Abstract
In recent years, a majority of young adults experience cohabitation. Nevertheless, cohabitation is a risk factor for intimate partner violence (IPV). Drawing on social exchange and commitment theory, we analyzed young adults’ IPV experiences using the recently collected (2011-2012) Toledo Adolescent Relationships Study data (n = 926). We found that sociodemographic characteristics, relationship commitment, quality, and constraints as well as prior experience with violence (in prior relationships and family of origin) were associated with IPV, but did not explain the association between cohabitation and IPV. We examined variation among individuals in cohabiting relationships to determine which cohabitors face the greatest risk of IPV. Serial cohabitors along with cohabitors who experienced both low commitment and high relational constraints experienced the greatest risk of IPV. These findings provided insights into the implications of cohabitation for the well-being of young adults.
Increasingly in the United States, cohabitation has become ubiquitous, as evidenced by over half of young adults reporting that they had cohabited (Manning, 2013). Although many cohabiting relationships are short term, lasting on average less than 2 years (Kennedy & Bumpass, 2011), researchers have documented that young adults in cohabiting, compared with dating or marital, relationships reported higher levels of intimate partner violence (IPV; e.g., Brown & Bulanda, 2008; Kenney & McLanahan, 2006; Magdol, Moffitt, Caspi, & Silva, 1998; Rhoades, Stanley, & Markman, 2012; Stets & Straus, 1990). Although previous work has examined differences in IPV across types of relationships, studies have not considered variation in reports of IPV among individuals in cohabiting relationships.
Drawing on the Toledo Adolescent Relationships Study (TARS), we used recently collected data from young men and women to examine factors associated with IPV in married, cohabiting, and dating relationships, with an eye toward understanding the greater levels of violence reported by individuals in cohabiting relationships. As such, this study moves beyond prior work in three key ways. First, we used recently collected data, which is important given the current prevalence of and continual increases in rates of cohabitation. The growth in cohabitation requires the use of recent data to reflect the contemporary context in which young adults make decisions about union formation. Second, we incorporated a wide range of theoretically relevant indicators of relationship quality, commitment, and constraints as well as socioeconomic background to explore potential factors associated with IPV. Third, we examined variation among individuals in cohabiting relationships to determine who faces the greatest risk of IPV. As cohabitation has become a majority experience, it is important to assess the implications of cohabitation for young adults’ well-being including experiences with IPV.
Background
A key developmental task following adolescence is learning to interact in healthy ways with intimate partners (Arnett, 2004; Clydesdale, 2007; Simon & Barrett, 2010). Yet compared with other stages in the life course, young adults have the highest risk for IPV (Halpern, Spriggs, Martin, & Kupper, 2009; W. L. Johnson, Giordano, Manning, & Longmore, 2015; Scott, Steward-Streng, Manlove, Schelar, & Cui, 2011). Nearly half (47%) of young men and women in the National Longitudinal Study of Adolescent to Adult Health (Add Health) reported IPV by young adulthood (Renner & Whitney, 2012). Research on IPV in adulthood often has focused on the experiences of married women with less attention to those relationships that exist outside the marital context (dating and cohabitation). Although virtually all young adults have nonmarital intimate relationships, we know surprisingly little about how these relationships are associated with IPV experiences. In recent decades, the levels of cohabitation have grown tremendously; in the late 1980s, 40% of women in their early 30s had cohabited, and in the late 2000s three quarters had done so (Manning, 2013). Thus, an analysis of IPV during young adult years requires attention to cohabiting, in addition to marital and dating relationships. Furthermore, given the increased levels of cohabitation, research must be based on contemporary cohorts of young adults.
Much prior research on cohabitation and IPV contrasts the experiences of married and cohabiting women. Researchers have consistently found higher rates of violence in cohabiting compared with marital relationships (e.g., Brown & Bulanda, 2008; Brownridge, 2008; Herrera, Wiersma, & Cleveland, 2008; Jackson, 1996; Magdol et al., 1998; Stets, 1991; Stets & Straus, 1989, 1990). Some of these studies, however, are based on cohabitation experiences that occurred over two decades ago (e.g., Jackson, 1996; Magdol et al., 1998; Stets, 1991; Stets & Straus, 1989, 1990), and do not reflect the current context in which the majority of young adults have cohabited. Similarly, much recent research that has relied on the third wave (2001-2002) of the Add Health, has found that cohabitation is associated with higher rates of IPV (e.g., Brown & Bulanda, 2008; Herrera et al., 2008; Renner & Whitney, 2012). To the extent that cohabitation is a more common experience in 2012 compared with 2001, and necessarily less selective, raises questions about whether individuals in current cohabiting unions continue to report higher odds of IPV compared with their married or dating counterparts. Brownridge (2008), in an analysis of Canadian couples argued that as cohabitation becomes more popular, fewer union status differentials should exist between cohabitation and marriage. Indeed, he reported a decline in the differentials in IPV experienced by Canadian cohabiting and married women.
