Neighborhood Norms, Disadvantage, and Intimate Partner Violence Perpetration

Abstract

Most theoretical treatments of intimate partner violence (IPV) focus on individual-level processes. More recently, scholars have begun to examine the role of macrolevel factors. Results of that research indicate that social ties facilitate the diffusion of cultural norms—including tolerance of deviance/violence—across neighborhoods. Yet the influence of the neighborhood normative climate extends beyond norms regarding the use of violence, shaping cultural understanding about dating and the opposite sex. Using data from the Toledo Adolescent Relationships Study (TARS), the current investigation examines the multilevel association between dating norms and IPV perpetration among a large, diverse sample of adolescents and young adults. Results indicate that individuals’ liberal dating attitudes are associated with IPV perpetration. Furthermore, this effect varies across levels of neighborhood disadvantage.

INTRODUCTION

In the past decade, increasing attention has been given to intimate partner violence (IPV), as a growing body of research has explored its risk factors, correlates, and consequences (see Capaldi et al. 2012 for a review). Although most of this work is restricted to adult populations, survey and official data indicate that IPV is particularly frequent among young adults (Brown and Bulanda 2008; Halpern et al. 2009). In fact, despite the focus on samples of older, married individuals, IPV risk seems to peak in late adolescence and young adulthood and then decline with age (Fritz and O’Leary 2004; Johnson et al. 2014; Kim et al. 2008). Overall, recent estimates from the National Longitudinal Study of Adolescent to Adult Health (Add Health) indicate that roughly two-fifths (39.8%) of young adults reported IPV exposure, including victimization and perpetration (Clark et al. 2016). The bulk of this involvement included perpetration (“perpetration only” and “perpetration and victimization”), such that nearly one-third (30.65%) of young men and women reported perpetrating violence within the relationship context. Notably, research suggests that patterns of IPV perpetration deviate from more general patterns of antisocial involvement across the young adult period, such that IPV perpetration among women increases alongside declines in other forms of antisociality (Johnson et al. 2014). Thus, it is important to focus empirical attention on this public health problem and particularly on factors contributing to IPV perpetration among this demographic group.

Most theoretical approaches to IPV have focused on individual-level correlates, including age, race, gender, and socioeconomic status (SES) (e.g., Capaldi, Kim, and Shortt 2007; Cunradi, Caetano, and Schafer 2002; Herrera, Wiersma, and Cleveland 2008; Kim et al. 2008), largely ignoring the role of structural or macrolevel factors. Recently, more scholarly attention has been paid to the influence of neighborhood context on IPV (e.g., Benson et al. 2004; Caetano, Ramisetty-Mikler, and Harris 2010; Emery, Jolley, and Wu 2011; VanderEnde et al. 2012; Wright 2012). Yet the theoretical focus of this growing body of scholarship has been narrow, drawing almost exclusively on aspects of social disorganization theory to unpack the role of social processes in the link between structural factors and IPV (Shaw and McKay 1942). Within this tradition, scholars have focused little attention on the role of cultural norms (Kubrin and Weitzer 2003).

Notwithstanding, a recent review of the literature on neighborhood context and IPV indicates that there is evidence that neighborhood norms are important predictors of IPV (Pinchevsky and Wright 2012), including cultural and social norms about family violence, “wife beating,” and deviance (Browning 2002; Koenig et al. 2006; Wright and Benson 2010). Thus, neighborhood social ties—including friends and more distal network members—can facilitate the diffusion of cultural norms (i.e., shared expectations and rules that guide group behavior) consistent with the use of violence (Kornhauser 1978; Warner 2003). However, whereas prior work has focused somewhat narrowly on norms about violence, recent evidence has established that outright endorsement of relationship violence is quite low (Copp et al. 2016). Accordingly, there is a need for work that examines the wider arena of dating and relationships, including important norms that may be drivers of conflict within young adult relationships. For example, liberal norms regarding relationships and sexuality (e.g., it’s okay to date more than one person at a time) may foster resentment or exacerbate concerns about partner infidelity, thus contributing to the use of violence within the dyadic context. That these broader patterns have been largely ignored is surprising given the increasing focus on relational factors with regard to IPV (e.g., Capaldi and Kim 2007; Ford 2017; Fritz, Slep, and O’Leary 2012; Giordano et al. 2010; Straus 2011), and the extensive amount of scholarship linking jealousy/cheating and gender attitudes to IPV at the individual level (e.g., Copp et al. 2015; Giordano et al. 2010; Giordano et al. 2015; Langhinrichsen-Rohling, McCullars, and Misra 2012; Reed et al. 2011). Furthermore, existing theorizing about these structural and social influence processes has not been fully incorporated into intervention and prevention strategies (Eckhardt et al. 2013; Whitaker et al. 2006).

Drawing on data from the Toledo Adolescent Relationships Study (TARS), the current investigation employs hierarchical generalized linear modeling (HGLM) techniques to examine the multilevel association between neighborhood normative climates and IPV perpetration among a diverse sample of adolescent and young adult men and women. We first explore the direct link between features of the neighborhood context, including neighborhood disadvantage and liberal dating norms, and IPV perpetration. Next, we assess the extent to which these neighborhood structural and cultural elements influence IPV perpetration indirectly via their impact on individual attitudes and beliefs. Finally, we examine potential moderating effects, including whether the influence of neighborhood- and individual-level normative beliefs on patterns of IPV perpetration are conditioned by neighborhood context.

