Skip to main content
PMC home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2018 Oct 1.
Published in final edited form as: J Fam Psychol. 2017 May 22;31(7):833–843. doi: 10.1037/fam0000327

The Legacy of Early Childhood Violence Exposure to Adulthood Intimate Partner Violence: Variable- and Person-Oriented Evidence

Angela J Narayan 1, Madelyn H Labella 2, Michelle M Englund 2, Elizabeth A Carlson 2, Byron Egeland 2
PMCID: PMC5662475  NIHMSID: NIHMS869993  PMID: 28530411

Abstract

This study examined prospective pathways from exposure to interparental violence (EIPV) during infancy (ages 0–24 months) and toddlerhood/preschool (ages 25–64 months) to intimate partner violence (IPV) perpetration and victimization in adulthood (ages 23, 26, and 32 years) using two complementary approaches. Building on past findings, a variable-oriented approach was used to examine the effects of developmental timing of EIPV in infancy versus toddlerhood/preschool to IPV involvement in early adulthood, at age 23 years. A person-oriented approach next examined whether continuity and change in IPV (persisting, increasing, and decreasing versus nonviolent patterns) across the transition from early adulthood to adulthood (ages 26 to 32 years) were predicted by developmental timing of EIPV within early childhood and/or contemporaneous adulthood factors (life stress and behavior problems). In this fully prospective longitudinal study beginning at birth, mothers reported on EIPV in infancy and toddlerhood/preschool, and participants (N = 179) reported on IPV and contemporaneous stress and behavior in early adulthood and adulthood. Results indicated that according to the variable-oriented approach, EIPV in toddlerhood/preschool but not in infancy predicted both IPV perpetration and victimization at age 23. The person-oriented approach revealed that, along with life stress and externalizing behavior, EIPV in toddlerhood/preschool, but not infancy, also differentiated patterns of IPV from ages 26 to 32. Findings converge on toddlerhood/preschool as a particular promising developmental period to intervene and deter long-term effects of EIPV on IPV across the transition from early adulthood to adulthood.

Keywords: intimate partner violence, infancy, toddlerhood/preschool, variable-oriented approach, person-oriented approach, developmental psychopathology

Children with exposure to interparental violence (EIPV) are at elevated risk for long-term maladaptation across multiple developmental domains (Davies & Woitach, 2008; Holt, Buckley, & Whelan, 2008; Kitzmann, Gaylord, Holt, & Kenny, 2003). Childhood EIPV is linked to more internalizing and externalizing behavior, criminal offending, involvement in relationship violence, and other adverse outcomes across the lifespan (Narayan, Englund, & Egeland, 2013; Bosquet Enlow, Egeland, Blood, Wright, & Wright, 2012; Fergusson & Horwood, 1998; Moffitt & the 2013 Klaus-Grawe Think Tank, 2013). Growing evidence suggests that EIPV during early childhood (birth to 5 years) is a particularly potent predictor of maladjustment because self-regulatory protective competencies are not yet fully developed (Howell, 2011; Kitzmann et al., 2003).

Clarifying the sequelae of early EIPV is crucial given the high prevalence of violence in families with young children. Children aged 0 to 5 are disproportionately likely to witness family violence and be directly involved (e.g., by intervening between parents or trying to call for help; Fantuzzo, Boruch, Beriama, Atkins, & Marcus 1997; Howell, 2011). In the National Survey of Children’s Exposure to Violence (NatSCEV), a nationally representative study of approximately 4,500 children, 17.2% of children aged 5 or younger were exposed to verbal, physical, or psychological violence involving parental figures (Hamby, Finkelhor, Turner, & Ormrod, 2011); this substantial percentage corresponds to approximately 4.2 million American children. Furthermore, 11.9% of children were exposed to physical violence alone, with an estimated 90% of these children directly seeing or hearing the events (Hamby et al., 2011).

Although all forms of violence (including psychological and sexual forms) between caregivers may be just as damaging for children to witness, the present study focuses on physical violence exposure, a particularly overt and threatening form of interparental conflict. In the short term, witnessing physical violence produces reactions of fear and dysregulation, and in the long run, it may be internalized as a dysfunctional model of conflict resolution and reactivated later in life (Davies & Woitach, 2008; Sroufe, Egeland, Carlson, & Collins, 2005). Indeed, early childhood EIPV is prospectively associated with relationship conflict in adolescence and intimate partner violence (IPV) perpetration and victimization in adulthood (Fite et al., 2008; Narayan et al., 2013; Narayan, Englund, Carlson, & Egeland, 2014). However, little is known about timing effects of EIPV within the early childhood period; for instance, it is unclear whether EIPV during infancy (birth-24 months) or toddlerhood/preschool (25–64 months) differentially predicts IPV in adulthood or contributes to changes in IPV across different adulthood periods.

The present study investigated developmental pathways linking exposure to physical EIPV during distinct stages of early childhood to involvement and change in IPV in adulthood. Following the NatSCEV survey (Hamby et al., 2011) and previous research by this team (Narayan et al., 2013, 2014), we adopt the term “exposure to interparental violence” (EIPV) to reflect physical violence (i.e., throwing something, pushing, slapping, kicking, hitting, beating up, threatening or using weapons; Straus, 1979) between a child’s parents or parental figures. Following the Center for Disease Control and Prevention (CDC), we use the term “intimate partner violence” to indicate the above defined physical violence occurring between adult romantic partners (CDC, 2012). As with EIPV, high rates of IPV have been documented in large-scale nationally representative studies. According to the National Intimate Partner and Sexual Violence Survey of over 16,500 adults, about one of four American women and one of seven American men has experienced severe physical violence (e.g., being hit with a closed fist; Black et al., 2011). Understanding pathways linking EIPV to IPV, including early developmental windows of particular salience, is critical for deterring intergenerational cycles of violence.

Developmental Timing of EIPV within Early Childhood

A handful of prospective studies have evaluated timing effects of EIPV across different periods of childhood, documenting long-term effects of EIPV early in development (Bosquet et al., 2012; Hibel, Granger, Blair, Cox, & the Family Life Key Project Investigators, 2011; Narayan et al., 2013; Yates, Dodds, Egeland, & Sroufe, 2003). Prospective research found that whereas EIPV in middle childhood (first through third grade) predicted contemporaneous behavior problems, EIPV in early childhood (birth to 5 years) predicted behavior problems in adolescence (Yates et al., 2003). EIPV across early childhood (but not middle childhood) was also found to directly predict IPV involvement between ages 21 and 23 years (Narayan et al., 2013).

Prospective research on developmental timing of EIPV within early childhood, however, is extremely limited. In a sample of predominantly low-income rural families, EIPV early in infancy was not related to adrenocortical functioning at seven or 15 months, but EIPV by 24 months predicted greater cortisol reactivity, suggesting the importance of cumulative violence exposure in the first two years of life (Hibel et al., 2011). Another study found that interpersonal trauma (i.e., EIPV, child abuse and/or neglect) occurring during infancy (0 to 2 years) but not preschool (2 to 5 years) predicted lower cognitive ability in middle childhood (Bosquet Enlow et al., 2012). Together, this evidence is mixed on whether EIPV is harmful for subsequent adaptation if it occurs during infancy or later in the early childhood period. Clarifying developmental timing effects of EIPV within early childhood would be particularly informative for policy and prevention efforts, as it would elucidate specific developmental windows to target.

A Developmental Psychopathology Perspective on EIPV and IPV

This study is grounded in the developmental psychopathology (DP) perspective, which emphasizes the role of early experiences in future adaptation and conceives of development as a hierarchically integrated progression through a series of stage-salient tasks. According to the DP perspective, developmental pathways reflect transactional relations between the individual and the environment, with early experiences carried forward in the form of internalized expectations about social relationships and the self (Cicchetti & Toth, 2009; Lieberman & Van Horn, 2008; Sameroff, 2000; Sroufe, 1979). Early childhood experiences, particularly those relevant to stage-salient tasks described below, set the stage for long-term social functioning and interpersonal relationships (Lieberman & Van Horn, 2008; Sroufe, 1979; Waters & Cummings, 2000).