By extension, as more young adults experience cohabitation in the United States, it is important to assess whether cohabitation is related to IPV in contrast to married young adults as well as individuals in dating relationships. Some prior studies have relied on college students, which represent a select sample of young adults who are most advantaged (Daley & Noland, 2001; Orcutt, Garcia, & Pickett, 2005; Smith, Thompson, Tomaka, & Buchanan, 2005; Stets & Straus, 1990). Researchers analyzing nationally representative data have found that cohabitors more often than daters report partner violence and that these differentials exist when considering either perpetration or victimization (e.g., Brown & Bulanda, 2008; Melander, Noel, & Tyler, 2010; Renner & Whitney, 2012; Rhoades et al., 2012; Scott et al., 2011). Some researchers explain the differential in IPV among dating and cohabiting couples with the inclusion of sociodemographic characteristics (e.g., Rhoades et al., 2012), while others cannot explain the gap in partner violence by union status (dating, cohabiting, marital) with extensive sets of control variables for family background, socioeconomic status, presence of children, and relationship characteristics (love, satisfaction, and duration; e.g., Brown & Bulanda, 2008; Melander et al., 2010). Thus, there is consensus that initial differences in IPV experiences exist with higher levels among cohabitors compared with daters, and some debate exists regarding explanatory variables.
Why Do Cohabitors Have Greater IPV?
Researchers have documented differentials in IPV according to union status, but have not directly explored why cohabiting, compared with dating and married, couples have higher incidences of IPV. One explanation is that the social rules governing cohabitation are not institutionalized (Nock, 1995), and because cohabitation is not a legally recognized union, individuals have to discover, define, and navigate their relationship roles without the benefit of social norms and institutionalized role expectations (such as those associated with the role of husband and wife). However, based on diffusion theory (Liefbroer & Dourleijn, 2006), as cohabitation becomes more common in a society, it will be less selective with greater acceptance and as a result will less often be associated with negative relational outcomes. Thus, our focus on a contemporary cohort of young adults is important.
A second reason that cohabitors experience higher risks of IPV than married or dating couples is grounded in commitment theory, cohabiting relationships more often contain relational precursors to IPV than dating or married relationships. Commitment theory as developed by Stanley and Markman (1992) focuses on two core features of commitment: interpersonal commitment and constraints. Interpersonal commitment or dedication represents the desire to be with a partner and have a long-term view of the relationship. We extend beyond commitment theory to include relationship quality measures that are consistent with prior work on IPV (Brown & Bulanda, 2008; Giordano, Copp, Longmore, & Manning, 2015; W. L. Johnson, Manning, Giordano, & Longmore, 2015). Measures of quality and commitment considered in this article include sexual exclusivity, intimate disclosure, dedication, and love. According to commitment theory, constraints are the factors that make it hard to break up (financial, few alternative partners) and are considered barriers to dissolving even dissatisfying relationships. Stanley and colleagues argue that constraints and commitment are separate constructs that operate together (Owen, Rhoades, Stanley, & Markman, 2011; Rhoades, Stanley, Kelmer, & Markman, 2010; Stanley & Markman, 1992). We expect that cohabiting couples may face high risks of IPV because, more often, these unions consist of a risky combination of low commitment/quality and high dissolution constraints. Cohabiting couples may confront more conflict because they have to negotiate everyday issues related to sharing a home (i.e., housework and money) without the support of a legal marriage as well as negotiating relationship tensions akin to dating couples (i.e., commitment and time with friends; Rhoades et al., 2012). Marital unions typically are characterized by both high dissolution constraints and high commitment/quality, while dating couples typically experience both low dissolution constraints and low commitment/quality.
One of the fundamental ways that cohabitation and dating relationships differ is that cohabiting young adults are sharing a residence indicating greater potential for interaction (both positive and negative) as well as greater opportunity for violence. Stets and Straus (1990) argued that daters, compared with cohabitors, have fewer conflicts and issues of control because their lives are not as intertwined. Additionally, dating, compared with cohabiting, couples have fewer instrumental, social, and emotional investments so the constraints to end a relationship are weaker among dating and greater among cohabiting couples (Rhoades et al., 2010). Cohabiting, compared with dating couples, for example, more often have children (Martinez, Daniels, & Chandra, 2012), report lower levels of instrumental support (Giordano, Manning, Longmore, & Flanigan, 2012), and report greater income pooling (Kenney, 2004). These factors represent relationship-specific investments or constraints that serve as reasons to avoid ending relationships. Thus, cohabitors (compared with daters) may be more constrained and less likely to end relationships that involve precursors to IPV, as such, they may experience higher rates of IPV. Furthermore, cohabitation may be higher risk relationship for IPV than dating because they have conflict over the typical topics of commitment and friends as well as the additional areas of conflict related to coresidence (e.g., household chores, money; Rhoades et al., 2012).