BACKGROUND

Contextual Influences on Adolescent and Young Adult Behavior

The effect of neighborhood residence on children and youth is well documented in the literature (Shaw and McKay 1942), and over the past few decades there has been a resurgence in research examining the role of neighborhood structure on delinquency and crime as well as other problem behaviors (e.g., Bursik 1988; Kornhauser 1978; Sampson 1992; Sampson and Groves 1989). This work suggests that there is an association between neighborhood SES, including neighborhood levels of poverty, welfare receipt, unemployment, and racial composition, and the well-being of children and adolescents (see Leventhal and Brooks-Gunn 2000 for a review).⁶ For example, scholars have identified a direct link between neighborhood levels of socioeconomic disadvantage and reported behavior problems—particularly externalizing behaviors (e.g., Chase-Lansdale et al. 1997; Duncan, Brooks-Gunn, and Klebanov 1994; Ludwig, Duncan, and Hirschfield 2001). Relatedly, others have found a significant association between socioeconomic disadvantage and delinquency, such that the frequency and severity of delinquency and crime was lower in more affluent communities (e.g., Beyers et al. 2001; Loeber and Wikstrom 1993; Sampson and Groves 1989). Yet much of the association between neighborhood disadvantage and a range of developmental outcomes appears to be indirect. Accordingly, scholars have increasingly looked to neighborhood organizational features to understand individual-level behaviors, including violence (Elliott et al. 1996; Haynie, Silver, and Teasdale 2006; Stewart, Simons, and Conger 2002).

Despite the well-documented associations between neighborhood characteristics and youth behavior, few empirical studies have examined contextual effects on relationship violence. In addition, of the existing research on neighborhoods and IPV, scholars have seldom focused specifically on adolescent and young adult samples. In the adult IPV literature, a limited body of work has examined the neighborhood context of IPV. Drawing on elements of social disorganization theory, this research examines whether the neighborhood structure influences participation in IPV (e.g., Benson et al. 2003; Browning 2002; Cunradi et al. 2000; Miles-Doan 1998). Findings indicate that neighborhood levels of concentrated disadvantage have significant effects on IPV net of individual-level characteristics (i.e., race, SES, and social support). In one of the few neighborhood-level investigations of IPV among adolescents and young adults, Jain and colleagues (2010) found that neighborhood social processes (i.e., collective efficacy) partially mediated the effect of neighborhood disadvantage on dating violence. More recently, Copp and colleagues (2015) found that the link between neighborhood disadvantage and IPV was partially explained by individual-level indicators of emotional distress (anger and depressive symptoms). Thus, based on the more general literature on neighborhood effects as well as the limited work on neighborhoods and IPV, evidence suggests that neighborhood disadvantage “matters,” but that it is insufficient as an explanation for the neighborhood variation in violence. There is a need for research examining other cultural and social processes in order to develop a more comprehensive understanding of the ways in which individuals are influenced by features of the broader context.

Community Norms and Violence

Criminological research on subcultural values indicates that pro-violence norms are more likely to exist in disadvantaged settings (e.g., Anderson 1999; Cohen 1955; Messerschmidt 1993; Willis 1977), and thus there is potential for wider attitudinal acceptance of IPV in high-poverty neighborhoods. Scholars have suggested that individuals in disadvantaged contexts may rely on a unique set of cognitive scripts to guide their behavior in social interactions (Luckenbill and Doyle 1989; Wilkinson and Fagan 1996; see also McGloin et al. 2011), including a heightened proclivity for aggression. Neighborhood norms have been examined in relation to individual-level behaviors (e.g., Anderson 1999; Brezina et al. 2004; Stewart and Simons 2010), and research findings are generally supportive of the notion that the neighborhood culture is a significant predictor of violence.

A similar line of theorizing has been extended to research on IPV as scholars have investigated whether neighborhood norms of violence are related to the use of violence between romantic partners (e.g., Browning 2002; Koenig et al. 2006; Wright and Benson 2010). Using multilevel regression techniques, Browning (2002) found that neighborhood-level attitudes about family violence (i.e., fighting between friends/families is a private matter) were positively associated with IPV. Similarly, Wright and Benson (2010) examined associations between the neighborhood culture and IPV and found that attitudes regarding whether family violence is “nobody else’s business” were associated with IPV risk, but a general intolerance of deviance was unrelated to IPV. A more direct measure of community endorsement of IPV—which included specific attitudes about “wife beating”—was a particularly salient predictor of IPV among a sample of Indian men (Koenig et al. 2006). A recent investigation using a sample of adolescent urban boys (aged 14–20 years) revealed that IPV perpetration was associated with perceptions of violent activity within the neighborhood (Reed et al. 2011). Overall, neighborhood norms regarding the use of violence generally, and IPV specifically, appear to influence involvement in partner violence across a diverse range of samples.

