Child Maltreatment, Youth Violence, and Intimate Partner Violence

Developmental Relationships

      Background

      Understanding the cycle of violence, from victimization to perpetration across the life span, is critical for designing successful prevention interventions. This study uses a nationally representative sample to examine the developmental relationships among three forms of child maltreatment, youth violence perpetration or victimization, and young adult intimate partner violence (IPV) perpetration or victimization.

      Methods

      Data describing self-reported youth violence perpetration (or victimization) from Wave I of the National Longitudinal Study of Adolescent Health (1994–1995) were matched with self-reported IPV perpetration (or victimization) in young adult sexual relationships and retrospective reports of child maltreatment collected during Wave III (2001–2002). Bivariate probit regression models were used to analyze the developmental relationships between child maltreatment, youth violence, and IPV. Analyses were completed in September 2006.

      Results

      Compared to nonvictims of child maltreatment, victims of child maltreatment are more likely to perpetrate youth violence (a likelihood increase ranging from −1.2% to 6.6% for females and 3.7% to 11.9% for males) and young adult IPV (an increase from 8.7% to 10.4% for females and from 1.3% to 17.2% for males), although the direct and indirect effects vary by type of child maltreatment experienced. Gender differences exist in the links between child maltreatment, youth violence and IPV, and in the effects of socioeconomic factors on youth violence and IPV.

      Conclusions

      Results suggest that it may be important to account for gender differences when designing violence prevention programs, and an integrative approach is critical for stopping the developmental trajectory of violence.

      Introduction

      The incidence and economic burden of violence is a national problem affecting millions of people each year in the United States.
      National Center for Injury Prevention and Control
      Web-based injury statistics query and reporting system (WISQARS).
      National Center for Injury Prevention and Control
      Costs of intimate partner violence against women in the United States.
      Violence can occur during childhood or adulthood and can be perpetrated by oneself, acquaintances, or strangers. Understanding the cycle of violence, from victimization to perpetration across the life span, is critical for designing successful prevention interventions. Research in this area has suggested that the victims of child maltreatment are at increased risk of perpetrating violence as well as being victims of violence as youths and during adulthood.
      • Briere J.
      • Runtz M.
      Childhood sexual abuse: long-term sequelae and implications for psychological assessment.
      • Capaldi D.M.
      • Clark S.
      Prospective family predictors of aggression toward female partners for at-risk young men.
      • Cicchetti D.
      • Toth S.L.
      A developmental psychopathology perspective on child abuse and neglect.
      • Ehrensaft M.K.
      • Cohen P.
      • Brown J.
      • Smailes E.
      • Chen H.
      • Johnson J.G.
      Intergenerational transmission of partner violence: a 20-year prospective study.
      • Fergusson D.M.
      • Lynskey M.T.
      Physical punishment/maltreatment during childhood and adjustment in young adulthood.
      • Ireland T.O.
      • Smith C.A.
      • Thornberry T.P.
      Developmental issues in the impact of child maltreatment on later delinquency and drug use.
      • Loos M.E.
      • Alexander P.C.
      Differential effects associated with self-reported histories of abuse and neglect in a college sample.
      • Magdol L.
      • Moffitt T.E.
      • Caspi A.
      • Silva P.A.
      Developmental antecedents of partner abuse: a prospective-longitudinal study.
      • Simons R.L.
      • Wu C.
      • Johnson C.
      • Conger R.D.
      A test of various perspectives on the intergenerational transmission of domestic violence.
      • Smith C.A.
      • Thornberry T.P.
      The relationship between childhood maltreatment and adolescent involvement in delinquency.
      • Stith S.M.
      • Rosen K.H.
      • Middleton K.A.
      • Busch A.L.
      • Lundeberg K.
      • Carlton R.P.
      The intergenerational transmission of spouse abuse: a meta-analysis.
      However, while connections among child maltreatment, youth violence, and intimate partner violence (IPV) have been documented, there are a number of issues requiring further examination.
      First, a majority of studies examining the developmental trajectories of violence have failed to adequately control for confounding individual socioeconomic, family background, and contextual factors. Evidence has suggested that child maltreatment, youth violence, and IPV are not uniformly distributed across the population and are more likely to occur in families characterized by social and economic disadvantage, parental separation and divorce, and families living in disadvantaged neighborhoods.
      • Smith C.A.
      • Thornberry T.P.
      The relationship between childhood maltreatment and adolescent involvement in delinquency.
      • Fergusson D.M.
      • Horwood L.J.
      • Lynskey M.T.
      Family change, parental discord and early offending.
      • Gelles R.J.
      • Cornell C.P.
      Intimate violence in families.
      • Mocan H.N.
      Economic conditions, deterrence and juvenile crime: evidence from micro data.
      • Sampson R.J.
      • Raudenbush S.
      • Earls F.
      Neighborhoods and violent crime: a multilevel study of collective efficacy.
      • Wikstrom P.H.
      • Loeber R.
      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.
      Without controlling for these factors, it is not clear whether the link among child maltreatment, youth violence, and IPV arises because of the socioeconomic and family context within which the violence occurs, or due to a cause-and-effect relationship in which experiencing child maltreatment increases the risk for later youth violence perpetration or victimization and/or IPV perpetration or victimization. Second, many studies in this field, especially in the area of IPV, have been based on relatively small or selected samples such as clinical and agency samples rather than general population samples. This limits the generalizability of their findings.
      Third, researchers have suggested that developmental trajectories of violence may operate differently for males and females.
      • Magdol L.
      • Moffitt T.E.
      • Caspi A.
      • Silva P.A.
      Developmental antecedents of partner abuse: a prospective-longitudinal study.
      • Stith S.M.
      • Rosen K.H.
      • Middleton K.A.
      • Busch A.L.
      • Lundeberg K.
      • Carlton R.P.
      The intergenerational transmission of spouse abuse: a meta-analysis.
      • Doumas D.
      • Margolin G.
      • John R.S.
      The intergenerational transmission of aggression across three generations.
      • Langhinrichsen-Rohling J.
      • Neidig P.
      • Thorn G.
      Violent marriages: gender differences in levels of current violence and past abuse.
      However, for the few studies that have investigated gender differences, especially those studying the gender differences in the link between child maltreatment and IPV, the findings are inconsistent. For example, Magdol et al.
      • Magdol L.
      • Moffitt T.E.
      • Caspi A.
      • Silva P.A.
      Developmental antecedents of partner abuse: a prospective-longitudinal study.
      found that the link between abusive discipline experienced as a child and IPV perpetration was stronger for females, while Doumas et al.
      • Doumas D.
      • Margolin G.
      • John R.S.
      The intergenerational transmission of aggression across three generations.
      and Langhinrichsen-Rohling et al.
      • Langhinrichsen-Rohling J.
      • Neidig P.
      • Thorn G.
      Violent marriages: gender differences in levels of current violence and past abuse.
      found that abused boys are at higher risk for perpetrating partner violence as adults. The differences in findings could be due to different specificities of the samples used in the studies. The samples used in both of the latter studies were small and selected in the U.S., which may suffer from sample selection biases. The sample for the study by Magdol et al.
      • Magdol L.
      • Moffitt T.E.
      • Caspi A.
      • Silva P.A.
      Developmental antecedents of partner abuse: a prospective-longitudinal study.
      was an unselected birth cohort born in Dunedin, New Zealand. Besides sample differences, research design and measurement problems are common in studies on the relationship between child maltreatment and later violent behaviors, and these problems may also explain, in part, why results differ.
      • Widom C.S.
      The cycle of violence.
      • Widom C.S.
      Does violence beget violence? A critical examination of the literature.
      Finally, it is important that research on developmental trajectories of violence from child maltreatment victimization to the victimization or perpetration of IPV, should explore whether youth violence perpetration mediates the effect of child maltreatment on IPV perpetration, or whether youth violence victimization mediates the effect of child maltreatment on IPV victimization. Studies linking child maltreatment and IPV have typically not included measures of youth violence, so it is unclear whether risk for IPV perpetration or victimization is a direct consequence of a history of child maltreatment or whether child maltreatment is really a marker for some other, more direct causal variable such as youth violence perpetration or victimization. Antisocial behavior theory proposes that child maltreatment might increase the risk for IPV perpetration through first promoting violent behavior in adolescence, with the aggressive behavior being carried through to adulthood and used against intimate partners.
      • Capaldi D.M.
      • Clark S.
      Prospective family predictors of aggression toward female partners for at-risk young men.
      • Simons R.L.
      • Wu C.
      • Johnson C.
      • Conger R.D.
      A test of various perspectives on the intergenerational transmission of domestic violence.
      • Herrenkohl T.I.
      • Mason W.A.
      • Kosterman R.
      • Lengua L.J.
      • Hawkins J.D.
      • Abbott R.D.
      Pathways from physical childhood abuse to partner violence in young adulthood.
      When applied to the link between child maltreatment and IPV victimization, learned helplessness theory suggests that exposure to child maltreatment may leave children with a sense of learned helplessness so that they do not develop appropriate skills to escape the violence in adolescence, which, in turn, increases the risk for later IPV victimization.
      • Walker L.
      Battered women and learned helplessness.
      • Walker L.
      The battered women syndrome study.
      These theories suggest that, for the connection between child maltreatment and IPV perpetration or victimization, it is important to consider whether youth violence perpetration or victimization plays a role in mediating the impact of child maltreatment on IPV perpetration or victimization.
      This study used a U.S. longitudinal and nationally representative sample to examine the direct relationships among three forms of child maltreatment (neglect, physical abuse, and sexual abuse), youth violence perpetration or victimization, and young adult IPV perpetration or victimization. This study expanded the authors’ previous work that examined the link between child maltreatment and IPV perpetration with different outcome measures of IPV perpetration.

