Abstract
Empirical knowledge regarding risk factors for intimate partner violence (IPV) from multiple partners (MP) is scarce and sought by clinicians and many women themselves for the prevention of future intimate partner violence relationships (IPVRs). Quantitative data were obtained through a structured interview with a stratified sample of help-seeking women (N = 154) with no (n = 48, 0IPVR), one (n = 55, 1IPVR), or multiple (n = 51, 2IPVR) IPVRs. This study investigated the association between (a) childhood family violence, (b) other childhood adversities, (c) victimization and perpetration of IPV in the last (index) relationship, and (d) controlling sociodemographic and contextual variables, and the following dependent variables: (a) women with 1IPVR and 2IPVR compared with 0IPVR and (b) women with 1IPVR compared with 2IPVR. Multivariate logistic regression analyses indicated that, compared with nonvictimized women, IPV victimized women were nearly three times more likely to report childhood sexual abuse. They also reported a higher frequency of peer victimization and a higher likelihood of having an immigrant partner. In addition, the length of the index relationship was shorter for IPV victimized women. Compared with women with 1IPVR, women with IPV by MP were more likely to report childhood emotional abuse and less education, and they were less likely to be immigrants. The two groups of IPV victimized women were indistinguishable regarding characteristics of victimization and perpetration of IPV. This study indicated that there were other risk factors for IPV by MP than for IPV in general and highlighted the importance of addressing parenting and emotional care in IPV families.
Keywords
Introduction
Intimate partner violence (IPV) is a serious, heterogenic, and complex issue associated with significant health, social, and economic costs to individuals and society (Cattaneo & Goodman, 2005; Cornelius & Resseguie, 2007; Costa et al., 2015; Mears, 2003). IPV comprises physical and sexual violence, stalking, and psychological aggression (including coercive tactics) by a current or former intimate partner (Breiding, 2015). An intimate partner is a person with whom one has a close personal relationship characterized by emotional connectedness, regular contact, and ongoing physical and/or sexual contact and with whom one shares identity as a couple and familiarity with each other’s lives (Breiding, 2015). Worldwide, almost one third of women who have been in a relationship have experienced physical and/or sexual violence by their intimate partner according to the World Health Organization (WHO; 2013). The National Intimate Partner and Sexual Violence Survey (Breiding, 2015) reported the following lifetime IPV experiences among women in the United States: rape (8.8%), sexual violence other than rape (15.8%), severe physical violence (22.3%), and stalking (9.2%). Nearly half of women in the United States had experienced psychological aggression by an intimate partner (Black et al., 2011). Lifetime prevalence in Norway showed that 14.4% of women had been victimized by minor physical violence from partners and 8.2% by severe physical violence (Thoresen & Hjemdal, 2014). The mental health issues associated with IPV include depression, posttraumatic stress disorder (PTSD), anxiety, self-harm, and sleep disorders; the physical health issues include poor functional health, somatic disorders, chronic disorders and chronic pain, gynecological problems, and increased risk of sexually transmitted infections (Dillon et al., 2013). Sizable proportions of women who terminate their relationship with a violent partner become involved with a subsequent violent partner: Ranging from 22.9% (Vatnar & Bjørkly, 2008) to 56% (Alexander, 2009), women in IPV victimized female samples had prior histories of intimate partner violence relationships (IPVRs). Accordingly, it is urgent to investigate risk factors for IPV by multiple partners (MP). Empirical knowledge concerning risk factors for IPV by MP, in particular, is sought by clinicians and many women themselves for the prevention of future IPVRs. Perpetrators must be held accountable for their violence, but focusing only on the perpetrator may distract attention from a possible vulnerability in some women for being revictimized. It would be troubling to overlook the experiences of a significant percentage of victimized women who remain vulnerable to violence even after they have succeeded in leaving one violent partner (Alexander, 2009).