Beyond norms of violence, the social environment in which individuals are embedded facilitates the transmission of messages regarding conduct in other facets of one’s life—including relationships with the opposite sex—and this nexus of norms, attitudes, and beliefs makes up the neighborhood normative climate (Butler 2002; Mollborn 2010; Warner et al. 2011). Although collective efficacy does involve a norm-based component (shared expectations), and research findings based on this social process are prolific, the broader normative climate remains vastly understudied (Sampson and Graif 2009). Instead, studies focus primarily on direct linkages between norms and behaviors that are often “verging on the tautological” (e.g., attitudes toward violence predict violence) (Giordano and Copp 2019:9.17). Yet as Butler (2002:299) notes, “norms may prescribe other behaviors that in turn influence the behavior of interest.” Thus, normative beliefs about dating and the opposite sex may indirectly affect IPV through an increase in unhealthy relationship behaviors (e.g., infidelity) or conflict related to perceived problems in the relationship—factors that have been linked to IPV perpetration across multiple studies (e.g., Giordano et al. 2015; Nemeth et al. 2012; see also Capaldi et al. 2012).

Particularly during the adolescent period, the neighborhood normative climate plays an important role in the socialization process, as individuals observe the behavior of the wider culture to which they are exposed (Harding 2009; Jencks and Mayer 1990). For example, research on “sex codes” indicates that subcultural values regarding the onset and frequency of sexual activity are transmitted through adolescent peer networks within the neighborhood (Anderson 1999). Consistent with this line of theorizing, Warner and colleagues (2011) found that sexual attitudes at the neighborhood level predicted adolescent sexual behaviors. A similar approach is generalizable to outcomes beyond sexual activity, including IPV, to allow for examination of the pathways through which social environments influence individual behavior, as well as the role of social and cultural factors. To date, however, few studies have examined how a broader range of neighborhood norms shape conduct in romantic relationships, including the use of IPV.

Theoretical Framework

The resurgence of research within the tradition of urban sociology has provided ample evidence that neighborhood effects are causal in nature. Less clear, however, are the mechanisms by which neighborhoods influence individual outcomes. In the criminological literature, early theorizing on adolescent decision making and behavior in disadvantaged neighborhoods often took the form of subcultural explanations (Cloward and Ohlin 1960; Cohen 1955; Miller 1958; Wolfgang and Ferracuti 1967), which argued that youth in such contexts adhered to beliefs and attitudes conducive to violence and delinquency. More recent research on the consequences of concentrated poverty and racial segregation has discussed youths’ development of unconventional norms and values as an adaptation to extreme isolation and blocked opportunities (Anderson 1999; Massey and Denton 1993). These perspectives are similar in that they both present the normative climates within disadvantaged neighborhoods in somewhat monolithic terms. Central to the views of these early criminologists and contemporary poverty scholars is the notion that an oppositional culture emerges within poor neighborhoods and that this oppositional culture, in turn, promotes behaviors at odds with mainstream society.

In contrast to the idea of a coherent value system or “oppositional culture” in poor communities, Shaw and McKay (1969:172) argued that youth residing in disadvantaged neighborhoods are “exposed to a variety of contradictory standards and forms of behavior rather than to a relatively consistent and conventional pattern.” Although more recent research in the social disorganization tradition has largely overlooked the subcultural aspects of Shaw and McKay’s original model (Bursik 1988), cultural heterogeneity has been examined on a very limited basis by urban ethnographers and cultural sociologists. In a recent example, Harding (2007) examined cultural frames and scripts in a nationally representative sample (Add Health) and found that there was greater cultural heterogeneity in disadvantaged settings. These findings question the view of disadvantaged neighborhoods as culturally homogeneous; however, a clear perspective on how cultural factors are shaped by social structural elements has yet to emerge (Lamont and Small 2008).

In the current investigation, we draw on a social learning perspective to further unpack the values and norms that guide individual behavior, and to situate these processes within the broader neighborhood context. According to social learning theory, individuals develop their own attitudes and behavioral repertoires through interaction with significant others (Bandura 1978). Neighborhoods represent a key context of social interaction in which individuals learn normative definitions through their personal experiences, and their observation of the behavior of adults in the community (Akers 2011). Yet unlike the cultural deviance models described above, a social learning approach recognizes that individuals are exposed to potentially conflicting definitions within the neighborhood context (Akers 1996). For example, neighborhood structural conditions, including poverty, blocked opportunities, and legal cynicism, may lead to the development of a worldview that emphasizes respect, toughness, and violence in public interactions. At the same time, violent behavior may be discouraged at home by parents who are strongly oriented to mainstream values (Anderson 1999). Accordingly, the influence of the neighborhood culture on individual outcomes is unlikely to be entirely direct, as there is variation in cultural norms—even within the context of a single neighborhood. Moreover, given that disadvantaged contexts are more likely to include a range of exigencies and associated survival strategies that support the use of violence under some circumstances, it is possible that the effect of liberal dating norms may be exacerbated by neighborhood structural features. That is, residents of disadvantaged neighborhoods may be more likely to use violence as an interpersonal response to relationship conflicts (e.g., related to infidelity) that are more likely when normative climates include permissive attitudes about dating and sexuality.

Thus, social learning theory is especially amenable to the proposed research, as it illuminates a potential pathway through which neighborhoods influence individual behavior, but also recognizes that neighborhood culture is not a “single entity” (Harding 2007). In addition, it remains one of the leading explanations of IPV. Most social learning approaches to IPV, however, focus exclusively on the social behavioral dimension of the theory. This work has led to the identification of key risk factors, including violence in the family of origin, witnessing interparental violence, and direct experiences of child abuse (e.g., Foshee et al. 2002; Kwong et al. 2003; White and Widom 2003). Yet within the IPV literature, and the criminological literature more generally, research on the definitional component of social learning theory remains largely underdeveloped (Giordano 2016). Even in Akers’s more recent formulations of the theory, definitions—whether more general or specific to a given situation—are based on the meanings that individuals attach to criminal or deviant behaviors (Akers 2011). Indeed, attitudes and beliefs about crime and violence are foundational to an understanding of antisociality. Nonetheless, in our view the definitions and meanings that are used to construct behavioral repertoires go beyond the scope of crime itself (Giordano and Copp 2019). There is a need for research that broadens the conceptualization of “definitions” to include attitudes in addition to those directly derivative of the outcome of interest (i.e., attitudes about violence predict violence).

