Abstract
A large body of criminological research often focuses on risk factors that propel individuals toward criminal activity rather than those that act as protective factors to reduce criminal involvement. In this study, we focus on a potentially important protective factor, empathy, which has long been considered an individual characteristic related to prosocial human behavior including non-offending. Specifically, we test the effects of empathy on juvenile recidivism and evaluate how empathy interacts with gender as a protective factor among a large sample of adjudicated youth followed for 1-year post-release from a residential treatment facility. Results show that empathy was associated with lower recidivism among the entire sample of youth, with a stronger protective effect against recidivism for justice-involved female youth. Findings contribute to the gendered literature on responses to crime and suggest empathy should be included in programming and interventions for youth involved in the juvenile justice system, particularly for females.
Empathy has often been considered an important characteristic in establishing prosocial human interaction and behaviors and is defined by an individual’s ability to understand another’s feelings or emotional experiences; it is considered a multidimensional construct, with emotional and cognitive components (Watt & Panksepp, 2016). Specifically, the emotional (i.e., affective) component of empathy addresses one’s ability to respond to, and share in, another person’s emotional state; the cognitive component focuses on the cognitive awareness of another person’s feelings through simple associations or perspective-taking (Cohen & Strayer, 1996).
Psychologists have posited that low empathy typically results in aggressive, antisocial, and deviant behaviors (Miller & Eisenberg, 1988). More recently, criminologists have begun to view empathy as an important characteristic in understanding criminal behavior (Jolliffe & Farrington, 2004, 2007, 2021; Narvey et al., 2021; van Langen et al., 2014). Individuals who possess high levels of empathy have the ability to understand and share in another’s emotional state or context, suggesting empathy may act as a protective factor against a variety of antisocial behaviors, including the likelihood of reoffending (Vachon et al., 2014). Those who can comprehend and share another’s emotions may be inhibited from wanting to be the cause of any negative emotional reactions and will therefore be less inclined to engage in behavior that results in these consequences. Those who exhibit less empathy, however, might be left incapable of understanding the consequences of their own negative behavior and therefore perceive fewer risks associated with their actions. As such, they are “unburdened by the experience or knowledge of the emotional consequences of their actions on others” (Jolliffe & Farrington, 2021, p. 2). For this reason, empathy has been regarded as one of the most valuable human resources (Pinker, 2011). This explains why both cognitive (d = .43) and affective (d = .19) empathy are associated with antisocial behavior (van Langen et al., 2014).
Empathy has been considered a protective factor against crime and is considered a female-coded prosocial characteristic (O’Neill, 2020). Empathic ability also varies between and within persons and can be expressed differently across several characteristics, including gender. O’Neill (2020) found relations between gender, empathy, and delinquency that may vary between childhood and adolescence were a result of gender socialization practices. These socialization practices may lead boys to experience a period of “anti-empathic development” after age 10 as they become aware of gender-normative behavior (p. 428), resulting in decreased empathic ability, while girls experience greater personal distress (i.e., greater affective empathy) when others suffer, increasing their likelihood of depression. Evidence suggests that exercising the ability to predict how one’s actions will result in another’s negative emotional reaction disinclines the individual from engaging in offending (O’Neill, 2020) and that this ability differs significantly between males and females (Broidy et al., 2003). The current study aims to expand on recent research by Narvey and colleagues (2021) that began to explore the interrelationships between risk factors, empathy, and recidivism among a sample of justice-involved juveniles. Findings from their research suggest that empathy acts as a buffer between the negative effects of adverse childhood experiences (ACEs) and later recidivism. While their study found that empathy was malleable during residential treatment, the authors did not explore the extent to which empathy may differentially affect the likelihood of recidivism between males and females. The recent gendered literature in criminology calls for differentiation in programming between males and females, espousing the argument that there are meaningful qualitative differences between genders that include some unique pathways to offending and to reoffending (Hubbard & Matthews, 2008; Rettinger & Andrews, 2010). While the gendered pathways to initial offending have been explored, less research has evaluated gender-specific protective factors against reoffending. This is especially important given the increase in the rate of justice-involved females in the last several decades (Kajstrura, 2019).
