Abstract
One overlooked result in a 1989 Science paper on the “cycle of violence” was a race-specific increase in risk for arrest for violence among Black maltreated children, but not White maltreated children. We examine whether race differences in the cycle of violence are explained by risk factors traditionally associated with violence. Using a prospective design, maltreated and non-maltreated children were matched on age, sex, race, and approximate family social class and interviewed at mean age 28.7 years (N = 1196). Arrest histories were obtained through age 50.5. Regression analyses included maltreatment, race, self-reported violent behavior, and risk factors (e.g., family, school, neighborhood variables). For arrests for violent crime, race was a significant predictor, whereas childhood maltreatment was not significant. For violent arrests, there was a significant race × maltreatment interaction when the total number of risk factors were included controlling for self-reported violent behaviors. For self-reported violent behaviors, childhood maltreatment remained significant for some risk factors. However, race did
In 1989, Science published a paper on the “cycle of violence” that has become a seminal citation for the belief that physically abused children grow up to become perpetrators of violence (Widom, 1989b). Those findings showed that being abused or neglected as a child increases a person’s risk for delinquency, adult criminal behavior, and violent criminal behavior, although the majority of abused and neglected children did not become delinquent, criminal, or violent. Several longitudinal investigations in different parts of the United States have documented a relationship between childhood abuse and neglect and juvenile and/or young adult crime (English et al., 2001; Lansford et al., 2007; Mersky & Reynolds, 2007; Smith et al., 2005; Stouthamer-Loeber et al., 2001; Zingraff et al., 1993). A meta-analysis of prospective studies (Fitton et al., 2020) found support for the relationship between childhood maltreatment and subsequent antisocial and violent behavior. Other meta-analyses and systematic reviews have reported associations between maltreatment and antisocial or aggressive behaviors in childhood, adolescence, or adulthood (Carr et al., 2020; Maas et al., 2008). Basto-Pereira et al. (2022) extended these findings to include cross-cultural studies and found similar relationships of physical abuse, sexual abuse, and neglect to criminal behavior. However, to our knowledge, no study has examined the extent to which the cycle of violence applies equally to children of different race or ethnic backgrounds.
Despite extensive media and public policy makers’ attention, one of the surprising race-specific results of the original Science (Widom, 1989b) article has been largely overlooked. Those results showed that Black abused and neglected children, but not White maltreated children, were at increased risk for an arrest for violence. Why should there be race differences in the cycle of violence? Do other risk factors explain the different impact on this outcome for Black and White maltreated children? This paper explores these questions.
Numerous explanations have been offered to explain the cycle of violence, including traditional theories, such as social learning (Bandura, 1973) and attachment (Bowlby, 1973; Madigan et al., 2016; Papalia & Widom, 2024; Savage, 2014; Sroufe et al., 2005). However, other explanations, including changes in self-concept, attitudes, or attributions (Bagley & Ramsay, 1986; Herman, 1981), maladaptive styles of coping (Spaccarelli, 1994; Widom, 2000), deficits in social information processing (Dodge et al., 1990), lack of prosocial socialization (e.g., antisocial peer) (Maas et al., 2008), strain (Agnew & Brezina, 2019), and negative emotionality and temperament (DeLisi & Vaughn, 2014; Garofalo & Velotti, 2017) have also received attention in the literature. Given this extensive body of theoretical and empirical literature on the cycle of violence, an examination of the extent to which there are racial disparities is important and needs attention.