EIPV in infancy and toddlerhood/preschool

EIPV that occurs in infancy would be expected to increase the risk for long-term relationship dysfunction by disrupting infants’ perceptions of basic interpersonal safety and interfering with the development of secure attachment bonds, a key developmental task (Davies & Woitach, 2008; Lieberman & Van Horn, 2008; Sroufe, 1979). Indeed, children who are exposed to interpersonal trauma in infancy (e.g., violence, maltreatment, dissociative parenting) are at higher risk for psychopathology in adulthood (Carlson, 1998; Sroufe et al., 2005); however, specific links between EIPV in infancy and IPV in adulthood remain untested.

EIPV that occurs during toddlerhood/preschool would be expected to increase the risk for IPV by disrupting different developmental tasks. EIPV in toddlerhood/preschool may undermine emotional and behavioral self-regulation, interfere with healthy models of conflict resolution, and compromise emerging social competence. Thus, EIPV during this period may portend dysregulation and relationship conflict in adulthood, rendering exposed children more vulnerable to IPV (Holt et al., 2008; Kitzmann et al., 2003; Sroufe et al., 2005; Waters & Cummings, 2000).

IPV in adulthood

In addition to examining the distinct contributions of EIPV in infancy versus toddlerhood/preschool, this study also focused on IPV occurring across age-salient adulthood transitions. Developmental scientists have begun to recognize that adulthood, like childhood, is composed of distinct developmental stages. Recent research differentiates early or emerging adulthood (comprising the late teens and early 20s) from adulthood proper, reflecting societal trends of extended identity exploration during early adulthood and delayed attainment of traditional adult milestones, such as marriage, co-parenting, and financial independence (Arnett, 2000; Collins & van Dulmen, 2006). Romantic relationships during these stages might be expected to serve different functions, with earlier relationships supporting identity exploration and later relationships supporting family life. Although not all individuals fit this pattern, the trend of developmental stages within adulthood raises important questions about continuity and change in violent behavior across transitional periods (Piquero, Carriaga, Diamond, Kazemian, & Farrington, 2012; Smith, Ireland, Park, Elwyn, & Thornberry, 2011). Whereas studies have documented that IPV during adolescence and young adulthood predicts continuations of both IPV perpetration and victimization into adulthood (Cui, Ueno, Gordon, & Fincham, 2013; Smith et al., 2011), they have not examined whether continuity and change in IPV (i.e., persisting versus increasing or decreasing involvement) stem from EIPV within early childhood.

In addition to emphasizing the salience of early experience, the DP perspective acknowledges the importance of contextual influences across the lifespan (Cicchetti & Toth, 2009; Sameroff, 2000). Existing evidence confirms the role of the current environment, including contextual stress, in predicting IPV. Chronic and acute stress levels account for substantial variation in violent behavior above and beyond an individual’s developmental history (Bell & Naugle, 2008; Bogat, Levendosky, Theran, Von Eye, & Davidson, 2003). Stress can amplify vulnerabilities in a relationship while taxing an individual’s self-regulatory capacity, increasing the likelihood of violent conflict (Bogat et al., 2003; Cano & Vivian, 2003).

Individual-level risk factors (e.g., history of EIPV, internalizing or externalizing psychopathology) and relationship-level risk factors (e.g., chronic relationship stress, prior IPV) can function independently or in conjunction with acute stress to predict continuity and change of IPV over time (Langer, Lawrence, & Barry, 2008; Roberts, McLaughlin, Conron, & Koenen, 2011). Thus, in addition to examining whether EIPV in infancy versus toddlerhood/preschool more strongly predicted IPV in adulthood, the present study investigated whether contemporaneous adulthood factors (i.e., co-occurring life stress and internalizing and externalizing behavior) predicted changing patterns of IPV from early adulthood (age 26) to adulthood (age 32). Identifying patterns of change within samples (e.g., persisting in vs. desisting from IPV) resonates with ongoing calls for person-oriented research to complement variable-oriented approaches in developmental science (Bergman & Magnusson, 1997; Bogat, Levendosky, & von Eye, 2005; Levendosky, 2013).

Variable-Oriented versus Person-Oriented Approaches

Variable-oriented analyses characterize associations between independent and dependent variables across an entire sample of interest, without regard to differing patterns in how these variables co-occur. In contrast, person-oriented analyses seek to identify inter-individual differences (between subgroups of a sample) or intra-individual differences (within individuals over time) and classify how individuals group together on meaningful characteristics (Bergman & Magnusson, 1997; Bogat et al., 2005; Nurius & Macy, 2008). Whereas variable-oriented analyses of violent behavior typically document group-level risk processes (e.g., the magnitude of direct or indirect effects on negative outcomes), person-oriented analyses can isolate subgroups of risk-exposed individuals who cluster on specific characteristics associated with ongoing vulnerability or resilience (Bergman & Magnusson, 1997; Nurius & Macy, 2008).

Person-oriented researchers argue that developmental, behavioral, and contextual factors should be considered when distinguishing person-specific profiles of violent behavior (Bell & Naugle, 2008; Bogat et al., 2005; Piquero et al., 2012). To date, however, person-oriented studies have mostly focused on concurrent rather than prospective predictors (Bell & Naugle, 2008; Bogat et al., 2005; Swartout & Swartout, 2012), and several are limited to classifying males as IPV perpetrators or females as IPV victims (Bogat et al., 2005; Holtzworth-Munroe & Meehan, 2004), despite evidence that both sexes play both roles (Archer, 2000; Cui et al., 2013). Here, the person-oriented approach examined developmental and contemporaneous factors, in addition to demographic factors such as sex, that accounted for within-sample change in patterns of IPV.

Study Overview, Aims and Hypotheses

The present study investigated EIPV within two early childhood periods, infancy (birth-24 months), and toddlerhood/preschool (25–64 months) to build on past research. In a previous study, Narayan and colleagues (2013) found that EIPV across the whole early childhood period (ages 0–64 months) predicted IPV involvement at age 23, whereas EIPV in middle childhood (grades 1–3) did not. Thus, the first aim was to use a variable-oriented approach to further hone in on long-term effects of EIPV within early childhood when predicting IPV at age 23. Given the enduring importance of interpersonal experiences in both infancy and toddlerhood/preschool, we hypothesized that EIPV during both periods would independently predict IPV at age 23.

The specific ages spanning the two early childhood periods were selected with reference to the unique age-salient tasks of infancy and toddlerhood/preschool. During the first two years of life, formation of attachment bonds, co-regulation of distress, and development of trust and security are the focal tasks. Alternatively, the period from 2 to 5 years involves a reorganization of parent-child relationships from co-regulation and dependence to self-regulation and emerging autonomy (and often begins with the “terrible twos”), and is also marked by increasing development of peer relationships (Erikson, 1963; Sroufe, 1979). Additionally, past research from the Minnesota Longitudinal Study of Risk and Adaptation (MLSRA), the larger study from which this study was drawn, has also differentiated the two periods in this way.

Consistent with a person-oriented approach, the second aim was to examine subgroups of individuals for continuity and change in patterns of IPV perpetration and victimization (persisting, increasing, decreasing, and non-violent patterns) across the transition from early adulthood to adulthood (ages 26 to 32). Given limited research in this area, exploratory person-oriented analyses were conducted with a focus on how EIPV in infancy and toddlerhood/preschool, as well as contemporaneous factors (life stress, internalizing and externalizing behavior), differentiated among patterns of IPV.