In contrast, married couples typically have high levels of commitment and quality along with high constraints to ending their relationships. Marriages, on average, are of longer duration compared with cohabitation, and provide greater opportunities or exposure to potentially negative interactions. The greater commitment and quality (defined as dedication, sexual exclusivity, love, and intimate disclosure) experienced by married compared with cohabiting couples (e.g., Brown & Bulanda, 2008; Giordano et al., 2012; Joyner, Manning, & Bogle, 2013; Maddox Shaw, Rhoades, Allen, Stanley, & Markman, 2013; Treas & Giesen, 2000) may explain, in part, why married couples report lower odds of relationship violence. Furthermore, married couples may less often be struggling with conflicts that are common to new relationships, such as establishing commitment and spending less time with friends (Rhoades et al., 2012). Along a similar vein, cohabitors often have lower levels of social support for their relationships than married couples resulting in higher levels of social isolation, which is associated with partner violence (Herrera et al., 2008; Stets, 1991; Van Wyk, Benson, Fox, & DeMaris, 2003).
A third explanation for higher rates of IPV among cohabitors is that selection or risk factors might explain differences in married and cohabiting couples’ violence. To the extent that individuals who are in cohabiting compared with dating or marital unions more often possess histories of violence and socioeconomic disadvantage, these selection factors may be important because a family history of violence along with socoiecomomic disadvantage are associated with IPV (Cui, Ueno, Gordon, & Fincham, 2013; Giordano, Johnson, Manning, & Longmore, 2016). Cherlin, Burton, Hurt, and Purvin (2004) reported that in a low-income sample, cohabiting compared with married women more often experienced sexual abuse in childhood. Furthermore, cohabitors, on average, compared with their married counterparts, have lower levels of education and more economically disadvantaged family backgrounds (Goodwin, Mosher, & Chandra, 2010; Joyner et al., 2013), background factors that closely align with other socioeconomic predictors of IPV. Thus, cohabitors more often than married individuals are characterized by factors that may translate into higher odds of experiencing violence in their own relationships. In the current study, we accounted for these selection factors including prior family of origin and relational experiences with violence as well as socioeconomic factors.
Current Investigation
Drawing on exchange and commitment theory and employing longitudinal data, the TARS, we examined partner violence across union types (cohabitation, dating, marital) and within cohabiting relationships. The TARS is well suited to our goals because of the in-depth measurement about relationship functioning and quality. The first research question assessed whether the known differences in self-reported IPV among cohabiting, married, and dating young adults can be explained with a rich set of covariates including relationship duration, prior IPV, relationship quality, constraints and commitment, social support, and selection factors (socioeconomic characteristics and family history). We considered whether low commitment/quality and high constraint relationships were most prone to IPV and whether they were more characteristics of cohabiting than married or dating relationships. The low commitment/quality and high constraint relationships are expected to generate more conflict and frustration with feeling trapped in relationships that do not have a future. The second goal was to focus on cohabiting unions and examine the heterogeneity within cohabitation by considering the types of cohabiting relationships associated with higher levels of self-reported relationship violence. This is not a new question, it was posed over 30 years ago by Yllo and Straus (1981) and warrants attention in the contemporary context. We expected that cohabiting young adults who experienced high relationship constraints and low commitment would face the greatest risk of IPV.
Data and Method
Data
We relied on the TARS, a longitudinal study of 1,321 Toledo area adolescents who were in 7th, 9th, and 11th grade in the fall of 2000 (first interview) and in their late 20s from 2011 to 2012 (fifth interview). The stratified random sample, drawn from school enrollment records of seven Toledo area school districts, totaling 62 schools in Lucas County, Ohio, oversampled Black and Hispanic students. Unlike other data sets that required students to be in school for inclusion in the sample, TARS only required students to be registered, allowing truant or otherwise absent students to participate. Respondents participated in structured in-home interviews through the computer-assisted personal interview technique using preloaded computer questionnaires. We interviewed respondents 1, 3, 5, and 10 years after the first interview for a total of five interviews. A total of 77% of the original sample participated in the fifth interview, the focus of the current study. Attrition analyses indicate that respondents retained at the fifth wave did not differ significantly on most dimensions (e.g., Wave 1 IPV report, parental coercion), but were somewhat more likely to be female (52% at Wave 1, 54% at Wave 5) and a slightly greater share were White (62% at Wave 1 and 65% at Wave 5). Most variables were drawn from the fifth interview except family structure (first interview), coercive parenting (first interview), and prior IPV experience (all interviews). These data offer an excellent lens on IPV as reported by young adults in dating, cohabiting, and marital unions, as well as a rich set of predictors found in previous studies to be associated with relationship violence.