Theoretical Background
One of the most widely used theories in the gendered literature is grounded in an interactionist framework, suggesting that offending is an action grounded in masculinity and therefore is incompatible with feminine traits (Messerschmidt, 2013; Steffensmeier & Allan, 1996). Interactional explanations of human behavior suggest that individuals attempt to take on imagined or perceived roles, thus adhering to the gender binary as part of this role taking and effort to pursue social acceptance and unity of self (Mead, 1934). Therefore, individuals will take on traits and behaviors that support their perspective of this imagined self (i.e., males use aggression to advance the idea of a masculinity). Empathy is a characteristic that aligns with girls’ socialization into care-oriented roles, pushing them to engage in behavior that advances emotional expression and perspective-taking (Giordano et al., 2007). This theory also helps to explain gender differences in types of crime. For instance, boys engage in more other-directed violence and behaviors that demonstrate masculinity (Sadeh et al., 2011); girls typically behave in more self-directed violence, demonstrating their socialization toward valuing care for others, which may help to explain offending that is more akin to defending valued relationships or survival crimes such as running away from home (Chesney-Lind & Pasko, 2013).
While there exists research linking empathy to offending, less work has focused on linking empathy to juvenile recidivism specifically. In this article, we add to the literature exploring the interrelationships between gender, empathy, and recidivism by further exploring gender differences in empathy in a sample of more than 11,000 juvenile offenders. Such an investigation is important because the (gendered) literature in criminology calls for a deeper exploration of the protective factors between genders which, in turn, could help identify unique individual characteristics worth specific (or more) attention within justice system settings to improve re-entry and reduce recidivism (Chesney-Lind, 1989; Covington & Bloom, 1999; Lowenkamp et al., 2009; Rettinger & Andrews, 2010). Doing so within a sample of justice-involved youth is also pertinent because they are the sample of interest to many policy makers and because recidivism rates for these individuals are high (Mulvey et al., 2004).
Empathy and Recidivism
Research has consistently highlighted the association between empathy and offending, demonstrating negative relations between empathy, aggression, and antisocial behaviors, where low empathy may encourage antisocial or aggressive behaviors as a result of one’s inability to understand the feelings of others (van Langen et al., 2014). Meta-analyses have reported that those with low levels of empathy were more likely to offend (see Jolliffe & Farrington, 2004, 2007) suggesting low empathy may contribute to chronic offending due to a poor emotional response to victims (Jolliffe & Farrington, 2007). In other words, empathy may play a role in habitual offending due to the emotional deficits experienced by individuals toward others, which may have implications for recidivism.
Only a handful of studies have investigated the association between empathy and recidivism among juvenile justice-involved samples (Bock & Hosser, 2014; Mulder et al., 2011; Narvey et al., 2021), and even less differentiate the effects of empathy between males and females. In a sample of adolescent and young adult justice-involved individuals, Bock and Hosser (2014) found that global empathy scores, empathic fantasy, and perspective-taking predicted recidivism, whereas those who scored lower in (cognitive) empathy were at an increased risk of reoffending. Low levels of empathy appear to share relations with habitual or high-frequency offending (Jolliffe & Farrington, 2007), suggesting empathy may be an important factor in explaining juvenile recidivism. In addition, Narvey and colleagues (2021) found that high empathy scores lessened the effects between ACEs and reoffending, identifying potential moderating effects of empathy among serious juvenile offenders and its role in re-offending. Furthermore, a lack of victim empathy has also been found to predict general recidivism among juvenile justice-involved males (Mulder et al., 2011). Jolliffe and Farrington (2007) hypothesize that low empathy may not have a significant effect on a single, isolated offense but rather reflects a characteristic that leads to repeat offending. As a result, it is hypothesized that empathy may act as a protective factor against recidivism.
Gender and Empathy
Empathic ability depends on various individual characteristics and complex interactions in how traits are expressed, that is, across age and between genders (Eisenberg et al., 2005). Males are more likely to commit more serious and/or violent crimes (Steffensmeier & Broidy, 2001), suggesting that they are either less likely to or less able to consider the impacts of their actions on their victims. This is supported by the gendered research on empathy, which has demonstrated that females report significantly higher empathy than males and that differences exist in the ways that empathy is expressed between genders (Broidy et al., 2003; Jolliffe & Farrington, 2007; O’Neill, 2020). Specifically, research finds that girls report significantly greater levels of empathy (particularly empathic concern) while it appears to be lower among boys (Van der Graaff et al., 2014). Van der Graaff and colleagues (2014) investigated the development of empathy among adolescents by gender. Findings revealed significant gender differences, where girls reported greater perspective-taking as well as higher levels of empathic concern that remained stable across adolescence compared to boys. Boys showed an initial decrease in empathic concern as they matured from early to mid-adolescence, suggesting gender differences exist that may vary over the life course and highlight unique relations between empathy and gender during adolescence, as well as implications on treatment and recidivism.