Risk Factors for Revictimization of IPV
The reported childhood risk factors for IPV revictimization are, in general, psychological (Cascio et al., 2017), sexual, and physical abuse (Barrios et al., 2015; Cascio et al., 2017; Coid et al., 2001; Stroem et al., 2019) and exposure to parental abuse (Ehrensaft et al., 2003; Krishnan et al., 2001; Trickett et al., 2011). The reported violence-related risk factor for IPV revictimization of a woman is that she initiates violence toward her male partner (Dixon & Graham-Kevan, 2011; Kuijpers et al., 2012a, 2012b, 2012c). Some resistance and coping strategies have been reported as risk factors for IPV revictimization (e.g., fighting back, mutual violence, manipulation, anger, intimacy, compliance, and refusing to do what he says; Goodman et al., 2005; Iverson et al., 2013).
Risk Factors for Revictimization of IPV by MP
The vast majority of studies on IPV revictimization do not distinguish between revictimization by the same partner and revictimization by MP, which implies mixed and inaccurate results. Risk factors related to IPV by MP may not be the same as the risk factors related to recurrent violence within a cycle of a single violent relationship. A systematic literature review regarding revictimization of IPV by MP in particular indicated that IPV by MP was significantly associated with childhood domestic trauma, drug abuse, characteristics of the partner violence, and attachment style. Regarding PTSD and personality disorders, the results were mixed and inconclusive, and depression did not appear as a salient risk factor for IPV by MP (Ørke et al., 2018). With only seven published studies, empirical research on risk for IPV revictimization by MP appeared to be scarce and had limited recent development, and the wide diversity in study designs, measurements, definitions, and variables in these studies precluded drawing firm conclusions about risk factors (Ørke et al., 2018). The review findings indicated that the vulnerability of women subjected to IPV by MP was associated with a history of childhood sexual abuse (Alexander, 2009; Cole et al., 2008; Stein et al., 2016; Vatnar & Bjørkly, 2008), in addition to emotional abuse (Cole et al., 2008), childhood domestic physical violence (Vatnar & Bjørkly, 2008), witnessing domestic IPV (Alexander, 2009; Vatnar & Bjørkly, 2008), and a history of torture and being held hostage (Stein et al., 2016). There was no evidence for one specific personality disorder typical of women at risk for IPV by MP (Ørke et al., 2018). Regarding characteristics of the partner violence, greater IPV severity (Bogat et al., 2003) and current psychological violence were associated with additional violent partners, and women with exposure to current sexual violence had a lower number of violent partners (Stein et al., 2016). Victims of IPV by MP scored higher on PTSD (Bogat et al., 2003). Long involvement with a new partner increased the likelihood of IPV revictimization by the new partner (Cole et al., 2008). Where age difference had been found between the two groups, women with IPV by MP were significantly younger than women with a single IPVR (Alexander, 2009; Testa et al., 2003). One American study reported that African American and White women had significantly more violent partners than their Latina counterparts (Stein et al., 2016).
Judging from a systematic review regarding revictimization of IPV by MP (Ørke et al., 2018), the use of a standard case definition would be one key factor needed to ensure that information is collected in a systematic fashion (German et al., 2001). Future studies could possibly benefit from more refined language that provided greater specificity in the labeling of some of the trauma history items (Stein et al., 2016). The risk of measurement bias increases when scales have been developed by authors but have not been adequately validated before investigations take place (Fellmeth et al., 2013). In this study, the abovementioned shortcomings in the existing literature were addressed.
Aims of the Study
The aims of the study were to investigate the association between women with no (0IPVR), one (1IPVR), or multiple (2IPVR) IPVRs and (a) childhood family violence and other childhood adversities and (b) IPV in the last (index) relationship. The analyses were adjusted for sociodemographic and contextual group differences. Research questions were as follows:
Method
Design and Settings
This study was part of a cross-sectional case–control study with two groups of help-seeking, IPV victimized women and a control group of help-seeking women not IPV victimized. To attain statistical power to compare subgroups, we conducted power analyses prior to the initiation of the project. One goal of the proposed study was to test the null hypothesis that the event rate is identical in the three groups (1IPVR, 2IPVR, or 0IPVR). The odds ratio (OR) for any comparison was 1.0, the log OR (β) was 0.0, and the relative risk was 1.0. Estimates for the alternate hypothesis were based on the following event rates: multiple-partner IPV = 0.50, one-partner IPV = 0.40, and no-partner IPV = 0.15. The study included a total of 120 subjects with 40 persons in each group. The alpha value was set at .05 (two tailed). For this distribution, the effect sizes were 0.50, 0.40, and 0.15, the sample size was 120, the alpha value was .05 (two tailed), and the power was 0.83. This means that 83% of studies would be expected to yield a significant effect, rejecting the null hypothesis that the OR is 1.0.