One would expect that the phenomenon of the cycle of violence would occur across maltreated children. But, child development and the consequences of child abuse and neglect do not occur in a vacuum. Other factors related to race likely play a key role in the racial differences observed in the original analyses (Bor et al., 2018; Braveman et al., 2022; Mallett, 2016). Within a sociopolitical framework, race is a proxy measure for social, environmental and structural -- not biological, factors (Boyd et al., 2020) and has important consequences for Black children’s involvement in the juvenile justice (Abrams et al., 2021; Crutchfield et al., 2009; Dragomir & Tadros, 2020) and child welfare (Dettlaff & Boyd, 2020; Maguire-Jack et al., 2020; Thomas et al., 2023) systems. Poverty, parental and neighborhood factors, and school discipline are potential causes for increased involvement in these systems. Furthermore, Black families and communities have higher governmental surveillance and therefore result in greater initial and subsequent contact with these systems (Baughman et al., 2021; Cénat et al., 2021; Homer & Fisher, 2020; Jahn et al., 2022; Shoub et al., 2020). Indeed, other factors related to racial bias and structural racism, including the school-to-prison pipeline and greater surveillance of Black families and communities, are likely be important in understanding the racial differences observed (Bor et al., 2018; Braveman et al., 2022; Mallett, 2016)
In a Department of Justice publication, Huizinga and colleagues (2007) used data from three community studies of delinquency to examine disproportionate minority contact (DMC) of adolescents in the juvenile justice system. They focused on factors that might affect DMC at the level of police contact or court referral and drew several conclusions from their results: (1) there was clear evidence of DMC, (2) the DMC was not explained by differences in the offending behavior of the different racial groups, and (3) DMC was substantially reduced by introducing a number of well-established risk factors for arrest. These well-established risk factors included family socioeconomic status (e.g., poverty), family structure (e.g., one or two parents), parenting (e.g., physical punishment), parent characteristics (e.g., parent education), youth characteristics (e.g., school performance) and neighborhood characteristics (e.g., neighbor poverty). Their overall conclusion was that the “effect of race/ethnicity on the chance of being contacted/referred is reduced but remains significant when both offending and risks are controlled”. These findings suggested that no single variable or combination of variables fully accounted for the disproportionate arrests and referrals for Black youth. More recently, Schleiden et al. (2020) found that Black young adults were arrested seven times more often than their White peers, despite controlling for contextual and behavioral factors, including neighborhood disadvantage, exposure to violence, and parent-child bond.
The results of these studies suggest that contextual factors, including individual and family characteristics, may partially explain the race differences in the cycle of violence findings. In this paper, we examine potential explanations for why childhood maltreatment might have a differential impact on the extent to which Black, but not White, maltreated children are at increased risk for arrest for violent crimes. We ask whether race differences in the cycle of violence can be explained or eliminated when risk factors traditionally associated with delinquency and crime are accounted for in statistical models. We have four research questions: 1. Does childhood maltreatment predict increased risk for being arrested for violent crimes and for self-reported violent behaviors? 2. Are there race differences in the cycle of violence? That is, are there race differences in the relationship between child maltreatment and arrests for violent crime and self-reported violent behaviors? 3. Are race differences in the cycle of violence explained by individual and/or cumulative risk factors? 4. Are race differences in the cycle of violence explained by self-reported violent behaviors?
Methods
Design
The methods and description of participants in this study draw heavily on earlier publications (Widom, 1989a; 1989b). This prospective cohort design study (Leventhal, 1982; Schulsinger et al., 1981) was initiated in 1986 with a large group of documented cases of childhood maltreatment (physical and sexual abuse and neglect; N = 908) and a comparison group of children matched on the basis of age, sex, race/ethnicity, and approximate family social class at the time of the childhood maltreatment (N = 667) (Widom, 1989a). Because of the matching procedure, the subjects are assumed to differ only in the risk factor; that is, having experienced childhood maltreatment. Since it is not possible to assign subjects randomly to groups, the assumption of equivalency for the groups is an approximation. The control group may also differ from the maltreated individuals on other variables nested with maltreatment, for example, parent history of psychiatric disorders or inherited characteristics. Characteristics of the design include: (1) court substantiated cases of child abuse and neglect (i.e., an unambiguous operationalization of maltreatment); (2) a prospective design; (3) a large sample; (4) a comparison group matched on age, sex, race and approximate social class background; and (5) assessment of the long-term consequences of maltreatment beyond childhood and adolescence into adulthood.