In the current investigation, we consider the role of normative beliefs about dating and sex. We focus particular attention on the young adult period, as it is during young adulthood that individuals become increasingly involved in more committed and long-term relationships (Giordano et al. 2012; Meier and Allen 2009). Indeed, the heightened levels of enmeshment and interdependence characteristic of young adult romantic relationships may “raise the stakes” and intensify partner concerns (e.g., partners’ perceived [lack of] commitment, infidelity)—particularly absent a codified set of social expectations guiding relationship conduct outside the marital context (Giordano et al. 2015). Accordingly, young adult couples must work together to co-construct their own relationship guidelines, which may lead to conflict should differences in viewpoints emerge. Although prior research has focused more specifically on attitudes toward IPV, here we widen the scope to consider a broader set of normative beliefs that reflect different worldviews/understandings of the dating realm that may be consequential for couple-level patterns of interaction, including conflict. For example, some normative climates may include greater tolerance for nonmarital sex and even having sex outside a traditional dating/romantic context. This broader set of attitudes are part of the background to more localized relationship dynamics such as mismatched levels of commitment and feelings of mistrust that have previously been linked to IPV.

CURRENT STUDY

Using data from TARS, we examine the multilevel association between neighborhood structural and cultural features and IPV perpetration, controlling for individual-level attitudes and beliefs, as well as a number of well-documented risk factors. Much of what we know about neighborhood context and IPV comes from work examining the role of neighborhood structure and social processes in IPV among older, married and cohabiting adults (e.g., Benson et al. 2003; Cunradi et al. 2000; Koenig et al. 2006). Our focus on the broader neighborhood normative climate, including dating norms, contributes beyond existing research in a number of key ways. In particular, the current study considers whether attention to the neighborhood cultural context helps explain variation in IPV perpetration, controlling for neighborhood structural factors. In addition, we consider whether the effect of the neighborhood normative climate operates indirectly, via its influence on individual-level attitudes and beliefs. Recognizing that disadvantaged neighborhoods may be characterized by greater levels of cultural heterogeneity, we examine whether the effect of individual-level attitudes and beliefs on IPV perpetration is moderated by the level of neighborhood disadvantage. Finally, we examine these associations among a large, diverse sample of young adults.

In multivariate models, we included a number of factors identified as correlates of normative beliefs and relationship violence in previous research. These included traditional social learning factors, such as coercive parenting, as well as a broad range of sociodemographic, adult status, and relationship characteristics. As highlighted above, there is a well-documented association between coercive parenting in the family of origin and IPV (e.g., Ehrensaft et al. 2003; Smith et al. 2011). Additionally, sociodemographic factors, including gender, race/ethnicity, age, and family background characteristics, are associated with IPV (see Capaldi et al. 2012 for a review of risk factors). Gainful activity (enrolled in school or employed) is negatively associated with IPV (Alvira-Hammond et al. 2013). Relationship characteristics, including relationship status and duration, have been linked to IPV (Brown and Bulanda 2008; Kenney and McLanahan 2006). Parenthood has also been shown to elevate IPV risk (Vest et al. 2002). We include these variables as controls in our models estimating IPV perpetration.

DATA AND METHODS

This research draws on data from TARS, which is based on a stratified random sample of 1,321 adolescents and their parents/guardians. The sampling frame of the TARS study encompassed 62 schools across 7 school districts. The initial sample was drawn from the enrollment records of registered students in the 7th, 9th, and 11th grades in Lucas County, Ohio. School attendance, however, was not a requirement for inclusion in the study. The stratified, random sample was devised by the National Opinion Research Center and includes oversamples of black and Hispanic adolescents. The TARS data were collected in the years 2001, 2002, 2004, 2006, and 2011. The analyses rely primarily on structured interviews conducted in connection with the fourth wave of interviews (2006), but a parent questionnaire administered at the first interview provided information about sociodemographic characteristics. Although the sample is regional, Lucas County shares similar sociodemographic characteristics as the nation. Based on analysis of U.S. Census data, the TARS sample mirrors similar age adolescents and young adults in the nation in terms of race, family status, parents’ education, and income. For example, among 12- to 18-year-olds in 2001, 19.5% in the nation, 22.4% in Lucas County, and 21.3% in TARS lived with a single mother. The sample retained this high level of national representativeness in the fourth interview wave in 2006/7.

To examine neighborhood effects, this study used the TARS contextual database, which was created by appending data from the 2000 U.S. Census. In the current investigation, we measure neighborhood as respondents’ census tract. Although there is some debate in the literature regarding the appropriate method for measuring neighborhoods, recent work acknowledges the utility of using the census tract for identifying neighborhood processes, as further aggregation reduces variation and potentially masks important relationships (e.g., Krieger et al. 2017; see also Hipp 2007).