The researchers cooperated with leaders of the nationwide agencies of women’s shelters, the Alternative to Violence (ATV) treatment agency, the police, and family counseling agencies in Norway to recruit participants for this study. These agencies were asked to invite all clients who met the inclusion criteria to participate in the study. The recruitment steps were as follows: (a) The initial recruitment of participants was conducted by agency personnel by presenting the study’s information consent letter, either in person or by phone, to all their female users who met the inclusion criteria; (b) after receiving written consent and contact information, the researcher sought contact with those recruited to discuss aspects of their participation in the study; and (c) the participating women came to a face-to-face interview with the same researcher, a female clinical psychologist, at the local recruitment or researcher’s office. To address diversity concerns, women who were not fluent in the Norwegian language were informed that a professional interpreter could be hired for the interview. Women were included regardless of the sex of their partner. There was no economic incentive for participation, but a refund for public transport was offered. All participants were given a sheet with the answer alternatives for the questions. Timed breaks were used when needed. The researcher registered the answers by hand in the codebook. The interviews lasted approximately 2 hr.
Inclusion and Exclusion Criteria
The inclusion criteria were as follows: The participant was at least 18 years old; had made contact with police, family counseling, women’s shelter, or the ATV for intimate partner–related problems; was in or had lately been in an intimate relationship that lasted at least 6 months; held either Norwegian citizenship or a residence permit; and had sufficient language fluency to understand the information letter and to make an appointment on the phone. The exclusion criterion was that the most recent IPVR ended more than 3 years ago.
Specific inclusion criteria for distinguishing the research categories were (a) women who had only experienced violence from one intimate partner within the last 3 years; (b) women who had experienced violence from an intimate partner within the last 3 years and in at least one previous intimate relationship; and (c) women who had currently or lately had an intimate relationship but had never been victims of IPV (control group).
All participants in the control group were recruited from family counseling offices. They shared with the study groups the characteristics of being adult women experienced with a recent intimate relationship and seeking help for intimate partner–related problems.
Dependent Variables
The women were recruited to the designated research category according to the definition of physical, psychological, and sexual violence in the information letter (Breiding, 2015). They were asked (both on the phone and, initially, in the interview) in how many adult intimate relationships they had experienced violence victimization. According to their self-reports, they were included in one of the following three research categories: 0IPVR, 1IPVR, or 2IPVR.
Procedures
Twenty-three local offices in rural as well as urban areas across Norway recruited participants for the study. Figure 1 depicts the recruitment outcomes. The total sample (N = 154) consisted of 36.4% (n = 56) recruited from family counseling offices, 35.1% (n = 54) from shelters, 24% (n = 37) from ATV, and 4.5% (n = 7) participants from the police. In five interviews (3.2%), a professional interpreter was hired. The 154 interviews were carried out between March 2018 and January 2019.

Recruitment process.
Subjects/Sample
The 154 women were between the ages of 20 and 69 (M = 39.85, SD = 10.28) and had a history of 1 to 13 intimate relationships (M = 2.97, SD = 1.824) (Table 1). There were women with no IPVR (31.2%, 48), women with one IPVR (35.7%, 55), and women with multiple IPVRs (33.1%, 51). Among the women with multiple IPVRs, the range was from two (62.7%, 32), three (23.5%, 12), four (7.8%, 4), and five (3.9%, 2) to six IPVRs (2%, 1). Most of the women (85.7%, 132) regarded themselves as native Norwegians, 4.5% (7) were immigrants with Norwegian citizenship, and 9.7% (15) were immigrants without Norwegian citizenship. Most women were mothers (90.3%, 139), and they had between one and six children (M = 2.29, SD = 1.030). Mean years of completed education was 14.96 years, ranging from 7 to 24 years (SD = 3.282). Significant sociodemographic and contextual group differences for 0IPVR, 1IPVR, and 2IPVR are listed in Table 1.