The rationale for identifying the maltreated group was that their cases were serious enough to come to the attention of the authorities. Only court-substantiated cases of child abuse and neglect were included here. Cases were drawn from the records of county juvenile and adult criminal courts in a metropolitan area in the Midwest during the years 1967 through 1971. To avoid potential problems with ambiguity in the direction of causality, and to ensure that temporal sequence was clear (that is, child abuse or neglect led to subsequent outcomes), maltreatment cases were restricted to those in which children were less than 12 years of age at the time of the abuse or neglect incident. Thus, these are cases of childhood abuse and/or neglect.
A critical element of the design involved the selection of a matched control group of children without documented histories of childhood maltreatment (N = 667). This matching was important because it is theoretically plausible that any relationship between childhood maltreatment and subsequent outcomes is confounded with or explained by social class differences (Bradley & Corwyn, 2002; Conroy et al., 2010; MacMillan et al., 2001; Widom, 1989b). The matching procedure used here is based on a broad definition of social class that includes neighborhoods in which children were reared and schools they attended. Similar procedures, with neighborhood school matches, have been used in studies of people with schizophrenia (Watt, 1972) to match approximately for social class. When random sampling is not possible, Shadish et al. (2002) recommend using neighborhood and hospital controls to match on variables that are related to outcomes. The control group establishes the base rates of health outcomes expected in a sample of adults from comparable circumstances who did not come to court attention in childhood as victims of maltreatment. Widom (1989a) provides greater description of the matching procedures.
Participants
The original sample (N = 1575) used administrative data from courts for child maltreatment- 908 maltreated and 667 demographically matched control children. Of the original sample of 1575 people, 1307 (83%) were located and 1196 were interviewed (76%) during 1989–1995. We conducted attrition analyses and found that there were no significant differences between the interviewed follow-up sample (N = 1196) and the original sample (N = 1575) in terms of demographic characteristics (male [p = .28]; white [p = .10]; poverty in childhood census tract [p = .44]; current age [p = .88]; death rates [p = .10] or group status (abuse or neglected vs. controls [p = .11]). Prior research has found no differences in mortality rates between the maltreated and control groups (White & Widom, 2003). Of the 1196 interviewed, 110 (9.2%) had substantiated cases of physical abuse, 96 (8.0%) sexual abuse, and 543 (45.4%) neglect.
The mean age of the sample at the time of the follow-up interview was 29.25 years (SD = 3.85, range = 19–37) and approximately half the sample is female (48.8%). To determine race and ethnicity, participants were shown a card with the names of racial and ethnic groups and asked to indicate which race or ethnic group best described them. For the present analysis, race was dichotomized into White, non-Hispanic and Black, non-Hispanic. Hispanic (n = 45), Native American, Pacific Islander, and other (n = 26) were excluded due to their small sample size. Race was thus coded as White (n = 735; 65.4%) and Black/African American (n = 389; 34.6%) and the analytic sample was 1124. Sample members completed an average of 11.47 (SD = 2.19) years of school.
Procedures
Study Timeline.
Note. Age = mean (standard deviation), range. Arrest records were searched for 1553 individuals. for the first phase of the study that involved the identification of cases, race was based on court records and information for four people was missing and there were 18 people who were not identified as White or Black. For the in-person interviews, the total sample was 1196. However, self-reports of race were collected during the first interview and 72 people self-identified as other than White or Black.
Measures and Variables
Child Maltreatment
Official reports of court substantiated cases of child abuse and/or neglect during the years 1967–1971 were used. Physical abuse cases included injuries such as bruises, welts, burns, abrasions, lacerations, wounds, cuts, and bone and skull fractures. Sexual abuse cases included felony sexual assault, fondling or touching, sodomy, incest, and rape. Neglect cases reflected a judgment that the parents’ deficiencies in childcare were beyond those found acceptable by community and professional standards at the time and represented extreme failure to provide adequate food, clothing, shelter, and medical attention to children. Child maltreatment is treated as a dichotomous variable to indicate cases (any physical or sexual abuse or neglect) and controls.
