Abstract
Intimate partner violence often begins during the courtship stage of romantic relationships. Although some relationships dissolve as a result of aggression, other relationships remain intact, increasing the risk for escalated violence. The present study identified factors predictive of individual differences in emerging adults’ commitment to physically aggressive or sexually coercive dating relationships. Specifically, Rusbult’s Investment Model of romantic relationships (e.g., investment, satisfaction, quality of alternatives, and commitment) was applied to a longitudinal sample of 148 young adult women who reported experiencing aggression or coercion from their current partners. To further explain commitment within aggressive or coercive dating relationships, rejection sensitivity and anxious and avoidant romantic relational styles were included as predictors of the Investment Model variables. A more avoidant romantic style indirectly predicted commitment through relationship satisfaction and investment. Both commitment and rejection sensitivity significantly predicted continuing an aggressive or coercive relationship 6 months later. The present study improves our understanding of the processes involved in relationship commitment. Continuing to understand these processes will inform interventions that seek to help women who have decided to end aggressive or coercive dating relationships.
Many young adults’ dating relationships involve physical aggression or sexual coercion (Hall-Smith, White, & Holland, 2003; Humphrey & White, 2000), and many continue to develop into long-term, committed relationships (Roscoe & Benaske, 1985) in which aggressive and coercive behavior tends to persist and escalate in frequency and intensity (Stets & Pirog-Good, 1987). Furthermore, the effects of ongoing aggression on women’s mental health worsen the longer aggression continues (Arata, 2002). Recent evidence suggests that women experience less physical aggression and improvements in psychological functioning after an aggressive relationship has dissolved (Young & Furman, in press 1 ). Unfortunately, aggressive and coercive relationships often do not dissolve (Griffing et al., 2002). While it is important to reduce partner aggression and coercion, it is also important to empower women to protect themselves. Thus, it is critically important to understand the factors that influence the dissolution of relationships that involve aggression or coercion.
Relationship Commitment
One factor that has emerged as a significant predictor of relationship dissolution is commitment to that relationship (Le & Agnew, 2003). Surprisingly, however, aggression does not consistently predict commitment (see Follingstad, 2009), suggesting that other factors may play an important role. Given that commitment is an important factor in deciding whether women remain in aggressive, coercive relationships, it is important to identify factors that influence commitment.
The Investment Model (Rusbult, 1980) has helped to explain commitment in relationships that involve aggression and coercion. In the Investment Model, commitment is primarily influenced by three important variables: satisfaction, investment, and quality of alternatives. An individual who perceives more positives than negatives within a relationship remains satisfied and thus more committed. Commitment also remains high when resources such as time, money, and emotional attachment have been invested into the relationship. On the other hand, when an individual sees opportunities to date others who are attractive, commitment declines.
Relationship satisfaction, investment, and quality of alternatives account for much of the variance in commitment in aggressive, coercive relationships (Le & Agnew, 2003; Rhatigan & Axsom, 2006). In turn, relationship commitment prospectively predicts the dissolution of aggressive relationships (Rusbult & Martz, 1995). Investment Model variables have been related to college-aged women’s intentions to leave their aggressive and coercive partners in cross-sectional studies (Choice & Lamke, 1999; Rhatigan & Street, 2005), and, in at least one study, have been shown to prospectively predict actual relationship outcomes (Katz, Kuffel, & Brown, 2006). Interestingly, in the latter study, verbally coerced sexual behavior was concurrently related to greater investment and greater commitment among college women. Greater commitment then predicted relationship maintenance at follow-up. This link bears replication given the relatively few women in this study who reported sexual coercion (n = 52) and a short follow-up period (6-8 weeks).
Although satisfaction, investment, and quality of alternatives are important in determining commitment, several questions remain unanswered. For example, why would satisfaction remain high for some women in the presence of aggressive or coercive behavior? Likewise, what are factors that would lead a woman to perceive that she has few dating alternatives outside of her current, aggressive partner, particularly among young adults who generally have considerable opportunity for social networking? And, given that young adults typically have lower external constraints that make leaving relationships difficult, why might feelings of investment in aggressive or coercive relationships remain high for some women? The present study addresses these questions to more fully explain individual differences in commitment for young adults in physically aggressive or sexually coercive dating relationships.