Sociodemographic and Contextual Group Differences Among Women With No (0IPVR), One (1IPVR) and Multiple IPVRs (2IPVR).
Note. The Kruskal–Wallis test was used to test for possible independent group differences for variables with nonparametric score distributions for more than two independent groups, and the Mann–Whitney U test for two independent groups. The Pearson chi-square test was used for nominal data and unrelated groups. Age, age of partner, age at the initiation of first intimate relationship, and time lapse since the last relationship were tested with nonsignificant results. IPVRs = intimate partner violence relationships.
Measures
Favoring validity, reliability, and results that can be compared with other international studies, we applied the following validated questionnaires. A modified version of UngVold2015 (Mossige & Stefansen, 2016) and three parts of the Childhood Trauma Questionnaire (CTQ-SF; Bernstein et al., 2003; Dovran et al., 2013) were used to measure childhood family violence and other childhood adversities.
The Revised Conflict Tactics Scale (CTS-2; Straus et al., 1996), the Psychological Maltreatment of Women Inventory (PMWI; Alsaker et al., 2011; Tolman, 1999), and the Spousal Assault Risk Assessment Guide, Version 3 (SARA-V3; Kropp & Hart, 2015) were used to measure victimization and perpetration of IPV and characteristics of IPV. With the PMWI, we added a list of mirrored questions to get a picture of violence inflicted by both parties throughout the relationship. All questionnaires had an authorized Norwegian version.
Demographic and contextual variables were drawn from UngVold2015 (Mossige & Stefansen, 2016) and Vatnar et al.’s study (Vatnar et al., 2017b) and are presented in Table 1.
In addition to the questionnaires, the following single items were developed especially for this study: “Do you have a person you can confide in?” (Yes/No); “Were there language misunderstandings during the interview?” (No/slight or considerable). The interview was prepared as a codebook with a structured assembly of the instructions and questionnaires.
Statistical Analyses
Univariate and bivariate analyses were conducted to compare the subgroups—(a) women with 1IPVR and 2IPVR (victimized) compared with those with 0IPVR (nonvictimized) and (b) women with 1IPVR compared with those with 2IPVR—and to inform the selection of variables to be included in the multivariate analysis. Multivariate logistic regression analyses were used to examine risk and protective factors associated with 1IPVR and 2IPVR. The stepwise options recommended for logistic regression for small samples were used (Altman, 1991; Pallant, 2010). In Step 1, as suggested by Altman and Pallant, initial comparisons of the two IPV groups were carried out by simple descriptive cross-tabulations with Pearson chi-square for categorical and nominal variables. For continuous variables, we used the Mann–Whitney U test, a nonparametric test for independent samples (Step 1, Tables 1 and 2). In the first multivariate logistic regression analyses (Step 2), variables with significant (p ≤ .05) or trend (p ≤ .10) group differences in bivariate analyses were adjusted for other significant differences within the same category. The following categories were analyzed in Step 2: (a) victimization and perpetration of IPV in the index relationship, (b) childhood family violence, (c) other childhood adversities, and (d) sociodemographic and contextual variables. Significant differences remaining after each of the four comparisons in Step 2 were forwarded to Step 3 (Tables 3 and 4) where we adjusted for all remaining group differences in Categories a, b, c, and d. Suitability for multivariate logistic regression analysis was investigated by the Hosmer–Lemeshow test. Cox and Snell R squared and Nagelkerke R squared were used to estimate the proportion of explained variance in the multivariate models that were tested. Values were estimated as model fit indices for the regression models (see Notes in Tables 3 and 4). Statistical analyses were performed using the statistical program package SPSS, version 25. A conventional p value of <.05 was used.
Prevalence of Childhood Family Violence and Other Childhood Adversities Among Women With No (0IPVR), One (1IPVR) and Multiple IPVRs (2IPVR).
Note. The Kruskal–Wallis test was used to test for possible independent group differences for variables with nonparametric score distributions for more than two independent groups, and the Mann–Whitney U test for two independent groups. The Pearson chi-square test was used for nominal data and unrelated groups. The following prevalence variables were tested with no significant results: mother victimized by IPV, father victimized by IPV, mother exposed to psychological and sexual violence from partner, father exposed to psychological, sexual, and other violence from partner or injured at home, and psychological violence from mother. IPVRs = intimate partner violence relationships; IPV = intimate partner violence.
