis of juvenile offenders. Criminal Justice and Behavior, 41, 1033–1056.

Baumer

E. P.

Messner

S. F.

Felson

(1998). Influence of crack cocaine on robbery, burglary, and homicide rates: A cross-city, longitudinal analysis. Journal of Research in Crime and Delinquency, 33, 316–340.

Braga

A. A.

Papachristos

A. V.

Hureau

D. M.

(2014). The effects of hot spots policing on crime: An updated systematic review and meta-analysis. Justice Quarterly, 31, 633–663.

Brooks-Gunn

Duncan

Kato

Sealand

(1993). Do neighborhoods influence child and adolescent behavior? American Journal of Sociology, 99, 353–395.

Carmines

E. G.

Zeller

R. A.

(1982). Reliability and validity assessment. Beverly Hills, CA: Sage.

Clark

L. A.

(2005). Temperament as a unifying basis for personality and psychopathology. Journal of Abnormal Psychology, 114, 505–521.

10.

Clark

L. A.

Watson

Mineka

(1994). Temperament, personality, and the mood and anxiety disorders. Journal of Abnormal Psychology, 103, 103–116.

11.

CrimeSolutions.gov. (2014). All programs & practices. Office of Justice Programs, National Institute of Justice. Retrieved from https://www.crimesolutions.gov/programs.aspx

12.

DeLisi

Vaughn

M. G.

(2014). Foundation for a temperament-based theory of antisocial behavior and criminal justice system involvement. Journal of Criminal Justice, 42, 10–25.

13.

DeLisi

Vaughn

M. G.

(2015). Ingredients for criminality require genes, temperament, and psychopathic personality. Journal of Criminal Justice. doi: 10.1016/j.jcrimjus.2015.05.005

14.

Eisenberg

Cumberland

Spinrad

T. L.

Fabes

R. A.

Shepard

S. A

Reiser

… Guthrie

I. K.

(2001). The relations of regulation and emotionality to children’s externalizing and internalizing problem behavior. Child Development, 72, 1112–1134.

15.

Eisenberg

Sadovsky

Spinrad

T. L.

Fabes

R. A.

Losoya

S. H.

Valiente

… Shepard

S. A.

(2005). The relations of problem behavior status to children’s negative emotionality, effortful control, and impulsivity: Concurrent relations and prediction of change. Developmental Psychology, 41, 193–211.

16.

Eisenberg

Spinrad

T. L.

Eggum

N. D.

(2010). Emotion-related self-regulation and its relation to children’s maladjustment. Annual Review of Clinical Psychology, 6, 495–525.

17.

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, 389–426.

18.

Enders

C. K.

Tofighi

(2007). Centering predictor variables in cross-sectional multilevel models: A new look at an old issue. Psychological Methods, 12, 121–138.

19.

Evans-Chase

(2014). Addressing trauma and psychosocial development in juvenile justice-involved youth: A synthesis of the developmental neuroscience, juvenile justice and trauma literature. Laws, 3, 744–758.

20.

Evans-Chase

Zhou

(2014). A systematic review of the juvenile justice intervention literature: What is can (and cannot) tell us about what works with delinquent youth. Crime & Delinquency, 60, 451–470.

21.

Farrington

D. P.

Ttofi

M. M.

Crago

R. V.

Coid

J. W.

(2014). Prevalence, frequency, onset, desistance, and criminal career duration in self-reports compared with official records. Criminal Behavior and Mental Health, 24, 241–253.

22.

Goldstein

A. P.

Glick

Reiner

Zimmerman

Coultry

T. M.

Gold

(1986). Aggression replacement training: A comprehensive intervention for the acting out delinquent. Journal of Correctional Education, 37, 120–126.

23.

Gottfredson

M. R.

Hirschi

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

24.

Gordon

J. A.

Diehl

R. L.

Anderson

(2012). Does ADHD matter? Examining attention deficit and hyperactivity disorder on the likelihood of recidivism among detained youth. Journal of Offender Rehabilitation, 51, 497–518.

25.

Heatherton

T. F.

Baumeister

R. F.

(1996). Self-regulation failure: Past, present, and future. Psychological Inquiry, 7, 90–98.

