Patterns of Dating Violence Perpetration and ...

579506 research-article2015

JIVXXX10.1177/0886260515579506Journal of Interpersonal ViolenceSpencer et al.

Article

Patterns of Dating Violence Perpetration and Victimization in U.S. Young Adult Males and Females

Journal of Interpersonal Violence 1–22 © The Author(s) 2015 Reprints and permissions: sagepub.com/journalsPermissions.nav DOI: 10.1177/0886260515579506 jiv.sagepub.com

Rachael A. Spencer, MPA,1 Lynette M. Renner, MSW, PhD,2 and Cari Jo Clark, ScD, MPH3

Abstract Dating violence (DV) is frequently reported by young adults in intimate relationships in the United States, but little is known about patterns of DV perpetration and victimization. In this study, we examined sexual and physical violence perpetration and victimization reported by young adults to determine how the violence patterns differ by sex and race/ethnicity. Data from non-Hispanic White, non-Hispanic Black, and Hispanic participants in Wave 3 of the National Longitudinal Study of Adolescent to Adult Health were analyzed. DV was assessed using responses to four questions focused on perpetration and four questions focused on victimization. The information on DV was taken from the most violent relationship reported by participants prior to Wave 3. Latent class analysis was first conducted separately by sex, adjusting for age, race/ethnicity, and financial stress, then by race/ethnicity, adjusting for age and financial stress. Relative model fit was established by comparing Bayesian Information Criteria (BIC), adjusted

1Gender-based

Violence Specialist, Atlanta, GA, USA of Minnesota School of Social Work St. Paul, MN, USA 3University of Minnesota Medical School, Minneapolis, MN, USA 2University

Corresponding Author: Rachael A. Spencer, Gender-based Violence Specialist, 116 Ponce de Leon Avenue, Atlanta, GA 30308. Email: [email protected]

Downloaded from jiv.sagepub.com by guest on April 7, 2015

2

Journal of Interpersonal Violence

BIC, entropy, interpretability of latent classes, and certainty of latent class assignment for covariate-adjusted models. The results indicate that patterns of violence differed by sex and for females, by race/ethnicity. A three-class model was the best fit for males. For females, separate four-class models were parsimonious for White, Black, and Hispanic females. Financial stress was a significant predictor of violence classification for males and females and age predicted membership in White and Black female models. Variations in DV patterns by sex and race/ethnicity suggest the need for a more nuanced understanding of differences in DV. Keywords dating violence, intervention, youth violence, bidirectional violence

Introduction Dating violence (DV), which is defined as physical, psychological, or sexual violence perpetrated by one or both intimate partners, is common during adolescence and young adulthood. DV is more likely to occur in relationships that are longer term and marked by frequent contact (Giordano, Soto, Manning, & Longmore, 2010) and, given its potential health consequences and associated risk factors (Ackard, Eisenberg, & Neumark-Sztainer, 2007; Clark et al., 2014; Silverman, Raj, Mucci, & Hathaway, 2001), the prevalence of DV in adolescence and early adulthood is alarming. In the National Intimate Partner and Sexual Violence survey, 22% of adult females and 15% of adult males reported experiencing some form of partner violence for the first time between the ages of 11 and 17 (Black et al., 2011). In a nationwide survey of high school students, approximately 10% of teens reported experiencing physical and 10% reported sexual victimization in the 12 months prior to the survey (Kann et al., 2014). In one nationally representative sample of 10th-grade students, 35% reported victimization and 31% reported perpetration of physical and/or psychological violence in a dating relationship (Haynie et al., 2013). Adolescents and young adults may be more likely to use violence to solve problems compared with adults (Cutter-Wilson & Richmond, 2011), and DV is often characterized by bidirectional violence (both partners perpetrate violence; Langhinrichsen-Rohling, Selwyn, & Rohling, 2012; Renner & Whitney, 2010). However, males and females experience partner violence differently; females report physical violence victimization and perpetration more frequently than males (Fiebert, 2004), and females report more sexual violence victimization and injury than males (Black et al., 2011). The


3

Spencer et al.

disparities in violence perpetration may be due to the use of violence as a form of self-defense among females (Swan & Snow, 2006) or as a response to the use of coercive tactics by males (Swan & Snow, 2002) although reports of female violence have generated debate about the characterization of males as primary aggressors in violent relationships (Hamby, 2009). Reports of DV perpetration and victimization not only differ by sex but they also differ across race/ethnicity. Black females are at increased risks of victimization and perpetration (Capaldi, Knoble, Shortt, & Kim, 2012; Foshee, McNaughton Reyes, & Ennett, 2010; Langhinrichsen-Rohling et al., 2012) and Blacks and Hispanic couples have been shown to experience bidirectional violence more frequently than Whites (Caetano, Ramisetty-Mikler, & Field, 2005; Champion, Wagoner, Song, Brown, & Wolfson, 2008). Although financial stress has been associated with recurrence of relationship violence in multiethnic samples (Caetano, Field, Ramisetty-Mikler, & McGrath, 2005), racial/ethnic differences in violence victimization and perpetration may be attenuated by individual and community measures of financial distress (Fox & Benson, 2006; Fox, Benson, DeMaris, & Van Wyk, 2002). This is especially true for Hispanics and Blacks who are disproportionately poor in the United States (DeNavas-Walt, Proctor, & Smith, 2013) and more likely to live in areas with generally higher concentrations of neighborhood disadvantage (Benson, Fox, DeMaris, & Van Wyk, 2000) compared with Whites. Despite a growing literature on the prevalence of and disparities in bidirectional violence in young adult relationships, there is little understanding of the back-and-forth nature of DV. Many researchers measure either perpetration or victimization and, among those who measure both, most do not incorporate measures of sexual violence or injury that are disproportionately reported by females (Hamby, 2009). Furthermore, with few exceptions (Orpinas, Hsieh, Song, Holland, & Nahapetyan, 2013), researchers do not explore differences in violence severity separately committed by one partner compared with the other. Similar research limitations exist in studies of violent adult relationships (see Catalano, 2011). Whereas bidirectional violence has been associated with more injury and severity compared with unidirectional violence (Whitaker, Haileyesus, Swahn, & Saltzman, 2007), sex and racial/ethnic disparities in violence experience suggest that bidirectionally violent relationships are likely characterized by various levels of severity and thus may be differently related to health. Therefore, clarifying differences in violence victimization and perpetration by sex and race/ethnicity is essential to developing nuanced interventions for at-risk young adults.