      Fang X, Corso P. Gender differences in the connections between violence experienced as a child and perpetration of intimate partner violence in young adulthood. J Fam Violence (to be published).

      This study further examined the indirect effects that child maltreatment has on future IPV perpetration or victimization through the presence of youth violence perpetration or victimization. Finally, this study assessed the impact of gender on these direct and indirect effects, and the role that socioeconomic factors play in the occurrence of violence.

      Methods

      Conceptual Framework

      This study examined two relationships: the relationship among child maltreatment, youth violence perpetration, and IPV perpetration (perpetration link), and the relationship between child maltreatment, youth violence victimization, and IPV victimization (victimization link). Based on previous literature, this study posited that three types of child maltreatment (neglect, physical abuse, and sexual abuse), family background factors, adolescent individual demographic factors, and adolescent community factors could affect the risk of youth violence perpetration. For the relationship between child maltreatment and IPV perpetration, child maltreatment could affect the risk of IPV perpetration directly and/or affect the risk of IPV perpetration indirectly through its effect on youth violence perpetration. Family background factors and young adult individual demographic factors could also directly affect the risk of IPV perpetration. It is posited that the pattern of relationship in the victimization link is similar to that in the perpetration link except that the intervening variable is youth violence victimization instead of youth violence perpetration, and the outcome variable is IPV victimization instead of IPV perpetration.

      Participants

      Data for this study came from the National Longitudinal Study of Adolescent Health (Add Health), a nationally representative study examining the health-related behaviors and outcomes of adolescents in grades 7 through 12, with follow-up outcomes collected in young adulthood.
      • Udry J.R.
      The National Longitudinal Study of Adolescent Health (Add Health), Waves I and II, 1994–1996, and Wave III, 2001–2002.
      Individual-, family-, school-, and community-level information was collected in two waves between 1994 (Wave I) and 1996 (Wave II). A third wave (Wave III) was conducted among Wave-I respondents in 2001 and 2002 to examine the effects of adolescent experiences and factors on subsequent adult outcomes. Details regarding sampling methods and interviewing techniques are described elsewhere.
      • Udry J.R.
      The National Longitudinal Study of Adolescent Health (Add Health), Waves I and II, 1994–1996, and Wave III, 2001–2002.
      • Harris K.M.
      • Florey F.
      • Tabor J.
      • Bearman P.S.
      • Jones J.
      • Udry J.R.
      The National Longitudinal Study of Adolescent Health: research design.
      Wave-III data collection involved 14,322 original Wave-I respondents, aged 18 to 26 at the time of the Wave-III interview. This study used data only for respondents included in both Waves I and III, and who reported being involved in at least one sexual relationship in the 2 years preceding the Wave-III survey. Nonsexual romantic relationships were excluded because the Add Health data do not include information on the ending dates of nonsexual relationships. This reduced the sample for analyses to 10,320 participants. An additional 952 respondents were excluded because they were missing information on outcome variables or covariates, resulting in a final study sample of 9368 participants. Of the study participants, 5179 (55.2%) were female. Data describing self-reported youth violence perpetration or victimization from Wave I of the Add Health study were matched with self-reported IPV perpetration or victimization within young adult sexual relationships and retrospective reports of child maltreatment collected during Wave III.