At wave 4, there are 1,092 valid respondents, or 83% of wave 1. Attrition analyses indicate that participation at wave 4 is not related to most characteristics, with a few exceptions. In particular, the analytic sample is more likely to be female and to report an “other” family structure. The analytic sample includes all those who participated in the wave 4 interview, but excludes individuals living in census tracts with fewer than five survey participants (241 respondents spread across 152 tracts). Additional exclusions included those who did not identify as black, white, or Hispanic (n = 22) and those who did not report on a current/most recent relationship (i.e., those who reported no romantic involvement in the last two years) (n = 85). Finally, we excluded those with missing information on variables from the contextual database (n = 10). The final analytic sample consists of 734 respondents distributed across 85 neighborhoods.

Measures

Dependent variable

Relationship violence perpetration is based on a revised version of the Conflict Tactics Scale (CTS2) (Straus et al. 1996), which includes the following four items: “thrown something at,” “pushed, shoved, or grabbed,” “slapped in the face or head with an open hand,” and “hit.” The questions refer to experiences at any time with the current/most recent partner, measured at the time of the wave 4 interview. We focus on relationship perpetration as we are interested in unpacking the ways in which the neighborhood influences individual attitudes and beliefs which, in turn, may influence individuals’ behavior—including the use of violence within the dyadic context. We used a dichotomous version of relationship violence perpetration (1 = any IPV perpetration; 0 = no IPV perpetration).

Neighborhood-level independent variables

To measure the neighborhood structure we included a summed scale of concentrated disadvantage, comprised of the following indicators from the U.S. Census (measured at the tract level at wave 4): proportion of households below the poverty line, proportion of households receiving public assistance, proportion of the population over 16 who are unemployed, and the proportion of female-headed households. In addition, we included a perceptual measure of neighborhood disorder, subjective disorder, based on seven items from the wave 4 questionnaire in which respondents were asked about potential problems in their neighborhoods, including “high unemployment,” “litter and trash on the sidewalk and streets,” “run-down and poorly kept buildings and yards,” “quarrels in which someone is hurt badly,” “drug use or drug dealing in the open,” “youth gangs,” and “graffiti.” Responses were first dichotomized to indicate whether these items posed a problem (1 = yes), and then summed. Summed individual scores were then aggregated to the neighborhood level. We assessed the neighborhood normative climate using six items tapping liberal dating norms from the wave 4 questionnaire including “A person should only have sex with someone they love” (reverse coded), “A person should only have sex if they are married” (reverse coded), “I would have to be committed to a girl/guy in order to have sex with her/him” (reverse coded), “It would be okay to have sex with someone I wasn’t dating,” “I would feel comfortable having sex with someone I was attracted to but did not know very well,” and “It’s okay to sometimes date more than one person at a time” (responses ranged from 1 “strongly disagree” to 5 “strongly agree”). Individuals’ responses were aggregated to the neighborhood level to capture the average level of endorsement of liberal dating attitudes within census tracts.

Individual-level independent variables

A disaggregated version of liberal dating norms was used to examine individuals’ attitudes, our focal level-one independent variable. Coercive parenting was based on a single item from the wave 1 adolescent report asking respondents, “When you and your parents disagree about things, how often do they push, slap, or hit you?” We dichotomized responses to indicate any coercive parenting. We include a series of sociodemographic indicators, including gender; age, measured in years using a continuous variable reported from respondent’s age at wave 4; gainful activity, a dichotomous variable indicating whether the respondent was enrolled in school or employed full time (wave 4); as well as three dummy variables to measure race/ethnicity including non-Hispanic white (reference category), non-Hispanic black, and Hispanic. Family structure (wave 1) includes the following categories: two biological parents (contrast category), stepfamily, single-parent family, and any “other” family type. To control for SES, we control for the highest level of education obtained by the respondent’s parent, including less than high school, high school (reference category), some college, and college or more. We refer to this as mother’s education, as the vast majority of parent questionnaires were completed by the respondent’s biological mother. We also measure the respondents’ relationship status at wave 4 using the following three dichotomous variables: dating (reference), cohabiting, and married. Relationship duration is taken from a single item, “How long have you been together?” referencing the focal relationship (responses range from 1 “less than a week” to 8 “a year or more”). We include a dichotomous variable to account for whether the respondent had any children at the time of the wave 4 interview. We control for the respondent’s prior IPV exposure based on a dichotomous indicator from the wave 1 adolescent questionnaire.

Analytic Strategy

We use hierarchical logistic regression models to examine the multilevel association between the neighborhood structure/culture, individual dating norms, and IPV perpetration. We estimate models using the PROC GLIMMIX function available in SAS 9.4. The analyses consist of several stages. First, we estimate an unconditional model to determine whether the between-neighborhood variation in IPV perpetration is significant. From this model, we compute the intraclass correlation (ICC) to determine the total variation that occurs between neighborhoods. In the current analyses, the ICC is 0.07, indicating that 7% of the variation in IPV perpetration is at the neighborhood level.

Descriptive statistics are presented in Table I. In Table II, we present the findings of models estimating IPV perpetration. Model 1 examines IPV perpetration, accounting for neighborhood concentrated disadvantage, neighborhood subjective disorder, and neighborhood dating norms. Next, we add the individual-level indicator of dating norms to determine whether its inclusion partially mediates the neighborhood effect on IPV perpetration. A third model adds the sociodemographic, adult status, and relationship characteristics. Supplemental analyses consider whether the effect of normative beliefs on IPV perpetration depends on the neighborhood structural/cultural context and gender.