Arrests for violence
Whether a participant had been arrested for a violent crime was determined from criminal history checks conducted during 1987–1988, 1994, and 2013 (see Table 1). The arrest records searches included information from both the Federal Bureau of Investigation’s National Crime Information Center and state law enforcement records for the Midwestern state in which childhood maltreatment cases were initially ascertained. This information was combined to create a lifetime criminal history record for each participant through mean age 50.54 years (SD = 3.49). Juvenile and adult arrests for violent crimes included murder and attempted murder, manslaughter and involuntary manslaughter, reckless homicide, rape, sodomy, robbery and robbery with injury, assault, assault and battery, aggravated assault, and battery and battery with injury. A dichotomous (yes/no) variable indicates “any violent arrest” and a continuous variable represents the number of violent arrests.
Self-Reported Violent Behaviors
During the interview, participants were asked to self-report whether they had ever engaged in a variety of violent behaviors (Wolfgang & Weiner, 1989), including (1) hurt someone badly enough they required medical attention, (2) threatened someone because they wouldn’t give you money or something else, (3) used a weapon to threaten someone, (4) attacked someone with the purpose of killing them, (5) used physical force to get money or drugs, (6) forced someone to have sex with you, and (7) shot someone. A dichotomous (yes/no) variable was created to reflect any self-reported violent behaviors and a continuous variable represents the number of violent behaviors reported. Previous research has found that self-report measures of delinquency are valid (Jolliffe et al., 2003; Thornberry & Krohn, 2000) and moderately associated with official records (Piquero et al., 2014). Other research has reported few racial differences in self-reports of violent behavior (Maxfield et al., 2000; Sohoni et al., 2021).
Risk Factors
Information about individual and family risk factors was obtained during the 1989-1995 interview. For individual risk factors, participants were asked about their lives as a child before they were 15 years old. Questions included whether they had ever repeated a grade, whether they frequently got into trouble with the teacher or principal for misbehaving in school, and whether they had ever been expelled or suspended from school. Participants were also asked whether they had grown up in a single-parent household or had lived with two parents until 18 years of age, whether the family had received welfare when the person was a child, and the family size (large family was defined as 5 or more children). Participants were also asked about their parents’ level of education and occupation. A composite family SES variable (Nikulina et al., 2011; Schuck & Widom, 2005) was created that included the average of two social (i.e., maternal and paternal levels of education) and four material (family’s welfare receipt when the participant was a child, paternal and maternal employment, and growing up in a single-parent household vs. living with two parents until 18 years of age) indicators. The childhood family SES variables were standardized and averaged across the number of variables for which each participant had data.
Neighborhood characteristics were based on 1970 U.S. Census data using the participant’s address during 1967–1971 to be consistent with the person’s childhood family neighborhood and included six variables: percentage of families in the tract living on public assistance, below the poverty line, in single-parent homes, in the same house for at least 5 years, and in owner-occupied homes, and percentage of people with at least a college degree. The neighborhood SES variable represents the average of the z-scores of the six items based on census data (Nikulina et al., 2011).
A cumulative or total number of risk factors score was created to determine whether the combination of factors, rather than a single variable, has the greatest explanatory power. This total number of risk factors score was calculated by summing the number of risk factors, including (1) not living with both parents, (2) large family (5+ kids), (3) repeat a grade, (4) expelled or suspended from school, and (5) misbehave in school as well as family and neighborhood SES variables. Since neighborhood SES and family SES were continuous variables, we dichotomized them by splitting them into half at the mean. Scores in the lower half were counted as risk indicators (coded 1) and scores above the mean were not counted as risk factors (coded 0).