Method

Participants

Participants (N = 179, 52% male; 67% Caucasian, 11% African-American, 17% biracial/multiracial, and 5% other/unreported) were a subset of the MLSRA followed from birth into adulthood. The MLSRA began in 1978 with 267 low-income mothers recruited during pregnancy, and their first-born children. At recruitment, 61% of mothers were unmarried, 50% were teenagers (mean age = 20.50, SD = 3.74, range = 12–34 years), and 41% had not completed high school. Their children were followed from birth with home, school, and lab visits occurring regularly from infancy through adolescence and periodically into adulthood. Participants were included here if they participated in at least one of three adulthood assessments at 23, 26, or 32 years (see Missing Data section for analyses of attrition and missing data). The Institutional Review Board at the University of Minnesota approved all procedures.

Measures–Primary Variables

Exposure to interparental violence

Throughout early childhood (ages 12, 18, 24, 30, 42, 48, 54, and 64 months), mothers completed interviews with open-ended questions about physical violence victimization in the home. Maternal responses to these questions, as well as their spontaneous remarks about the presence of physical violence across the entire interview, were later reviewed and coded by two trained raters for EIPV. Trained coders rated all interviews on a 0–7 point Likert scale from 0-No evidence of violence to 7-Most severe form of violent interaction (see Yates et al., 2003 for full scale description). Ratings were collapsed across the periods within infancy and toddlerhood/preschool, and the highest score was used.. Of note, this rating scale took into account severity, injury, and chronicity of ratings, with higher scores reflecting maternal victimization resulting in greater injuries, and/or more chronic victimization in the home. Raters achieved excellent reliability [intraclass correlation coefficients (ICCs) = .93 to .99]. Previous research has shown these EIPV ratings to also have predictive validity for later cognitive and behavioral functioning (Bosquet Enlow et al., 2012; Yates et al., 2003).

Intimate partner violence

A ten-item Conflict Tactics Scale (CTS; Straus, 1979), including two verbal and eight physical items, was administered to participants at ages 23, 26, and 32 years. This study focused on eight physical aggression items (i.e., threw something at partner, pushed, slapped, kicked, hit, beat up, threatened with a gun or knife, and used a gun or knife). At each assessment, participants were asked whether they had experienced each behavior as perpetrator or victim with a current or past partner. At each subsequent age, they were asked about any violence with their current or past partner since the prior assessment. The IPV variables at age 23 reflected any IPV since age 21, the IPV variables at age 26 reflected any IPV since age 23, and the IPV variables at age 32 reflected any IPV since age 26. Positively endorsed items on the physical aggression subscale were summed to create two scores (0 to 8) for perpetration and victimization during each of the three periods. This computational method, known as the variety score, is commonly used with the CTS because of its favorable psychometric properties: it minimizes skew, reduces estimation errors in frequency reporting, and eliminates the need for presumed severity ratings (Iverson et al., 2013; Moffitt et al., 1997).

Life stress

At ages 26 and 32 years, participants completed the Life Events Scale (Egeland, Breitenbucher, & Rosenberg, 1981), a semi-structured, interviewer-administered questionnaire. The 41 items reflect stressful events or transitions (e.g., physical or mental health problems, financial transitions or strain, personal or interpersonal stressors) occurring over the past year. For each item, participants were first asked a yes/no question (e.g., “Did you experience trouble with bosses or continued tension at work?”), which, if positively endorsed, was followed by a probe for more details. Responses were coded for severity of resulting stress on weighted 0–3 point scales from No Disruption to Highly Disruptive. At each assessment period, a subset of 50 cases was coded by two independent raters to establish reliability, which was excellent (ICCs = .98 and .96 for ages 26 and 32, respectively; overall α = .63, 68). Analyses used the total life stress score, minus one item pertaining to IPV that overlapped with the CTS.

Externalizing and internalizing behavior

Participants completed the Young Adult Self Report (YASR; Achenbach, 1997) and the Adult Self Report (ASR; Achenbach & Rescorla, 2003) at ages 26 and 32 years, respectively. These self-report forms contain 132 items (YASR) and 126 items (ASR) of adaptive/maladaptive behaviors on 3-point scales, 0-Not True to 2-Very True or Often True. T-scores were derived for Internalizing Problems (e.g., depressed or anxious behaviors) and Externalizing Problems (e.g., aggressive or hyperactive behaviors).

Measures–Control Variables

Demographic factors

Control variables were participant sex, maternal age at participants’ birth, and SES during the prenatal period. SES T-scores were calculated using the revised Duncan Socioeconomic Index based on highest educational attainment, annual income, and head of household occupational status (Duncan, 1961; Stevens & Featherman, 1981).

Adverse caregiving: Abuse and neglect

The MLSRA uses the rubric childhood experiences of adverse caregiving as an umbrella term to refer to a variety of atypical parent-child experiences that were prospectively measured in the MLSRA cohort. The present study included information collected about MLSRA participants’ adverse caregiving experiences of physical abuse, sexual abuse, and physical neglect. Physical abuse was defined as the intentional use of physical force by a caregiver against a child resulting in or having potential to result in physical injury. Sexual abuse, was defined as sexual contact (e.g., molestation, rape) or noncontact exploitation (e.g., intentional exposure of child to pornography) by a custodial caregiver or by a perpetrator five or more years older than the target child. Physical neglect included the child’s basic physical and cognitive needs and was defined as a caregiver’s failure to provide adequate hygiene, shelter, clothing, medical care, supervision, or education.

Information on abuse and neglect was collected from birth to 17.5 years (up to 25 assessments); obtained from parent-child observations, caregiver interviews, reviews of available child protection and medical records, adolescent reports, and teacher interviews; and reviewed by trained coders to determine the presence or absence of each type of adverse caregiving. Two coders reviewed each case and demonstrated good to excellent reliability: kappa coefficients were all between .80 and .98 for presence or absence of physical abuse, sexual abuse, and physical neglect. All discrepancies were resolved by consensus. (See Raby et al., in press, for additional information about how all abuse and neglect variables were coded). Several studies from the MLSRA have documented strong predictive validity of abuse and neglect with negative outcomes spanning from childhood to adulthood, including poorer attachment relationships, emotional regulation skills, and self-esteem; lower school engagement; less effective peer and romantic relationships, and increased psychopathology (see Sroufe et al., 2005 for a review). The present study included any physical or sexual abuse, or physical neglect occurring from birth to 64 months (concurrent with early childhood EIPV). Within this subsample, 37.4% of participants had experienced any physical abuse, sexual abuse, and/or neglect during early childhood.

Data Analytic Plan

Analyses proceeded in two parts in SPSS version 22. The variable-oriented approach examined whether timing of EIPV in infancy versus toddlerhood/preschool predicted IPV by age 23 years. Two hierarchical linear regressions were conducted with IPV perpetration and victimization at age 23 as the outcome variables. Covariates (maternal age, SES, abuse/neglect, and child sex) were included as controls in the first step. Younger maternal age and lower SES are known risks for EIPV (Fantuzzo et al., 1997; Fergusson & Horwood, 1998; Hibel et al., 2011; Yates et al., 2003), and child abuse/neglect often covaries with EIPV, prompting calls to isolate EIPV from direct abuse and neglect (Holt et al., 2008; Lieberman & Van Horn, 2008). EIPV during infancy was entered in the second step, and EIPV during toddlerhood/preschool was entered in the third step. Regressions were examined for influential cases using Cook’s d > 4/n (Cook & Weisberg, 1982; Rawlings, 1988), and none affected the findings.