4


Materials and Method Sample Participants in this study were part of the National Longitudinal Study of Adolescent to Adult Health (Add Health), which is a longitudinal study of adolescents in Grades 7 to 12 in the United States during the 1994-1995 school year drawn from a nationally representative sample (Harris et al., 2009). Of the 15,197 participants in Wave 3 (2001-2002), data for this analysis include 10,665 Wave 3 participants who self-identified as non-Hispanic White (72.3%; n = 6,484), non-Hispanic Black (15.6%; n = 2,335), or Hispanic (12.1%; n = 1,846), had valid sampling weights, and reported on at least one relationship in which violence was assessed. Wave 3 participants were chosen for this study because it assessed relationships that had occurred since baseline and assessed a greater number of relationships over a longer period of time than at either Wave 2 or 4. The parent study was approved by the institutional review board (IRB) of the University of North Carolina, Chapel Hill. The University of Minnesota IRB determined that the study did not meet the regulatory definition of human subjects research due to the sole use of de-identified data.

Measures Violence. Participants were considered to have perpetrated DV or experienced DV victimization in the modeled relationship if they answered affirmatively to any of four measures of perpetration or any of the four measures of victimization, respectively. The items focused on perpetration and victimization were based on the Revised Conflict Tactics Scale (Straus, Hamby, Boney-McCoy, & Sugarman, 1996) and included acts referred to as (a) pushing (i.e., pushing, shoving, or threats of violence), (b) hitting (i.e., hitting, slapping, or kicking), (c) sexual (i.e., insisting on or making the partner have sexual relations), and (d) injury (i.e., injury as a result of fighting with a partner). The most violent relationship was chosen for analysis. This was defined as the relationship in which the participant reported the highest score across violence measurement items—from among the participant’s serious relationships reported since Wave 1. Relationship seriousness was defined by Add Health as those being the most important (i.e., “the relationship whose end would be most painful for you or which you would be happiest to continue”), longer term (i.e., duration of 3 months or longer), recent (i.e., current or the most recent), or involved marriage, cohabitation, or resulted in a pregnancy.


5

Spencer et al.

Sociodemographic variables. At Wave 1, self-reports of race/ethnicity were recorded as non-Hispanic White, non-Hispanic Black, and Hispanic. Age at Wave 3 was modeled as a continuous variable. A variable for financial stress at Wave 3 was created using seven items that assessed the study participant’s inability to pay for basics in the prior 12 months, such as an absence of phone service and experiencing eviction. Financial stress at Wave 3 was modeled as a dichotomous variable. Financial stress and age were incorporated in the models because financial stress is associated with partner violence (Fox et al., 2002), and the chances of having entered into a relationship and been exposed to a violent relationship increase across adolescence and into young adulthood (Orpinas et al., 2013).

Statistical Analysis Latent class analysis (LCA) was performed to develop patterns of young adults’ experiences of DV perpetration and victimization. LCA is one example of a pattern-centered or person-centered analysis, which allows individuals to be classified into subgroups. The focus of LCA is the discovery of homogeneous subgroups, heterogeneity across subgroups, and the test of associations between subgroups. Taking DV experiences into account, the aim of this study was to maximize homogeneity within classes of respondents and to maximize heterogeneity between classes while accounting for sex, race/ethnicity, and financial stress. Utilizing LCA allowed for the exploration of diversity among populations of adolescents’ and young adults’ experiences of DV while highlighting similarities within subgroups. Such information is essential to developing intervention strategies that complement the diversity of experiences and needs among individuals who have experienced violence (Nurius & Macy, 2008). LCA is predicated on the assumption that two or more observed variables create patterns that can be explained by unobserved or latent variables plus error (Muthén, 2004). Class membership probabilities and item-response probabilities, which are conditional on class membership, are calculated based on responses to observed variables. Models were identified by comparing parameter estimates using multiple starting estimates to avoid the use of local maxima. For this study, relative model fit was established by comparing Bayesian Information Criteria (BIC) and adjusted BIC (Schwarz, 1978) (i.e., generally a lower value for each estimate indicates a better model fit; Collins & Lanza, 2010), entropy value (i.e., closer to 1 is more desirable), interpretability of latent classes (Collins & Lanza, 2010), and certainty of latent class assignment (i.e., >.80; Rost, 2006) for covariate-adjusted models.


6


Models were initially examined separately by sex based on robust literature indicating that males and females report perpetration and victimization differently (Archer, 2000; Black et al., 2011). Models were then stratified by race/ethnicity due to evidence suggesting that Blacks and Hispanics are more likely to be in bidirectionally violent relationships compared with Whites (Caetano, Ramisetty-Mikler, et al., 2005; Champion et al., 2008). For male models stratified by race/ethnicity, entropy, information criteria, and interpretability pointed to three-class models for each race/ethnicity. The similarity of item-response probabilities indicated that males could be combined into one model and latent class prevalence could be compared across race/ ethnicity. For female models stratified by race/ethnicity, entropy, information criteria, interpretability, and marked differences in item-response probabilities for the most parsimonious models pointed to the identification of three separate four-class models. For female models, latent class prevalence cannot be compared across race/ethnicity due to the specification of separate models. Multinomial logistic regression was performed to account for differences by race/ethnicity, financial stress, and age for males and financial stress and age for females, with the classes as the outcome variables. To avoid removing participants with missing values on covariates of interest, maximum likelihood estimation with robust standard errors was performed for all models. Covariate models were compared with corresponding models without covariates to ensure continuity in item-response probabilities and latent class prevalence. Models were specified in Mplus Version 7.2 (Muthén & Muthén, 1998-2012), and data were weighted using sampling weights and a clustering variable was incorporated according to Add Health study guidance (Chen & Chantala, 2014).