      Measures

      The measures used to define the outcomes of interest are described in Table 1. Both IPV and youth violence measures were dichotomized for two reasons. First, the majority of the items used to define IPV and youth violence had a low base rate of occurrence resulting in highly skewed distributions. Second, they allowed for the use of bivariate probit regression that can decompose the effects of child maltreatment on IPV into direct effects and indirect effects in the multivariate setting. IPV perpetration was defined by a respondent answering “Yes” to any one of three questions in Wave III reflecting physical and sexual violence toward a partner during the preceding 2 years. IPV victimization was defined by a respondent answering “Yes” to any one of three items in Wave III reflecting experience of physical or sexual violence from a partner during the preceding 2 years. Youth violence perpetration was defined by a respondent answering “Yes” to any one of five questions in Wave I reflecting violence committed during the preceding year. Youth violence victimization was defined by a respondent answering “Yes” to any one of four questions in Wave I reflecting violence experienced during the preceding year.
      Table 1Measures used to define outcomes
      OutcomeMeasuresSurvey wave
      IPVPerpetrationVictimizationIII
      • 1
        Threatened partner with violence, pushed, shoved, or threw something at partner that could hurt.
      • 2
        Slapped, hit, or kicked partner.
      • 3
        Insisted on or made partner have sexual relations with me when [he/she] didn’t want to.
      • 1
        Partner threatened me with violence, pushed, shoved, or threw something at me that could hurt.
      • 2
        Partner slapped, hit, or kicked me.
      • 3
        Partner insisted on or made me have sexual relations with him/her when I didn’t want to.
      Youth violencePerpetrationVictimizationI
      • 1
        Took part in a group fight.
      • 2
        Hurt someone badly enough to need bandages or care from a doctor or nurse.
      • 3
        Used or threatened to use a weapon to get something from someone.
      • 4
        Pulled a knife or gun on someone.
      • 5
        Shot or stabbed someone.
      • 1
        I was jumped.
      • 2
        Someone pulled a knife or gun on me.
      • 3
        Someone shot me.
      • 4
        Someone cut or stabbed me.
      Child maltreatment—neglect1. How often had your parents or other adult care-givers left you home alone when an adult should have been with you?III
      2. How often had your parents or other adult care-givers not taken care of your basic needs, such as keeping you clean or providing food or clothing?
      Child maltreatment--physical abuse1. How often had your parents or other adult care-givers slapped, hit, or kicked you?III
      Child maltreatment--sexual abuse1. How often had one of your parents or other adult care-givers touched you in a sexual way, forced you to touch him or her in a sexual way, or forced you to have sexual relations?III
      IPV, intimate partner violence.
      The measures of the three types of child maltreatment were also dichotomized on the basis of respondents’ retrospective reports on the time before starting 6th grade (Table 1). The measure of neglect was constructed using two questions from Wave III. Following the work of Leeb et al.,
      • Leeb R.T.
      • Mueller M.
      • Swahn M.H.
      • Feldman M.
      • Melanson C.
      Sugar and Spice, Snips and Snails: gender effects on risk factors for delinquency and violence in youth.
      neglect was defined by a respondent reporting that care-givers had left him/her home alone as a child more than five times when an adult should have been with the child, or care-givers had not taken care of his/her basic needs at least once. Physical abuse was defined by a report that caregivers had slapped, hit, or kicked respondent more than 5 times.
      • Leeb R.T.
      • Mueller M.
      • Swahn M.H.
      • Feldman M.
      • Melanson C.
      Sugar and Spice, Snips and Snails: gender effects on risk factors for delinquency and violence in youth.
      Sexual abuse was defined by a report that caregivers had touched respondent in a sexual way, forced the respondent to touch him/her in a sexual way, or forced the respondent to have sexual relations at least once.
      To study the link between child maltreatment and youth violence perpetration or victimization, other control variables included adolescent individual demographic factors (age, age squared, gender, and race/ethnicity), community factors, and family background characteristics found to be related to child maltreatment and/or later violent behaviors in previous research.
      • Mocan H.N.
      Economic conditions, deterrence and juvenile crime: evidence from micro data.
      • Sampson R.J.
      • Raudenbush S.
      • Earls F.
      Neighborhoods and violent crime: a multilevel study of collective efficacy.
      • Gottfredson D.C.
      • McNeil III, R.J.
      • Gottfredson G.D.
      Social area influences on delinquency: a multilevel analysis.
      • Brown S.E.
      Social class, child maltreatment, and delinquent behavior.
      • Coleman P.K.
      • Maxey C.D.
      • Rue V.M.
      • Coyle C.T.
      Associations between voluntary and involuntary forms of perinatal loss and child maltreatment among low-income mothers.
      Family background characteristics included parent (mother or the main care-giver) education, family poverty, and whether the participant was from a two-parent family.
      Previous research suggests that neighborhood characteristics, measured independently of the characteristics of the individuals, contribute to the explanation of individual youth violence behavior.
      • Mocan H.N.
      Economic conditions, deterrence and juvenile crime: evidence from micro data.
      • Sampson R.J.
      • Raudenbush S.
      • Earls F.
      Neighborhoods and violent crime: a multilevel study of collective efficacy.
      • Gottfredson D.C.
      • McNeil III, R.J.
      • Gottfredson G.D.
      Social area influences on delinquency: a multilevel analysis.
      Two community variables were included as covariates in the study. First, an index of community economic disadvantage (CED) was constructed using the following three county-level items drawn from the 1990 U.S. Census: proportion of people with income below poverty level in 1989, proportion aged >25 without high school diploma or equivalency, and unemployment rate. Each item was standardized individually and then the mean was used to construct the CED index. The overall Cronbach alpha for this scale is 0.79. Second, following Levitt’s work,
      • Levitt S.D.
      Juvenile crime and punishment.
      the study also included the total number of crimes in the county per 100,000 population to control for the impact of omitted factors that may influence youth violence.
      To examine the link between child maltreatment and IPV perpetration or victimization, in addition to controlling for youth violence perpetration or victimization, other covariates included individual demographic characteristics during young adulthood (age, age squared, gender, race/ethnicity, marital status, school enrollment, and employment status) and family background characteristics.

      Statistical Analyses

      Following Greene’s work,
      • Greene W.H.
      Gender economics courses in liberal arts colleges: further results.
      • Greene W.H.
      Econometric analysis.
      bivariate probit regression models were used to analyze the developmental relationships among child maltreatment, youth violence, and IPV. In these models, youth violence perpetration (or youth violence victimization) was not only a dependent variable on its own, but also an independent variable for IPV perpetration (or IPV victimization), which allowed testing for indirect effects. The recursive simultaneous equation models were estimated by maximum likelihood methods. Direct effects and/or indirect effects of a variable which might be a determinant of youth violence and/or IPV perpetration or victimization were calculated based on the formulas appearing in Greene.
      • Greene W.H.
      Gender economics courses in liberal arts colleges: further results.
      In Greene’s paper,
      • Greene W.H.
      Gender economics courses in liberal arts colleges: further results.
      full details are provided on how the coefficients are scaled to produce the marginal effects (direct and indirect effects) for both continuous and binary variables. The direct and indirect effects of a given variable were evaluated at sample means. Bootstrapping techniques were used to estimate the standard errors for all indirect effects. All analyses were conducted with STATA SE, version 9 (Stata Corp, College Station TX, 2005).
      To test the significance of gender, this study included the interaction terms between the gender dummy and each predictor variable for IPV and youth violence perpetration (or victimization) in the pooled regressions, and then tested the hypothesis that these interactions were jointly significantly different from zero.
      • Greene W.H.
      Econometric analysis.