Findings provide some explanation for the gender differences in delinquent behaviors and identify the importance of empathy deficits that may differentially affect males and females, leading to serious offending and recidivism. Females may be less likely to commit crimes that result in serious and/or violent victimization because they more readily comprehend and think about the effects of their behavior on others (Steffensmeier & Allan, 1996). Much of these behaviors are reflective of the female socialization process that prioritizes caring for others and valuing relationships, which is intimately connected to empathic ability. In fact, research on the gendered pathways to recidivism has found that family support and interpersonal relationships act as a stronger protector against recidivism for girls than for boys (Taylor, 2015). Empathy, then, might operate as a better protective factor for females than males. In short, the interrelationships between empathy and recidivism might be further illuminated by the ways in which this characteristic is expressed by each gender. This is especially important when we consider gender-neutral or gender-specific programming and interventions designed to facilitate rehabilitation (Belisle & Salisbury, 2021; Lowenkamp et al., 2009).
Early arguments suggest that factors implicated in criminal activity are applicable across gender and are, therefore, sufficient in guiding correctional policy and treatment planning to reduce recidivism across genders (Lowenkamp et al., 2009). Others, however, contend that the failure to account for gender differences in the correlates of crime, whether they are differences in degree or differences in kind, undermines intervention efforts (Belisle & Salisbury, 2021; Hubbard & Matthews, 2008). This research is important given that females who engage in crime often do so for reasons that differ from males, such as a result of trauma or abuse and/or in defense of valued relationships, both reflective of females’ stronger emphasis on care and empathy (Chesney-Lind & Pasko, 2013). At the same time, less research has investigated the extent to which individual-level risk/protective factors may differentially relate to recidivism across gender. Therefore, exploring differences in empathy is critical to (a) further advance our understanding of protective factors, particularly among female offenders and (b) target specific needs between genders to appropriately respond and reduce recidivism.
Current Study
Although extant research has documented empathy differences between males and females, and that these differences are associated with differences in offending, much less research in this area has examined recidivism specifically, and even less has done so with a sample of juvenile offenders (Broidy et al., 2003; Chesney-Lind & Pasko, 2013). The present study extends the recent work of Narvey and colleagues (2021) who found that empathy increased for many during residential treatment, thus demonstrating the potential malleability of the characteristic (similar to what has been found with other individual characteristics, such as self-control; Piquero et al., 2010). We now aim to consider the extent to which empathy differentially relates to recidivism across gender. An assessment of the empathy–(re)offending association among juveniles who have committed an offense is critical as it may help to provide useful information to juvenile justice agencies and treatment providers when assigning individuals to intervention programming. Accordingly, this study adds to the extant literature by (a) exploring the role of empathy in reoffending among a statewide sample of youth returning to the community following a juvenile justice residential placement and (b) investigating whether the association between empathy and recidivism operates similarly for both male and female youth. This research will help contribute to the growing body of literature surrounding gender-based programming and rehabilitative interventions, specifically for female offenders who remain largely understudied.
Method
Sample
We leverage a sample of all youth who completed a juvenile justice long-term residential program in Florida during the three fiscal years of July 1, 2009, through June 30, 2012, to examine the relationship between empathy and recidivism across gender. Two data exclusions are noteworthy in that (a) the sample was limited to the first residential placement of each unique youth within the study period and (b) the analysis necessitated that each youth was assessed using the Residential Positive Achievement Change Tool (R-PACT) risk/need assessment used by the Florida Department of Juvenile Justice (FDJJ), which was implemented in 2009 meaning only new admissions during our study timeframe were included. Importantly, excluding the sample to only a youth’s first residential placement between July 1, 2009, and June 30, 2012, results in each youth being included only once. However, a youth may have experienced a residential placement prior to the study timeframe (before July 1, 2009). As such, analyses controlled for those prior placements (see Study Measures below). The two exclusions resulted in a final sample of 11,053 (14% female) unique youth who completed a FDJJ residential program between 2009 and 2012. As the vast majority of prior work examining the effects of empathy on juvenile recidivism examined males exclusively (Bock & Hosser, 2014; van der Helm et al., 2012), our sample of more than 11,000 serious juvenile offenders including more than 1,000 females offers us a unique opportunity to contribute to the knowledge base in this area.