Results On average, the sample was 21.8 (95% confidence interval [CI] = [21.6, 22.1]) years old and 50.4% female (n = 5,780). Of the 10,655 participants in the study, 72.3% (n = 6,484) were White, 15.6% (n = 2,335) were Black, and 12.1% (n = 1,846) were Hispanic. Table 1 presents demographic characteristics of the sample by DV perpetration and victimization measures. Except for injury victimization, Whites were the least likely to report any form of violence victimization or perpetration compared with same-sex Blacks and Hispanics. Compared with females, males of the same race were less likely to report any type of violence perpetration or victimization with the exception of injury victimization and sexual violence perpetration.



7

22.0 (21.7, 22.3) 41 284 (18.5)

68

383 (13.2)

209 (22.9)

161 (18.5)

22.0 (21.7, 22.3) 31 410 (26.6)

58

549 (18.2)

240 (28.0)

197 (20.6)

168 (18.3)

184 (21.6)

570 (17.9)

48

22.1 (21.8, 22.4) 21 367 (24.94)

89 (9.6)

86 (10.7)

141 (5.7)

51

22.3 (21.9, 22.6) 24 136 (9.8)

45 (5.1)

83 (3.8)

161 (6.0)

47

22.0 (21.6, 22.4) 20 134 (9.2)

36 (3.7)

66 (8.2)

109 (4.0)

43

22.0 (21.6, 22.4) 17 96 (6.2)

98 (9.4)

126 (11.3)

221 (7.8)

49

21.9 (21.6, 22.2) 23 126 (8.0)

items include pushing, shoving, or threats of violence. items include hitting, slapping, or kicking. items include injury as a result of fighting with a partner. dSexual items include insisting on or making the partner have sexual relations.

cInjury

bHit

aPush

Missing (n) Financial Stress, n (%) Missing (n) Race Non-Hispanic White Non-Hispanic Black Hispanic

Age, M (SD)

Females (n= 5,780)

66 (5.8)

83 (8.3)

143 (5.1)

50

21.9 (21.5, 22.3) 22 187 (12.4)

315 (34.8)

521 (38.5)

903 (26.2)

59

21.7 (21.4, 22.0) 35 728 (36.0)

281 (31.7)

386 (29.1)

904 (25.9)

74

21.7 (21.4, 21.9) 50 800 (39.6)

182 (20.7)

251 (19.1)

542 (15.5)

56

21.8 (21.5, 22.1) 32 473 (23.7)

259 (30.7)

387 (29.9)

736 (21.2)

55

21.6 (21.3, 21.9) 31 640 (32.2)

114 (14.7)

143 (9.3)

345 (9.9)

57

21.7 (21.4, 22.1) 33 311 (15.2)

75 (8.6)

117 (8.3)

174 (5.1)

58

21.8 (21.4, 22.2) 34 189 (9.0)

148 (17.3)

219 (14.8)

443 (12.9)

66

21.7 (21.4, 22.0) 42 369 (17.0)

55 (6.1)

115 (8.1)

63 130 (3.8)

21.9 (21.5, 22.2) 39 153 (7.0)

Push Push Hit Hit Injury Injury Sexual Sexual Push Push Hit Hit Injury Injury Sexual Sexual Victimizationa Perpetrationa Victimizationb Perpetrationb Victimizationc Perpetrationc Victimizationd Perpetrationd Victimizationa Perpetrationa Victimizationb Perpetrationb Victimizationc Perpetrationc Victimizationd Perpetrationd

Males (n = 4,885)

Table 1. Participant Characteristics by Violence Perpetration and Victimization, Males and Females (N = 10,665).

8


Table 2. Information Criteria for Three-, Four-, and Five-Class Models With Covariates, Males and Females (N = 10,665). Class

BIC

Adjusted BIC

Non-Hispanic White, Non-Hispanic Black, and Hispanic males (n = 4,885) Three-class 25,131.08 25,016.68 Four-class 25,077.46 24,921.75 Five-class 25,019.44 24,822.42 Non-Hispanic White females (n = 3,498) Three-class 21,260.32 21,158.64 Four-class 21,130.66 20,994.03 Five-class 21,215.86 21,044.28 Non-Hispanic Black females (n = 1,337) Three-class 9,329.83 9,228.18 Four-class 9,294.38 9,157.79 Five-class 9,310.58 9,139.04 Non-Hispanic Hispanic females (n = 945) Three-class 6,733.83 6,632.20 Four-class 6,718.67 6,582.10 Five-class 6,705.71 6,534.21

Entropy .91 .85 .85 .82 .85 .87 .84 .90 .91 .79 .88 .91

Note. BIC = Bayesian Information Criteria.

Model Selection Information criteria for three-, four-, and five-class, covariate-adjusted models for males and females are indicated in Table 2. Because one model was selected for males and three separate models were selected for females, each model is described separately below. White, Black, and Hispanic males. Both the three- and four-class models could be considered an appropriate fit for males; however, the three-class male model was selected based on its higher entropy value and interpretability. The average latent class assignment probabilities for the three-class model indicated excellent precision of the classification (i.e., >.90). White females. A four-class model was parsimonious for White females. BIC and adjusted BIC dipped to the lowest point for the four-class model and rose thereafter. Average latent class assignment probabilities for the four-class model indicated a good precision of the classification (i.e., >.85). Black females. Black females produced a parsimonious model at four latent classes. The lower BIC and adjusted BIC for the four-class compared with Downloaded from jiv.sagepub.com by guest on April 7, 2015

9

Spencer et al.