Statistical Analyses
Bivariate t-tests and chi-square tests were used to detect significant differences between the control and maltreated groups. Negative binomial regressions with continuous variables were used to examine predictors using the number of arrests for violent crimes and number of self-reported violent behaviors. In order to eliminate possible bias caused by missing data, we used multiple imputation with chained equation (Azur et al., 2011) to replace the missing values. Multiple imputed datasets were used and pooled (“averaged” or combined) results from those imputed datasets were reported. All analyses were carried out in R (version 4.4.0).
Results
Demographic Characteristics, and Risk Factors for Maltreated and Control Groups Overall and Separately for Black and White Individuals
Demographic Characteristics and Descriptive Statistics for the Maltreated and Control Groups Overall and White and Black Individuals Separately.
Note. Degrees of freedom for chi-square (χ2) tests = 1. The neighborhood SES variable represents the average of the z-scores of the six items from census information: percent receive public assistance, below poverty line, single parent, same house 5y+, owner-occupied house, college degree. A higher value means lower SES. Family SES represents the average of the z-scores of the five items including family receiving welfare, mother/father’s year of school and type of jobs. Higher value means lower SES. The total number of risk factors was calculated by summing the number of risk factors, including (1) not living with both parents, (2) large family (5+ kids), (3) repeat a grade, (4) expelled or suspended from school, and (5) misbehave in school as well as family and neighborhood SES variables. To create dichotomous indicators of neighborhood SES and family SES to be used to calculate the total number of risk factors, the composite scores were split into half at the mean and scores in the lower half were counted as risk indicators (1) and scores above the mean were coded as zero.
Table 2 also shows that both Black and White maltreated children were less likely to be living with both parents, more likely to report having trouble in school (i.e., repeating a grade, being expelled or suspended from school, and misbehaving in school) and more likely to come from families with lower SES compared to Black and White controls, respectively. Compared to White controls, White maltreated children lived in lower SES neighborhoods and had larger families, whereas Black maltreated children and controls did not differ in terms of neighborhood SES or growing up in a large family.
Childhood Maltreatment and Violence Overall and for Black and White Individuals Separately
Descriptive Statistics for Arrests for Violence and Self-Reported Violent Behaviors for Maltreated and Control Groups Overall and White and Black Individuals Separately.
Note. Degrees of freedom for chi-square (χ2) tests = 1. Self-reported violent behaviors include the total number of violent behaviors before and after age 18.
For any self-reported violent behavior and the number of self-reported violent behaviors, the pattern of results was reversed. Significantly more White individuals with histories of childhood maltreatment reported violent behavior (33.7%), compared to the White controls (20.4%). In contrast, for the Black youth, the maltreated and control groups did not differ in the percent who reported engaging in violent behaviors. Similarly, White maltreated individuals reported significantly more violent behaviors compared to White controls, whereas Black maltreated individuals did not differ from Black controls in the number of self-reported violent behaviors.
Childhood Maltreatment and Violence Controlling for Race and Other Risk Factors
Negative Binomial Regressions Predicting Arrests for Violent Crimes and Self-Reported Violent Behaviors With Each of the Risk Factors Traditionally Associated With Crime and Violence.
Notes: SE = Standard error. Multiple imputation with chained equation was used for imputed datasets. Self-reported violent behaviors include the total number of violent behaviors before and after age 18. The neighborhood SES variable represents the average of the z-scores of the six items from census information: percent receive public assistance, below poverty line, single parent, same house 5y+, owner-occupied house, college degree. A higher value means lower SES. Family SES represents the average of the z-scores of the five items including family receiving welfare, mother/father’s year of school and type of jobs. A higher value means lower SES. The total number of risk factors was calculated by summing the number of risk factors, including (1) not living with both parents, (2) large family (5+ kids), (3) repeat a grade, (4) expelled or suspended from school, and (5) misbehave in school as well as family and neighborhood SES variables. To create dichotomous indicators of neighborhood SES and family SES to be used in calculating the total number of risk factors, the composite scores were split into half at the mean and scores in the lower half were counted as risk indicators (1) and scores above the mean were coded as 0. R2 was computed using the McFadden pseudo-R2.