Interpersonal Cognitions in Relationship Commitment
The aforementioned questions suggest that additional factors may be important in the process of forming commitment. Recently, Rhatigan, Shorey, and Nathanson (2011) found that sexually victimized women’s symptoms, self-efficacy, and shame indirectly predicted commitment to a hypothetical relationship through their influence on satisfaction, investment, and perceived quality of alternatives, but as yet, nobody has examined the role of such interpersonal factors in actual commitment. Thus, the present study examined three factors related to women’s interpersonal cognitions about romantic relationships. Each factor was selected for its association with increased risk of experiencing physical aggression (Vezina & Hebert, 2007) or sexual coercion (Young & Furman, 2008). In addition, each of the factors is thought to influence the way in which women understand and approach romantic relationships and the way in which interactions with a romantic partner are interpreted, thereby influencing satisfaction, investment, and perception of available alternatives.
Romantic Relational Style
The first two variables pertain to women’s romantic relationship styles. Attachment theorists have conceptualized romantic styles as representations of oneself, the partner, and the relationship that influence one’s expectations and behavior within a romantic relationship (Furman & Wehner, 1994). Romantic styles are often measured in terms of the two dimensions of anxious and avoidant styles (Brennan, Clark, & Shaver, 1998). Individuals high on the anxiety dimension tend to intensely desire intimacy and closeness and may become more emotionally invested, finding it difficult to dissolve the relationship. In fact, among adult battered women, those with more anxious styles report greater feelings of love and desire to return to their partner (Henderson, Bartholomew, & Dutton, 1997). It is expected that a more anxious romantic style will be associated with higher relationship investment.
In contrast, women with more avoidant romantic styles tend to minimize the affective importance of relationships, minimize emotional intimacy, and emphasize their own strength and independence (Cassidy & Kobak, 1988). Women who are more avoidant are likely to invest fewer emotional resources into an intimate relationship and are unlikely to remain in a problematic relationships (Henderson et al., 1997). Moreover, women who are more avoidant move from partner to partner and are thus likely to perceive greater quality of alternative relationship partners. It is expected that a more avoidant romantic style will be associated with lower relationship investment and higher perceived quality of alternatives.
Rejection Sensitivity
Individuals high on rejection sensitivity place a high degree of importance on intimate relationships but fear rejection and abandonment from their partners (Downey & Feldman, 1996). Individuals high on rejection sensitivity have been shown to tolerate unwanted sexual activity and abusive partner behavior to maintain their relationships (Downey, Bonica, & Rincon, 1999). Given their expectation of interpersonal rejection, women high on rejection sensitivity are expected to perceive that they have few alternatives to their current dating partners; thus, they may strive to maintain their current relationships despite experiencing aggression. Indeed, high rejection sensitivity prospectively increases risk for sexual victimization (Young & Furman, 2008).
Rejection sensitivity may have different effects on satisfaction and investment than on perceptions of alternatives. Because individuals high on rejection sensitivity anticipate rejection, they may not achieve high levels of intimacy and closeness with their partners (Downey, Freitas, Michaelis, & Khouri, 1998). Thus, high rejection sensitivity is expected to be related to low investment and low satisfaction.
Present Study
This study sought to understand factors that may influence commitment when physical aggression or sexual coercion is present. The Investment Model was applied to a longitudinal sample of young adult women who have experienced physical aggression or sexual coercion from their current partners. We expected that aggressive or coercive dating relationships with higher commitment would be more likely to remain intact over 6 months. Anxious and avoidant relationship styles and rejection sensitivity were examined as predictors of the Investment Model variables. We anticipated that these variables would indirectly predict commitment through satisfaction, investment, and quality of alternatives.
Method
Participants
College women from two universities completed an online survey about their dating relationships. Women were eligible to participate in the 6-month longitudinal study if they were between the ages of 18 and 25 years, experienced physical aggression or sexual coercion from their current dating partners, and were not married or engaged to their partners. Of the 2,358 female college students who completed the screening questionnaire (approximately 12% of the total female student body across both universities), 430 women were eligible. Of these, 148 women completed the initial longitudinal assessment. Among eligible women, no differences were found between those who participated and those who did not on any screening questionnaire variables, including age, ethnicity, physical aggression, or sexual coercion.
On average, participants were 21.52 years old (SD = 2.77 years). Most participants were Caucasian (83.1%); smaller proportions were from ethnic minority backgrounds (6.8% Hispanic, 4.1% Asian American, 1.4% African American, 2.0% Other). Participants self-defined whether they were in a dating relationship to ensure that we did not exclude relationships they considered to be dating relationships or include ones they did not consider to be dating relationships. The average length of participants’ current relationships was 23.14 months (SD = 17.62 months, range = 0.50-72.75 months). Nineteen participants (12.8%) were living with their current partners.