the three-class model and the higher entropy indicated a better fit for the fourclass model. The four-class model was considered superior to the five-class model because the BIC increased for the five-class model and the average latent class assignment probabilities for the four-class model indicated a high precision of the classification (i.e., >.80), whereas the five-class model fell below this threshold. Hispanic females. Hispanic females produced a parsimonious four-class model. A decrease in BIC and significant increase in entropy indicated a better fit for the four-class compared with the three-class model. The five-class model demonstrated a leveling off of change in entropy and also was not chosen because the additional class showed insignificant difference in itemresponse probabilities from the class in the four-class model from which it was derived. Average latent class assignment probabilities for the four-class model indicated a good precision of the classification (i.e., >.85). Tables 3 to 6 present the latent class models for males and females. To simplify the naming process for this study, the same labels were applied to classes with generally similar structures. However, models are significantly unique in item-response probability patterns and are not nested; therefore, comparisons across models are interpreted with caution and are limited to the Unexposed class because item-response probabilities vary by less than .07 across all models. Table 3 presents the three latent classes for males, which are labeled Unexposed (low probability of reporting any violence perpetration or victimization), Severe Bidirectional (high probability of reporting perpetration and victimization for all perpetration and victimization items), and Moderate Victim (high probability of reporting pushing and hitting victimization and low probability of reporting sexual violence and injury victimization and perpetration). The term moderate was used to indicate that the violence experienced was limited to items considered to be less severe. Seventyseven percent (n = 3,758) of males were categorized as Unexposed, 19% (n = 933) were Severe Bidirectional, and 4% (n = 194) were categorized as Moderate Victim. Compared with Unexposed White males, Black males were 207% (odds ratio [OR] = 3.07, 95% CI = [1.80, 5.22]) more likely to be a Moderate Victim and 37% (OR = 1.37, 95% CI = [1.03, 1.81]) more likely to be classified as Severe Bidirectional. Financial stress was associated with a 77% (OR = 1.77, 95% CI = [1.11, 2.81]) increase in odds of being a Moderate Victim and 83% (OR = 1.83, 95% CI = [1.52, 2.19]) higher odds of being a Severe Bidirectional compared with males who were in the Unexposed class. Age was not a significant predictor of class membership for males (Table 3).


10


Table 3. Item-Response Probabilities and Predictors of Class Membership, ThreeClass Non-Hispanic White, Non-Hispanic Black, and Hispanic Males (n = 4,885). Moderate Victim (n = 194)a Push perpetrationc Push victimizationc Hit perpetrationd Hit victimizationd Sexual perpetratione Sexual victimizatione Injury perpetrationf Injury victimizationf Race White Black Hispanic Age Financial stress

.04 .45 .80 .18 .67 .12 .18 .05 .15

Severe Bidirectional (n = 933)a

Latent class prevalence .19 Item-response probabilitiesb .93 .90 .81 .93 .50 .51 .82 .86

Unexposed (n = 3,758)a .77 .04 .01 0 .02 .02 .04 0 .01

Odds Ratio (95% CI)g



Ref 3.07 [1.80, 5.22]* 1.12 [0.53, 2.35] 0.98 [0.87, 1.11] 1.77 [1.11, 2.81]*

Ref 1.37 [1.03, 1.81]* 1.21 [0.88, 1.66] 1.05 [0.98, 1.12] 1.83 [1.52, 2.19]*

Ref Ref Ref Ref Ref

Notes. CI = confidence interval; Item-response probabilities appear in bold to facilitate interpretation of latent class patterns. aAdjusted for age, race, and financial stress. bIndicates the probability of endorsing the violence item. cPush items include pushing, shoving, or threats of violence. dHit items include hitting, slapping, or kicking. eSexual items include insisting on or making the partner have sexual relations. fInjury items include injury as a result of fighting with a partner. gEstimate relative to men’s Unexposed class. *p < .05.

Latent class membership is presented separately for White (Table 4), Black (Table 5), and Hispanic (Table 6) females due to differences in each of the fourclass compositions. Three classes, including an Unexposed class (low probability of reporting any violence perpetration or victimization), Moderate Perpetrator class (high probability of reporting hitting and pushing perpetration and low probability of reporting sexual violence and injury victimization and perpetration), and a Severe Bidirectional class (high probability of reporting Downloaded from jiv.sagepub.com by guest on April 7, 2015

11

Spencer et al.

Table 4. Item-Response Probabilities and Predictors of Class Membership, FourClass Non-Hispanic White Females (n = 3,498). Moderate Perpetrator (n = 667)a Push perpetrationc Push victimizationc Hit perpetrationd Hit victimizationd Sexual perpetratione Sexual victimizatione Injury perpetrationf Injury victimizationf Age Financial stress

.19 .68 .43 .42 .08 .06 .17 .03 .03

Severe Victim (n = 124)a

Severe Bidirectional Unexposed (n = 416)a (n = 2,291)a

Latent class prevalence .04 .12 Item-response probabilitiesb 0 .92 1.00 .99 .11 .88 .92 .93 .08 .10 .59 .34 .10 .36 .69 .60

Odds Ratio (95% CI)

Odds Ratio (95% CI)

Odds Ratio (95% CI)

0.91 [0.85, 0.98]* 2.19 [1.68, 2.86]*

0.94 0.94 [0.81, 1.08] [0.87, 1.03] 2.16 3.27 [1.17, 3.97]* [2.35, 4.56]*

.65 .02 .03 .03 0 .02 .05 0 0 Odds Ratio (95% CI) Ref Ref

Notes. CI = confidence interval; Item-response probabilities appear in bold to facilitate interpretation of latent class patterns. aAdjusted for age and financial stress. bIndicates the probability of endorsing the violence item. cPush items include pushing, shoving, or threats of violence. dHit items include hitting, slapping, or kicking. eSexual items include insisting on or making the partner have sexual relations. fInjury items include injury as a result of fighting with a partner. *p < .05.

perpetration and victimization for all perpetration and victimization items with the exception of sexual violence perpetration), were identified for all models. Similar to labels used for the male model, the term, moderate, was used to indicate that the violence experienced was limited to the less severe categories. The additional class for Black females was labeled Moderate Bidirectional (high probability of reporting hitting and pushing victimization and perpetration and low probability of reporting sexual violence and injury victimization and perpetration). Violence patterns for the remaining class for Hispanic females were labeled Sexual Violence Bidirectional (high probability of reporting sexual violence victimization and perpetration and low probability of Downloaded from jiv.sagepub.com by guest on April 7, 2015

12


Table 5. Item-Response Probabilities and Predictors of Latent Class Membership, Four-Class Non-Hispanic Black Females (n = 1,337). Severe Bidirectional (n = 163)a Push perpetrationc Push victimizationc Hit perpetrationd Hit victimizationd Sexual perpetratione Sexual victimizatione Injury perpetrationf Injury victimizationf Age Financial stress