Childhood Maltreatment and Violence, Race, and Self-Reported Offending
Negative Binomial Regressions Predicting Arrests for Violent Crimes, Controlling for the Total Number of Risk Factors and Self-Reported Violent Behaviors.
Notes: SE = Standard error. Multiple imputation with chained equation was used for imputed datasets. Self-reported violent behaviors include the total number of violent behaviors before and after age 18. The total number of risk factors was calculated by summing the number of risk factors, including (1) not living with both parents, (2) large family (5+ kids), (3) repeat a grade, (4) expelled or suspended from school, and (5) misbehave in school as well as family and neighborhood SES variables. To create dichotomous indicators of neighborhood SES and family SES to be used to calculate the total number of risk factors, the composite scores were split into half at the mean and scores in the lower half were counted as risk indicators (1) and scores above the mean were coded as 0. R2 was computed using the McFadden pseudo-R2.
Results of regression predicting arrests for violent crime. Note: This bar graph shows the results of binomial regressions predicting arrests for violent crimes in a model including maltreatment, race, the total number of risk factors, maltreatment × race interaction, and self-reported violent behaviors. The Y axis shows the number of arrests for violence. Results are expressed as means and standard errors. The maltreated group is represented with hatched bars and the controls are represented by the solid gray bars. Scores for Black and White individuals are presented separately. For arrests for violent crimes, there was a significant race by maltreatment interaction (β = 0.63, SE = 0.29, p = .031).
Discussion
Despite the widespread assumption that child maltreatment increases a person’s risk for violence, we find a more nuanced relationship. These results show that childhood maltreatment predicts self-report of violent behaviors and that Black maltreated children and controls were more likely to be arrested for violent crimes than White maltreated children and controls. However, our findings also show that race modifies the relationship between child maltreatment and arrests for violent crimes and Black maltreated children are most likely to be arrested, followed by Black non-maltreated children. We then sought to examine whether individual risk factors traditionally associated with crime and delinquency and their cumulative effects could reduce or eliminate these disparities. Similar to the results of the Huizinga et al. (2007) study, in which the disproportionate contact of minority youth with law enforcement and the courts could not be explained by differences in offending behavior, our findings suggest that disproportionate arrests among Black maltreated children could not be explained by offending behavior alone. Indeed, even with the introduction of all risk factors and self-reported violent behavior in the model, the race by maltreatment interaction was significant.
Some of our traditional risk factors were significantly associated with arrests for violent crime and self-report of violent behaviors. We found that being suspended or expelled from school and misbehaving in school were significantly associated with arrests for violent crimes and self-reports of violent behaviors. Students who are suspended and/or expelled from school are likely to be disconnected (Mizel et al., 2016, p. 102), and this disconnection with school has been indirectly associated with antisocial behaviors and contact with juvenile justice (Monahan et al., 2014). Furthermore, school misbehavior can lead directly to law enforcement and justice system involvement (Barrett & Katsiyannis, 2015). Thus, school systems may serve as an early entrance for later violent behavior.
There is also evidence that Black children who have behavioral problems are often referred to law enforcement, whereas White children are provided mental health care (Breland-Noble, 2004). Furthermore, although a police presence in schools increases the likelihood of arrests for all groups, the association is stronger for Black youth (Homer & Fisher, 2020). This raises questions about whether Black youth, particularly maltreated youth who may present with more behavioral problems associated with their maltreatment, are more likely to be ‘pushed out” of school and into more restrictive systems (e.g., criminal legal system), instead of being provided resources and less restrictive responses (Skiba et al., 2011). Once individuals become involved with the juvenile or criminal justice system, stigma can negatively impact their lives, leading to difficulties in obtaining an education, finding employment, recidivism, and marital problems (Pratt et al., 2016). In addition to these individual costs, there are consequences of arrest that affect families and communities (Kamalu et al., 2010).