The analytic sample consisted of 926 respondents who reported on a current or most recent different sex intimate partner relationship at the time of the fifth interview (2011-2012). We next limited the sample to only Black, White, and Hispanic respondents, and excluding excluded 22 who reported “other” as their race because there were too few cases to analyze. These data do not include a multiracial category. These restrictions resulted in a final analytic sample of 904 respondents, aged 22 to 29 years, with a mean age of 25 years. The TARS data were from a stratified, random sample; thus, each respondent had a unique probability of inclusion. The distribution of the variables as well as means and standard deviations are reported in Table 1.
Distribution of Dependent and Independent Variables.
Note. IPV = intimate partner violence. Percentages or means reported depending on the nature of the variable. The standard deviations are presented in parentheses. Data source Toledo Adolescent Relationship Study.
Cohabitors significantly different from married or dating respondents, p < .05. bMarried and dating respondents significantly different from one another, p < .05.
IPV referred to the frequency of any physical victimization or perpetration in the current or most recent relationship, based on 12 items from the Revised Conflict Tactics Scale (Straus, Hamby, Boney-McCoy, & Sugarman, 1996). The prompt stated, “During this relationship, how often has/did [partner]”: “throw/n something at you”; “push/ed, shove/d, or grab/bed you”; “twist/ed your arm or hair”; “use/d a knife or gun on you”; “punch/ed or hit you with something that could hurt”; “choke/d you”; “slam/med you against a wall”; “beat you up”; “burn/ed or scald/ed you on purpose”; “kick/ed you”; “slap/ped you in the face or head with an open hand”; or “hit you.” Another set of similar questions were asked about the frequency that the respondent enacted any of the behaviors. These questions were asked at the time of the fifth interview in relation to experiences with the current or most recent partner and referenced both victimization and perpetration experiences with that partner. The responses to each measure are quite skewed because the majority of respondents report never experiencing any of these events. Respondents were coded 1 if they reported having experienced any of these acts and 0 otherwise, resulting in a binary response variable. Similar results are obtained when we rely on a logged frequency indicator. The literature and our community-based sample suggest that the self-reported violence we assessed often reflected violence that emerged from more general patterns of conflict and negative communications or “situational couple violence” rather than the more serious and gendered form known as “intimate terrorist violence” (M. P. Johnson & Leone, 2005). Situational couple violence occurs at substantially higher rates and is often mutual (individuals are both perpetrators and victims). At the time of interview, over half (55%) of respondents reported the violence in their relationship was classified as mutual. Therefore, our work encapsulates a broad definition of IPV experience and our measure is not limited to just perpetration or victimization.
Union status was based on relationship histories and included dating (35.81%), cohabiting (32.36%), and married (23.30%). Cohabitation status was established based on the following question, “Have you ever lived with a boy/girlfriend (not as a roommate) without being married?” Current or most recent relationship indicated that the respondent reported on the current (79.94%) or the most recent (20.06%) relationship. Relationship duration ranged from 0.5 (half a month or less) to 14 years. The duration was based on the current or most relationship state (dating, cohabiting, or married); however, similar results were obtained when estimating duration as the total length of the relationship. The mean relationship duration was approximately 2.5 years with 1.5 years for daters, 2.6 years for cohabitors, and 4.4 years for married individuals.
Prior IPV was measured by assessing whether the respondent had reported experience with IPV (identically as measured above) in a prior relationship to the fifth interview. The assessment of IPV was asked at every prior interview wave and measured the same at each wave of data collection and is coded as 1, any violence, and 0, no violence.
Coercive parenting was measured using a single item from the first interview when respondents were adolescents. We asked, “When you and your parents disagree about things, how often do they push, slap, or hit you?” Responses ranged from 1 (never) to 6 (two or more times a week). Due to the highly skewed nature of responses, we dichotomized this variable to indicate any reports of parental coercion (1 = yes).
Relationship history included measures of prior marriage and prior cohabitation based on the dates of marriage and cohabitation. Given the young age of the sample relatively few respondents were previously married, but nearly half had been in a prior cohabiting relationship.