Procedure
All students enrolled at a private university and at a large public university in the western United States received an email inviting them to take a brief Internet survey on their dating relationships. Students who completed the screening questionnaire were given the opportunity to enter a US$100 raffle. Eligible women were then invited by email to participate in the longitudinal phase of the study, which included an initial assessment and a 6-month follow-up. Both assessments were administered online and took approximately 1.5 hours to complete. Participants in the longitudinal study were paid US$20 for each assessment. The institutional review boards at both universities approved the protection of human subjects in this study.
Measures
Screening questionnaire
Participants were asked if they had experienced each of six types of physical aggression and five types of sexual coercion from their current partners. Items were taken from the Conflict Tactics Scale—Revised (CTS-2; Straus, Hamby, Boney-McCoy, & Sugarman, 1996). Physical aggression items included how often (1 = never, 4 = many times) a partner threw something at you; pushed, grabbed, or shoved you; slapped or hit you; kicked, bit, or hit you with a fist; hit or tried to hit you with something; and beat you up. Sexual coercion items included how often (1 = never, 4 = many times) a partner insisted on sex play when you did not want to; tried to make you have sex without a condom; insisted on intercourse, or oral or anal sex when you did not want to; used threats to make you have intercourse, or oral or anal sex when you did not want to; and used force (like hitting or holding you down) to make you have intercourse, or oral or anal sex when you did not want to. Endorsement of at least one of these items was taken as indication of being in an aggressive or coercive relationship and eligibility for the longitudinal study.
Physical aggression and sexual coercion
Women completed the CTS-2 (Straus et al., 1996) that is a widely used measure of aggression within intimate relationships. It includes a 12-item physical aggression scale and a 7-item sexual coercion scale (physical: α = .68; sexual: α = .71). In addition, women completed the Sexual Experiences Survey (SES; Koss & Gidycz, 1985). The SES consists of 11 items that measure specific types of sexual coercion (α = .68). The SES and the sexual coercion scale of the CTS were standardized separately and then combined to provide a more complete assessment of sexual coercion (α = .79).
Investment Model
Participants’ relationship satisfaction, investment, perceived quality of alternatives, and commitment to their romantic partners were assessed with the Investment Model Scale (IMS; Rusbult, Martz, & Agnew, 1998). Ten items assess relationship satisfaction (α = .96), and 10 items describe emotional investment in the relationship (α = .86). The perceived quality of alternatives scale consists of nine items (α = .85) that assess the extent to which a participant perceives appealing romantic options outside of the current relationship. Finally, seven items (α = .91) assess commitment to a current partner.
Rejection sensitivity
Participants completed the nine-item romantic relationship scale of the Rejection Sensitivity Questionnaire (RSQ; Downey & Feldman, 1996). Using 6-point scales, each participant first rated the degree to which she would be anxious or concerned about the outcome of a situation and then rated the extent to which she would expect her partner to respond in an accepting manner. These two components of each situation were multiplied into a single score; these nine scores were then averaged into a single scale (α = .85).
Romantic relational styles
The 36-item romantic partner version of the Behavioral Systems Questionnaire (BSQ; Furman & Wehner, 1999) assessed participants’ romantic relational style. Using 5-point scales, each participant rated her agreement with statements related to four behavioral systems (attachment, affiliation, caretaking, and sexuality). These items are divided into three scales that assess secure, dismissing (avoidant), or preoccupied (anxious) styles. Previous factor analyses of the BSQ have found two underlying dimensions: anxious and avoidant romantic relational styles. Thus, two relational style scores were used in all analyses, both with good internal reliability (αs = .91 and .85).
Relationship outcome
At follow-up, each participant answered a single question indicating whether she was still dating the partner about whom she had answered questions 6 months earlier. Relationships that remained intact were coded as 1, and relationships that dissolved were coded as 2.
Results
Preliminary Analyses
Data preparation and missing data
Variables in the data set were assessed for normality of distribution and the presence of outliers. No violations of normality were noted; all values for both skew and kurtosis were within normal limits (skew, −3 to +3; kurtosis, −10 to +10). Outliers were identified and corrected by equating extreme values to scores of ±1.5 times the interquartile range below the 25th percentile or above the 75th percentile.