Multivariate Logistic Regression Analyses: Victimized (n = 105) Compared With Nonvictimized Women (Baseline) (n = 48).
Note. The results were from multivariate binary logistic regression, forward stepwise (Wald). Model 1 without length of the last relationship: Cox & Snell R2 = .205, Nagelkerke R2 = .289, Hosmer and Lemeshow test = .937. Model 2 without sexual abuse: Cox & Snell R2 = .218, Nagelkerke R2 = .306, Hosmer and Lemeshow test = .216. All women who reported forced penetration in childhood (17%, 26) were victimized by IPV in adulthood. This variable was not entered in logistic regression analysis. CI = confidence interval; IPV = intimate partner violence; n.s. = nonsignificant; prev. = prevalence, freq. = frequency.
Multivariate Logistic Regression Analyses (n = 105): Women With Multiple IPVRs (n = 50) Compared With Women With One IPVR (Baseline) (n = 55).
Note. The results were from multivariate binary logistic regression, forward stepwise (Wald). Cox & Snell R2 = .246, Nagelkerke R2 = .328, Hosmer and Lemeshow test = .324. IPVR = intimate partner violence relationship; IPV = intimate partner violence; CI = confidence interval; prev. = prevalence; n.s. = nonsignificant; freq. = frequency.
Ethical Aspects
The study was approved by the Regional Norwegian Ethics Committee. All ethical and safety recommendations from the WHO (2001) were observed. An information letter informed the participants about the study objectives and that some questions were of an intimate nature. They were assured that their participation was voluntary, that they were free to withdraw from the study at any time, that withdrawal would not affect the services they received at the recruitment office, that information would be stored confidentially, and that they were welcome to call the researcher on a given phone number. All cases were included irrespective of socioeconomic status, race, ethnicity, language, nationality, sex, gender identity, sexual orientation, religion, geography, ability, and age.
Results
Prevalence and Characteristics Among Women With 0IPVR, 1IPVR, and 2IPVR
In general, 82.2% of all participants confirmed one or more incidents of mild or severe childhood family violence. Prevalence of mother victimized by IPV was confirmed by 46.1% of the women, and that of father victimized by IPV was confirmed by 27.3%. There was no significant group difference concerning these variables. However, several bivariate group differences regarding the prevalence and frequency of childhood family violence and other childhood adversities were found. Significant prevalence variables are presented in Table 2.
Regarding the characteristics of the IPV victimization, there were no significant group differences between women with 1IPVR and 2IPVR. Women with 1IPVR or 2IPVR reported victimization of threats (84.3%), physical violence (89.1%), sexual violence (52%), severe partner violence (68.3%), violence persistency (95.1%), violence escalation (83.3%), and violence-related restriction disobedience (31.6%). The most prevalent (100%) was minor and severe psychological violence. The victimized women (1IPVR and 2IPVR) reported minor physical IPV (84.9%) and severe physical IPV (68.9%). More than half (57.5%) of the women had experienced minor and 43.4% had experienced severe sexual coercion. Women in both groups (1IPVR and 2IPVR) had experienced minor (54.7%) or severe (35.8%) injury from partners. Victimization by emotional/verbal maltreatment was reported by 98.1% of the women, and dominance/isolation by 83.7% of the women.
Regarding the women’s perpetration of violence, two trends for group differences were registered: prevalence of minor physical IPV from her (40% 1IPVR and 56.9% 2IPVR, p = .083) and total number of minor physical IPV from her (mean rank 1IPVR 49.08, 2IPVR 58.26, p = .097). The victimized women reported perpetration of minor (84%) and severe psychological IPV (48.1%), severe physical assault (18.9%), minor injury (16%), severe injury (3.8%), minor or severe sexual coercion (1.9%), emotional/verbal maltreatment (17.9%), and dominance/isolation (5.8%).