Commitment and quality was measured by relying on four separate indicators of sexual exclusivity, love, intimate disclosure, and dedication. These concepts drew on the work by Stanley, Whitton, and Markman (2004) as well as Prager (2000) and Christopher and Sprecher (2000). Sexual exclusivity was assessed by an affirmative answer to the question, “Are you and [name] sexually exclusive (not having sex with other people)?” and the response, “never” to the question, “How often have you gotten physically involved (“had sex”) with other guys [girls]?” Love was measured using the following three items: (1) “How much do you love [
Constraints were assessed with four separate indicators: children, pooling income, instrumental support, and lack of relationship alternatives and were consistent with indicators developed by Stanley and colleagues (Owen et al., 2011; Rhoades et al., 2010; Stanley & Markham, 1992). Children indicated whether the respondent had reported having any biological children (41%). Instrumental support was based on items about frequency of receiving or providing help with paying for groceries, clothes, activities, rent and bills, and gifts as well as counting on each other financially if the need arises. The responses for these 12 items range from 1 to 5 and the alpha was .85. Income pooling was based on questions about pooling of money or a shared bank account with values of “1” indicating pooling and “0” no income pooling. Lack of relationship alternatives was based on a single item asking extent of agreement on a 5-point scale with the item, “Sometimes I think I stay with [name] because I might not be able to get another boyfriend [girlfriend].”
For illustrative purposes, we generated summary indicators, one that encapsulated the commitment and quality indicators and another summarizing constraints. The variables were rescaled and summed with responses ranging from 1 to 20. Respondents were recoded as 1, indicating high levels (at or above the median) and otherwise coded as 0 indicating low levels. Dichotomizing these variables allows up to report the proportion of respondents belonging to one of the four categories: high commitment/quality and high constraints, low commitment/quality and high constraints, high commitment/quality and low constraints, and low commitment/quality and low constraints. Thus, we used this strategy to identify the share of respondents who may be in the hypothesized risky subset of relationships that have high constraints and low commitment/quality. We recognize this is not a validated approach and there are issues related to equally weighting each indicator. This strategy is used for descriptive purposes and has the advantage of characterizing the relationship rather than focusing on the mean levels of the individual measures. This aligns with discourse in commitment theory to consider more complex ways of characterizing relationships (Knopp, Rhoades, Stanley, & Markman, 2015).
Social support included closeness to parents and friends. The parental closeness measure was based on a single item indicating level of agreement with the statement, “I feel close to my parents.” Closeness to friends was based on three items asking about how often the respondent spoke to friends about “something really great that happened,” “something really bad that happened,” and “private thoughts or feelings.” The alpha was .78.
Gender, a dichotomous variable, indicated whether the respondent was female. Age was the difference between date of birth and the fifth interview date, with a mean of 25. Race/ethnicity consisted of three categories: non-Hispanic White, non-Hispanic Black, and Hispanic. Family structure, from the respondent’s first interview asked, “During the past 12 months, who were you living with most of the time?” Respondents selected one of 25 categories, which we collapsed into four categories: two biological parents, single parent, stepparents, or “other family” including living with other family members or foster care. Mother’s education was measured at the fifth interview and included less than high school, high school graduate, some college, and college or more. Employment at the fifth interview included three categories: part-time, full-time, and not employed.
Method
We first presented descriptive analyses showing the distribution of respondents across the dependent and independent variables with reports of the percentages and means with standard deviations. Logistic regression models were then estimated that predicted the odds of experiencing IPV. The results are presented in the tables as odds ratios and for ease of interpretation are translated to percentage change in the odds in the body of the article (100-exponential value of the coefficient; DeMaris, 1992). We present the zero-order association between union status and IPV. The first model includes the traditional indicators including sociodemographic indicators, relationship factors including duration and relationship history, and prior violence (family and relational) measures. The second model adds in the quality, commitment, and constraints indicators along with social support measures. Next, we limit our focus to cohabiting couples and present an initial model including the traditional predictors, sociodemographic, relationship duration and history, and prior violence experience. The second model includes the quality, commitment, and constraints indicators as well as social support measures. Additional models were estimated, but not shown, that substituted the specific constraints and commitment measures for the four-category summary indicator.
Results
Similar to prior studies, we found, at the bivariate level, that young adults in cohabiting, compared with dating or marital, relationships more often reported any IPV (Table 1). About one third (31%) of young adults in cohabiting relationships, 23% in marriages, and 18% in dating relationships experienced any violence in their current or most recent relationship. These prevalence rates were similar to prevalence rates based on nationally representative samples such as the Add Health (Cui et al., 2013). Young adults in cohabiting relationships experienced statistically greater rates of IPV than those in married or dating relationships. The higher prevalence of partner violence in cohabiting unions was observed for both male and female respondents.
Table 1 also showed the distribution of the covariates separately for young adults in married, cohabiting, and dating unions. Consistent with prior literature, marital unions, compared with dating and cohabiting unions, were of longer duration. Nearly all marriages were current or ongoing relationships as were two thirds of dating relationships. Relatively few respondents were previously married (3.5%), while half had cohabited previously. Nearly three quarters of married young adults had cohabited prior to marriage, most often with their spouse. About 40% lived with a different partner before this most recent or current cohabiting relationship (serial cohabitation), and 44.5% of daters reported some prior experience with cohabitation. About half of respondents, regardless of union status, had experienced any IPV in a prior relationship. Regarding coercive parenting while growing up, prevalence rates did not differ for married, cohabiting, and dating respondents.