      Results

      Rates (in proportion forms) with 95% confidence intervals for all dichotomous variables and means with standard deviations for all continuous variables are presented in Table 2. All statistics listed in Table 2 and all of the following analyses used sample weights to adjust for stratification and over-sampling of under-represented groups. After the adjustment of sample weights, the study sample had very similar gender and racial/ethnic distributions to the nation. Table 2 shows a larger percentage of females than males reported both IPV perpetration and victimization, while a larger percentage of males than females reported both youth violence perpetration and victimization. These numbers are consistent with evidence suggesting that females perpetrate some forms of violence at rates proportional or higher to males.
      • Magdol L.
      • Moffitt T.E.
      • Caspi A.
      • Silva P.A.
      Developmental antecedents of partner abuse: a prospective-longitudinal study.
      • Archer J.
      Sex differences in physically aggressive acts between heterosexual partners: a meta-analytical review.
      • Moffitt T.E.
      • Caspi A.
      • Rutter M.
      • Silva P.A.
      Sex differences in physical violence and sex similarities in partner abuse.
      • Morse B.J.
      Beyond the conflict tactics scale: assessing gender differences in partner violence.
      • Straus M.A.
      • Gelles R.J.
      Physical violence in American families: risk factors and adaptations to violence in 8,145 families.
      In terms of child maltreatment, 21.1% and 15.3% of males and females, respectively, reported having experienced neglect during childhood. For both genders, approximately 8% reported being physically abused and 4% reported sexual abuse during their childhood.
      Table 2Descriptive statistics for sample
      VariableFemaleMale
      Rate/meanCI/SDRate/meanCI/SD
      Independent variablesIPV perpetration (%)31.830.5–33.117.616.4–18.7
      IPV victimization (%)27.826.6–29.125.724.4–27.1
      Youth violence perpetration (%)23.422.2–24.537.836.3–39.3
      Youth violence victimization (%)10.69.8–11.527.526.2–28.9
      Child maltreatmentNeglected (%)15.314.3–16.321.119.8–22.4
      Physically abused (%)8.37.5–9.08.07.1–8.8
      Sexually abused (%)4.33.7–4.84.53.8–5.1
      Demographic characteristicsWhite (%)71.670.4–72.970.569.1–72.0
      Black (%)14.213.2–15.213.412.3–14.4
      Asian (%)2.52.0–2.92.41.9–2.9
      Native (%)0.50.3–0.61.00.7–1.3
      Hispanic (%)10.59.7–11.412.011.0–13.0
      Adolescent demographic variables
       Age in Wave 115.291.7515.501.82
      Young adult demographic variables
       Age in Wave 321.671.7821.891.86
       Married (%)23.121.9–24.216.115.0–17.3
       School enrollment (%)39.337.9–40.633.231.7–34.6
       Employed (%)68.667.3–69.974.372.9–75.7
      Community factors during adolescenceEconomic disadvantage index
      The community economic disadvantage index was constructed using three county-level items drawn from the 1990 U.S. Census. For details, see text.
      −0.0290.911−0.0490.897
      County crime rate (per 100,000 population)5585.882661.995568.872596.27
      Family background factorsParent education
      Parent education was divided into six categories (coded 1 to 6): 8th grade or less (includes never went to school), more than 8th grade but did not graduate from high school, high school graduate (or equivalent), some post---high school training/college, college graduate, and postgraduate or professional training.
      3.611.223.661.24
      Family poverty (%)
      A dichotomous family poverty indicator was constructed using the household income measure such that a value of 1 represented annual household income of ≤$20,000, and zero otherwise.
      21.620.4–22.721.520.2–22.8
      Two-parent family (%)73.872.6–75.075.273.9–76.6
      Note: Rate (in proportion form) with 95% confidence interval (CI) is presented for a dichotomous variable, while mean with standard deviation (SD) is presented for a continuous variable.
      CI, confidence interval; IPV, intimate partner violence; SD, standard deviation.
      a The community economic disadvantage index was constructed using three county-level items drawn from the 1990 U.S. Census. For details, see text.
      b Parent education was divided into six categories (coded 1 to 6): 8th grade or less (includes never went to school), more than 8th grade but did not graduate from high school, high school graduate (or equivalent), some post---high school training/college, college graduate, and postgraduate or professional training.
      c A dichotomous family poverty indicator was constructed using the household income measure such that a value of 1 represented annual household income of ≤$20,000, and zero otherwise.