For the person-oriented approach, an exploratory two-step cluster analysis was conducted. The two-step approach minimizes the distances between cases by placing them in preliminary clusters and then uses hierarchical clustering to aggregate the preclusters into the best-fitting number of groups. Goodness of fit was determined by using the log-likelihood procedure and examining the silhouette measure of cohesion and separation for the final solution. Good fit was considered ≥.50, at which point the clusters show adequate within-group similarity (cohesion) and between-group difference (separation; Norusis, 2005–2012).

Four variables (continuous IPV perpetration and victimization scores at ages 26 and 32) were entered as indicators for cluster membership. First a series of multinomial logistic regressions were conducted to examine whether EIPV in infancy versus toddlerhood/preschool predicted cluster membership after accounting for covariates. Second, mean level differences in contemporaneous variables (life stress, internalizing and externalizing behavior at ages 26 and 32) across each of the clusters were examined using analysis of covariance (ANCOVA) tests. Significant ANCOVAs were further examined for significant pairwise contrasts. All p-values were corrected using the Bonferroni approach (α’ = .05/6 = .008), and only pairwise contrasts meeting this corrected alpha are reported. The same four covariates were used in all analyses.

Missing Data

Of the 179 participants, 142 completed all three adulthood assessments. The remainder did not participate in one or two of the adult assessment waves; 11.2% (n = 20) did not participate at age 23, 9.5% (n = 17) did not participate at age 26, and 8.9% (n = 16) did not participate at age 32. There were no significant differences on covariates between participants who had complete versus incomplete adulthood data, or between the current sample and the original MLSRA study participants. For variable-oriented analyses (n = 179), missing data were minimal at 4.5% across the entire dataset, including EIPV, IPV, and covariates. Person-oriented analyses (n = 151) used the subsample of individuals who participated in assessments specifically at ages 26 and 32. Data were imputed across 20 datasets using fully conditional specification and results were pooled across all imputed sets (Rubin, 1987; Shafer & Graham, 2002). Pooled results did not differ from the raw data, so results from the raw data are reported.

Results

Descriptive Analyses

Table 1 shows the bivariate correlations and descriptive statistics for all variables in the hierarchical regression analyses. Rates of EIPV in infancy were 19.6% (n = 35), of whom 34.3% (n = 12) witnessed EIPV in infancy only, but not in toddlerhood/preschool. Rates of EIPV in toddlerhood/preschool were 31.8% (n = 57), of whom 59.6% (n = 34) witnessed EIPV only in toddlerhood/preschool, but not in infancy; 12.8% of participants (n = 23) witnessed EIPV across both periods. Rates of IPV perpetration were 27.4% (n = 49) at age 23, 31.8% (n = 57) at age 26, 33.0% (n = 59) age 32, and 53.6% (n = 96) across any period. Rates of IPV victimization were 35.8% (n = 64) at age 23, 38.0% (n = 68) at age 26, 34.6% (n = 62) at age 32, and 63.1% (n = 113) across any period. Infancy EIPV was not related to age 23 IPV, but toddlerhood/preschool EIPV was significantly associated with age 23 perpetration (r = .27, p < .01) and victimization (r = .24, p < .01).

Table 1.

Bivariate Correlations Between All Variables in the Variable-Oriented Hierarchical Regressions

Primary Variables 1 2 3 4 5 6 7 8
1. Infancy EIPV --
2. Preschool EIPV .40** --
3. IPV Perpetration 23y .11 .27** --
4. IPV Victimization 23y .06 .24** .56** --
Control Variables
5. Child Sex (Female) .08 .14 .17* −.03 --
6. SES −.14 −.11 −.06 −.12 −.14 --
7. Maternal Age −.15* −.19* −.12 −.25** −.03 .44** --
8. Abuse/neglect .29** .12 .05 .09 −.06 −.24** −.22* --
Mean (or % if binary) 0.98 1.57 .58 .97 48.04% 50.57 20.79 37.43%
Standard Deviation 2.10 2.47 1.08 1.62 NA 9.93 3.76 NA
Range 0–7 0–7 0–4 0–8 NA 32–106 15–34 NA
Open in a new tab

Note.

**

p < .01,

*

p < .05.

Variable-Oriented Approach: Regression Analyses for Developmental Timing of EIPV

Results of both hierarchical regressions are presented in Table 2. Higher levels of EIPV in toddlerhood/preschool but not in infancy significantly predicted higher levels of IPV perpetration at age 23 (β= .26, p < .01), providing a significant increase in variance over prior steps (ΔR2 = .06, p < .01). Similarly, higher levels of EIPV in toddlerhood/preschool but not infancy significantly predicted higher levels of IPV victimization at age 23 (β = .25, p < .01), also accounting for a significance increase in variance (ΔR2 = .05, p < .01).

Table 2.

Regressions of IPV Perpetration and Victimization at 23 Years on Developmental Timing of EIPV in Infancy and Toddlerhood/Preschool

IPV Perpetration
IPV Victimization
B SE β R2 F ΔR2 B SE β R2 F ΔR2
Step 1: Covariates .03 1.20 .03 .07 2.62* .07
Child Sex (Female) .29 .18 .14 −.18 .27 −.06
SES .01 .01 .05 .00 .02 .00
Maternal Age −.04 .03 −.14 −.11 .04 −.24**
Abuse/neglect .07 .19 .03 .15 .28 .05
Step 2: Infancy EIPV .04 1.23 .01 .07 2.12 .00
Child Sex (Female) .28 .18 .13 −.19 .27 −.06
SES .01 .01 .05 .00 .02 .00
Maternal Age −.04 .03 −.13 −.11 .04 −.24**
Abuse/neglect .01 .20 .01 .12 .29 .04
Infancy EIPV .05 .04 .10 .03 .07 .03
Step 3: Preschool EIPV .10 2.47* .06** .12 3.20** .05**
Child Sex (Female) .21 .18 .10 −.28 .26 −.09
SES .00 .01 .03 .00 .02 −.02
Maternal Age −.03 .03 −.09 −.09 .04 −.21*
Abuse/neglect .00 .19 .00 .10 .28 .03
Infancy EIPV .00 .05 .00 −.05 .07 −.07
Preschool EIPV .11 .04 .26** .16 .06 .25**
Open in a new tab
*

p < .05,

**

p < .01

Person-Oriented Approach: Cluster Analysis for Patterns of IPV across Adulthood

The final cluster solution indicated four clusters and showed good fit with the data (average silhouette = .60). No cases were excluded as outliers. All four of the IPV variables had substantial predictive importance in the final cluster solution: age 32 IPV victimization (1.0), age 32 IPV perpetration (.77), age 26 IPV perpetration (.57), and age 26 IPV victimization (.52). All IPV variables showed significant mean differences across clusters: age 26 perpetration, F(3, 150) = 58.72, p < .01; age 26 victimization, F(3, 150) = 51.51, p < .01; age 32 perpetration, F(3, 150) = 91.68, p < .01; and age 32 victimization, F(3, 150) = 140.84, p < .01.

The four clusters had the following IPV characteristics (see Table 3): Cluster 1 (n = 8) had the highest mean levels on all four IPV variables (age 26 perpetration and victimization and age 32 perpetration and victimization) and was labeled “IPV Persisting.” Cluster 2 (n = 34) had low levels of perpetration and victimization at age 26 but moderately high levels of IPV perpetration and victimization at age 32 and was labeled “IPV Increasing.” Cluster 3 (n = 28) had moderately high levels of IPV perpetration and victimization at age 26 but low levels of perpetration and victimization at age 32 and was labeled “IPV Decreasing.” Cluster 4 (n = 81) had low levels of IPV perpetration and victimization at both ages and was labeled “Non-Violent.” There were no significant differences across clusters for any of the covariates.

Table 3.