Women With IPVRs Compared With Women With No IPV
The univariate analyses gave 10 variables with significant group differences or trends among women with IPVRs compared with women with no IPV regarding childhood family violence. The 10 variables were as follows: prevalence of mother victimized by other violence from partner and father victimized by physical violence from partner; frequency of mother victimized by other violence from partner; prevalence of mild and severe physical violence from mother and psychological violence from father; and frequency of psychological and mild and severe physical violence from mother and psychological violence from father. In addition, univariate analyses gave six variables with significant or trend group differences regarding other childhood adversities (physical neglect, emotional neglect, and emotional abuse; prevalence of sexual abuse and forced penetration; frequency of peer victimization; the prevalence variables with significant group differences are displayed in Table 2). Also, seven sociodemographic and contextual variables (Table 1) showed significant or trend group differences in the initial analyses. In Step 2, 11 of the differences from the initial analyses remained significant or with a trend. After controlling for all remaining group differences, the following differences between women from IPVRs and women with no IPVR remained significant in the multivariate logistic regression models (Table 3). IPV victimized women were nearly three times more likely to report childhood sexual abuse. They also reported a higher frequency of peer victimization and a higher likelihood of having an immigrant partner (Model 1). In addition, the length of the index relationship was shorter for IPV victimized women (Model 2). All women who reported forced penetration (17%, n = 26) as a part of childhood sexual abuse were victimized by IPV in adulthood. Accordingly, this variable of severe childhood sexual abuse was not suitable for the multivariate logistic regression model. To attain goodness of fit, the variables were analyzed in two models (see Note in Table 3).
Women With 2IPVR Compared With Women With 1IPVR
The univariate analyses indicated two significant group differences or trends between 1IPVR and 2IPVR regarding IPV (prevalence of minor physical assault from her and total number of minor physical assaults from her). Initially, there were 12 univariate differences regarding childhood family violence (prevalence of mother victimized by physical IPV, prevalence of father victimized by physical IPV, mother injured at home; frequency of mother victimized by physical IPV, frequency of father victimized by physical IPV; prevalence of mild physical violence from mother, psychological violence from father, mild physical violence from father, and severe physical violence from father; frequency of mild physical violence from mother, psychological violence from father, and mild physical violence from father). There were seven univariate differences regarding other childhood adversities (frequency of physical neglect, emotional neglect and emotional abuse; parent’s alcohol/drug use; forced penetration; prevalence and frequency of peer victimization) and five regarding sociodemographic and contextual variables (see Table 1 and 2). In Step 2, eight of these univariate differences remained significant or showed trends when adjusted for the other included variables within the same aim categories. In Step 3, after controlling for all remaining group differences in a multivariate logistic regression model, the following three variables remained explanatory for the two groups (Table 4): Women with IPV by MP were more likely to report childhood emotional abuse (frequency), had less education, and were less likely to be immigrants.
Discussion
Main Findings
Compared with nonvictimized, IPV victimized women were nearly three times more likely to report childhood sexual abuse, reported a higher frequency of peer victimization, and had a greater likelihood of having an immigrant partner. In contrast, compared with women victimized by one partner, women with IPV by MP, in particular, had an increased likelihood of reporting childhood emotional abuse and of having less education, and they were less likely to be immigrants. Childhood family violence and characteristics of the IPV did not remain as risk factors for IPV by MP.
Childhood Emotional Abuse as a Risk Factor for IPV by MP
Cole et al. (2008) initially reported a significantly higher prevalence of childhood emotional abuse among women with IPV by MP, but this result was not sustained in multivariate analysis. Some related experiences like subjection or exposure to a parent’s physical IPV (Alexander, 2009; Vatnar & Bjørkly, 2008), multiple forms of childhood trauma (Alexander, 2009), and psychological childhood victimization (Vatnar & Bjørkly, 2008) have been reported to be associated with IPV by MP. This study included both witnessing and victimization of childhood physical and psychological violence and childhood emotional abuse concurrently. This allowed for an improved understanding of explicatory factors. Although the measures of psychological violence pertained to acts, the measures of childhood emotional abuse were broader and could perhaps be understood as describing the family atmosphere. Definitions of violence include behavior, intentions, and consequences (Kropp & Hart, 2015), whereas childhood emotional abuse also encompasses humiliating or hostile behavior not included in definitions of violence.