Table I.

Means/Percentages and Standard Deviations for All Study Variables (n = 734)

	Full Sample		No Perpetration (n = 540)			Perpetration (n = 194)
	Mean/Percentage	SD	Mean/Percentage	SD		Mean/Percentage	SD
Dependent Variable
IPV Perpetration	26.43%		0.00%			100.00%
Independent Variables
Neighborhood Structure
Concentrated disadvantage
Percent below poverty line	14.78%		13.14%		***	18.11%
Percent receiving pub asst	4.94%		4.35%		***	6.05%
Percent unemployed	7.16%		6.46%		***	8.87%
Percent female-headed hh	10.10%		9.23%		***	11.65%
Neighborhood Culture
Liberal dating norms	2.72	0.31	2.71	0.32		2.74	0.29
Individual-Level
Liberal dating norms	2.74	0.83	2.71	0.83		2.83	0.82
Subjective disorder	2.37	2.62	2.06	2.50	***	3.23	2.77
Coercive parenting	21.80%		17.59%		***	33.50%
Female	53.08%		48.79%		***	64.95%
Age	20.32	1.79	20.26			20.48%
Gainful activity	72.48%		75.19%		**	64.95%
Race (white)
Black	23.02%		20.93%		*	28.87%
Hispanic	11.72%		9.26%		***	18.56%
Family structure
Single parent	23.84%		22.41%			27.84%
Stepparent	13.08%		11.30%		*	18.04%
Other family	12.67%		10.93%		*	17.53%
Mother’s education (high school)
Less than high school	12.53%		10.00%		***	19.59%
Some college	34.33%		34.26%			34.54%
College or more	21.11%		23.70%		**	13.92%
Relationship status (dating)
Cohabiting	22.21%		17.96%		***	34.02%
Married	6.40%		5.74%			8.25%
Children	21.80%		18.15%		***	31.96%
Relationship duration	6.70	1.74	6.54		***	7.16
Prior IPV (wave 1)	20.98%		16.48%		***	33.51%

^*p < .05;

^**p < .01;

^***p < .001.

Table II.

Multilevel Logistic Regression of Relationship Violence on Neighborhood and Individual Dating Norms (n = 734)

	Zero Order		Model 1		Model 2		Model 3
Variable	OR	SE	OR	SE	OR	SE	OR	SE
Neighborhood Structure
Concentrated disadvantage	1.118***	(0.03)	1.069	(0.05)	1.070	(0.05)	1.044	(0.05)
Subjective disorder	1.265***	(0.06)	1.124	(0.10)	1.123	(0.10)	1.035	(0.11)
Neighborhood Culture
Liberal dating norms	1.271	(0.32)	1.000	(0.31)	0.852	(0.33)	0.819	(0.34)
Individual-Level
Liberal dating norms	1.184	(0.10)			1.181	(0.11)	1.401*	(0.13)
Coercive parenting	2.275***	(0.19)					2.063***	(0.21)
Female	1.901***	(0.18)					2.271***	(0.21)
Age	1.061	(0.05)					0.941	(0.06)
Gainful activity	0.664*	(0.19)					1.064	(0.22)
Race (white)
Black	1.671*	(0.21)					1.096	(0.28)
Hispanic	2.566***	(0.25)					1.629	(0.29)
Family structure
Single parent	1.748*	(0.22)					1.038	(0.25)
Stepparent	2.338***	(0.26)					1.738*	(0.28)
Other family	2.292**	(0.26)					1.318	(0.28)
Mother’s education (high school)
Less than high school	1.927*	(0.26)					1.431	(0.29)
Some college	1.027	(0.21)					1.014	(0.23)
College or more	0.641	(0.27)					0.891	(0.30)
Relationship status (dating)
Cohabiting	2.400***	(0.20)					1.530	(0.24)
Married	1.804	(0.33)					1.632	(0.40)
Children	2.003***	(0.20)					0.893	(0.25)
Relationship duration	1.249***	(0.06)					1.156*	(0.07)
Prior IPV (wave 1)	2.426***	(0.20)					2.213***	(0.22)
Variance component, Intercept
T			0.105	(0.10)	0.109	(0.10)	0.054	(0.11)

^*p < .05;

^**p < .01;

^***p < .001.

RESULTS

We present all means and standard deviations for the pooled sample and by perpetration status in Table I. Roughly a quarter (26.43%) of the sample reported IPV perpetration at the time of the fourth interview. Examination of differences between the violent and nonviolent subgroups across the broad range of study variables reveals a number of notable differences. In particular, individuals reporting IPV perpetration resided in neighborhoods with higher levels of concentrated disadvantage. Somewhat surprisingly, there was no bivariate difference between neighborhood and individual levels of dating norms across these two groups. Individuals reporting IPV perpetration also indicated higher levels of subjective disorder within their neighborhoods. A greater proportion of those reporting IPV perpetration were female and belonged to a racial minority group (black, Hispanic), and a lower proportion reported involvement in gainful activity. As compared to those reporting no IPV perpetration, a greater portion of the IPV perpetration subgroup reported coercive parenting. In addition, those belonging to the subgroup reporting IPV perpetration were more likely to grow up in a stepparent or “other” family structure. Overall, IPV perpetrators reported lower levels of parental education, as a greater share had mothers with less than a high school education and fewer had mothers who were college educated. As compared to their nonviolent counterparts, a greater proportion of respondents reporting IPV perpetration were in cohabiting relationships. These individuals also reported involvement in relationships of longer duration, and a greater share had children at the time of the wave 4 interview. Finally, a higher percentage of those reporting IPV perpetration also indicated prior exposure to IPV (wave 1).