.12 .87 .96 .83 .89 .31 .57 .60 .73

Moderate Bidirectional (n = 112)a

Moderate Perpetrator (n = 321)a

Latent class prevalence 0.08 .24 Item-response probabilitiesb .73 .83 .74 .40 .88 .41 1.00 0 .05 .07 .06 .13 0 .01 0 0

Unexposed (n = 741)a .55 0 .01 .03 0 .04 .07 .01 0

Odds Ratio (95% CI)

Odds Ratio (95% CI)

Odds Ratio (95% CI)

Odds Ratio (95% CI)

1.09 [0.95, 1.25] 2.04 [1.25, 3.32]*

1.04 [0.90, 1.21] 1.96 [1.17, 3.29]*

0.84 [0.75, 0.93]* 1.84 [1.17, 2.89]*

Ref Ref


reporting all other violence items). For White females, the additional class was labeled as Severe Victim (high probability of reporting all victimization items and low probability of reporting all perpetration items). Overall, the models for females were similar in that most White (65%; n = 2,291), Black (55%; n = 741), and Hispanic (62%; n = 584) females were classified as Unexposed. Moderate Perpetrator and Severe Bidirectional were the second and third most common classes for each female model, respectively. Financial stress was a significant predictor of class membership in each of the three female models with the exception of Sexual Bidirectional


13

Spencer et al.

Table 6. Item-Response Probabilities and Predictors of Class Membership, FourClass Hispanic Females (n = 945). Severe Bidirectional (n = 134)a Push perpetrationc Push victimizationc Hit perpetrationd Hit victimizationd Sexual perpetratione Sexual victimizatione Injury perpetrationf Injury victimizationf Age Financial stress

.14 .72 .96 .74 .91 .17 .51 .52 .87

Moderate Perpetrator (n = 213)a

Sexual Violence Bidirectional (n = 14)a

Latent class prevalence .23 0.01 Item-response probabilitiesb .84 0 .59 0 .67 0 .27 0 .09 1.00 .23 1.00 .02 0 .04 0.10

Unexposed (n = 584)a .62 .07 .05 .06 0 0 .04 0 0

Odds Ratio (95% CI)

Odds Ratio (95% CI)

Odds Ratio (95% CI)

Odds Ratio (95% CI)

1.02 [0.88, 1.12] 2.39 [1.32, 4.32]*

0.97 [0.83, 1.13] 2.09 [1.17, 3.76]*

1.11 [0.89, 1.37] 0.85 [0.18, 4.09]

Ref Ref


compared with Unexposed Hispanic females. Financial stress increased the odds of classification as a Moderate Perpetrator by between 84% and 119% and of being Severe Bidirectional by between 104% and 237% compared with their respective Unexposed class for White (Table 4), Black (Table 5), and Hispanic (Table 6) females. Age predicted membership in the Moderate Perpetrator class among White (OR = .91, 95% CI = [.85, .98]) and Black (OR = .84, 95% CI = [.75, .93]) females with younger participants being more likely to be Moderate Perpetrator than older participants. Age was not a significant predictor of class membership for Hispanic females.


14


Discussion The patterns of DV identified in this study shed light on the diversity of experiences encountered by adolescents and young adults in violent dating relationships and move DV research forward by the identification of patterns of violence that may be differentially related to health. The prominence of moderate forms of violence (i.e., hitting and pushing) in primarily female-perpetrated DV relationships—a consistent finding across models for males and females—provides important clarifications to researchers’ prior findings that females are more likely to be DV perpetrators compared with males. Furthermore, this study’s findings not only confirm the existence of racial/ ethnic disparities in DV experience but also demonstrate that financially stressed individuals of all race/ethnicities are more likely to experience most forms of DV. Thus, we strengthen the argument not only for global DV prevention but also for interventions tailored to address the diversity of DV experienced by young adults of different race/ethnicity and sex. Examination of measures of sexual and physical DV victimization and perpetration and financial stress in a nationally representative sample of White, Black, and Hispanic young adults strengthens these findings. Overall, the patterns of DV perpetration and victimization identified in this study are similar to previous researchers who found that nonexposure is the most frequent experience of young adults; yet, bidirectional violence is common in young adult violent relationships (Gray & Foshee, 1997; Renner & Whitney, 2010), and reports of violence differ by sex (Archer, 2000; Black et al., 2011; Hamby, 2009) and race/ethnicity (Caetano, Ramisetty-Mikler, et al., 2005; Renner & Whitney, 2010). Our findings highlight commonalities of DV experience across biological sex by indicating that, in general, bidirectional violence is characterized by sexual violence and injury. These commonalities extend to reports of primarily female-perpetrated violence; when females are the primary perpetrators of violence, moderate violence (i.e., little to no sexual violence or injury) is perpetrated. Although bidirectional violence has been associated with greater injury and severity (Whitaker et al., 2007), each pattern of bidirectional violence identified in this study is likely differentially related to health. Further investigation into the health consequences associated with different patterns of perpetration and victimization is required to identify those at greatest health risk. While this study cannot speak to the intent behind White, Black, and Hispanic females’ Moderate Perpetrator classes, they might be considered “false positives” or “horseplay” that has been described by Hamby (2009) as “pushing and sometimes even hitting or kicking [which] are not necessarily violent acts . . . horseplay is common and probably serves both social and


15

Spencer et al.