26.

Holmqvist

Hill

Lang

(2009). Effects of aggression replacement training in young offender institutions. International Journal of Offender Therapy and Comparative Criminology, 53, 74–92.

27.

Kagan

Snidman

(2009). The long shadow of temperament. Cambridge, MA: Harvard University Press.

28.

Kochanska

Knaack

(2003). Effortful control as a personality characteristic of young children: Antecedents, correlates, and consequences. Journal of Personality, 71, 1087–1112.

29.

Kubrin

C. E.

Stewart

E. A.

(2006). Predicting who reoffends: The neglected role of neighborhood context in recidivism studies. Criminology, 44, 165–197.

30.

Land

K. C.

McCall

P. L.

Cohen

L. E.

(1990). Structural covariates of homicide rates: Are there any invariances across time and social space? American Journal of Sociology, 4, 922–963.

31.

Lengua

L. J.

West

Sandler

(1998). Temperament as a predictor of symptomology in children: Addressing contamination of measures. Child Development, 69, 164–181.

32.

Lynam

D. R.

Caspi

Moffitt

T. E.

Wikstr öm

P. O.

Loeber

Novak

(2000). The interaction between impulsivity and neighborhood context on offending: The effects of impulsivity are stronger in poorer neighborhoods. Journal of Abnormal Psychology, 109, 563–574.

33.

MacDonald

K. B.

(2008). Effortful control, explicit processing, and the regulation of human evolved predispositions. Psychological Review, 115, 1012–1031.

34.

Massey

D. S.

(2001). The prodigal paradigm returns: Ecology comes back to sociology. In Booth

Crouter

(Eds.), Does it take a village? Community effects on children, adolescents, and families (pp. 41–48). Mahwah, NJ: Lawrence Erlbaum.

35.

Moffitt

T. E.

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

36.

Moffitt

T. E.

Arseneault

Belsky

Dickson

Hancox

R. J.

Harrington

… Caspi

(2011). A gradient of childhood self-control predicts health, wealth, and public safety. PNAS, 108, 2693–2698.

37.

Monahan

Steinberg

Piquero

A. R.

(2015). Juvenile justice policy and practice: A developmental perspective. Crime and Justice: A Review of Research, 44, 577–619.

38.

Morenoff

J. D.

Sampson

R. J.

Raudenbush

S. W.

(2001). Neighborhood inequality, collective efficacy, and the spatial dynamics of urban violence. Criminology, 39, 517–559.

39.

Morizot

(2015). The contribution of temperament and personality traits to criminal and antisocial behavior development and desistance. In Morizot

Kazemian

(Eds.), The development of criminal and antisocial behavior (pp. 137–165). New York, NY: Springer.

40.

Olds

D. L.

(2007). Preventing crime with prenatal and infancy support of parents: The nurse-family partnership. Victims & Offenders, 2, 205–225.

41.

Piquero

A. R.

Gomez-Smith

Langton

(2004). Discerning unfairness where others may not: Low self-control and unfair sanction perceptions. Criminology, 42, 699–733.

42.

Piquero

A. R.

Jennings

Farrington

D. P.

(2010). On the malleability of self-control: Theoretical and policy implications regarding a general theory of crime. Justice Quarterly, 27, 803–834.

43.

Piquero

A. R.

Moffitt

T. E.

Lawton

(2005). The contribution of individual, familial, and neighborhood-level risk factors to life-course-persistent offending. In Hawkins

Kempf-Leonard

(Eds.), Our children, their children: Confronting racial and ethnic differences in American juvenile justice (pp. 202–244). Chicago, IL: University of Chicago Press.

44.

Pratt

T. C.

Cullen

F. T.

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

45.

Raudenbush

S. W.

Bryk

A. S.

(2002). Hierarchical linear models. Thousand Oaks, CA: Sage.

46.

Reddy

L. A.

Goldstein

A. P.

(2001). Aggression replacement training: A multimodal intervention for aggressive adolescents. Residential Treatment for Children & Youth, 18, 47–62.

47.

Roose

Bijttebier

Van der Oord

Claes

Lilienfeld

S. O.

(2013). Psychopathic traits in youth and associations with temperamental features. Journal of Individual Differences, 34, 1–7.