We present the results of multivariate, multilevel analyses in Table II. At the zero order, neighborhood concentrated disadvantage and subjective disorder are associated with IPV perpetration; however, liberal dating norms at the neighborhood level are not associated with IPV. Model 1 includes the neighborhood-level factors as a block. Controlling for the other neighborhood structural and cultural considerations, the effects of concentrated disadvantage and subjective disorder are no longer significant. Similar to the bivariate findings, the link between neighborhood dating norms and IPV perpetration was not significant. Thus, contrary to study expectations, there does not appear to be an aggregate effect of the neighborhood culture with respect to liberal dating norms on IPV perpetration at the individual level. Individual liberal dating norms are introduced in Model 2. Net of aggregate- and individual-level dating norms, the effects of neighborhood concentrated disadvantage and subjective disorder on IPV perpetration remain nonsignificant. Further, individual-level dating norms are not associated with IPV perpetration, net of neighborhood characteristics. Accordingly, we find no support for the notion of an indirect pathway, as the neighborhood normative climate neither exerts a direct effect on IPV perpetration, nor does it operate indirectly via its influence on individual-level attitudes.

Model 3 includes controls for sociodemographic, adult status, and relationship characteristics. Net of these factors, the effects of neighborhood-level indicators of concentrated disadvantage, subjective disorder, and normative beliefs on IPV perpetration remain nonsignificant. In addition, coercive parenting, gender, family structure (stepparent), relationship duration, and prior IPV were all positively associated with IPV perpetration, such that respondents indicating exposure to coercive parenting, female respondents, respondents raised in stepparent households, respondents in relationships of longer duration, and those with prior IPV exposure, are at heightened risk of IPV perpetration. Notably, following the introduction of the full roster of study variables, the association between liberal dating norms and IPV perpetration increased in magnitude (p < .05). This suppression effect was due to the addition of gender to the model. That is, female respondents report significantly lower levels of liberal dating norms as compared to their male peers, yet are more likely to report IPV perpetration. Despite these differences, supplemental analyses revealed that the effect of liberal dating norms on IPV perpetration was similar for male and female respondents.

In a final set of analyses, we interacted the neighborhood-level indicators of concentrated disadvantage and subjective disorder with liberal dating norms to determine whether the association between individual dating norms and IPV perpetration was moderated by the neighborhood structure. The cross-product of individual-level dating norms and concentrated disadvantage was not significant, suggesting that the effect of liberal dating norms on IPV perpetration is similar across levels of concentrated disadvantage (results not shown). However, the effect of liberal dating norms interacts with subjective disorder, such that the effect of liberal dating norms is more pronounced at higher levels of subjective disorder (Fig. 1). Notably, the predicted probability of IPV perpetration among those exhibiting high levels of liberal dating attitudes nearly doubles as neighborhood levels of subjective disorder increase, from 19.6% in areas characterized by low disorder to 39.4% in areas high in subjective disorder.

Fig. 1.

Effects of liberal dating norms on the predicted probability of relationship violence perpetration across levels of subjective disorder.

CONCLUSION

Research over the past several decades has made great strides in identifying individual-level risk factors for IPV; however, we know little about whether contextual forces contribute to the use of relationship violence. A handful of recent studies have directed attention to neighborhood effects, yet this work overwhelmingly concentrates on the role of social processes (i.e., collective efficacy, social ties) to explain the link between neighborhood disadvantage and IPV. Few studies have examined other mechanisms by which neighborhoods contribute to variation in IPV perpetration. Using data from TARS, this investigation builds on the growing body of literature on neighborhoods and IPV. TARS is particularly well suited for these analyses as it includes detailed information about the family backgrounds and sociodemographic characteristics of respondents, as well as the neighborhood context within which respondents reside (based on data from the U.S. Census). Additionally, TARS provides information about the broader relationship context within which IPV occurs, and directs empirical attention to young adulthood, the period of the life course during which rates of IPV peak.

Prior neighborhood-level analyses of IPV and other forms of violence have tended to focus on either structural or cultural explanations (e.g., Browning 2002; Caetano et al. 2010). Researchers have found support for the general association between neighborhood levels of disadvantage and IPV (Browning 2002; Caetano et al. 2010; Frye et al. 2008; Wright and Benson 2011). Cultural explanations are less common in the IPV literature as compared to other areas of criminological research. Yet in general, these approaches highlight the values and norms that direct behavior—most commonly examining whether attitudes and beliefs about the acceptability of violence correspond to the use of violence (e.g., Copp et al. 2016). In the current investigation, we considered the role of structural and cultural influence processes in tandem. Informed by a social learning perspective, we expanded the theoretical lens by taking into consideration normative beliefs reflecting liberal dating attitudes which, although not directly related to IPV perpetration, likely shape how couples interact with each other. We first examined the direct association between neighborhood structural and cultural characteristics and IPV perpetration. Next, we considered whether the influence of the neighborhood context on IPV is indirect, operating via individual-level dating attitudes.