Descriptive Statistics and Pairwise Contrasts for IPV Clusters

Cluster 1 (n = 8)Persisting Cluster 2 (n = 34)Increasing Cluster 3 (n = 28)Decreasing Cluster 4 (n = 81)Non-Violent Total Sample Significant Contrast Sample Range
IPV Variables*
Perpetration 26y 3.50 (1.77) .32 (.59) 2.00 (1.59) .19 (.39) .73 (1.31) 1 v 2,4; 2 v 3,4 0–6
Victimization 26y 3.38 (2.93) .71 (.84) 3.00 (1.91) .21 (.49) 1.04 (1.68) 2 v 3; 3 v 4 0–7
Perpetration 32y 4.25 (1.17) 1.26 (1.08) .61 (.83) .11 (.32) .71 (1.21) 1 v all; 2 v 4 0–6
Victimization 32y 5.00 (1.77) 2.53 (1.35) .54 (.74) .04 (.19) 1.01 (1.64) 1 v 3,4; 2 v 3,4 0–7
Contemp.Variables†
Life Stress 26y 19.85 (2.14) 9.79 (1.01) 12.39 (1.16) 8.23 (.64) 10.03 (6.12) 1 v all; 3 v 4 0–30
Externalizing 26y 63.49 (3.34) 51.62 (1.58) 53.52 (1.82) 48.95 (1.00) 51.22 (9.15) 1 v 2,4 30–75
Internalizing 26y 57.20 (3.79) 48.34 (1.79) 52.02 (2.06) 46.91 (1.14) 48.74 (10.28) n.s. 26–77
Life Stress 32y 20.54 (2.23) 10.45 (1.05) 9.94 (1.21) 6.19 (.67) 8.47 (6.81) 1 v all; 2 v 4 0–36
Externalizing 32 y 61.56 (3.64) 54.02 (1.72) 51.62 (1.98) 47.98 (1.10) 50.66 (10.08) 1 v 4; 2 v 4 30–82
Internalizing 32y 55.38 (4.31) 50.13 (2.03) 51.30 (2.35) 45.79 (1.30) 48.64 (11.96) n.s. 30–87
Open in a new tab

Note. Contemp. = Contemporaneous. Bolded items reflect significant contrasts. All significant contrasts p < .008 following Bonferroni correction.

*

Descriptive statistics for IPV variables are means and standard deviations.

†

Descriptive statistics for contemporaneous variables are estimated marginal means and standard errors.

Four multinomial regressions were conducted to allow each cluster to be a reference group and examine the six possible pairwise contrasts (Table 4). Results from the first regression indicated that compared to the Non-Violent group, individuals in the IPV Increasing group had significantly higher levels of EIPV in toddlerhood/preschool (adjusted odds ration (OR) = 1.41, 95% CI = [1.15–1.75], p < .01). Results from the second regression also revealed that compared to the IPV Decreasing group individuals in the IPV Increasing group also had significantly higher levels of EIPV in toddlerhood/preschool (OR = 1.37, 95% CI = [1.07–1.76], p < .05). Results from the third and fourth regressions with the IPV Increasing and IPV Persisting groups as the reference groups revealed no significant pairwise contrasts involving EIPV.

Table 4.

All Pairwise Contrasts from Multinomial Regressions

Cluster (C) Contrasts (C1 = Persisting, C2 = Increasing, C3 = Decreasing, C4 = Non-Violent)
Early EIPV C1 vs C4 OR (95% CI) C2 vs C4 OR (95% CI) C3 vs C4 OR (95% CI) C1 vs C3 OR (95% CI) C2 vs C3 OR (95% CI) C1 vs C2 OR (95% CI)
Infancy EIPV 1.24 (.87–1.77) 1.03 (.81–.1.31) 1.06 (.84–1.35) 1.17 (.79–1.72) .97 (.73–1.29) 1.20 (.81–1.77)
Todd./Preschool EIPV 1.01 (.72–1.43) 1.41**(1.15–1.72) 1.03 (.83–1.27) .99 (.68–1.43) 1.37*(1.07–1.760 .72 (.50–1.04)
Covariates
Child Sex (Female) 1.68 (.29–9.65) .36*(.14–.94) .79 (.30–2.04) 2.13 (.33–13.94) .46 (.14–1.49) 4.64 (.71–30.24)
SES .97 (.87–1.09) .94 (.88–1.01) 1.03 (.97–1.09) .95 (.84–1.07) .92*(.84–.99) 1.03 (.91–1.17)
Maternal Age 1.02 (.80–1.31) 1.12 (.97–1.28) .88 (.74–1.04) 1.17 (.87–1.55) 1.27*(1.04–1.56) .92 (.71–1.19)
Abuse/neglect .79 (.13–4.79) .46 (.15–1.31) 1.14 (.41–3.20) .69 (.10–4.86) .40 (.11–1.45) 1.73 (.25–12.10)
Open in a new tab
*

p < .05,

**

p < .01

A series of ANCOVAS examined whether cluster membership was associated with mean level differences in life stress and behavior problems at ages 26 and 32 (Table 3). The association between cluster membership and life stress at age 26 was significant, F(3, 132) = 10.99, p <.01, η2 = .20. The IPV Persisting cluster had significantly higher life stress than all other clusters, and the IPV Decreasing group had significantly higher life stress than the Non-Violent cluster. The association between cluster membership and externalizing behavior at age 26 was also significant, F(3, 132) = 6.76, p < .01,η2 = .13; the IPV Persisting cluster had significantly higher externalizing behavior than the Increasing and Non-Violent clusters. The association between cluster membership and internalizing behavior at age 26 was also significant, F(3, 132) = 3.56, p < .05, η2 = .08, but no Bonferroni-corrected contrasts remained.

At age 32, the association between cluster membership and life stress was significant, F(3, 132) = 15.40, p < .01, η2 = .26. The IPV Persisting cluster had significantly higher levels of life stress than all other clusters, and the IPV Increasing group had significantly higher levels than the Non-Violent cluster. The association between cluster membership and externalizing behavior at age 32 was also significant, F(3, 132) = 6.36, p < .01, η2 = .13. The IPV Persisting cluster had significantly higher externalizing behavior than the Non-Violent cluster, and the IPV Increasing cluster had significantly higher externalizing behavior than the Non-Violent cluster. Finally, the association between cluster membership and internalizing behavior at age 32 was significant, F(3, 132) = 2.97, p < .05, η2 = .06, but no corrected contrasts remained significant.

Discussion

This study used both variable-oriented and person-oriented approaches to provide novel evidence that developmental timing of EIPV within early childhood is crucial to understanding the etiology of IPV involvement in adulthood. Results demonstrated that EIPV timing, as well as contemporaneous factors, accounted for continuity and change of IPV across the transition from early adulthood to adulthood. According to the variable-oriented linear regression analyses, higher severity EIPV in toddlerhood/preschool, but not infancy, predicted IPV perpetration and victimization by age 23. Holding the effects of EIPV in infancy constant, EIPV in toddlerhood/preschool predicted IPV perpetration and victimization in adulthood over and above the effects of EIPV in infancy. Extending past research (Fite et al., 2008; Narayan et al., 2013; Yates et al., 2003) these findings indicate that EIPV during the toddlerhood/preschool period is a particularly salient prospective predictor of both IPV perpetration and victimization in early adulthood.

According to the person-oriented cluster analyses, EIPV during toddlerhood/preschool also predicted change in IPV involvement from early adulthood to adulthood. EIPV in toddlerhood/preschool, but not in infancy, significantly differentiated cluster membership: the IPV Increasing group (who exhibited low IPV perpetration and victimization at age 26 but high perpetration and victimization at age 32) had significantly higher levels of toddlerhood/preschool EIPV than individuals who did not engage in IPV during ages 26 to 32 (the Non-Violent group) and individuals who declined in IPV from ages 26 to 32 (the Decreasing group). Thus, EIPV in toddlerhood/preschool predicted new-onset cases of IPV across early adulthood to adulthood and also differentiated individuals who increased versus decreased in IPV during this transition. These patterns would not have been evident with variable-oriented analyses, demonstrating the utility of person-oriented approaches (Bogat et al., 2005; Nurius & Macy, 2008).