The average rate of missing data at the initial assessment was low (M = 1.4%, range = 0%-2.7%). Relationship outcome data were not available for 22 participants (14.9%) who did not complete the follow-up assessment. A series of independent samples t tests assessed for differences between participants who completed both longitudinal assessments and those who only completed the initial assessment. No differences were found between these groups by demographic characteristics (e.g., age, ethnicity, year in school) or by dating characteristics (number of different people dated in the past 12 months, length of time spent dating current partner, age of current partner). No differences were found for most of the primary variables of interest, including level of physical aggression and sexual coercion, rejection sensitivity, avoidant and anxious romantic styles, and relationship investment, satisfaction, quality of alternatives, and commitment. Of the 16 comparisons made, only one comparison yielded a significant difference. Women who did not complete the follow-up assessment reported having experienced sexual coercion from their current partner more recently (M = 6.11 months, SD = 4.81 months; M = 8.71 months, SD = 4.71 months; t(136) = −2.23, p = .03). Missing data were estimated using full information maximum likelihood (FIML) procedures.
Physical aggression and sexual coercion
As would be expected among a sample of young adult dating couples, participants reported a variety of physically aggressive and sexually coercive behaviors that ranged in severity. At the initial assessment, 60 (40.5%) women reported having experienced some form of physical aggression from their current romantic partners, mainly involving pushing or shoving (42.9%). Other forms of aggression reported by women included having something thrown at them (31.7%), having their arms or hair twisted (18.3%), or having been slammed against the wall (18.3%). Some women did report injuries as a result of the physical aggression, including a sprain, bruise, or cut (26.7%) and pain lasting until the next day (14.3%). Sexual coercion was more common, with 126 women (85.1%) reporting unwanted sexual contact with their current dating partners; the majority of sexually coercive acts were described as unwanted sexual play (e.g., fondling, kissing, petting; 50.9%) or unwanted intercourse due to verbal pressure (75.5%). However, some women reported more serious acts, including receiving threats to coerce sex (36.5%) or having been made to have sex without a condom (34.1%). A few women reported having been physically forced to have sex (11.9%).
Investment model
Table 1 presents the means, standard deviations, and correlations among the investment model and interpersonal cognition variables at the initial assessment. The majority of relationships remained intact at follow-up (68.3%). Significant differences were found between women whose relationships dissolved and women whose relationships remained intact. Those whose relationships remained intact reported significantly higher satisfaction (intact: M = 5.89, SD = 0.94; dissolved: M = 5.35, SD = 1.07; t(124) = 2.85, p = .005), higher investment (intact: M = 5.46, SD = 0.83; dissolved: M = 4.89, SD = 1.15; t(124) = 3.19, p = .002), and lower perceived quality of alternatives (intact: M = 3.54, SD = 1.16; dissolved: M = 4.13, SD = 1.27; t(124) = −2.54, p = .012). Women whose relationships remained intact reported significantly higher commitment than those whose relationships dissolved (intact: M = 5.91, SD = 1.15; dissolved: M = 4.86, SD = 1.67; t(124) = 4.10, p < .001). Neither physical aggression nor sexual coercion was directly related to commitment at the initial assessment or relationship outcome at follow-up.
Means (SDs) and Pearson Correlations for the Investment Model, and Interpersonal Cognition Variables.
Note. Ms (SDs) are presented separately for participants whose relationships remained intact or dissolved by follow-up. Correlations are presented for the entire sample.
p < .05. **p < .01.
Primary Analyses
Path analysis
Path models were estimated in Amos 7.0 (Arbuckle, 2006), and model fit was assessed using chi-square, the comparative fit index (CFI), and the root mean square error of approximation (RMSEA; see Kline, 2005). The first step was to replicate the original Investment Model to concurrently predict relationship commitment and to extend that model to prospectively predict the relationship outcome. This model fit the data well, χ2(3, N = 148) = 1.71, p = .64; CFI = 1.00; RMSEA = .00. As expected, higher satisfaction and greater investment predicted more commitment. Higher perceived quality of alternatives predicted less commitment. These variables explained approximately 59% of the variance in commitment (R2 = .59). Furthermore, relationships in which women reported more commitment were more likely to remain intact at follow-up (R2 = .12).