sexual desires for newcomers to adult relationships” (p. 29). This may be especially relevant for White and Black females because these classes are associated with younger age in both models. An alternative theory is that these classes represent scuffling types of relationships (Draucker et al., 2010) involving a “series of minor arguments or altercations that have a ‘back and forth’ quality without a major escalation. Both partners exhibit aggression. The aggression is usually verbal but might include minor physical abuse” (p. 521). We suggest this as a possibility because, although females appear to be the primary physical aggressors in these classes, there may be bidirectional verbal abuse and coercive control occurring in the partnership that our analysis does not capture. In studies of psychological violence, male and female adolescents and young adults report perpetrating verbal abuse in almost equal numbers (Renner & Whitney, 2010) and males may use coercive control more successfully than females based on their ability to instill fear in females (Swan & Snow, 2006). Including the use of psychological violence and coercive control in future analyses would help to shed light on the intent underlying higher female reports of perpetration compared with males. Furthermore, measuring levels of fear is important to understanding the power dynamics inherent in females’ reports of violence (Swan & Snow, 2006). Person-centered analyses are gaining momentum among interpersonal violence researchers (Swartout & Swartout, 2012). These types of analyses highlight meaningful commonalities within subgroups and distinguish between subgroups. Within this sample of young adults, our findings reveal that there are distinctions in DV experiences between classes, especially with respect to race/ethnicity among males. Similar patterns of victimization and perpetration among males allowed for comparisons across race/ethnicity and provide more insight into existing findings that compared with Whites males, Black males were more likely to be in a violent relationship (Foshee et al., 2010; Renner & Whitney, 2010). Whereas some researchers have found that Hispanic couples report higher rates of bidirectional violence compared with Whites (Caetano, Ramisetty-Mikler, et al., 2005), this study and others (Klevens, 2007) found no significant difference in bidirectional violence rates for Hispanic males compared with White males. We found that patterns of victimization and perpetration differed for White, Black, and Hispanic females. Overall, the findings for Black females are in line with extant literature indicating that Black females report high rates of both physical perpetration and victimization compared with White females (Capaldi et al., 2012; Foshee et al., 2010; Langhinrichsen-Rohling et al., 2012). Unique patterns of violence reported by Black females may be reflective of the use of self-defense among Black females and acceptance of “horseplay” among Black youth (Love & Richards, 2013). Another potential


16


explanation for the reports of physical violence perpetration by Black females is the high barrier to formal help seeking. Black females are often pressured not to reveal problems to police or social service organizations and to be strong enough to deal with problems on their own; therefore, Black females may rely on violence to resolve conflicts as a last resort (Swan & Snow, 2006). Extant literature has been less consistent about Hispanic females’ experience with intimate partner violence. With few exceptions (Caetano, Ramisetty-Mikler, et al., 2005), community samples indicate similar rates of violence among Hispanic and non-Hispanic Whites (Langhinrichsen-Rohling et al., 2012) while clinical samples suggest that Hispanics experience violence differently based on their place of birth (i.e., inside vs. outside the United States; Capaldi et al., 2012) and level of acculturation stress (Caetano, Ramisetty-Mikler, Vaeth, & Harris, 2007). Research including clinical samples of female partner violence victims also indicates that, based on the patriarchal nature of some Hispanic cultures, Latinas experience particularly high rates of sexual violence victimization compared with non-Latinas (Glass et al., 2009; Klevens, 2007). Studies of community samples may obscure the differences in violence exposure identified in clinical samples by combining physical and sexual acts of violence into a “severe” violence category for the purpose of analysis. The present study makes an initial effort to clarify differences in violent experiences by race/ethnicity but clearly more research with community samples is needed to inform prevention efforts. Financial stress predicted class membership across race/ethnicity and sex, with the exception of the class of Hispanic female Sexual Violence Bidirectional. Experience of financial stress is a risk factor for partner violence exposure in adulthood (Capaldi et al., 2012), in part due to the importance that intimate partners place on financial well-being and their partner’s economic productivity (Fox et al., 2002). Other measures of financial stress, including negative feelings about one’s financial status and a couple’s economic risk profile (family debt load, male’s employment, female’s and male’s feelings of economic well-being, and the family’s income-to-needs ratio) have similarly been related to partner victimization and perpetration (Fox & Benson, 2006; Fox et al., 2002). Because young adults often lack the mechanisms to cope with stress (Cutter-Wilson & Richmond, 2011) and emotional distress is a risk factor for DV in adolescence (Vagi et al., 2013; Yahner, Dank, Zweig, & Lachman, 2014), they may be more likely to use violence when facing financial stress compared with adults. Indicators of community economic disadvantage have also been related to partner violence (Fox & Benson, 2006; Pinchevsky & Wright, 2012), suggesting that partner violence is not only an individual problem among financially stressed young adults but rather a community resource issue. This has implications for DV prevention


17

Spencer et al.

programs that operate nationwide but may need to be targeted toward economically disadvantaged communities. Future research should include more robust examination of how the patterns of violence identified in this study relate to macro-level indicators of community economic disadvantage.

Limitations The findings of this study were limited in that they model violence experienced in the most violent relationship prior to Wave 3 and the participant’s classification does not account for victimization or perpetration in less violent relationships. Furthermore, we did not incorporate any measure of psychological violence that has been associated with mental and physical health sequelae and found to be differentially reported by sex (Coker et al., 2002) and race/ethnicity (Orpinas, Nahapetyan, Song, McNicholas, & Reeves, 2012; Renner, Whitney, & Vasquez, 2015).

Conclusion The differences in patterns of DV by race/ethnicity and sex in this study shed light on the back-and-forth dimensions of DV. While prior researchers indicate that bidirectional DV is associated with greater frequency and injury compared with unidirectional violence, the unique latent classes identified in this study suggest that patterns of bidirectional violence are likely to differ in their relation to health. Examining the nuanced nature of bidirectional violence as it is experienced by individuals of different race/ethnicity and sex is not only important to contextualizing findings that females are more likely to perpetrate DV compared with males but also to tailoring DV interventions to address the diversity of DV in young adulthood. Authors’ Note Information on how to obtain the Add Health data files is available on the Add Health website (http://www.cpc.unc.edu/addhealth).

Declaration of Conflicting Interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Support was provided by National


18


Center for Advancing Translational Sciences of the National Institutes of Health Award No. 8UL1TR000114-02/KL2TR000113 and Grant R03HD068045 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development. Additional support was provided by the Program in Health Disparities Research and the Applied Clinical Research Program at the University of Minnesota. This research uses data from Add Health, a program project directed by Kathleen Mullan Harris and designed by J. Richard Udry, Peter S. Bearman, and Kathleen Mullan Harris at the University of North Carolina at Chapel Hill, and funded by Grant P01-HD31921 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development, with cooperative funding from 23 other federal agencies and foundations. Special acknowledgment is due to Ronald R. Rindfuss and Barbara Entwisle for assistance in the original design. No direct support was received from Grant P01-HD31921 for this analysis. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