Childhood Sexual Abuse as a Risk Factor for IPV by MP
The results of multivariate logistic regression reinforced the body of literature pointing to childhood sexual abuse as a risk factor for IPV victimization in general (e.g., Barrios et al., 2015; Coid et al., 2001; Stroem et al., 2019; Vézina & Hébert, 2007; Whitfield et al., 2003). This was also supported by the descriptive finding that every woman who had been victimized by forced penetration in childhood was later revictimized by IPV. However, in contrast to earlier findings (Alexander, 2009; Cole et al., 2008; Stein et al., 2016; Vatnar & Bjørkly, 2008), our study did not point to childhood sexual abuse as a risk factor for IPV by MP in particular.
Immigration as a Risk Factor for IPV by MP
Our study indicated that women with an immigrant partner, regardless of the partner’s country of origin, ran a greater risk for IPV in general, but not for MP. An American study (Gupta et al., 2010) found that, among immigrant men, those who were nonrecent immigrants and reported limited English-speaking ability were at the highest risk for IPV perpetration, compared with recent immigrants with high English-speaking ability. Studies have shown that immigrants were proportionally overrepresented in intimate partner homicide (IPH) statistics as both victims and perpetrators (see, for example, Campbell et al., 2007; Dobash & Dobash, 2015; Garcia & Hurwitz, 2007; Vatnar et al., 2017a), and it has been reported that the overrepresentation was attributed to social and economic disadvantage (particularly unemployment), rather than immigration and ethnicity per se (Vatnar et al., 2017a). Such information about the partner was not controlled for in this study.
Regarding the immigration status of the women themselves, on the other hand, our study showed an opposite pattern. Women with IPV by MP were less likely to be immigrants. An American study reported that African American and White women had significantly more violent partners than their Latina counterparts, but these results only specified ethnicity and not immigrant status (Stein et al., 2016). Another study (Alexander, 2009) failed to find any association between ethnicity and IPV by MP. In sum, this issue is not settled yet and more research is needed.
Sociodemographic Risk Factors for IPV by MP
In contrast to two studies that found no association (Alexander, 2009; Stein et al., 2016), this study found that women with IPV by MP had less education than women with 1IPVR. An association between economic hardship and economic dependency on a romantic partner and IPV exposure has been reported (Golden et al., 2013). It has been reported that actions by violent partners negatively affected the women’s ability to be and stay employed (Alsaker et al., 2014). Less education may be a risk factor or a consequence of violence in multiple relationships, and our results do not show the direction of that association. There might even be an interactional association.
Victimization and Perpetration of IPV as Risk Factors for IPV by MP
Based on earlier studies, we would expect to find a higher IPV severity score (Bogat et al., 2003), less exposure to sexual IPV, and more exposure to psychological IPV (Stein et al., 2016) among women with IPV by MP. However, we found no significant group differences regarding any kind of IPV victimization in this study.
Regarding perpetration, men with a violent family background have been found to be more violent as adults (Askeland et al., 2011). It has been claimed that violence between partners is likely to have a common etiology across genders and that individuals prone to acting-out behavior had been more exposed to violence than others (Pape, 2011). Our initial findings did indicate that women with IPV by MP reported more childhood family violence. Thus, one might speculate that these women would exhibit an acting-out pattern similar to men and be at greater risk for perpetrating retaliation and escalating violence. However, our results indicated that the women with IPV by MP were not more likely to perpetrate IPV than were women with 1IPV. A gender difference has been reported from a birth cohort study where childhood maltreatment was associated with an increased risk of later delinquency for young adult males, but not for females (Abajobir et al., 2017). We have not discovered any previous study that included female perpetration of violence in the comparison of 1IPVR and 2IPVR. Earlier research has indicated that victim-perpetrated IPV was a risk factor for being revictimized (Kuijpers et al., 2012b). However, because the Kuijpers study did not differentiate between revictimization of IPV within and across relationships, it is likely that their investigation may have included recurrent violence within a cycle of a single violent relationship. In sum, this issue is not settled yet and more research is needed.
Peer Victimization as a Risk Factor for IPV
Although consequences of childhood sexual abuse have been studied extensively, concurrent childhood emotional abuse and peer victimization have received less attention in IPV risk studies. An association has been reported between the perpetration of both bullying and IPV (Corvo & Delara, 2010), but to our knowledge our study is the first to report a link between peer victimization and IPV victimization.