      Bivariate Probit Models

      First, the significance of a gender effect in the perpetration link and victimization link was tested. Results indicated that the regression models (results not shown) for males and females were systematically different for both the perpetration link (χ2=28, df=15, p<0.05) and the victimization link (χ2=25.9, df=15, p<0.05). Therefore, subsequent regressions were estimated separately for males and females. This study also considered the possibility that the variables used in the models might interact with each other. Because regression results revealed that none of the interactions among these factors were significant for either the perpetration link or the victimization link (results available upon request), the subsequent regression models did not include any interaction terms.
      The estimated marginal effects (direct effects and indirect effects) of the main predictor variables for IPV and youth violence perpetration or victimization are presented in Table 3, Table 4. The marginal effects of a predictor variable refer to changes in probabilities of an outcome event given a unit change in a continuous predictor variable, or change of state in a categorical variable under consideration. For example, for females, the direct effect of neglect on IPV perpetration 0.08 indicates that being neglected as a child directly increases the likelihood of IPV perpetration by 8%, and the indirect effect 0.007 indicates that being neglected as a child indirectly increases the likelihood of IPV perpetration by 0.7% by increasing the likelihood of youth violence perpetration.
      Table 3Estimated direct and indirect effects on IPV and youth violence perpetration by gender
      VariablesYouth violence perpetration
      The effects of predictor variables on youth violence perpetration are not only direct but also total effects. For example, for males, the direct (or total) effect of neglect on youth violence perpetration 0.1025 indicates that controlling for other variables, being neglected as a child increases the likelihood of youth violence perpetration by 10.25%.
      Marginal effects (robust SE)
      IPV perpetration
      The total effect of each predictor variable on IPV perpetration is decomposed into direct effect and indirect effect. The direct effect represents the effect of a predictor variable on IPV perpetration, not through the presence of youth violence perpetration. The indirect effect represents the effect of a predictor variable on IPV perpetration through the presence of youth violence perpetration. For example, for females, the direct effect of neglect on IPV perpetration 0.0795 indicates that being neglected as a child directly increases the likelihood of IPV perpetration by 7.95%, and the indirect effect 0.007 indicates that being neglected as a child indirectly increases the likelihood of IPV perpetration by 0.7% by increasing the likelihood of youth violence perpetration.
      Probit regression coefficient (robust SE)
      FemalesMalesFemalesMales
      Direct (total)Direct (total)DirectIndirectDirectIndirect
      Youth violence perpetration
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0949
      p<0.01 (all bolded).
      (0.0217)
      0.0772
      p<0.01 (all bolded).
      (0.0213)
      Neglected
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0662
      p<0.05;
      (0.0273)
      0.1025
      p<0.01 (all bolded).
      (0.0324)
      0.0795
      p<0.05;
      (0.0333)
      0.0070
      p<0.05;
      (0.0031)
      0.0037 (0.0240)0.0116
      p<0.01 (all bolded).
      (0.0033)
      Physically abused
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0655
      p<0.05;
      (0.0345)
      0.1188
      p<0.05;
      (0.0475)
      0.0971
      p<0.05;
      (0.0426)
      0.0070 (0.0040)0.0042 (0.0370)0.0092
      p<0.05;
      (0.0047)
      Sexually abused
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      −0.0120 (0.0443)0.0370 (0.0694)0.1025 (0.0588)−0.0010 (0.0042)0.1695
      p<0.01 (all bolded).
      (0.0683)
      0.0020 (0.0084)
      Parent education−0.0324
      p<0.01 (all bolded).
      (0.0081)
      −0.0311
      p<0.01 (all bolded).
      (0.0111)
      −0.0152 (0.0094)−0.0031
      p<0.05;
      (0.0012)
      −0.0177
      p<0.05;
      (0.0080)
      −0.0024
      p<0.05;
      (0.0011)
      Family poverty
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0743
      p<0.01 (all bolded).
      (0.0298)
      −0.0341 (0.0405)−0.0082 (0.0233)0.0070
      p<0.05;
      (0.0034)
      0.0151 (0.0237)−0.0029 (0.0031)
      Two-parent family
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0010 (0.0192)−0.0349 (0.0347)−0.0290 (0.0257)−0.0001 (0.0022)−0.0026 (0.0234)−0.0027 (0.0028)
      Married
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      −0.0069 (0.0229)0.0441 (0.0268)
      School enrollment
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      −0.0316 (0.0218)−0.038 (0.0194)
      Employed
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0228 (0.0195)−0.0139 (0.0233)
      Economic disadvantage index−0.0138 (0.0093)0.0043 (0.0107)–0.0013 (0.0010)0.0003 (0.0010)
      County crime rate (per 100 population)0.0027 (0.0042)0.0109
      p<0.05;
      (0.0047)
      0.0003 (0.0004)0.0008 (0.0005)
      Note: All results were adjusted for age and race/ethnicity. Bootstrap estimates of the standard errors for indirect effects came from 1000 bootstrap samples.
      IPV, intimate partner violence; SE, standard error.
      low asterisk p<0.05;
      low asterisklow asterisk p<0.01 (all bolded).
      a The effects of predictor variables on youth violence perpetration are not only direct but also total effects. For example, for males, the direct (or total) effect of neglect on youth violence perpetration 0.1025 indicates that controlling for other variables, being neglected as a child increases the likelihood of youth violence perpetration by 10.25%.
      b The total effect of each predictor variable on IPV perpetration is decomposed into direct effect and indirect effect. The direct effect represents the effect of a predictor variable on IPV perpetration, not through the presence of youth violence perpetration. The indirect effect represents the effect of a predictor variable on IPV perpetration through the presence of youth violence perpetration. For example, for females, the direct effect of neglect on IPV perpetration 0.0795 indicates that being neglected as a child directly increases the likelihood of IPV perpetration by 7.95%, and the indirect effect 0.007 indicates that being neglected as a child indirectly increases the likelihood of IPV perpetration by 0.7% by increasing the likelihood of youth violence perpetration.
      c The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      Table 4Estimated direct and indirect effects on IPV and youth violence victimization by gender
      VariablesYouth violence victimization
      The effects of predictor variables on youth violence victimization are not only direct but also total effects. For example, for males, the direct (or total) effect of neglect on youth violence victimization 0.10 indicates that controlling for other variables, being neglected as a child increases the likelihood of youth violence victimization by 10%.