Regarding the sociodemographic characteristics, there were union status differences by age, race/ethnicity, family background, and employment. Dating respondents were younger than cohabiting or married young adults. Married respondents were more often non-Hispanic White (79%) compared with cohabitors (66%) and daters (63%). Cohabiting, compared with married or dating, respondents were more likely to have lived in single mother or stepfamilies while growing up. Married and dating respondents more often had mothers with college degrees than their cohabiting respondents. Respondent’s full-time employment was more common among married and cohabiting than dating respondents.
The measures of quality and commitment were, on average, higher among married than cohabiting or dating respondents. A significantly higher share of married respondents (80%) reported that they were sexually exclusive, compared with 68% of cohabitors and 65% of daters. The levels of dedication, on average, were higher for married than cohabiting or dating respondents. The levels of love and intimate disclosure were, on average, higher for married than cohabiting respondents. Constraint indicators followed a similar pattern in which married, compared with cohabiting and dating, respondents more often had children, received instrumental support, and pooled their incomes. Lack of relationship alternatives were quite similar for cohabiting and married respondents, and dating respondents scored higher on lack of alternatives compared with cohabiting or married respondents. For illustrative purposes, we estimated summary measures of constraints and commitment. While most (63.0%) of the married respondents reported high commitment/quality and high constraints, only 27.8% of cohabitors, and 8.7% of daters did so (results not shown). The modal category for cohabitors is the same one that was hypothesized to be most highly associated with IPV, low commitment/quality and high constraints (34% of cohabitors). As expected, daters most often were in the low commitment/quality and low constraints category (44%). The indicators of parental and friend closeness were similar across union status.
Table 2 presented three logistic regression models, with cohabitation as the referent union status category, predicting the odds of any IPV. The first model presented the zero-order results. The second model added relationship duration, relationship history, prior violence experience (family and relational), and sociodemographic measures. The second model included the indicators of constraints, quality, commitment, and social support. The fundamental question is whether cohabitation is associated with greater odds of IPV and whether this association persists with the inclusion of the covariates.
Logistic Regression Odds Ratio of IPV (n = 904).
Note. IPV = intimate partner violence. Reference category in parentheses. Data source Toledo Adolescent Relationship Study.
p < .10. *p < .05. **p < .01. ***p < .001.
At the bivariate level, dating and married, compared with cohabiting, respondents reported significantly lower odds of experiencing any IPV. Married individuals had a 0.65 lower odds ratio of experiencing IPV compared with cohabitors, and daters had a 0.48 lower odds ratio of experiencing IPV compared with cohabitors. The difference in the odds of IPV for daters and married individuals was not statistically significant (results not shown).
The second model showed that daters and married individuals, compared with cohabitors, continued to experience significantly lower odds of IPV controlling for relationship duration, current status of relationship, relationship history, prior family or relational violence, and sociodemographic characteristics. The significance of the marital indicator decreased to marginal significance (p = .08) due in part to the inclusion of family history and race/ethnicity in the model. The dating coefficients remained significantly associated with lower odds of IPV. Consistent with prior work, relationship duration was associated with higher odds of IPV. Respondents who reported on a recent, compared with current, relationship reported lower odds of IPV, demonstrating the importance of including current and most recent relationships. Prior marital and cohabitation experience were not associated with IPV. As expected, respondents with a history of IPV in a prior relationship had twice the odds of experiencing IPV in their current or most recent relationship. Reports of coercive parenting during adolescence were positively associated with the odds of IPV. The next set of variables included socioeconomic status and family background. The demographic characteristics showed no association with IPV in the full model, but in zero-order models, Hispanic and Black young adults reported higher odds of IPV (results not shown). Young adults who were raised in single parent, compared with two biological, families had higher odds of IPV. Mother’s education was not associated with IPV in the multivariate model, but at the zero-order respondents with highly educated mothers (college graduates) reported lower odds of IPV (results not shown). Respondents who reported either part-time or full-time employment had lower odds of IPV compared with unemployed respondents.