The developmental salience of EIPV in toddlerhood/preschool for involvement in IPV at 23 years and changes in IPV from 26 to 32 years is striking in contrast to EIPV in infancy, which did not significantly predict IPV during any of these adulthood periods. Potentially, the developmental tasks of toddlerhood/preschool (i.e., increasing autonomy, forming social relationships, and using emotional and behavioral self-regulation skills), are more directly relevant to the specific outcome of interest, IPV, than the developmental tasks of infancy (i.e., establishing sense of safety and developing secure attachment). EIPV during toddlerhood /preschool may model poor conflict management when children are first developing self-regulation skills, which may be a particularly strong predictor of increased behavioral dysregulation during relational conflict in adulthood. Alternatively, it is plausible that EIPV during infancy would be more strongly associated with attachment-specific outcomes in adulthood (e.g., perceived security and trust in romantic relationships) rather than IPV (Davies & Woitach, 2008; Waters & Cummings, 2000). Given that EIPV in infancy is associated with other outcomes, such as poorer cognitive and neurobiological functioning (Bosquet Enlow et al., 2013; Hibel et al., 2011), different aspects of EIPV (e.g., timing, severity, frequency and chronicity) likely have different implications for different functional domains.

Methodologically, it is also possible that EIPV in infancy did not significantly predict IPV in adulthood because of the relatively small subsample of children with EIPV in infancy (approximately 20% of participants). Thus, we cannot conclude that EIPV in infancy does not affect the risk for IPV in adulthood. Regardless, a closer examination of the specific developmental tasks of infancy and toddlerhood/preschool that may be disrupted by EIPV and the mediating processes accounting for the links from toddlerhood/preschool EIPV to adulthood IPV represent viable future research.

IPV clusters were also differentiated by contemporaneous factors. The IPV Persisting group, who had the highest mean levels of perpetration and victimization at both time periods, had the highest levels of life stress and externalizing behavior at ages 26 and 32, suggesting that high contextual stress and behavioral dysregulation may closely correspond to changes in IPV. The IPV Increasing group, who displayed low levels of perpetration and victimization at age 26 but moderately high levels of both at age 32, had high life stress and externalizing behavior at age 32, corresponding with their uptick in IPV. Conversely, the IPV Decreasing group, had high levels of life stress at age 26 but not at age 32. Finally, the Non-Violent group, who did not have high levels of perpetration or victimization at either time period, had the lowest levels of life stress and externalizing behavior at both ages.

There are several plausible interpretations for this correspondence between IPV, and life stress and externalizing behavior. On the one hand, increases in life stress and externalizing behavior may precipitate increases in IPV. Alternatively, changes in IPV may in turn affect levels of life stress or externalizing behavior. In either case, findings should be interpreted cautiously due to the small cell size of clusters and the correlational analyses.

Findings from the cluster analyses collectively support the importance of examining both early EIPV and contemporaneous factors to inform understanding of unique patterns of continuity and change in physical IPV in adult romantic relationships (Bell & Naugle, 2008; Bogat et al., 2005; Piquero et al., 2012). EIPV during toddlerhood/preschool and contemporaneous life stress and externalizing behavior were the strongest correlates of cluster membership, echoing past research on the importance of considering early adversity, contextual strain, and psychopathology in understanding violent behavior (Bogat et al., 2003; Cano & Vivian, 2003; Langer et al., 2008; Roberts et al., 2011). Conversely, participant sex did not differentiate cluster membership, which supports past research in community samples that females may perpetrate IPV at comparable rates to males (Archer, 2000; Langer et al., 2008).

Strengths and Limitations

This is the first study to prospectively identify how EIPV within sensitive periods of early childhood prospectively predicts relationship functioning across important adulthood transitions. The prospective longitudinal design afforded the unique opportunity to examine the effects of EIPV, documented at the time periods it occurred in early childhood, across more than thirty years of development. The prospective design guarded against the high potential for recall bias in EIPV, particularly for witnessing trauma. Another notable strength is the use of multi-informant data, including maternal reports for EIPV and participant reports for IPV.

Despite these strengths, limitations exist. The EIPV variable only included mothers’ reports of physical victimization, as they were not asked about physical perpetration or psychological or sexual violence at the time when the data was originally collected. Although rates of EIPV were fairly high, they may have been underestimates due to lack of information on maternal perpetration. Additionally, the EIPV rating only reflected the most severe incident of violence during each period. It was also not completely known whether children had directly witnessed each episode of maternal-reported violence. However, several considerations support the likelihood that children were indeed exposed. First, EIPV was only rated if it occurred in the home. Higher scores on the EIPV rating scale reflected more severe injury to the mother and greater chronicity, making it likely injuries would be visible and exposure would be repeated for children with higher scores. Second, previous research suggests that children are direct witnesses in the majority of family violence episodes (Fantuzzo et al., 1997; Hamby et al., 2011; Howell, 2011). Even in cases of indirect exposure or very young children, research also suggests that infants’ brains respond differently to negative/conflictive adult voices compared to neutral voices even when infants are sleeping (Graham, Fisher, & Pfeifer, 2013), making it plausible that even very young infants could be affected by exposure even if they didn’t directly see it.

Additional limitations include that IPV variables were gathered by self-report and involved some retrospective reflection on past relationships, potentially increasing error or bias. This study was also relatively small, predominantly Caucasian, and conducted in one Midwestern metropolitan area, so findings may not generalize to all families. In addition, the cell sizes of the clusters, and particularly the IPV Persisting group, were small, making replication in a larger sample essential. Finally, this study would have been improved by a fourth time point that extended farther into adulthood. Over a longer period of time across adulthood proper, it would be important to see whether individuals steadily increase or decrease IPV involvement as they age or experience ups and downs in IPV. These patterns may also be tied to partner characteristics, relational instability, and substance use.

Implications and Conclusions

EIPV in toddlerhood/preschool showed enduring effects on development by contributing to a higher risk for IPV involvement in early adulthood and predicting increases in IPV during the transition to adulthood proper for a subgroup of individuals. Furthermore, patterns of IPV across important adulthood transitions were closely linked to changes in contextual stress and externalizing behavior. These findings support the legacy of early experiences across development and the ongoing relevance of the contemporaneous context (Sameroff, 2000; Sroufe et al., 2005). Notably, within the Non-Violent group, one-third of individuals (n = 27) had witnessed EIPV during infancy, toddlerhood/preschool, or both. This suggests resilience processes at work, buffering the effects of early EIPV on adulthood maladaptation. Future research should continue to use person-oriented approaches to understand what helps individuals overcome risks associated with EIPV and follow pathways to positive relational adaptation.

The current findings have several implications for policy, prevention, and practice. Given toxic effects of emotional trauma on early development (Moffitt et al., 2013; Sroufe et al., 2005), policy and prevention efforts might wish to focus on trauma screening throughout development, beginning in infancy, with particular attention to the toddlerhood/preschool period. Family-focused intervention efforts, such as child-parent psychotherapy (CPP), should be considered to deter long-term involvement in IPV by repairing self-regulation skills and interpersonal expectations disrupted by family violence (Egeland & Erickson, 2004; Lieberman & Van Horn, 2008). Adulthood prevention efforts could also focus on reducing contextual stress.