Next, the interpersonal cognition variables were added to the model. These variables were included as indirect predictors of commitment through relationship investment, satisfaction, and perceived quality of alternatives. This model provided adequate fit to the data, χ2(9, N = 148) = 18.60; CFI = .97; RMSEA = .09. More avoidant romantic styles were directly related to lower investment, lower satisfaction, and higher perceived quality of alternatives. Higher rejection sensitivity also was directly related to lower satisfaction. Adding these variables to the model did not alter the path coefficients from the previous step (e.g., Investment Model variables predicting commitment and relationship outcome); thus the final model is presented in Figure 1.

The investment model with interpersonal cognition variables.
Several post hoc analyses were performed based on strong bivariate correlations among several variables (see Table 1). First, a direct path was estimated in the model from rejection sensitivity to the relationship outcome. Higher rejection sensitivity prospectively predicted relationship dissolution 6 months later. Adding this path provided a significant improvement in the model fit, Δχ2(1, N = 148) = 11.47, p = .001. In addition, adding this direct effect increased the amount of variance predicted in relationship outcome (R2 = .20). The final model provided an excellent fit to the data, χ2(8, N = 148) = 7.13; CFI = 1.00; RMSEA = .00.
Second, rejection sensitivity and anxious romantic styles were also strongly related. To assess whether rejection sensitivity may have masked an effect for anxious romantic styles, the model was estimated without rejection sensitivity, χ2(7, N = 148) = 8.45; CFI = 1.00; RMSEA = .04. Interestingly, higher anxious romantic styles significantly predicted lower satisfaction.
Indirect effects
The interpersonal cognition variables were hypothesized to exert an indirect effect on relationship commitment. To test this, a distribution-of-products approach (MacKinnon, Lockwood, & Williams, 2004) was taken to construct confidence intervals around the indirect effects in the path model using the Prodclin software program (MacKinnon, Fritz, Williams, & Lockwood, 2007). An avoidant romantic style was found to have significant indirect effects on commitment separately through satisfaction (αβ = −.65; 95% CI [−.42, −.92]), investment (αβ = −.44; 95% CI [−.23, −.68]), and perceived quality of alternatives (αβ = −.206; 95% CI [−.06, −.39). The indirect effect of rejection sensitivity on relationship commitment through satisfaction was also significant (αβ = −.03; 95% CI [−.05, −.003]). Anxious romantic style did not have a significant effect on relationship commitment.
Discussion
When dating relationships that involve aggression dissolve, young adult women subsequently report experiencing less physical aggression and improvements in psychological functioning (blinded citation); yet, these relationships often do not dissolve (Griffing et al., 2002), and the factors influencing commitment in these dating relationships are not well understood. The present study sought to better understand factors that contribute to individual differences in young adults’ commitment to physically aggressive or sexually coercive dating relationships.
Previous findings regarding the Investment Model were replicated among a sample of young adult women who experienced physical aggression or sexual coercion from a current dating partner. Both the direction and strength of relations among the Investment Model variables in the present sample were similar to previous studies (e.g., Le & Agnew, 2003; Rhatigan & Street, 2005; Rusbult et al., 1998), and these variables predicted nearly 60% of the variance in women’s commitment to their aggressive, coercive partners. In addition, to our knowledge, the present study and Katz et al. (2006) are the only studies to demonstrate that earlier commitment would prospectively predict the actual relationship outcome in dating relationships that involve physical aggression or sexual coercion. Less than two thirds of women had ended their relationships by the 6-month follow-up, underscoring the propensity for relationships to remain intact despite aggression or coercion. Women whose relationships remained intact also reported significantly higher satisfaction and greater investment than women whose relationships dissolved, and they perceived themselves as having fewer quality alternatives to their current dating partners. Together, these findings provide further evidence in support of the Investment Model as an explanatory mechanism in the development of commitment and the dissolution of aggressive, coercive dating relationships.
The present study also examined three variables related to women’s interpersonal cognitions as predictors of individual differences in commitment. Women who were more avoidant in their romantic style reported less investment and less satisfaction in their relationships, and they perceived greater availability of romantic opportunities outside their current relationships. Furthermore, having a more avoidant style indirectly predicted lower commitment. Discomfort with emotional intimacy and lower expectations of a partner’s availability for support may serve to diminish enjoyment of a relationship and orient women toward eventually ending the relationship (see a review by Mikulincer & Shaver, 2007). Aggression or coercion may further diminish satisfaction and investment in women who are more romantically avoidant (relative to less romantically avoidant women), thereby reducing commitment and increasing the likelihood of the relationship dissolving.