      Marginal effects (robust SE)
      IPV victimization
      The total effect of each predictor variable on IPV victimization is decomposed into direct effect and indirect effect. The direct effect represents the effect of a predictor variable on IPV victimization, not through the presence of youth violence victimization. The indirect effect represents the effect of a predictor variable on IPV victimization through the presence of youth violence victimization. For example, for females, the direct effect of neglect on IPV victimization 0.0004 indicates that being neglected as a child directly increases the likelihood of IPV victimization by 0.04%, and the indirect effect 0.0072 indicates that being neglected as a child indirectly increases the likelihood of IPV victimization by 0.72% by increasing the likelihood of youth violence victimization.
      Marginal effects (robust SE)
      FemalesMalesFemalesMales
      Direct (total)Direct (total)DirectIndirectDirectIndirect
      Youth violence victimization
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0442 (0.0305)0.0706
      p<0.01 (all bolded).
      (0.0254)
      Neglected
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0242 (0.0177)0.0999
      p<0.01 (all bolded).
      (0.0336)
      0.0439 (0.0305)0.0015 (0.0013)0.0004 (0.0289)0.0072
      p<0.05;
      (0.0034)
      Physically abused
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0253 (0.0251)0.0189 (0.0417)0.0634 (0.0388)0.0017 (0.0016)0.0163 (0.0409)0.0013 (0.0034)
      Sexually abused
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0119 (0.0268)0.0455 (0.0638)0.0660 (0.0509)0.0011 (0.0018)0.1174 (0.0677)0.0030 (0.0059)
      Parent education−0.0094 (0.0061)−0.0177 (0.0102)−0.0030 (0.0095)−0.0004 (0.0005)−0.0115 (0.0091)−0.0013 (0.0009)
      Family poverty
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0254 (0.0207)−0.0084 (0.0347)0.0046 (0.0237)0.0012 (0.0014)0.0453 (0.0289)−0.0010 (0.0024)
      Two-parent family
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      −0.0218 (0.0143)−0.0634
      p<0.05;
      (0.0289)
      −0.0200 (0.0249)−0.0012 (0.0013)0.0190 (0.0256)−0.0046 (0.0027)
      Married
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      −0.0495
      p<0.05;
      (0.0238)
      0.0451 (0.0306)
      School enrollment
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      −0.0630
      p<0.01 (all bolded).
      (0.0185)
      −0.0569
      p<0.05;
      (0.0257)
      Employed
      The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      0.0009 (0.0186)−0.0471 (0.0276)
      Economic disadvantage index−0.0050 (0.0074)0.0275
      p<0.05;
      (0.0131)
      −0.0002 (0.0004)0.0020 (0.0012)
      County crime rate (per 100 population)0.0051 (0.0027)0.0087 (0.0048)0.0002 (0.0002)0.0006 (0.0004)
      Note: All results were adjusted for age and race/ethnicity. Bootstrap estimates of the standard errors for indirect effects came from 1000 bootstrap samples.
      IPV, intimate partner violence; SE, standard error.
      low asterisk p<0.05;
      low asterisklow asterisk p<0.01 (all bolded).
      a The effects of predictor variables on youth violence victimization are not only direct but also total effects. For example, for males, the direct (or total) effect of neglect on youth violence victimization 0.10 indicates that controlling for other variables, being neglected as a child increases the likelihood of youth violence victimization by 10%.
      b The total effect of each predictor variable on IPV victimization is decomposed into direct effect and indirect effect. The direct effect represents the effect of a predictor variable on IPV victimization, not through the presence of youth violence victimization. The indirect effect represents the effect of a predictor variable on IPV victimization through the presence of youth violence victimization. For example, for females, the direct effect of neglect on IPV victimization 0.0004 indicates that being neglected as a child directly increases the likelihood of IPV victimization by 0.04%, and the indirect effect 0.0072 indicates that being neglected as a child indirectly increases the likelihood of IPV victimization by 0.72% by increasing the likelihood of youth violence victimization.
      c The reference category for each dichotomous variable is the absence of the condition (e.g., neglected compared to not neglected; married compared to not married).
      For perpetration (Table 3), childhood physical abuse and childhood neglect were significant predictors of youth violence perpetration for both genders; however, they were more significantly predictive of youth violence perpetration for males than for females. Family poverty and low levels of parent education significantly increased the risk of perpetrating youth violence for females, while for males, only low parent education significantly increased the likelihood of youth violence perpetration. Living in a high-crime neighborhood significantly increased the risk of perpetrating youth violence for males, but this effect was not significant for females.
      Significant direct predictors of IPV perpetration in males were youth violence perpetration and childhood sexual abuse, while for females, youth violence perpetration, physical abuse, and childhood neglect were significant direct predictors. For both genders, childhood neglect had a significant indirect effect on IPV perpetration through the presence of youth violence perpetration. Childhood physical abuse had a significant indirect effect on IPV perpetration for males; however, for females, the indirect effect was only marginally significant (p<0.10). Almost no family background characteristics included in this study were directly and significantly associated with IPV perpetration for either males or females; except that for males, low parent education was directly and significantly associated with IPV perpetration.
      For victimization (Table 4), child neglect was a significant predictor of youth violence victimization for males. For females, there were no child maltreatment outcomes that significantly predicted youth violence victimization. No family background characteristics were significantly associated with youth violence victimization for females. For males, however, being from a two-parent household significantly decreased the risk of being a victim of youth violence, and living in an economically disadvantaged area increased the risk.
      Youth violence victimization was significantly predictive of IPV victimization for males, but not for females. No child maltreatment outcomes were significant direct predictors of IPV victimizations for either gender. There were no child maltreatment outcomes that had significant indirect effects on IPV victimization through the presence of youth violence victimization for females. For males, only childhood neglect had a significant indirect effect on IPV victimization. Being married was significantly associated with the higher risk of IPV victimization for females, but not for males. School enrollment significantly decreased the risk of IPV victimization for both genders.