In conclusion, both person-oriented and variable-oriented approaches have value for understanding the effects of EIPV within early childhood on the etiology of violent behavior. This prospective evidence illustrates that EIPV in toddlerhood/preschool may activate a maladaptive pathway towards elevated risk for IPV perpetration and victimization in adulthood, and that life stress and externalizing behavior may also contribute to continuity and change in adulthood IPV over time. This intergenerational cycle of violence begins with maternal IPV victimization during toddlerhood/preschool and extends to participant IPV in adulthood. Whether continuation of this cycle extends farther into adulthood remains unknown, as are the reasons why some individuals with EIPV do not engage in IPV. There is much more to be done to deepen and sharpen our understanding of the developmental processes by which violence is internalized and reactivated over time and across generations.

Acknowledgments

This research was supported in part by grants from the National Institute of Child Health & Human Development (HD054850) and the National Institute of Mental Health (MH40864), and graduate fellowships to the first author from the University of Minnesota and to the second author from the National Science Foundation. Data from this study were previously disseminated at the Society for Research in Child Developmental (SRCD) conference held in Philadelphia in 2015. The authors declare that they have no conflicts of interest.

References

  1. Achenbach TM. Manual for the young adult self-report and young adult behavior checklist. Burlington: University of Vermont, Department of Psychiatry; 1997. [Google Scholar]
  2. Achenbach TM, Rescorla LA. Manual for the ASEBA Adult Forms & Profiles. Burlington, VT: University of Vermont, Research Center for Children, Youth, & Families; 2003. [Google Scholar]
  3. Archer J. Sex differences in aggression between heterosexual partners: A meta-analytic review. Psychological Bulletin. 2000;126:651–680. doi: 10.1037/0033-2909.126.5.651. [DOI] [PubMed] [Google Scholar]
  4. Arnett JJ. Emerging adulthood: A theory of development from the late teens through the twenties. American Psychologist. 2000;55:469–480. doi: 10.1037/0003-066X.55.5.469. [DOI] [PubMed] [Google Scholar]
  5. Bergman LR, Magnusson D. A person-oriented approach in research on developmental psychopathology. Development and Psychopathology. 1997;9:291–319. doi: 10.1017/S095457949700206X. [DOI] [PubMed] [Google Scholar]
  6. Black MC, Basile KC, Breiding MJ, Smith SG, Walters ML, Merrick MT, … Stevens MR. The National Intimate Partner and Sexual Violence Survey (NISVS): 2010 summary report. Atlanta, GA: National Center for Injury Prevention and Control; 2011. [Google Scholar]
  7. Bell KM, Naugle AE. Intimate partner violence theoretical considerations: Moving towards a contextual framework. Clinical Psychology Review. 2008;28:1096–1107. doi: 10.1016/j.cpr.2008.03.003. [DOI] [PubMed] [Google Scholar]
  8. Bogat GA, Levendosky AA, Theran S, Von Eye A, Davidson WS. Predicting the psychosocial effects of interpersonal partner violence (IPV) How much does a woman's history of IPV matter? Journal of Interpersonal Violence. 2003;18:1271–1291. doi: 10.1177/0886260503256657. [DOI] [PubMed] [Google Scholar]
  9. Bogat GA, Levendosky AA, Von Eye A. The future of research on intimate partner violence: Person-oriented and variable-oriented perspectives. American Journal of Community Psychology. 2005;36:49–70. doi: 10.1007/s10464-005-6232-7. [DOI] [PubMed] [Google Scholar]
  10. Bosquet Enlow M, Egeland B, Blood EA, Wright RO, Wright RJ. Interpersonal trauma exposure and cognitive development in children to age 8 years: a longitudinal study. Journal of Epidemiology and Community Health. 2012;66:1005–1010. doi: 10.1136/jech-2011-200727. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cano A, Vivian D. Are life stressors associated with marital violence? Journal of Family Psychology. 2003;17:302–314. doi: 10.1037/0893-3200.17.3.302. [DOI] [PubMed] [Google Scholar]
  12. Carlson EA. A prospective longitudinal study of attachment disorganization/disorientation. Child Development. 1998;69:1107–1128. doi: 10.1111/j.1467-8624.1998.tb06163.x. [DOI] [PubMed] [Google Scholar]
  13. Center for Disease Control and Prevention. Understanding intimate partner violence fact sheet. Atlanta, GA: National Center for Injury Prevention and Control; 2012. [Google Scholar]
  14. Cicchetti D, Toth SL. The past achievements and future promises of developmental psychpathology: The coming of age of a discipline. The Journal of Child Psychology and Psychiatry. 2009;50:16–25. doi: 10.1111/j.1469-7610.2008.01979.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Collins A, van Dulmen M. Friendships and romance in emerging adulthood: Assessing distinctiveness in close relationships. In: Jenson J, Tanner JL, editors. Emerging adults in America: Coming of age in the 21st century. Washington, D. C: American Psychological Association; 2006. pp. 3–20. [Google Scholar]
  16. Cook DR, Weisberg S. Residuals and influence in regression. London, UK: Chapman & Hall; 1982. [Google Scholar]
  17. Cui M, Ueno K, Gordon M, Fincham FD. The continuation of intimate partner violence from adolescence to young adulthood. Journal of Marriage and Family. 2013;75:300–313. doi: 10.1111/jomf.12016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Davies PT, Woitach MJ. Children’s emotional insecurity in the interparental relationship. Current Directions in Psychological Science. 2008;17:269–274. doi: 10.1111/j.1467-8721.2008.00588.x. [DOI] [Google Scholar]
  19. Duncan OD. A socioeconomic index for all occupations. In: Reiss JAJ, editor. Occupations and social status. New York: Free Press; 1961. [Google Scholar]
  20. Egeland B, Breitenbucher M, Rosenburg D. Prospective study of the significance of life stress in the etiology of child abuse. Journal of Consulting and Clinical Psychology. 1981;48:195–205. doi: 10.1037//0022-006X.48.2.195. [DOI] [PubMed] [Google Scholar]
  21. Egeland B, Erickson MF. Lessons from STEEP: Linking theory, research, and practice on the well-being of infants and parents. In: Sameroff A, McDonough S, Rosenblum K, editors. Treating parent–infant relationship problems. New York: Guilford Press; 2004. pp. 213–242. [Google Scholar]
  22. Ehrensaft MK, Cohen P, Brown J, Smailes E, Chen H, Johnson JG. Intergenerational transmission of partner violence: A 20-year prospective study. Journal of Consulting and Clinical Psychology. 2003;71:741–753. doi: 10.1037/0022-006X.71.4.741. [DOI] [PubMed] [Google Scholar]
  23. Erikson EH. Childhood and society. 2. New York: Norton; 1963. [Google Scholar]
  24. Fantuzzo J, Boruch R, Beriama A, Atkins M, Marcus S. Domestic violence and children: Prevalence and risk in five major US cities. Journal of the American Academy of Child & Adolescent Psychiatry. 1997;36:116–122. doi: 10.1097/00004583-199701000-00025. [DOI] [PubMed] [Google Scholar]
  25. Fergusson DM, Horwood LJ. Exposure to interparental violence in childhood and psychosocial adjustment in young adulthood. Child Abuse and Neglect. 1998;22:339–357. doi: 10.1016/s0145-2134(98)00004-0. [DOI] [PubMed] [Google Scholar]