48.

Rothbart

M. K.

(1989). Temperament and development. In Kohnstamm

G. A.

Bates

J. E.

Rothbart

M. K.

(Eds.), Temperament in childhood (pp. 187–247). Chichester, England: John Wiley.

49.

Rothbart

M. K.

(2007). Temperament, development, and personality. Current Directions in Psychological Science, 16, 207–212.

50.

Rothbart

M. K.

(2011). Becoming who we are: Temperament and personality in development. New York, NY: Guilford Press.

51.

Rothbart

M. K.

Evans

D. E.

Ahadi

S. A.

(2000). Temperament and personality: Origins and outcomes. Journal of Personality and Social Psychology, 78, 122–135.

52.

Sampson

R. J.

Morenoff

J. D.

Gannon-Rowley

(2002). Assessing “neighborhood effects”: Social processes and new directions in research. Annual Review of Sociology, 28, 443–478.

53.

Sampson

R. J.

Raudenbush

S. W.

Earls

(1997). Neighborhoods and violent crime: A multilevel study of collective efficacy. Science, 277, 918–924.

54.

Sawyer

A. M.

Borduin

C. M.

(2011). Effects of multisystemic therapy through midlife: A 21.9 year follow-up to a randomized clinical trial with serious and violent juvenile offenders. Journal of Consulting and Clinical Psychology, 79, 643–652.

55.

Sexton

T. L.

Turner

C. W.

(2010). The effectiveness of functional family therapy for youth with behavioral problems in a community practice setting. Journal of Family Psychology, 3, 339–348.

56.

Sherman

L. W.

Gartin

P. R.

Buerger

M. E.

(1989). Hot spots of predatory crime: Routine activities and the criminology of place. Criminology, 27, 27–56.

57.

Stevenson

Goodman

(2001). Association between behavior at age 3 years and adult criminality. The British Journal of Psychiatry, 179, 197–202.

58.

Sulik

M. J.

Huerta

Zerr

A. A.

Eisenberg

Spinrad

T. L.

Valiente

… Taylor

H. B.

(2010). The factor structure of effortful control and measurement invariance across ethnicity and sex in a high-risk sample. Journal of Psychopathology and Behavioral Assessment, 32, 8–22.

59.

Tillyer

M. S.

Vose

(2011). Social ecology, individual risk, and recidivism: A multilevel examination of main and modeling influences. Journal of Criminal Justice, 39, 452–459.

60.

U.S. Census Bureau. (2014). American community survey 2008-2012 5-year summary file technical documentation. Washington, DC: U.S. Department of Commerce. Retrieved from http://www2.census.gov/acs20135yr/summaryfile/ACS2013SFTechDoc.pdf.

61.

Weisburd

(2015). The 2014 Sutherland address: The law of concentration and the criminology of place. Criminology, 53, 133–157.

62.

Wikström

P. O. H.

Loeber

(2000). Do disadvantaged neighborhoods cause well-adjusted children to become adolescent delinquents—A study of male juvenile serious offending, individual risk and protective factors, and neighborhood context. Criminology, 38, 1109–1142.

63.

Wilson

J. Q.

Herrnstein

R. J.

(1985). Crime and human nature: The definitive study of the causes of crime. New York, NY: The Free Press.

64.

Wolff

K. T.

Baglivio

M. T.

Intravia

Piquero

A. R.

(2015). The protective effect of immigrant concentration on juvenile recidivism: A statewide analysis of youth offenders. Journal of Criminal Justice. doi: https://dx-doi-org.web.bisu.edu.cn/10.1016/j.jcrimjus.2015.05.004

65.

Wright

K. A.

Kim

Chassin

Losoya

S. H.

Piquero

A. R.

(2014). Ecological context, concentrated disadvantage, and youth reoffending: Identifying social mechanisms in a sample of serious adolescent offenders. Journal of Youth and Adolescence, 43, 1781–1799.

66.

Wright

K. A.

Rodriguez

(2014). A closer look at the paradox: Examining immigration and youth reoffending in Arizona. Justice Quarterly, 31, 882–904.