To provide context, all FDJJ residential programs are operated by private providers contracted by the State, and all programs are “specialized” such that youth placed in a given program have similar overarching treatment needs (e.g., substance abuse treatment, comprehensive mental health needs, prior sexual offending behavior), with services individually tailored within the program. A disposition to a residential program can only come from juvenile court judges (who do not operate under the FDJJ) and is for indeterminate lengths of stay, based on the completion of individualized treatment and performance plans, which are developed from R-PACT risk/needs and clinical diagnoses/symptomology. All youth are required to attend school year-round (or until completion of a high school diploma/equivalent), and all youth are assigned a case manager and a therapist. FDJJ Administrative Code requires any youth with a mental health or substance abuse condition to receive treatment for those issues. 1 All residential programs provide individual counseling, family therapy, and treatment groups, provided by licensed (or supervised) mental health professionals, to every youth admitted based on their individualized assessed needs (as per the R-PACT and clinical assessments). Group therapy includes primarily cognitive–behavioral therapies (CBT), social skills training, substance abuse prevention or treatment (dependent on histories and clinical diagnoses), services related to sexual offending (if applicable), and anger management, discussion of healthy relationships, and trauma-specific treatment (as applicable). The services provided at each facility are dictated by contract (which specific brand name services are to be provided, such as Aggression Replacement Training, and number of required days/hours per week of each). Mental health groups are provided a minimum of 5 days per week across all programs in Florida. Service provision is monitored by FDJJ through both assigned contract monitors and a team of monitoring and quality improvement staff; deficiencies in service provision result in mandatory corrective action and contract actions as per FDJJ policy (Florida Department of Juvenile Justice, 2000).
Study Measures
Study measures are extracted from the FDJJ centralized information system, which captures complete demographic, offense, placement, and risk/need assessment information of all youth arrested within Florida. With the exception of demographic, placement, and recidivism measures (described below), all other measures come from the R-PACT assessment. Each youth’s R-PACT administered prior to release from the residential placement was used, as it captures the most accurate snapshot of the youth’s risk and protective factors at the beginning of the recidivism follow-up period. The R-PACT assesses items across the following domains: criminal history, education/vocation, use of free time, employment, relationships/peer associations, family, substance use, mental health, attitudes, aggression, and social skills. The validity of the R-PACT has demonstrated higher domain risk scores at release increase the odds of subsequent recidivism (e.g., Hay et al., 2018). 2
Dependent Variable
The study is concerned with recidivism, and its principal dependent variable was rearrest and is measured as a subsequent referral or adult arrest for a new-law offense committed within 365 days of the release date for a given participant. Both juvenile referrals and adult arrests are included as some individuals were (or turned) 18 years of age (the age of majority in Florida) during the 1-year tracking period. Importantly, rearrest only includes new law violations. While some youth who completed placement under the age of 18 may have been under post-release probation supervision (dictated by the juvenile judge at the disposition hearing where residential placement was determined), technical violations were not included in our measure of rearrest. 3 As shown in Table 1, 58.2% (SD = 0.493) of the sample was rearrested within 1 year of release.
Descriptive Statistics for the Analysis of Empathy, Sex, and Juvenile Recidivism
Independent Variable
Empathy is the central independent variable of the current study, created from four R-PACT items standardized and combined into an empathy index (α = .746). All four items are coded such that higher values indicate greater empathy, capturing specifically cognitive empathy (the ability to understand the emotions of others). First, the extent to which the youth has empathy, remorse, sympathy, or feelings for the victims is classified as youth according to whether the youth does not have empathy for victims, has some empathy for victims, or has empathy for victims (coded 1–3; see also Piquero, 2017). Second, respect for the property of others classifies youth as having no respect for the property of others, conditional respect for personal property of others, respects others’ personal property but not public property, and respecting the property of others (coded 1–4). Third, respect for authority figures distinguishes whether the youth defies or is hostile toward most authority, resents most authority, does not respect authority figures and may resent some, or respects most authority figures (coded 1–4). The fourth item assessed whether the youth accepts responsibility for his or her antisocial behavior as proud of antisocial behavior, accepts antisocial behavior as okay, minimizes or justifies antisocial behavior, or accepts responsibility for antisocial behavior (coded 1–4). Although positively skewed, the resulting measure varies substantially between youth, with males exhibiting less empathy, on average, than females (see Table 1). Of note, this standardized index of empathy has been used in prior work (Narvey et al., 2021).
Control Variables
Demographics
Demographic controls included biological sex (as assigned at birth; male = 1; 85.8% male), age at time of R-PACT assessment (average age = 16.4, SD = 1.31.), and dichotomous indicators of race/ethnicity (Black = 1; Hispanic = 1; with White representing the reference category). Notably, the current study follows FDJJ protocol such that ethnicity supersedes race where all Black and White youth are non-Hispanic, while Hispanic youth may be Black persons or White persons. Furthermore, FDJJ places youth in residential programs based on biological sex and all programs serve exclusively either female or male (biological sex) youth.