  26. Fite JE, Bates JE, Holtzworth-Munroe A, Dodge KA, Nay SY, Pettit GS. Social information processing mediates the intergenerational transmission of aggressiveness in romantic relationships. Journal of Family Psychology. 2008;22:367–376. doi: 10.1037/0893-3200.22.3.367. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Graham AM, Fisher PA, Pfeifer JH. What sleeping babies hear: A functional MRI study of interparental conflict and infants’ emotion processing. Psychological Science. 2013;24:782–789. doi: 10.1177/0956797612458803. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Hamby SL, Finkelhor D, Turner HA, Ormrod R. Children's exposure to intimate partner violence and other family violence. US Department of Justice, Office of Justice Programs, Office of Juvenile Justice and Delinquency Prevention; 2011. [Google Scholar]
  29. Hibel LC, Granger DA, Blair C, Cox MJ the Family Life Key Project Investigators. Maternal sensitivity buffers the adrenocortical implications of intimate partner violence exposure during early childhood. Development and Psychopathology. 2011;23:689–701. doi: 10.1017/S0954579411000010. [DOI] [PubMed] [Google Scholar]
  30. Holt S, Buckley H, Whelan S. The impact of exposure to domestic violence on children and young people: a review of the literature. Child Abuse and Neglect. 2008;32:797–810. doi: 10.1016/j.chiabu.2008.02.004. [DOI] [PubMed] [Google Scholar]
  31. Holtzworth-Munroe A, Meehan JC. Typologies of men who are maritally violent: Scientific and clinical implications. Journal of Interpersonal Violence. 2004;19:1369–1389. doi: 10.1177/0886260504269693. [DOI] [PubMed] [Google Scholar]
  32. Howell KH. Resilience and psychopathology in children exposed to family violence. Aggression and Violent Behavior. 2011;16:562–569. doi: 10.1016/j.avb.2011.09.001. [DOI] [Google Scholar]
  33. Iverson KM, Litwack SD, Pineles SL, Suvak MK, Vaughn RA, Resick PA. Predictors of intimate partner violence revictimization: The relative impact of distinct PTSD symptoms, dissociation, and coping strategies. Journal of Traumatic Stress. 2013;26:102–110. doi: 10.1002/jts.21781. [DOI] [PubMed] [Google Scholar]
  34. Kitzmann KM, Gaylord NK, Holt AR, Kenny ED. Child witness to domestic violence: a meta-analytic review. Journal of Consulting and Clinical Psychology. 2003;71:339–352. doi: 10.1037/0022-006X.71.2.339. [DOI] [PubMed] [Google Scholar]
  35. Langer A, Lawrence E, Barry RA. Using a vulnerability-stress-adaptation framework to predict physical aggression trajectories in newlywed marriage. Journal of Consulting and Clinical Psychology. 2008;76:756–768. doi: 10.1037/a0013254. [DOI] [PubMed] [Google Scholar]
  36. Levendosky AA. Drawing conclusions: An intergenerational transmission of violence perspective. Psychodynamic Psychiatry. 2013;41:351–360. doi: 10.1521/pdps.2013.41.2.351. [DOI] [PubMed] [Google Scholar]
  37. Lieberman AF, Van Horn P. Psychotherapy with infants and young children. New York: Guilford; 2008. [Google Scholar]
  38. Moffitt TE the Klaus-Grawe 2012 Think Tank. Childhood exposure to violence and lifelong health: Clinical intervention science and stress-biology research join forces. Development and Psychopathology. 2013;25:1619–1634. doi: 10.1017/So954579413000801. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Moffitt TE, Caspi A, Krueger RF, Magdol L, Margolin G, Silva PA, Sydney R. Do partners agree about abuse in their relationship? A psychometric evaluation of interpartner agreement. Psychological Assessment. 1997;8:47–56. doi: 10.1037/1040-3590.9.1.47. [DOI] [Google Scholar]
  40. Narayan AJ, Englund MM, Egeland B. Developmental timing and continuity of exposure to interparental violence and externalizing behavior as prospective predictors of dating violence. Development and Psychopathology. 2013;25:973–990. doi: 10.1017/S095457941300031X. doi: 10.1017=S095457941300031X. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Narayan AJ, Englund MM, Carlson EA, Egeland B. Adolescent conflict as a developmental process in the pathway from exposure to interparental violence to dating violence. Journal of Abnormal Child Psychology. 2014;42:399–350. doi: 10.1007/s10802-013-9782-4. doi: 10.1007=s10802-013-9782-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Norusis M. Cluster analysis. [accessed May 13, 2014];IBM SPSS statistical guides. 2005–2012 http://www.norusis.com/pdf/SPC_v13.pdf.
  43. Nurius PS, Macy RJ. Heterogeneity among violence-Exposed women: Applying person-oriented research methods. Journal of Interpersonal Violence. 2008;23:389–415. doi: 10.1177/0886260507312297. [DOI] [PubMed] [Google Scholar]
  44. Piquero AR, Carriaga ML, Diamond B, Kazemian L, Farrington DP. Stability in aggression revisited. Aggression and Violent Behavior. 2012;17:365–372. doi: 10.1016/j.avb.2012.04.001. [DOI] [Google Scholar]
  45. Raby KL, Labella MH, Martin J, Carlson EA, Roisman GI. Childhood abuse and neglect and insecure attachment states of mind in adulthood: Prospective, longitudinal evidence from a high-risk sample. Development and Psychopathology. doi: 10.1017/S0954579417000037. in press. [DOI] [PubMed] [Google Scholar]
  46. Rawlings JO. Applied regression analysis: A research tool. Pacific Grove, CA: Wadsworth & Brooks=Cole; 1988. [Google Scholar]
  47. Roberts AL, McLaughlin KA, Conron KJ, Koenen KC. Adulthood stressors, history of childhood adversity, and risk of perpetration of intimate partner violence. American Journal of Preventative Medicine. 2011;40:128–138. doi: 10.1016/j.amepre.2010.10.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Rubin DB. Multiple imputation for nonresponse in surveys. New York, NY: Wiley & Sons; 1987. [Google Scholar]
  49. Sameroff AJ. Developmental systems and psychopathology. Development and Psychopathology. 2000;12:297–312. doi: 10.1017/S0954579400003035. [DOI] [PubMed] [Google Scholar]
  50. Schafer JL, Graham JW. Missing data: Our view of the state of the art. Psychological Methods. 2002;7:147–177. doi: 10.1177=0049124102030003008. [PubMed] [Google Scholar]
  51. Smith CA, Ireland TO, Park A, Elwyn L, Thornberry TP. Intergenerational continuities and discontinuities in intimate partner violence: A two-generational prospective study. Journal of Interpersonal Violence. 2011;26:3720–2752. doi: 10.1177/0886260511403751. [DOI] [PubMed] [Google Scholar]
  52. Sroufe LA. The coherence of individual development: Early care, attachment, and subsequent developmental issues. American Psychologist. 1979;34:834–841. doi: 10.1037/0003-066X.34.10.834. [DOI] [Google Scholar]
  53. Sroufe LA, Egeland B, Carlson EA, Collins WA. The development of the person. New York: Guilford; 2005. [Google Scholar]
  54. Stevens G, Featherman DL. A revised socioeconomic index of occupational status. Social Science Research. 1981;10:364–395. doi: 10.1016/0049-089X(81)90011-9. [DOI] [Google Scholar]
  55. Straus MA. Measuring intrafamilial conflict and violence: The Conflict Tactics (CT) Scales. Journal of Marriage and the Family. 1979;41:75–88. http://www.jstor.org/stable/351733. [Google Scholar]
  56. Swartout KM, Swartout AG. Shifting perspectives: Applying person-centered analyses to violence research. Psychology of Violence. 2012;2:309–312. doi: 10.1037/a0029910. [DOI] [Google Scholar]
  57. Waters E, Cummings EM. A secure base from which to explore close relationships. Child Development. 2000;71:164–172. doi: 10.1111/1467-8624.00130. [DOI] [PubMed] [Google Scholar]
  58. Yates TM, Dodds MF, Sroufe LA, Egeland B. Exposure to partner violence and child behavior problems: A prospective study controlling for physical abuse and neglect, child cognitive ability, socioeconomic status, and life stress. Development and Psychopathology. 2003;15:199–218. doi: 10.1017/s0954579403000117. doi: 10.1017.20954579403000117. [DOI] [PubMed] [Google Scholar]

RESOURCES