Higher rejection sensitivity and higher anxious romantic styles were both related to lower satisfaction, though only rejection sensitivity was a significant predictor when both variables were in the model. The degree of shared variance between these variables seems to have masked the effect of anxious romantic styles on satisfaction. Both measures assess worry and negative anticipation of being rebuffed by a partner. Thus, it seems that the fear of rejection and worry about a partner’s responsiveness captured by these measures were negatively related to satisfaction, which in turn predicted lower relationship commitment. In addition, women with higher rejection sensitivity tend to react angrily and defensively toward others when they perceive interpersonal rejection (Downey et al., 1998). Thus, it may be that women with higher rejection sensitivity are more likely to interpret a partner’s aggressive behavior as a sign of potential rejection, thereby reducing satisfaction and commitment. This explanation is supported by the finding that higher rejection sensitivity directly predicted relationship dissolution.
Several interesting implications follow from this study’s findings. First, individuals who are high on rejection sensitivity and avoidant and anxious romantic style may actually be at lower risk for the adverse mental health effects associated with ongoing aggression or coercion, as they are more likely to dissolve the relationship. This conclusion seems somewhat counterintuitive; after all, rejection sensitivity and insecure romantic styles are generally considered risk factors for relationship instability and problems in psychological adjustment and interpersonal relationships (Ayduk, Downey, & Kim, 2001; Creasey & Hesson-McInnis, 2001). However, when continued interaction with aggressive or coercive partners places women at risk of further aggression and worse mental health (Arata, 2002; Stets & Pirog-Good, 1987), it stands to reason that any features that serve to reduce commitment and hasten the dissolution of aggressive, coercive relationships—even rejection sensitivity and insecure romantic styles—may actually serve to protect against these effects. Conversely, features that would promote intimacy or commitment, such as secure romantic styles or low rejection sensitivity, could have the unhealthy effect of increasing the likelihood that the relationship continues. Alternatively, the partners of women who are high in rejection sensitivity or insecure style may be more likely to dissolve the relationship. Either way, it appears that such characteristics may protect against the deleterious effects of aggressive, coercive relationships.
An important caveat to this conclusion, however, is the extent to which the findings would also apply to women who experience more severe forms of aggression or coercion. It seems unlikely that women who have less rejection sensitivity and who are more secure in their romantic style would remain more committed to a partner if the aggression or coercion they were experiencing was more severe in nature. Nevertheless, it is interesting to consider the present results in light of the fact that interpersonal violence generally begins at lower levels of severity, escalating and intensifying as the relationship continues to develop. Similarly, commitment also begins low and increases over time. Thus, if both aggressive, coercive behavior and commitment begin low but increase over the course of a relationship, at what point does the balance shift such that negative aspects of the relationship outweigh the good and commitment is reduced? Where is the tipping point?
Another interesting consideration in the present study is the extent to which rejection sensitivity and insecure romantic styles are precursors of relationship aggression and coercion or are sequelae of these experiences. Theoretically, rejection sensitivity and insecure romantic styles are conceptualized as fairly stable characteristics that begin forming from earlier interactions with parents and friends and are then carried forward into romantic relationships (W. A. Collins & Sroufe, 1999; Downey et al., 1998). Thus, it seems likely that features of these constructs were already present in participants before they began experiencing relationship aggression and coercion. On the other hand, it seems plausible that experiencing aggression or coercion from a dating partner would create feelings of rejection, doubts concerning a partner’s supportiveness, and questions about the stability of the relationship.
In reality, these features are probably both precursor and consequence of relationship aggression and coercion. Individuals who present with features of rejection sensitivity and insecure romantic styles are at greater risk of relationship aggression and coercion. In turn, experiencing aggression or coercion reinforces expectations for rejection and relationship insecurity, creating a self-fulfilling feedback loop. Indeed, empirical support has been demonstrated for a similar feedback loop involving an intensification of adolescents’ risk behaviors after experiencing sexual coercion (blinded citation).
Limitations and Future Directions
Several limitations are important to acknowledge. First, causal inferences cannot be made about the dissolution of aggressive or coercive dating relationships. In particular, we do not know who decided to end the relationships. In some cases, partners may have initiated the break-up, and some women may have preferred for the relationships to continue.