References Ackard, D. M., Eisenberg, M. E., & Neumark-Sztainer, D. (2007). Long-term impact of adolescent dating violence on the behavioral and psychological health of male and female youth. The Journal of Pediatrics, 151, 476-481. doi:10.1016/j. jpeds.2007.04.034 Archer, J. (2000). Sex differences in aggression between heterosexual partners: A meta-analytic review. Psychological Bulletin, 126, 651-680. Benson, M., Fox, G., DeMaris, A., & Van Wyk, J. (2000). Violence in families: The intersection of race, poverty, and community context. In G. Fox & L. Benson (Eds.), Families, crime and criminal justice (pp. 91-110). London, England: Elsevier. Black, M. C., Basile, K. C., Breiding, M. J., Smith, S. G., Walters, M. L., Merrick, M. T., . . . Stevens, M. R. (2011). The national intimate partner and sexual violence survey (NISVS): 2010 summary report. Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention. Caetano, R., Field, C. A., Ramisetty-Mikler, S., & McGrath, C. (2005). The 5-year course of intimate partner violence among White, Black, and Hispanic couples in the United States. Journal of Interpersonal Violence, 20, 1039-1057. doi:10.1177/0886260505277783 Caetano, R., Ramisetty-Mikler, S., & Field, C. A. (2005). Unidirectional and bidirectional intimate partner violence among White, Black, and Hispanic couples in the United States. Violence and Victims, 20, 393-406. Caetano, R., Ramisetty-Mikler, S., Vaeth, P. A. C., & Harris, T. R. (2007). Acculturation stress, drinking, and intimate partner violence among Hispanic couples in the U.S. Journal of Interpersonal Violence, 22, 1431-1447. Capaldi, D. M., Knoble, N. B., Shortt, J. W., & Kim, H. K. (2012). A systematic review of risk factors for intimate partner violence. Partner Abuse, 3, 231-280.


19

Spencer et al.

Catalano, S. (2011). Intimate partner violence in the United States (12/12/2011). Retrieved from http://bjs.ojp.usdoj.gov/content/intimate/ipv.cfm#contents Champion, H., Wagoner, K., Song, E. Y., Brown, V. K., & Wolfson, M. (2008). Adolescent date fighting victimization and perpetration from a multi-community sample: Associations with substance use and other violent victimization and perpetration. International Journal of Adolescent Medicine and Health, 20, 419-429. Chen, P., & Chantala, K. (2014). Guidelines for analyzing Add Health data. Chapel Hill, NC: Carolina Population Center, University of North Carolina at Chapel Hill. Clark, C. J., Spencer, R. A., Everson-Rose, S. A., Brady, S. S., Mason, S. M., Connett, J. E., . . . Suglia, S. F. (2014). Dating violence, childhood maltreatment, and BMI from adolescence to young adulthood. Pediatrics, 134, 678-685. doi:10.1542/ peds.2014-1179 Coker, A., Davis, K., Arias, I., Desai, S., Sanderson, M., Brandt, H., & Smith, P. (2002). Physical and mental health effects of intimate partner violence for men and women. American Journal of Preventive Medicine, 23, 260-268. doi:10.1016/ S0749-3797(02)00514-7 Collins, L., & Lanza, S. (2010). Latent class and latent transition analysis: With applications in the social, behavioral, and health sciences. Hoboken, NJ: Wiley. Cutter-Wilson, E., & Richmond, T. (2011). Understanding teen dating violence: Practical screening and intervention strategies for pediatric and adolescent healthcare providers. Current Opinion in Pediatrics, 23, 379-383. doi:10.1097/ MOP.0b013e32834875d5 DeNavas-Walt, C., Proctor, B., & Smith, J. (2013). Income, poverty, and health insurance coverage in the United States: 2012—Current population reports. Washington, DC: U.S. Census Bureau, Government Printing Office. Draucker, C. B., Martsolf, D. S., Stephenson, P., Heckman, T., Ferguson, C. P., Perkins, S., & Cook, C. B. (2012). Types of aggressive relationships in adolescent dating violence. Journal of Aggression, Maltreatment & Trauma, 21, 516-539. Draucker, C. B., Martsolf, D., Stephenson, P., Risko, J., Heckman, T., Sheehan, D., . . . Ferguson, C. (2010). Aggressive events in adolescent dating violence. Issues in Mental Health Nursing, 31, 599-610. Fiebert, M. (2004). References examining assaults by women on their spouses or male partners: An annotated bibliography. Sexuality & Culture, 8, 140-177. Foshee, V. A., McNaughton Reyes, H. L., & Ennett, S. T. (2010). Examination of sex and race differences in longitudinal predictors of the initiation of adolescent dating violence perpetration. Journal of Aggression, Maltreatment & Trauma, 19, 492-516. doi:10.1080/10926771.2010.495032 Fox, G. L., & Benson, M. L. (2006). Household and neighborhood contexts of intimate partner violence. Public Health Reports, 121, 419-427. Fox, G. L., Benson, M. L., DeMaris, A. A., & Van Wyk, J. (2002). Economic distress and intimate violence: Testing family stress and resources theories. Journal of Marriage and Family, 64, 793-807. doi:10.1111/j.1741-3737.2002.00793.x