Practitioners and policymakers should make resources available for all maltreated children, with the use of the most restrictive means (i.e., arrests) being the last resort. Further, law enforcement personnel should be encouraged to be more sensitive in their responses to minoritized maltreated children and to provide resources and other alternatives to assist maltreated children rather than resorting to arrests.
Our findings suggest that the disproportionate arrests among Black children, especially Black maltreated children, could not be explained by offending behavior, individual risk factors, or cumulative risk factors. Indeed, this pattern is likely the result of a complex process influenced by systemic racial bias and structural racism that results in increased likelihood of arrest for Black youth (Andersen, 2015; Schleiden et al., 2020). There are several potential explanations for our findings including greater surveillance of neighborhoods within which Black individuals reside (Shoub et al., 2020), longstanding discriminatory policies implemented by the government such as redlining (Powell & Porter, 2023; Rothstein, 2017), and racially biased policies within the criminal legal system (Mitchell & Caudy, 2015).
Limitations to the Study
Since our maltreatment cases were based on substantiated cases from the courts, these findings are not generalizable to unreported or unsubstantiated cases of child maltreatment. The sample was predominantly children from lower socioeconomic backgrounds; thus, the findings might not be generalizable to cases of maltreatment that occurred in middle- or upper-income families. This study represents the experiences of children growing up in the late 1960s and early 1970s in the Midwest part of the United States and may raise concerns about applying these findings to current cases or cases from different parts of the country. However, the maltreatment cases studied here are quite similar to cases currently being processed by the child protection system and the courts. It is also worth noting that some prior research has found that arrest trajectories vary by birth cohorts and may not generalize to more recent cohorts (Neil et al., 2021). Nonetheless, we note the value of longitudinal studies in establishing clear temporal sequencing in these relationships. From a pragmatic standpoint, it would take 40–50 years to collect similar data in another cohort.
We did not have complete matching for all maltreated children with control children, and thus, it is possible that this would affect the findings. However, re-analyses of earlier findings were conducted using only matched pairs and the results did not change with the smaller sample size (Currie & Widom, 2010; Widom, 1989b; Widom et al., 2007).
Finally, several of the family and school measures were based on self-report. While many studies use self-report measures for family size and school behaviors, there may be limitations in retrospective recollection and perceptions of behaviors. This limitation, particularly in perception, may explain the unexpected results that suspension and expulsion were reported at higher rates compared to these individuals’ reports of their own misbehavior in schools. It is also unknown whether the juvenile arrests led to school expulsions. Future research should explore these issues.
Conclusions
These results reveal that applying a racial “lens” to the cycle of violence findings leads to the conclusion that many traditional risk factors did not eliminate the racial differences, and that other more systemic factors that we were not able to examine may be at play. However, future research should operationalize and directly examine racial biases and structural racism to explain this relationship. Applying a racial lens to the cycle of violence also requires that we consider the overrepresentation of Black children in the child welfare system (Newman, 2022). Indeed, systemic biases similar to those for arrests may be operating that may account for higher representation of referred and substantiated cases of child maltreatment among Black children (Dettlaff & Boyd, 2020; Thomas et al., 2023). If possible, researchers should revisit their data and apply a racial lens to prior results to not miss important differences that may have been obscured by failure to examine them. Through deeper examinations and testing of competing explanations, future research may yield powerful implications for policies, practices, and interventions.
Footnotes
Acknowledgments
Dr. Maureen Allwood, our beloved colleague and friend, passed away unexpectedly in March 2024. Dr. Allwood was a clinical psychologist committed to improving the lives of young people of color by understanding the disproportionate impact of trauma on racial and ethnic minority youth and thinking about interventions that might mitigate these effects. We acknowledge her major contribution to this work and honor her memory.
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) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported in part by National Institute of Mental Health (MH49467 and MH58386) and National Institute of Justice (86-IJ-CX-0033, 89-IJ-CX-0007, and 2011-WG-BX-0013).