Second, several limitations in the data may affect the generalizability of the present results. Only 34% of women who were eligible according to the screening questionnaire chose to participate in the longitudinal study. No differences were, however, found between those who participated and those who did not on any screening questionnaire variables, including age, ethnicity, physical aggression, or sexual coercion. In addition, 22 of the 148 women who initially enrolled did not provide 6-month follow-up data. These women reported having experienced sexual coercion more recently than women who completed the follow-up. No other group differences were found on demographics, dating experiences, or the primary variables of interest in the study (including physical aggression or level of sexual coercion) that would suggest ways in which the findings may apply differently. Results may also not generalize beyond the college setting, in which the social milieu represents a unique set of circumstances (e.g., dormitories or shared housing, large-group social gatherings, extensive social networks, etc.) in which patterns of relationship aggression may unfold differently.
Finally, aggressive or coercive behavior of a partner may be interpreted or experienced differently among women in same-gender relationships or women from a different cultural or socioeconomic background. It will be important to replicate the present results in other populations, in particular adolescents and young adults not in a college setting.
Despite these limitations, these findings suggest several avenues for future work. The present study extended previous cross-sectional findings to include prospective prediction of relationship outcomes, and future work should follow these relationships beyond 6 months to assess patterns of continued interaction, aggression, and relationship development. Although some women’s aggressive or coercive relationships dissolved, the long-term health of these women’s romantic relationship experiences remains unclear, particularly given that insecure relationship styles and rejection sensitivity are risk factors for poor relationship outcomes in general (Downey et al., 1998). A longer follow-up period would help to understand what dating experiences women may have after their relationships dissolved. For example, did the relationships remain dissolved or become reinstated? Did women enter new relationships, and did those new relationships involve aggression or coercion? Similarly, continued follow-up would also enable the continued assessment of relationships that remain intact, including changes that may occur in aggression, coercion, and the interpersonal variables.
It will also be important to continue identifying other predictors of commitment. For example, commitment to a dating relationship may be influenced by social pressure to stay together, concern for the partners’ welfare, perceived difficulty in dissolving the relationship, and various structural constraints such as sharing finances, signing a lease, owning furniture or a pet together, and having made future plans (Rhoades, Stanley, & Markman, 2010). Furthermore, identifying skill-based characteristics that influence relationship dissolution such as assertiveness, communication, and problem-solving and emotion regulation skills may offer specific targets of intervention. Relatedly, it seems plausible that these interpersonal variables would play an important role in relationships that do not involve aggression or coercion but that contain other negative features. For example, the present model may also help to explain how commitment operates to influence outcomes in relationships with a high level of verbal conflict, emotional jealousy, or infidelity.
Finally, this study focused on only one partner’s perceptions and commitment to the relationship. Relationships are dyadic by nature, however, and the characteristics of both individuals combine and interact to influence each other. Relatedly, physical aggression within dating relationships is often reciprocal in nature (see Langhinrichsen-Rohling, Selwyn, & Rohling, 2012). The present study did not specifically address aggressive behavior on the part of both partners or how this may impact commitment and relationship outcomes. Aggressive behavior by women is thought to serve several functions within dating relationships (Foshee, Bauman, Linder, Rice, & Wilcher, 2007) and conceivably may influence one or both partners’ satisfaction, investment, and commitment. Thus, it will be important for future work to consider the characteristics and commitment of both partners, as well as the contribution made by each to aggressive behavior in the relationship.
The present study findings provide further evidence in support of the Investment Model as an explanatory mechanism in the development of commitment and the dissolution of dating relationships that involve physical aggression or sexual coercion. Anxious and avoidant romantic styles as well as rejection sensitivity elaborate the model and help explain differences in the relationship outcome when aggression or coercion is present. Continuing to identify factors that influence satisfaction, investment, quality of alternatives, and ultimately commitment will improve our understanding of these processes and inform interventions that seek to help women who have decided to end their aggressive or coercive relationships.
Footnotes
Acknowledgements
Appreciation is expressed to the staff of the Gender Violence Education and Support Center at the University of Denver, the Rape and Gender Education Program at the University of Colorado-Boulder, the Assault Survivors Advocacy Program at the University of Northern Colorado, and the Project STAR staff at the University of Denver for their collaboration and assistance in data collection.
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: Preparation of this manuscript was supported by Ruth Kirschstein F31 Predoctoral Fellowship (5F31MH085395) through the National Institute of Mental Health. In addition, support was provided by a grant (HD049080) from the National Institute of Child Health and Human Development (W. Furman, principal investigator).