Criminal History Indicators
To control for the extent and severity of prior offending, criminal history indicators were included measuring age at first arrest (12 and under, 13–14, 15, 16, over 16, as per the R-PACT captured categories), prior felony arrests (none, 1, 2, or 3 or more), against person/violent felony arrests (none, 1–2, or 3 or more), prior sexual felony arrests (none, 1 or more), and prior long-term FDJJ residential placements (= 1 for youth with residential placements prior to the current one; 10.0% of youth having a prior residential placement).
Self-Control
Self-control was measured as a four-item standardized index, where all four items were coded such that higher values were indicative of greater self-control. Each of the following items was first standardized, and an index of self-control was created by summing the standardized scores (α = .787). A categorical measure of the ability to exhibit self-control distinguished whether the youth lacks techniques to use self-control, rarely uses such techniques, sometimes uses techniques, and often uses techniques for exhibiting self-control (coded 1–4). Second, an item assessed whether the youth believed he or she can avoid/stop engaging in antisocial behavior, somewhat believes antisocial behavior is controllable, or believes his or her antisocial behavior is out of his or her control (coded 1–3). Third, impulsivity was assessed as using self-control/usually thinking before acting, sometimes thinks before acting, impulsive, or highly impulsive/usually acts before thinking (coded 1–4, with higher values indicative of higher impulsivity). Finally, the ability to control impulsive behavior distinguishes between lacking techniques to control impulsive behavior, rarely using techniques to control impulsive behavior, sometimes using such techniques, or youth having no problems with impulsivity/often using techniques to control impulsive behavior (coded 0–3). An identical index of self-control is created from R-PACT items among FDJJ residential youth has been used in prior work (Wolff et al., 2020). Some readers will note that empathy and self-control are likely to occupy common-ground, which is the case in our study (r = .66) but not of sufficient association to warrant strong concerns over multicollinearity (which was assessed using variance inflation factors). Yet, given the importance of self-control for predicting criminal behavior, it is important to include it in our analysis. In this regard, if empathy is relevant over and above self-control (regardless of the specific self-control effect), then this would imply that empathy stands as a unique correlate of criminal behavior.
Gang Involvement
While the current sample is exclusively serious juvenile offenders in residential placement, a dichotomous measure of whether the youth has a history of gang membership/affiliation was included (= 1). The measure is self-reported affiliation yet corroborated with the FDJJ information system which includes alerts for any history of FDJJ suspected or law enforcement certified gang membership/association. An affirmative self-report was indicated as gang membership/association, and a denial in the youth’s self-report but a FDJJ information system alert was also included as a gang membership/association.
Past Drug Use/Abuse
The youth’s extent of prior drug use is classified as no past use, past use, or past use which has caused problems for the youth coded 0–2, with higher values indicating a greater extent/repercussion from drug use. Areas in which drug use could have caused problems for the youth included drug use disruptions in education (attendance and/or grades), family conflict, maintaining prosocial friends, health problems, contributing to criminal behavior, and increased tolerance and/or withdrawal problems. As shown (Table 1), 42.6% of the sample had prior drug use while 44.8% had use that caused problems for the youth.
Child Welfare Involvement
The extent of youth’s child welfare system placement history is included as a dichotomous indicator of 1 or more prior out-of-home child welfare placements (= 1, with 34% of the sample having at least one such prior placement, indicative of a high proportion of “crossover youth”).
Mental Health Problems
A dichotomous indicator of no history of mental health problems (= 0) versus those with mental health problems (= 1) was included. Mental health problems included schizophrenia, bipolar, mood, thought, personality, and adjustment disorders. Conduct disorder, oppositional defiant, attention-deficit disorder (ADD)/attention-deficit hyperactivity disorder (ADHD), and substance abuse disorders were excluded as per the R-PACT protocol. All mental health problems must have been confirmed by a professional qualified to do so (e.g., psychologist and licensed mental health counselor).
Runaway History
The youth’s history of running away was categorized as no history, 1 instance, or 2 or more instances, with higher values indicating a greater history of running away for at least 24 hr (coded 0–2, respectively).
Analytic Strategy
After providing a description of the sample, the relationship between empathy and recidivism (12-month rearrest) is explored through multivariable methods. First, we examine the bivariate correlations among all variables included in the analysis to assess the focal associations and to ensure that collinearity between our independent variables is not an issue. Subsequently, logistic regression was used to examine the independent effect of our empathy index on rearrest, independent of all other independent measures considered. We further explore whether the association between empathy and recidivism is conditional on the gender of the juvenile. This will be done by including an interaction between our empathy index and the indicator of gender in the model, thus making it possible to estimate the effect of empathy for both males and females independently while controlling for all other variables considered.