      Discussion

      Using a U.S. nationally representative sample, the present results demonstrate that, in general, victims of child maltreatment are more likely to perpetrate future violence in the form of youth violence and IPV, while there is less of an effect of child maltreatment on future victimization of youth violence or IPV. These findings reinforce the commonly held views that preventing child maltreatment may be key to preventing the perpetration of youth violence, and that interventions targeting youth violence may also serve to prevent later IPV perpetration or concurrent dating violence.
      • Capaldi D.M.
      • Clark S.
      Prospective family predictors of aggression toward female partners for at-risk young men.
      • Cicchetti D.
      • Toth S.L.
      A developmental psychopathology perspective on child abuse and neglect.
      • Ehrensaft M.K.
      • Cohen P.
      • Brown J.
      • Smailes E.
      • Chen H.
      • Johnson J.G.
      Intergenerational transmission of partner violence: a 20-year prospective study.
      • Fergusson D.M.
      • Lynskey M.T.
      Physical punishment/maltreatment during childhood and adjustment in young adulthood.
      • Ireland T.O.
      • Smith C.A.
      • Thornberry T.P.
      Developmental issues in the impact of child maltreatment on later delinquency and drug use.
      • Simons R.L.
      • Wu C.
      • Johnson C.
      • Conger R.D.
      A test of various perspectives on the intergenerational transmission of domestic violence.
      • Smith C.A.
      • Thornberry T.P.
      The relationship between childhood maltreatment and adolescent involvement in delinquency.
      Although child maltreatment was found to be connected to youth violence and IPV perpetration, the direct and indirect effects varied by type of child maltreatment experienced. Childhood neglect and physical abuse were significantly associated with youth violence perpetration for both genders. The indirect effects of neglect/physical abuse on IPV perpetration through their effects on youth violence perpetration were also significant for both genders, supporting the theory that antisocial behavior may partly explain the developmental relationship between child maltreatment and later violent behaviors for both genders.
      • Capaldi D.M.
      • Clark S.
      Prospective family predictors of aggression toward female partners for at-risk young men.
      • Simons R.L.
      • Wu C.
      • Johnson C.
      • Conger R.D.
      A test of various perspectives on the intergenerational transmission of domestic violence.
      After controlling for youth violence perpetration, neglect/physical abuse remained a strong, direct predictor of IPV perpetration for females but not for males. These results support the findings of Magdol et al.
      • Magdol L.
      • Moffitt T.E.
      • Caspi A.
      • Silva P.A.
      Developmental antecedents of partner abuse: a prospective-longitudinal study.
      that the link between child maltreatment and IPV perpetration is stronger for females than for males. The effects of childhood sexual abuse on future perpetration of violence differ from those for neglect/physical abuse. Childhood sexual abuse was not significantly associated with youth violence perpetration either for males or females. And while childhood sexual abuse was the strongest predictor of IPV perpetration for males, it was not significant for females at all. Although previous research suggests that prevention of IPV perpetration should begin as early as adolescence,
      • Magdol L.
      • Moffitt T.E.
      • Caspi A.
      • Silva P.A.
      Developmental antecedents of partner abuse: a prospective-longitudinal study.
      • Avery-Leaf S.
      • Cascardi M.
      • O’Leary K.D.
      • Cano A.
      Efficacy of a dating violence prevention program on attitudes justifying aggression.
      findings from this study suggest that children who have been maltreated, especially girls who are victims of physical abuse/neglect and boys who are victims of sexual abuse, may be good candidates for IPV prevention.
      The study found less of an effect of child maltreatment on future youth violence or IPV victimization. For example, only males experiencing child neglect were at increased risk for IPV victimization, but only through the effect of increased risk for youth violence victimization which in turn increased the risk for IPV victimization. However, youth violence victimization by itself was found to be a significant independent predictor of IPV victimization for males. Thus, preventing youth violence victimization may be a key to preventing future IPV victimization for males.
      There are a number of limitations in this study. First, as an analysis of secondary data, this study is limited in the number of survey questions available to define child maltreatment. For example, physical abuse is defined by one question, neglect by two questions, and sexual abuse by one question. As such, there could exist wide exposure variance within each category of child maltreatment that is not accounted for in the model. However, many other surveys and subsequent studies on retrospective reports of child maltreatment are similarly limited,
      • Ehrensaft M.K.
      • Cohen P.
      • Brown J.
      • Smailes E.
      • Chen H.
      • Johnson J.G.
      Intergenerational transmission of partner violence: a 20-year prospective study.
      • Doumas D.
      • Margolin G.
      • John R.S.
      The intergenerational transmission of aggression across three generations.
      • Langhinrichsen-Rohling J.
      • Neidig P.
      • Thorn G.
      Violent marriages: gender differences in levels of current violence and past abuse.
      • Hussey J.M.
      • Chang J.J.
      • Kotch J.B.
      Child maltreatment in the United States: prevalence, risk factors, and adolescent health consequences.
      • Leeb R.T.
      • Barker L.E.
      • Strine T.W.
      The effect of childhood physical and sexual abuse on adolescent weapon carrying.
      • Cohen P.
      • Brown J.
      • Smailes E.
      Child abuse and neglect and the development of mental disorders in the general population.
      • Felitti V.J.
      • Anda R.F.
      • Nordenberg D.
      • et al.
      The relationship of adult health status to childhood abuse and household dysfunction.
      • Dube S.R.
      • Anda R.F.
      • Felitti V.J.
      • Chapman D.
      • Williamson D.F.
      • Giles W.H.
      Childhood abuse, household dysfunction and the risk of attempted suicide throughout the life span: findings from Adverse Childhood Experiences Study.
      and researchers have argued that using a few questions instead of a large protocol to define child maltreatment is intended to maximize specificity (few false positives) at the expense of possibly low sensitivity (high rates of false negatives).
      • Ehrensaft M.K.
      • Cohen P.
      • Brown J.
      • Smailes E.
      • Chen H.
      • Johnson J.G.
      Intergenerational transmission of partner violence: a 20-year prospective study.
      • Cohen P.
      • Brown J.
      • Smailes E.
      Child abuse and neglect and the development of mental disorders in the general population.
      • Cohen P.
      The effect of instruments and informants on ascertainment.
      Low sensitivity does not significantly bias statistical estimates of rare phenomena in a general population study, although it does reduce statistical power.
      • Cohen P.
      • Brown J.
      • Smailes E.
      Child abuse and neglect and the development of mental disorders in the general population.
      • Cohen P.
      The effect of instruments and informants on ascertainment.
      But the limitation does suggest that caution should be taken in interpreting these results, and that future research is needed to validate these results with a more complete assessment of child maltreatment.
      Other definitional concerns include the retrospective nature of the child maltreatment reports, which may underestimate the prevalence of child maltreatment. Individuals may not report child maltreatment due to memory or social desirability biases. Ideally, child maltreatment experiences would be assessed throughout childhood. However, this design faces some insurmountable practical difficulties. For instance, since child maltreatment is generally intrafamilial, there are clear difficulties in assessing abuse on the basis of reports of family members or children. Next, even if such reports could be collected in a valid way, disclosure of child maltreatment during childhood would pose severe ethical problems, since investigators would be obligated to report or intervene to reduce child maltreatment.
      • Fergusson D.M.
      • Lynskey M.T.
      Physical punishment/maltreatment during childhood and adjustment in young adulthood.
      Due to these ethical issues, retrospective reports are commonly used to measure child maltreatment. Furthermore, evidence indicates that self-report validity is increased when data are collected anonymously as was done in this survey.
      • Brener N.D.
      • Collins J.L.
      • Kann L.
      • Warren C.W.
      • Williams B.I.
      Reliability of the youth risk behavior survey questionnaire.
      Additionally, the Add Health study omits any questions about childhood experience of sexual abuse by any people other than the participant’s parents or other adult care-givers, thereby excluding a measure of sexual abuse by a stranger.
      While this study has controlled the connections between child maltreatment and later violent behaviors for a wide range of confounding factors, the possibility remains that the apparent links between child maltreatment and later violent behaviors may be due to the effects of uncontrolled confounding factors. One factor that the Add Health Study did not ask but which clearly requires further examination is the role of childhood exposure to IPV in these associations, because witnessing IPV may be related to exposure to child maltreatment and later youth violence and IPV.
      Another limitation is the inclusion of IPV perpetration or victimization that occurs only in sexual relationships. Excluding nonsexual romantic relationships may alter the results; however, this should be minimal as 86% of romantic relationships were defined as sexual relationships in the Add Health data. There is further potential for selection bias in that the study focuses on young adults who reported sexual relationships in the preceding 2 years. However, the study examined the distribution of child maltreatment and the perpetration or victimization of youth violence between those who reported sexual relationships and those who did not, and found that the percent of those reporting child maltreatment and youth violence perpetration or victimization were approximately equal (no statistically significant difference). Moreover, participants who remained in the sample did not differ substantially from those who were excluded with respect to average age (21.8 years vs 22.0 years), gender (51.7% female vs 51.2% female), and race/ethnicity (71% white vs 68% white).
      Despite these limitations, the study presents strong evidence of the relationship between child maltreatment, youth violence perpetration, and IPV perpetration, and to a lesser extent, evidence of a relationship among child maltreatment, youth violence victimization, and IPV victimization. The study results indicate current trends in prevention efforts that focus on perpetration are appropriate given that the developmental association is stronger in that link. The results also suggest that prevention efforts should occur early in life and should focus on multiple types of violence that occur in the developmental trajectory. It is not enough to focus exclusively on preventing (or treating) one form of violence in adolescence, for example, without recognizing that this population may already have increased exposure to violence as a child that will carry through to adulthood if not addressed accordingly.
      There are other important public health implications as well. Because there are clear direct and/or indirect links among child maltreatment, youth violence, and IPV, this suggests that there are even more health benefits for prevention programs and policies aimed at reducing child maltreatment beyond the health benefits of reducing child maltreatment independent of youth violence and IPV. From an economic viewpoint, this means that assessing the lifetime benefits of preventing child maltreatment would necessarily include the decreased costs associated with the reduction in future youth violence and IPV. These study results provide the necessary probabilities that could be used in future models to assess these lifetime benefits.
      As well, these study findings indicate that gender differences not only exist for the developmental relationship between child maltreatment and future violence perpetration or victimization, but also exist for the role that socioeconomic factors have on future youth violence and IPV perpetration or victimization. Thus, it may be important to account for gender differences when designing the optimal time and setting for violence prevention programs. Without exception, an integrative approach to preventing violence is critical for stopping the developmental trajectory of violence.
      This research uses data from Add Health, a program project designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris, and funded by the National Institute of Child Health and Human Development (grant P01-HD31921), with cooperative funding from 17 other agencies. Special acknowledgment is due Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. To obtain data files from Add Health, contact Add Health, Carolina Population Center, 123 W. Franklin Street, Chapel Hill NC 27516-2524 ( [email protected] ).
      The authors would like to thank the following individuals for their helpful comments on this manuscript: Daniel Whitaker, James Mercy, Janet Saul, Linda Dahlberg, Qian An, Jennifer Wyatt, Rebecca Leeb, and John R. Lutzker.
      The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.
      No financial disclosures were reported by the authors of this paper.