The third column of Table 2 included constraints, quality, commitment, and social support measures. Based on contrasts of the log likelihood ratios, the constraints and commitment indicators contributed to the fit of the model (p < .01). Both dating and married respondents had lower odds of experiencing IPV with the inclusion of constraints (children, lack of alternatives, and instrumental support) as well as the quality and commitment (sexual exclusivity, love, intimate disclosure, and dedication) measures compared with their cohabiting counterparts. Thus, these indicators did not fully explain the differences in IPV by union status. In terms of quality and commitment measures, sexual exclusivity, dedication, and intimate disclosure were associated with lower odds of IPV. The constraint indicator that was associated positively with IPV is income pooling. Although positively associated with IPV in zero-order models (results not shown), in multivariate models, children, instrumental support, and relationship alternatives were not significantly associated with IPV. Taken together, low commitment and high constraints were associated with twice the odds of IPV relative to any other combination of constraints and commitment. Given that cohabiting individuals more often experienced relationships characterized by low commitment and quality along with high constraints (34%) compared with dating or married individuals (Table 1), this finding offered insights into potential explanations for why cohabitors may experience higher odds of IPV. Finally, the social support indicators showed that closeness to family and friends was associated with significantly lower odds of IPV.
Although IPV rates were higher in young adult cohabiting relationships, not all individuals in cohabiting experience IPV. Table 1 indicated that the majority (69%) of cohabiting individuals did not experience any IPV. The second key research question was to assess variation in IPV experiences among cohabiting couples (Table 3). In the first model, longer duration cohabiting unions were associated with higher odds of IPV. Prior experience with cohabitation (or serial cohabitation) was associated with two times higher odds of IPV compared with cohabitors who were in their first cohabiting union. This indicator was not associated with IPV in the prior analyses that included dating and married respondents. A history of relationship violence was associated with higher odds of IPV in the current or most recent cohabiting union. In the multivariate model, coercive parenting during adolescence was not associated with IPV among cohabitors, but was associated with IPV at the zero order (results not shown). In the bivariate models, but not multivariate models, cohabiting Black and Hispanic, compared with White, respondents reported higher odds of IPV. There were no gender or age differences in rates of IPV among this sample of cohabiting young adults. Mother’s higher education level was associated with lower odds of IPV, specifically attaining some post–high school education. Among cohabiting young adults full-time employment was associated with lower odds of IPV compared with unemployed respondents.
Logistic Regression Odds Ratio of IPV Among Cohabitors (n = 288).
Note. IPV = Reference category in parentheses. Data source Toledo Adolescent Relationship Study.
p < .10. *p < .05. **p < .01. ***p < .001.
The second model included the quality, commitment, constraints, and social support measures. These indicators contributed to the fit of the model (p < .01). Generally, among cohabitors quality and commitment indicators were negatively associated with IPV. In the zero-order model, respondents in sexually exclusive relationships were less likely to experience IPV, but in the multivariate model sexual exclusivity was not associated with IPV. Higher levels of dedication and intimate self-disclosure were associated with lower odds of IPV. In terms of constraints, children were associated with lower odds of IPV. Thus, children did not operate as a traditional relationship constraint and appeared to function as a form of commitment. Income pooling, instrumental support and lack of relationship alternatives were not associated with IPV in Table 3, but were in zero-order models. Among cohabitors, it appears the combination of low commitment and quality along with high constraints is associated with higher odds of IPV. Finally, social support from friends was associated with marginally (p = .055) lower odds of IPV, but in the multivariate model parental support was not related to IPV.
Discussion
Consistent with prior studies, in our contemporary sample of young adults’ cohabitation was associated with higher odds of IPV. Based on diffusion theory, we expected that as cohabitation became a common feature of the relationship landscape for young adults, it would be more weakly associated with IPV and result in fewer differences between cohabiting, married, and dating individuals. In Canada, the marriage and cohabitation gap in relationship violence has diminished as the prevalence of cohabitation has risen (Brownridge, 2008). Our results were not consistent with this approach and we found that cohabiting men and women experienced higher rates of IPV compared with dating or married young adults. The results indicated that 3 in 10 cohabiting couples experienced IPV in their relationship.
The findings in this article are consistent with commitment theory and prior empirical results (Brown & Bulanda, 2008; Rhoades et al., 2010). We find that couples with greater relationship constraints and lower levels of interpersonal commitment and relationship quality experienced higher levels of IPV. While cohabiting couples more often had low commitment/quality and high constraints, the inclusion of these relational indicators did not fully explain the higher levels of IPV experienced by cohabiting couples. Researchers studying commitment reference that high constraints and low commitment may signal feeling trapped in a relationship, but to our knowledge no one has included these indicators in one model. Our results support considering these constructs in conjunction with one another and not simply as separate bivariate constructs (e.g., Knopp et al., 2015). Furthermore, this work demonstrates the importance of sexual exclusivity as a domain of relationship quality and predictor of IPV. Most theory and studies on relationship quality ignore the sexual basis of relationships when in fact this is of central importance to couples (Christopher & Sprecher, 2000).