Importantly, recent methodological advances suggest that some caution is necessary when interpreting the significance of interactions in nonlinear regression models (Long & Mustillo, 2018). Following recent recommendations (Mize, 2019), we evaluate our results based on adjusted predictions rather than significance of the coefficient of the product term alone. More specifically, when examining the interaction between a categorical (e.g., gender) and a continuous (e.g., empathy) variable, we assess the substantive and statistical significance of this interaction by examining the “second differences” (i.e., second derivatives) of the marginal effects of the continuous variable (e.g., increases in empathy) across different values of the categorical variable (Long & Mustillo, 2018; Mize, 2019). This approach has been used in recent juvenile justice research examining the effect of race across urban and rural areas (Pupo & Zane, 2021) and allows for the results to be presented graphically so that any differences in the association between empathy and recidivism can be assessed with greater confidence.
Results
Table 2 displays the bivariate association between each of the measures included in the current study for both males and females separately. The results shown in Table 2 suggest that our measure of empathy was negatively and significantly related to being rearrested within a year of placement completion for both male and female youth (r = −.069 and r = −.133, respectively; p < .001). A number of demographic measures were also significantly associated with rearrest including, age at release, and race, with younger youth and Black youth being more likely to be rearrested than their older and White or Hispanic counterparts. Prior felony referrals were also positively related to recidivism among both male and female youth (r = .111 and r = .107, respectively; p < .001), while prior gang involvement was only a significant predictor for males (r = .041; p < .001). As could be anticipated, higher levels of self-control, as indicated by higher scores our index, were negatively associated with rearrest among males (r = −.076; p < .001) and females (r = −.131; p < .001).
Bivariate Assessment of Empathy and Juvenile Recidivism
p < .05. **p < .01. ***p < .001.
Table 3 displays the most central results of our study. There are two logistic regression models shown. In the first model, the association between empathy and recidivism is established among the full sample of youth, controlling for relevant confounding variables to account for their association with juvenile recidivism. The second model shown in Table 3 includes an interaction term, allowing for the estimation of gender-specific effects of empathy on rearrest. Looking at the first model, the odds ratio (OR) of <1.0 suggests that youth with higher levels of empathy upon release from residential placement are less likely to be rearrested within a year. More specifically, a one-unit increase in empathy upon release was associated with a 7.6% reduction in the odds or rearrest (OR = .924; 95% confidence interval [CI]: 0.860, 0.992) net of all other factors considered. Results also suggest, consistent with prior research, that males were more than twice as likely as females to be rearrested within the follow-up period (OR = 2.13; 95% CI: [1.88, 2.41]). Black persons (OR = 1.68; 95% CI: [1.54, 1.84]) and Hispanic persons (OR = 1.29; 95% CI: [1.12, 1.47]) were also more likely than White persons to recidivate as were those with a greater number of felony referrals (OR = 1.22; 95% CI: [1.17, 1.27]). Past substance abuse (OR = 1.19; 95% CI: [1.11, 1.26]) and a history of running away (OR = 1.12; 95% CI: [1.06, 1.18]) were also positively associated with rearrest following completion of a residential placement, while higher levels of self-control was protective against recidivism (OR = .891; 95% CI: [0.832, 0.954]).
Analysis of Empathy, Sex, and Juvenile Recidivism (n = 11,053)
Note. OR and logistic coefficients with 95% CIs shown. OR = odds ratio; CI = confidence interval.
p < .05. **p < .01. ***p < .001.
In the second model shown in Table 3, we assess the possibility that the association between empathy and recidivism varies between male and female youth. This was done by entering an interaction term between those two measures into the regression equation. To facilitate interpretation of the moderated effect of empathy on rearrest, we present predicted probabilities of rearrest for both male and female youth across different levels of empathy (low, average, and high), setting all other variables at their means (see Figure 1). Using the methods outlined by Mize (2019), we found that the association between empathy and rearrest was significantly stronger (more negative) among female youth than among male youth (second difference = 0.031, p = .022) which is consistent with the interpretation of the coefficient of the interaction term. Substantively, the interpretation here is that while empathy represents a significant protective factor for female youth returning home from residential placement, its effect is much less pronounced (less negative) among male youth. 4 In fact, one of the most important features of the results displayed in Figure 1 concerns the relationship between different levels of empathy and the predicted probability of recidivism among males and females. For example, at the lowest level of empathy (−3SD), the risk of recidivism for both males and females is at their high point—and quite similar to one another. Conversely, when empathy is at its highest level (+3SD), empathy has its most protective effect on recidivism for both males and females, but most importantly, the risk of recidivism among males is still double (.6) than what it is among females (.3). As can be seen, the risk of recidivism for males is virtually the same regardless of the level of empathy, while it is clear that among females, as empathy increases, the risk of recidivism decreases quite sharply. These findings and their implications for research, policy, and treatment are discussed in greater detail below.