      References

        • National Center for Injury Prevention and Control
        Web-based injury statistics query and reporting system (WISQARS).
        Centers for Disease Control and Prevention, Atlanta GA2006
        • National Center for Injury Prevention and Control
        Costs of intimate partner violence against women in the United States.
        Centers for Disease Control and Prevention, Atlanta GA2003
        • Briere J.
        • Runtz M.
        Childhood sexual abuse: long-term sequelae and implications for psychological assessment.
        J Interpers Violence. 1993; 8: 312-330
        • Capaldi D.M.
        • Clark S.
        Prospective family predictors of aggression toward female partners for at-risk young men.
        Dev Psychol. 1998; 24: 1175-1188
        • Cicchetti D.
        • Toth S.L.
        A developmental psychopathology perspective on child abuse and neglect.
        J Am Acad Child Adolesc Psychiatry. 1995; 34: 541-565
        • Ehrensaft M.K.
        • Cohen P.
        • Brown J.
        • Smailes E.
        • Chen H.
        • Johnson J.G.
        Intergenerational transmission of partner violence: a 20-year prospective study.
        J Consult Clin Psychol. 2003; 71: 741-753
        • Fergusson D.M.
        • Lynskey M.T.
        Physical punishment/maltreatment during childhood and adjustment in young adulthood.
        Child Abuse Negl. 1997; 21: 617-630
        • Ireland T.O.
        • Smith C.A.
        • Thornberry T.P.
        Developmental issues in the impact of child maltreatment on later delinquency and drug use.
        Criminology. 2002; 40: 359-399
        • Loos M.E.
        • Alexander P.C.
        Differential effects associated with self-reported histories of abuse and neglect in a college sample.
        J Interpers Violence. 1997; 12: 340-360
        • Magdol L.
        • Moffitt T.E.
        • Caspi A.
        • Silva P.A.
        Developmental antecedents of partner abuse: a prospective-longitudinal study.
        J Abnorm Psychol. 1998; 107: 375-389
        • Simons R.L.
        • Wu C.
        • Johnson C.
        • Conger R.D.
        A test of various perspectives on the intergenerational transmission of domestic violence.
        Criminology. 1995; 33: 141-172
        • Smith C.A.
        • Thornberry T.P.
        The relationship between childhood maltreatment and adolescent involvement in delinquency.
        Criminology. 1995; 33: 451-481
        • Stith S.M.
        • Rosen K.H.
        • Middleton K.A.
        • Busch A.L.
        • Lundeberg K.
        • Carlton R.P.
        The intergenerational transmission of spouse abuse: a meta-analysis.
        J Marriage Fam. 2000; 62: 640-654
        • Fergusson D.M.
        • Horwood L.J.
        • Lynskey M.T.
        Family change, parental discord and early offending.
        J Abnorm Child Psychol. 1992; 33: 1059-1075
        • Gelles R.J.
        • Cornell C.P.
        Intimate violence in families.
        2nd ed. Sage, Newbury Park CA1990
        • Mocan H.N.
        Economic conditions, deterrence and juvenile crime: evidence from micro data.
        Am Law Econ Rev. 2005; 7: 319-349
        • Sampson R.J.
        • Raudenbush S.
        • Earls F.
        Neighborhoods and violent crime: a multilevel study of collective efficacy.
        Science. 1997; 277: 918-924
        • Wikstrom P.H.
        • Loeber R.
        Do disadvantaged neighborhoods cause well-adjusted children to become adolescent delinquents?.
        Criminology. 2000; 38: 1109-1142
        • Doumas D.
        • Margolin G.
        • John R.S.
        The intergenerational transmission of aggression across three generations.
        J Fam Violence. 1994; 9: 157-175
        • Langhinrichsen-Rohling J.
        • Neidig P.
        • Thorn G.
        Violent marriages: gender differences in levels of current violence and past abuse.
        J Fam Violence. 1995; 10: 159-176
        • Widom C.S.
        The cycle of violence.
        Science. 1989; 244: 60-166
        • Widom C.S.
        Does violence beget violence?.
        Psychol Bull. 1989; 106: 3-28
        • Herrenkohl T.I.
        • Mason W.A.
        • Kosterman R.
        • Lengua L.J.
        • Hawkins J.D.
        • Abbott R.D.
        Pathways from physical childhood abuse to partner violence in young adulthood.
        Violence Vict. 2004; 19: 123-136
        • Walker L.
        Battered women and learned helplessness.
        Victimology. 1977; 2 (/78): 525-534
        • Walker L.
        The battered women syndrome study.
        in: Finkelhor D. Gelles R.J. Hotaling G.T. Straus M.A. The dark side of families. Sage, Beverly Hills CA1983: 31-48
      1. Fang X, Corso P. Gender differences in the connections between violence experienced as a child and perpetration of intimate partner violence in young adulthood. J Fam Violence (to be published).

        • Udry J.R.
        The National Longitudinal Study of Adolescent Health (Add Health), Waves I and II, 1994–1996, and Wave III, 2001–2002.
        Carolina Population Center, University of North Carolina at Chapel Hill, Chapel Hill2003
        • Harris K.M.
        • Florey F.
        • Tabor J.
        • Bearman P.S.
        • Jones J.
        • Udry J.R.
        The National Longitudinal Study of Adolescent Health: research design.
        University of North Carolina, Chapel Hill2003 (Available at: www.cpc.unc.edu/projects/addhealth/design.)
        • Leeb R.T.
        • Mueller M.
        • Swahn M.H.
        • Feldman M.
        • Melanson C.
        Sugar and Spice, Snips and Snails: gender effects on risk factors for delinquency and violence in youth.
        2004 (Paper presented at 2004 Victimization of Children and Youth: An International Research Conference, Portsmouth NH.)
        • Gottfredson D.C.
        • McNeil III, R.J.
        • Gottfredson G.D.
        Social area influences on delinquency: a multilevel analysis.
        J Res Crime Delinq. 1991; 28: 197-226
        • Brown S.E.
        Social class, child maltreatment, and delinquent behavior.
        Criminology. 1984; 22: 259-278
        • Coleman P.K.
        • Maxey C.D.
        • Rue V.M.
        • Coyle C.T.
        Associations between voluntary and involuntary forms of perinatal loss and child maltreatment among low-income mothers.
        Acta Paediatrica. 2005; 94: 176-183
        • Levitt S.D.
        Juvenile crime and punishment.
        J Pol Econ. 1998; 106: 1156-1185
        • Greene W.H.
        Gender economics courses in liberal arts colleges: further results.
        J Econ Educ. 1998; 29: 291-300
        • Greene W.H.
        Econometric analysis.
        5th ed. Prentice Hall, Upper Saddle River NJ2003
        • Archer J.
        Sex differences in physically aggressive acts between heterosexual partners: a meta-analytical review.
        Aggression Violence Behav. 2002; 7: 313-351
        • Moffitt T.E.
        • Caspi A.
        • Rutter M.
        • Silva P.A.
        Sex differences in physical violence and sex similarities in partner abuse.
        in: Moffitt T.E. Caspi A. Rutter M. Silva P.A. Sex differences in antisocial behavior: conduct order, delinquency, and violence in the Dunedin Longitudinal Study. Cambridge University Press, New York2001: 53-70
        • Morse B.J.
        Beyond the conflict tactics scale: assessing gender differences in partner violence.
        Violence Vict. 1995; 10: 251-272
        • Straus M.A.
        • Gelles R.J.
        Physical violence in American families: risk factors and adaptations to violence in 8,145 families.
        Translation Publishers, New Brunswick NJ1990
        • Avery-Leaf S.
        • Cascardi M.
        • O’Leary K.D.
        • Cano A.
        Efficacy of a dating violence prevention program on attitudes justifying aggression.
        J Adolesc Health. 1997; 21: 11-17
        • Hussey J.M.
        • Chang J.J.
        • Kotch J.B.
        Child maltreatment in the United States: prevalence, risk factors, and adolescent health consequences.
        Pediatrics. 2006; 118: 933-942
        • Leeb R.T.
        • Barker L.E.
        • Strine T.W.
        The effect of childhood physical and sexual abuse on adolescent weapon carrying.
        J Adolesc Health. 2007; 40: 551-558
        • Cohen P.
        • Brown J.
        • Smailes E.
        Child abuse and neglect and the development of mental disorders in the general population.
        Dev Psychopathol. 2001; 13: 981-999
        • Felitti V.J.
        • Anda R.F.
        • Nordenberg D.
        • et al.
        The relationship of adult health status to childhood abuse and household dysfunction.
        Am J Prev Med. 1998; 14: 245-258
        • Dube S.R.
        • Anda R.F.
        • Felitti V.J.
        • Chapman D.
        • Williamson D.F.
        • Giles W.H.
        Childhood abuse, household dysfunction and the risk of attempted suicide throughout the life span: findings from Adverse Childhood Experiences Study.
        JAMA. 2001; 286: 3089-3096
        • Cohen P.
        The effect of instruments and informants on ascertainment.
        in: Dunner D.L. Gershon E.S. Barrett J.E. Relatives at risk for mental disorder. Raven Press, New York1988: 31-52
        • Brener N.D.
        • Collins J.L.
        • Kann L.
        • Warren C.W.
        • Williams B.I.
        Reliability of the youth risk behavior survey questionnaire.
        Am J Epidemiol. 1995; 141: 575-580