Our work is consistent with arguments that histories of violence and socioeconomic disadvantage are associated with IPV. We found that although prior experience with IPV was associated with higher prevalence of any IPV, it did not explain the union status differentials in IPV. Similarly, coercive parenting during adolescence was associated with IPV, but did not explain why cohabitors have higher rates of IPV. The sociodemographic indicators that select individuals into certain types of unions accounted for some of the cohabitation differential in IPV. Specifically, race and family background accounted for some of the IPV differential between married and cohabiting young adults. Taken together, the selection into cohabitation based on both the sociodemographic factors and history of violence measures does not explain the strong association between cohabitation and IPV.
Although cohabitation was associated with higher odds of IPV, not all cohabiting young adults experienced IPV. This finding aligns with conclusions made over 30 years ago, Yllo and Straus (1981) stated that “cohabitation should not be viewed as a unitary phenomenon” (p. 346). As noted above, we find that 3 in 10 cohabitors had some IPV experience in their current or most recent relationship representing a substantial minority of cohabiting couples. Among cohabitors, some of the significant correlates of IPV were relationship duration, relationship, and IPV history along with constraints, quality, and commitment. While growing shares of young adults have entered more than one cohabiting union (serial cohabitation), little attention has been paid to the implications of serial cohabitation for well-being. Our finding that serial cohabitors were more prone to IPV provides new evidence about the potential negative implications of serial cohabitation. An unexpected finding was that cohabitors with children experienced lower odds of IPV than those without, suggesting that children in cohabiting unions may operate as indicators of commitment rather than relational constraints. The summary indicator of high constraints and low commitment, representing 34% of cohabitors, was significantly associated with higher odds of IPV. IPV levels were about twice as high among cohabiting young adults who faced low commitment and high constraints compared with those with high commitment and high constraints. These findings illustrated the importance of considering the variation in the experiences of cohabiting individuals. In short, this is a constellation of relationship dynamics that matter for understanding variability across union types as well as observed differences within the subgroup of cohabitors.
While this study has contributed to our understanding of cohabitation and IPV, there were a few limitations. The analyses were based on a regional sample of young adults who grew up in Toledo, Ohio. Although the sample was similar in terms of sociodemographic characteristics of young adults across the United States, these analyses should be replicated with representative samples. Second, this study relied on a static measure of union status and did not address transitions, such as movement from singlehood into cohabitation or from cohabitation into marriage. Further attention to how transitions into and out of cohabitation are associated with IPV is warranted. Relationship “churning” (reconciliations after breakups) is more common among those relationships with higher levels of conflict (Halpern-Meekin, Manning, Giordano, & Longmore, 2013). Third, this study focused on IPV at one point in the relationship and did not analyze how or why individuals started or stopped IPV with a particular partner and did not access transitions into and out of cohabiting unions. Additionally, the study relied on the responses from one partner and a couple-based study may advance our understanding of relationship quality. Our study focused on situational couple violence, and as such did not differentiate perpetration from victimization as most individuals in relationships with IPV were both perpetrators and victims of violence. Further attention to new classifications of IPV experiences may be warranted. While our work rests on a select set of measures of quality, commitment, and constraints, there are additional indicators that align with theoretical traditions (commitment or social exchange) that could be important to consider in future work. Finally, this work was limited to only one measure of negative relationship dynamics, IPV, and further important measures that should be assessed in future studies include psychological abuse, sexual coercion, as well as threats of violence.
Cohabitation is a common experience in young adulthood, but one that has the potential to result in negative relationship functioning, specifically IPV. Our work demonstrates that cohabitors have higher rates of IPV even after accounting for indicators of quality, commitment and constraints, histories of family and relationship violence, and sociodemographic measures. Yet our findings cannot be generalized to all cohabiting couples. A subset of cohabitors face the greatest risk of IPV, those who have had a prior cohabiting partner (serial cohabitors) and were in relationships with both low levels of commitment and high constraints. These findings may help refine interventions that target couples who possess the combination of low commitment and high constraints and broaden the scope to consider the full range of potentially unhealthy relationship dynamics. Our work moved forward assessments of cohabitation and IPV by acknowledging the variation in experiences of cohabiting couples, but further work on the specific mechanisms explaining differentials in IPV is warranted.
Footnotes
Authors’ Note
The opinions, findings, and conclusions or recommendations expressed in this publication/program/exhibition are those of the authors and do not necessarily reflect the official views of the National Institutes of Health, Department of Health and Human Services, or Department of Justice.
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: This research was supported by grants from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (HD036223 and HD044206), the Department of Health and Human Services (5APRPA006009), the National Institute of Justice, Office of Justice Programs, U.S. Department of Justice (Award Nos. 2009-IJ-CX-0503 and 2010-MU-MU-0031), and in part by the Center for Family and Demographic Research, Bowling Green State University, which has core funding from the Eunice Kennedy Shriver National Institute of Child Health and Human Development (R24HD050959).