Empathy and Probability of Rearrest for Males and Females
Discussion
Criminologists have long been interested in understanding the protective and risk factors that decrease or increase reoffending behaviors among justice-involved persons. This basic knowledge, of course, is critical for the development of prevention and intervention strategies aimed at reducing the persistence of criminal activity. It is also important for treatment programming to provide offenders with the requisite programming to aid in their re-entry into society. All of this is magnified among juvenile offenders, who are at a turning point in their criminal careers, with some desisting and others continuing (Piquero et al., 2012). Therefore, taking stock of these risk and protective factors is important; just as relevant is assessing the extent to which they are invariant across demographic groups, including gender, the focus of the current study. This study focused on one of these potential protective factors, empathy, its relation to re-offending in a sample of juvenile offenders, and whether its relationship with re-offending operated in a similar manner for males and females.
Two key findings emerged from our study. First, empathy emerged as a significant protective factor against recidivism—a finding that was observed even after extensive controls for a range of variables known to predict criminal activity, including previous offending and justice system placements, substance use, gang involvement, and self-control. Second, and most important, we detected a significant interaction between gender and empathy such that empathy emerged as much more strongly protective against recidivism for females than males. In fact, the difference in the predicted probability of recidivism among males remained twice as high as it did for females at the highest level of empathy. It is important to note that when we estimated separate models for males and females, we found that the relationship between empathy and offending was significant in the expected direction for females but not for males.
These findings underscore the need to include more empathy-based interventions as part of a more inclusive set of treatment initiatives, such as substance abuse programming or anger management. Treatment efforts with youth scoring high in psychopathy have demonstrated effectiveness in reducing violence and recidivism when targeting changes in callous/unemotional traits—characteristics that are closely related to empathy. Thus, including empathy-based targets in larger treatment programs might be effective and especially so for females. Treatment programming as well as risk and need assessments have remained male-focused, based largely on research derived from samples of boys and men (Chesney-Lind & Pasko, 2013; Rettinger & Andrews, 2010). Advocates for gender-responsive programming have cited research demonstrating that girls respond best to a strong therapeutic bond grounded in trust, respect, and empathy (Morgan & Patton, 2002). Hence, the valid assessment of gender-based needs is pertinent to the development and implementation of appropriate programming designed with reduced reoffending as a primary objective.
Only recently have justice system programs started to consider the role of empathy in reducing reoffending. The current most widely used empathy training programs involve providing justice-involved persons with a clear description of the known harmful effects of criminal behaviors on victims, having the accused write apology letters to the victim(s) and take responsibility for their actions, reading or recorded messages of victim impact statements, as well as practicing perspective-taking techniques. Fortunately, recent work has indicated that, on average, empathy improves during residential placement, although, unfortunately, the extent of improvements was not homogeneous across the sample (Narvey et al., 2021).
Findings from the current study, however, unpack some of these differences in empathy improvement, suggesting that they may differ across gender. This is in line with recent calls for gender-based programming for justice-involved persons. Research on empathy training programs has found that program effectiveness is significantly improved through tailoring to the individuals’ specific needs and individual differences (Mayer et al., 2018). Therefore, we recommend that programs should consider adopting more tailored empathy training practices across genders that could improve their empathic abilities and thus potentially decrease risks of recidivism. Gender-based programming may provide more effective responses and improve services that target behavior for females within the criminal justice system as they have some unique needs that are different from males (Bloom et al., 2003). For example, Bloom and Covington (1998) suggested that gender-specific treatment programs for females should consider context and environment. An environment must be supportive, where justice-involved women and girls can feel safe, connected to others within treatment, and empowered. Furthermore, the facilitator of these groups should model the desired behavior, including respect and empathy (Bloom & Covington, 1998).