Our findings suggest that neighborhood levels of concentrated disadvantage are positively associated with IPV perpetration. In addition, perceptions of disadvantage (subjective disorder) are positively associated with IPV perpetration. Yet these associations are fairly tenuous and are easily explained by other features of the neighborhood context or neighborhood compositional features. Further, the aggregate effect of liberal dating norms on IPV perpetration was not significant in our analyses—neither in models controlling for the neighborhood structure nor at the zero order. Beyond the direct association between the neighborhood context and IPV perpetration, we posited that a potential mechanism driving neighborhood effects on IPV perpetration was individual-level normative beliefs. Yet as we described above, neighborhood levels of concentrated disadvantage and subjective disorder were not associated with IPV after controlling for other factors. Further, there was no aggregate effect of dating norms on IPV perpetration. Thus, given the lack of a contextual effect on IPV perpetration, we found no evidence of an indirect pathway via individual-level liberal dating norms.

Interestingly, although the effect of individual dating norms was not associated with IPV perpetration at the bivariate level, the effect of dating norms on IPV perpetration was significant in our full model. Supplemental analyses revealed that gender (female) suppresses the effect of dating norms on IPV perpetration. More specifically, due to the positive correlation between female and IPV perpetration, and the negative correlation between female and dating norms, failure to account for gender results in a null association between individuals’ liberal dating attitudes and IPV perpetration. In line with our final objective, we also found that the effect of liberal dating attitudes on IPV perpetration varied across neighborhood contexts. In particular, liberal dating norms have a direct effect on IPV perpetration, but this effect is magnified by subjective disorder. Thus, individuals who strongly endorse liberal dating norms are at greater risk of engaging in IPV, and this association is especially pronounced among those residing in neighborhoods characterized by greater levels of disorder.

Although this study offers new contributions, there are some limitations that are worth noting. First, the measure of IPV perpetration used in the current investigation is derived from the CTS2. Despite being a well-validated and commonly used measure of relationship violence—particularly in studies that rely on survey data—it does suffer from some important weaknesses, including a lack of attention to the context in which the violence occurs, as well as the motivations underlying the resort to violence (see Dobash and Dobash 1992). This may complicate a more nuanced understanding of certain dynamics, such as female perpetration which, similar to other community surveys, was more common than male perpetration in our sample. Importantly, we control for prior IPV exposure; however, given the high level of mutuality observed in young adult samples (see, e.g., Copp et al. 2016), our dependent variable, IPV perpetration, likely includes a certain level of victimization. Also, the current investigation relied on data from a regional sample of young adults. Nevertheless, the sample mirrored similar-aged young adults in the United States in terms of their sociodemographic profiles. Finally, this study focused exclusively on a discrete set of liberal dating norms among a sample of young adults. Future work should examine a broader range of attitudes and beliefs that may influence relationship conduct during the young adult period to further develop our understanding of the full range of “definitions” that contribute to relationship violence during this key phase of the life course.

Taken together, these findings provide support for a social learning approach to IPV. In particular, we found that the attitudes or meanings that individuals attach to dating have consequences for IPV perpetration. Whereas prior research on definitions has focused rather narrowly on attitudes “favorable and unfavorable” to a particular behavior, we argued that greater endorsement of liberal dating norms potentially reflects more permissive sexual behavior, including casual sex involvement and/or infidelity, which may contribute to relationship discord. Indeed, infidelity has been identified as a particularly salient “domain of contestation” in relationship disputes that escalate to the point of violence (Giordano et al. 2015). Accordingly, efforts to target the attitudes and beliefs that undergird such behaviors have the potential to lessen violent interactions at the couple level. These findings also underscore the need for additional research on a broader range of normative beliefs or “definitions”—both in studies at the individual and neighborhood levels. That the neighborhood normative climate was not associated with IPV perpetration in our analyses does not suggest that neighborhood norms are inconsequential for shaping individuals’ attitudes and guiding their behaviors. Future research is needed that provides a more nuanced examination of the role of culture in explaining behavior, including IPV.

In particular, it is likely that individuals are presented with conflicting cultural definitions, and this may particularly be the case in certain contexts (i.e., high-poverty neighborhoods). Thus, the extent to which the neighborhood presents a coherent set of values and norms, as well as the extent to which individuals’ attitudes and beliefs reflect these broader cultural patterns, is likely to vary. As was exemplified in the current investigation, for example, there is an association between neighborhood structure (based on objective and subjective criteria) and IPV perpetration. Further, individuals’ views of relationships have implications for their use of violence in the dyadic context. Importantly, residents of neighborhoods characterized by high levels of subjective disorder who endorsed liberal dating norms were at heightened risk of perpetrating violence. We suspect that this may be the case as rates of extramarital sex, nonmarital childbearing, and multiple partner fertility are higher in disadvantaged contexts (Anderson 1999; Furstenberg 1995; Manlove et al. 2008), and thus individuals who reside in such contexts may be especially attuned to particular relationship concerns, including infidelity. In addition, residents of disadvantaged neighborhoods are taught to “be tough” and to respond to affronts with violence. Thus, relationships involving individuals who (1) endorse liberal dating attitudes and (2) adopt a worldview that supports the use of violence as an interpersonal coping strategy are at heightened risk of relationship conflicts that include violence. Future research is needed that examines the processes by which individuals come to acquire definitions, the extent to which definitions underlie behaviors, and the types of definitions that individuals draw upon to make behavioral choices.