20


Giordano, P. C., Soto, D. A., Manning, W. D., & Longmore, M. A. (2010). The characteristics of romantic relationships associated with teen dating violence. Social Science Research, 39, 863-874. Glass, N., Perrin, N., Hanson, G., Mankowski, E., Bloom, T., & Campbell, J. (2009). Patterns of partners’ abusive behaviors as reported by Latina and non-Latina survivors. Journal of Community Psychology, 37, 156-170. doi:10.1002/jcop.20286 Gray, H. M., & Foshee, V. (1997). Adolescent dating violence: Differences between one-sided and mutually violent profiles. Journal of Interpersonal Violence, 12, 126-141. Hamby, S. (2009). The gender debate about intimate partner violence: Solutions and dead ends. Psychological Trauma: Theory, Research, Practice and Policy, 1, 24-34. Harris, K. M., Halpern, C. T., Whitsel, E., Hussey, J., Tabor, P., Entzel, P., & Udry, J. R. (2009). The National Longitudinal Study of Adolescent to Adult Health: Research Design. Retrieved from http://www.cpc.unc.edu/projects/addhealth/ design Haynie, D. L., Farhat, T., Brooks-Russell, A., Wang, J., Barbieri, B., & Iannotti, R. J. (2013). Dating violence perpetration and victimization among U.S. adolescents: Prevalence, patterns, and associations with health complaints and substance use. Journal of Adolescent Health, 53, 194-201. doi:10.1016/j.jadohealth.2013.02.008 Kann, L., Kinchen, S., Shanklin, S. L., Flint, K. H., Hawkins, J., Harris, W. A., . . . Zaza, S. (2014). Youth risk behavior surveillance—United States, 2013. MMWR Surveillance Summaries, 63(Suppl. 4), 1-168. Klevens, J. (2007). An overview of intimate partner violence among Latinos. Violence Against Women, 13, 111-122. Langhinrichsen-Rohling, J., Selwyn, C., & Rohling, M. L. (2012). Rates of bidirectional versus unidirectional intimate partner violence across samples, sexual orientations, and race/ethnicities: A comprehensive review. Partner Abuse, 3, 199-230. doi:10.1891/1946-6560.3.2.199 Love, S. R., & Richards, T. N. (2013). An exploratory investigation of adolescent intimate partner violence among African American youth: A gendered analysis. Journal of Interpersonal Violence, 28, 3342-3366. doi:10.1177/0886260513496898 Muthén, B. O. (2004). Latent variable analysis. Growth mixture modeling and related techniques for longitudinal data. In K. David. (Ed.), The SAGE handbook of quantitative methodology for the social sciences (pp. 345-370). Newbury Park, CA: SAGE. Muthén, B. O., & Muthén, L. K. (1998-2012). Mplus user’s guide (7th ed.). Los Angeles, CA: Author. Nurius, P. S., & Macy, R. J. (2008). Heterogeneity among violence-exposed women applying person-oriented research methods. Journal of Interpersonal Violence, 23, 389-415. Orpinas, P., Hsieh, H. L., Song, X., Holland, K., & Nahapetyan, L. (2013). Trajectories of physical dating violence from middle to high school: Association with relationship quality and acceptability of aggression. Journal of Youth and Adolescence, 42, 551-565. doi:10.1007/s10964-012-9881-5


21

Spencer et al.

Orpinas, P., Nahapetyan, L., Song, X., McNicholas, C., & Reeves, P. M. (2012). Psychological dating violence perpetration and victimization: Trajectories from middle to high school. Aggressive Behavior, 38, 510-520. doi:10.1002/ ab.21441 Pinchevsky, G. M., & Wright, E. M. (2012). The impact of neighborhoods on intimate partner violence and victimization. Trauma, Violence, & Abuse, 13, 112-132. doi:10.1177/1524838012445641 Renner, L. M., & Whitney, S. D. (2010). Examining symmetry in intimate partner violence among young adults using socio-demographic characteristics. Journal of Family Violence, 25, 91-106. Renner, L. M., Whitney, S. D., & Vasquez, M. (2015). Individual and interpersonal factors for physical intimate partner violence perpetration by sex and ethnicity. Violence and Victims, 30(1), 97-119. Rost, J. (2006). Latent-Class-Analyse [Latent class analysis]. In F. Petermann & M. Eid (Eds.), Handbuch de Psycologischen Diasnostik [Handbook of Psychological Assessment] (pp. 275-287). Gottingen, Germany: Hogrefe. Schwarz, G. (1978). Estimating the dimension of a model. The Annals of Statistics, 6, 461-464. doi:10.1214/aos/1176344136 Silverman, J. G., Raj, A., Mucci, L. A., & Hathaway, J. E. (2001). Dating violence against adolescent girls and associated substance use, unhealthy weight control, sexual risk behavior, pregnancy, and suicidality. Journal of the American Medical Association, 286, 572-579. Straus, M., Hamby, S. L., Boney-McCoy, S., & Sugarman, D. B. (1996). The Revised Conflict Tactics Scales (CTS2): Development and Preliminary Psychometric Data. Journal of Family Issues, 17, 283-316. doi:10.1177/019251396017003001 Swan, S. C., & Snow, D. L. (2002). A typology of women’s use of violence in intimate relationships. Violence Against Women, 8, 286-319. Swan, S. C., & Snow, D. L. (2006). The development of a theory of women’s use of violence in intimate relationships. Violence Against Women, 12, 1026-1045. doi:10.1177/1077801206293330 Swartout, K. M., & Swartout, A. G. (2012). Shifting perspectives: Applying personcentered analyses to violence research. Psychology of Violence, 2(4), 309-312. Vagi, K., Rothman, E., Latzman, N., Tharp, A., Hall, D., & Breiding, M. (2013). Beyond correlates: A review of risk and protective factors for adolescent dating violence perpetration. Journal of Youth and Adolescence, 42, 633-649. doi:10.1007/s10964-013-9907-7 Whitaker, D. J., Haileyesus, T., Swahn, M., & Saltzman, L. S. (2007). Differences in frequency of violence and reported injury between relationships with reciprocal and nonreciprocal intimate partner violence. American Journal of Public Health, 97, 941-947. doi:10.2105/AJPH.2005.079020 Yahner, J., Dank, M., Zweig, J. M., & Lachman, P. (2015). The Co-Occurrence of Physical and Cyber Dating Violence and Bullying Among Teens. Journal of Interpersonal Violence, 30(7), 1079-1089. doi:10.1177/0886260514540324


22


Author Biographies Rachael A. Spencer, MPA, is a gender-based violence specialist and research consultant. Her research interests include reproductive and physical health outcomes associated with partner violence. Recently, she has worked with the United Nations Development Programme–Iraq, the New York City Mayor’s Office to Combat Domestic Violence, and the University of Minnesota. Lynette M. Renner, MSW, PhD, is an associate professor at the University of Minnesota School of Social Work. Her research interests include mental health outcomes for children and women who experience family violence. Her recent projects focused on reporting patterns for perpetration and victimization and help-seeking strategies among women in rural areas. Cari Jo Clark, ScD, MPH, is an assistant professor at the University of Minnesota Medical School. Her research examines the linkages between violence and health, with a focus on the long-term health effects of exposure to intimate partner violence and the development of primary and secondary prevention interventions.