Limitations
Due to cultural and social differences, the findings from our sample of IPV help-seeking women do not necessarily generalize to a general community sample nor to women outside of Norway. This calls for careful interpretation of the generalizability of our findings. As in several studies of IPV help-seeking women, a significant number of the women who were invited to participate declined to do so. We have no information about these women concerning group differences. However, all the women invited were help seeking at the time and in stressful and demanding life situations that may have made it difficult for them to participate in a research project.
This investigation was based on retrospective self-reports of IPV experiences. In other studies, recall bias has been associated with underreporting of IPV (Schwartz, 2005). Because the reports of IPV vary in terms of recency, they may also vary in salience (Bogat et al., 2003). It has been claimed that retrospective questions about childhood events cannot differentiate between an actual history of severe abuse and selective recollection (Dovran et al., 2013). It is possible that more severe events were recalled with greater frequency than less severe events. Costa et al. (2015) stated that recall of childhood experiences results in a substantial rate of false negatives, measurement error, and bias that could elevate Type I errors, with higher well-being linked to retrospective forgetting and lower well-being tied to greater retrospective reporting.
There are methodological limitations to interviewing only the victims, risking over- and underreporting according to social desirability. However, the purpose of this study was to understand the experiences of the victimized and not only to capture the objective “fact.” One researcher conducted all the interviews which may have increased the risk of systematic measurement error. However, reliability was strengthened by having only one interviewer, thereby avoiding low interrater reliability. The structured interview with behavior-specific questions and fixed-response options modified this risk. When investigating private and potentially traumatizing matters, one may anticipate underreporting. We consider this adjusted for by introducing the objectives of the study in an information letter, by securing confidentiality, and by pacing the interview according to the needs of the participant. One review reported that higher disclosure rates were found in studies using in-person interviews conducted by a “skilled and trained” clinician and in studies that included specific questions about the different types of IPV (Taillieu & Brownridge, 2010).
The OR of immigrant partner was high for IPV. Still, the wide confidence intervals indicate that this finding should be interpreted with caution.
Finally, the cross-sectional design has limitations concerning the measurement of the causality and temporal ordering between variables.
Conclusion
In light of the paucity of empirical studies of IPV by MP and the significant diversity in study designs, measurements, definitions, and variables, this study explored previous findings and related themes with standardized measures that can be replicated. Childhood physical domestic violence has been hypothesized to put women at risk for IPV by MP. On the contrary, one main finding in the multivariate logistic regression analysis was that childhood emotional abuse was the major risk factor for IPV by MP in particular. However, this study supported childhood sexual abuse as a predictive factor for IPV in general. As well, to our knowledge, peer victimization as a predictive factor for IPV had not been described before. Our research extended our knowledge on immigration, indicating that women with IPV had a greater likelihood of having an immigrant partner, whereas women with IPV by MP were less likely to be immigrants.
Clinical and Policy Implications
Risk of revictimization by future partners should be discussed with all women who have been in previous violent relationships (Cole et al., 2008). As part of risk assessment for IPV among adults, screening for a history of childhood emotional abuse, childhood sexual abuse, and peer victimization is needed. Health professionals are urged to recognize that women with a history of childhood emotional abuse need special attention. Our study supported the high importance of addressing parenting beyond the termination of physical violence in IPV families. Having less education needs attention among victimized women as it constitutes a risk of IPV by MP.
Research Implications
First, future studies would benefit from a prospective, longitudinal design to allow for the exploration of characteristics that predict and prevent women’s involvement with multiple violent partners. Second, childhood emotional abuse should be assessed when exploring the characteristics of women associated with IPV by MP. Third, research is needed to understand the etiology of the connection between childhood emotional abuse and risk for multiple violent relationships. Fourth, further research is needed to explore sociodemographic factors such as how being an immigrant may be associated with reduced risk of IPV by MP and the impact of education for IPV by MP. Finally, additional research is needed with more diverse samples and in different contexts.
Footnotes
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) received no financial support for the research, authorship, and/or publication of this article.