Future work would significantly contribute to our knowledge of intervention efficacy in examining the extent to which specific treatment models enhance empathy and whether those improvements reduce subsequent reoffending. A prior random assignment study analysis of FDJJ’s restorative justice/victim impact intervention, Impact of Crime, demonstrated improvements in empathy, remorse, sympathy, and feelings for the victims of one’s antisocial behavior as well as skills in understanding the feelings of others, though primarily among males (Baglivio & Jackowski, 2015). Unfortunately, that study did not assess post-release recidivism. Nonetheless, this does point to the possibility that restorative justice programs foster empathic ability through perspective taking and learning the emotions of the victim. Empathy allows for emotional experiences grounded in social relationships through experience and interaction with others. Therefore, fostering relationships through empathic connection could also increase social bonds, thus offering further protective factors against recidivism. This might be especially true for females who exhibit stronger preferences for social connections and value intimacy and closeness more than males, thus potentially explaining, in part, why empathy seems to be more important for girls (Hall, 2011). While social bonds were not included in the current study, research has supported the positive effects of interpersonal relationships on reducing recidivism (Taylor, 2015).
To be sure, while we believe that our results are important to the knowledge base on empathy generally, and to the literature on gender differences in protective factors in particular, some limitations need to be highlighted. First, our recidivism data were officially based for a 1-year period. Although we do not think that the role of empathy would be sensitive to how crime data were collected—or for a longer duration of time—it remains to be seen if this is the case. In addition, data were not available for controlling for any post-release probation supervision for youth released prior to age 18 and who received such orders during their dispositional hearing. However, we mitigated some such effects as the rearrest measure did not include technical violations for supervised youth, which likely involve juvenile probation officer discretion. Second, we captured empathy in adolescence and as a result we do not know when and why empathy may have already shown differences by the time the youth entered the justice system. We suspect that, like self-control, empathy likely has its origins in socialization practices early in the life course and whether males and females are taught to be similarly empathic is an important consideration. It may be, that, much like Anderson’s street code thesis, males are taught to espouse tougher and less empathic attitudes in the public sphere, as empathy has been rooted in culture to some degree (Cassels et al., 2010). If so, then it is especially important for treatment and intervention programming to include empathy training and encourage perspective-taking. Furthermore, we note that the current study leveraged data on juveniles completing their residential placement disposition between July 1, 2009 and June 30, 2012. The extent to which more recent cohorts of youth completing placement differed in levels of empathy at release from such programs may led to different associations from those found in the current study.
In addition, individual-level information on specific services received within the facility, the duration/dosage of each service, nor the quality/fidelity of services provided were not available in the data. Such treatment components (fidelity, treatment quality, duration, dosage, whether services were appropriately matched to risk/needs assessment results) have been demonstrated as critical components in recidivism reduction among youth completing residential programs in Florida (Baglivio et al., 2021; Baglivio, Wolff, Howell, et al., 2018; Baglivio, Wolff, Jackowski, et al., 2018). However, we note that our purposive decision to use the R-PACT assessment just prior to release (therefore just prior to the beginning of the recidivism follow-up), by default controls for differences in services received during placement as that assessment is a snapshot of each youth’s risk/needs profile after services have been provided.
Finally, it is worth noting that our study measured primarily cognitive empathy (understanding another’s emotional state and taking their perspective). It is possible, although unlikely, that results may differ between those who exhibit high levels of cognitive versus affective empathy (sharing the emotions of others). However, individuals tend to share similar levels of both types of empathy (Ze et al., 2014). In addition, meta-analyses have highlighted that the relationship between cognitive empathy and criminal and/or delinquent behavior may not be as strong as affective empathy. For example, van Langen and colleagues (2014) found that only cognitive empathy was associated with offending, while emotional (i.e., affective) empathy did not significantly predict offending. Furthermore, Jolliffe and Farrington (2004) identified a stronger negative relationship between cognitive empathy and offending compared with affective empathy. Taken together, findings suggest cognitive empathy may play a greater role in offending and recidivism, which provides evidence in support of the need for offender-based treatment programs targeting self-conscious emotions and perspective taking.
Moving forward, we believe that empathy has a seat at the criminological table as well as a potentially even more important one in the prevention and intervention arena. Criminological research is ripe with research on the risk factors that potentially contribute to reoffending, but the discussion on the role of empathy is somewhat underwhelming. Rehabilitative programming, as well as other alternatives to incarceration such as restorative justice, would benefit from including empathic training and development in their offerings and might consider how to differently apply programming principles to boys versus girls, depending on specific risks and needs. The salience of empathy in the mitigation of delinquent behavior, as a protective factor, has been established, and we suspect that supporting its development is likely to lead to better interpersonal relationships and success in other life domains.
Footnotes
Authors’ Note:
We have no conflicts of interest to disclose. The analysis and conclusions presented here are those of the authors and should not be attributed to the Bureau of Justice Statistics or the U.S. Department of Justice.
