The limits of ‘adaptive’ coping: Well-being and mood reactions to stressors among women in abusive dating relationships

Abstract

Coping is typically thought to be adaptive if it reduces immediate distress and promotes well-being. However, coping strategies might appear beneficial in a given situation, but when considered in the broader stressor context, those situational benefits may actually undermine well-being. Two studies (N = 473 and N = 80 women) demonstrated that, in the context of psychologically or physically abusive dating relationships, coping orientations were rooted in women's stressor history (prior assault trauma) and elevated emotion-focused and lower problem-focused efforts were predictive of greater depressive symptoms. Yet, in response to a stressor video that acted as a reminder of women's abusive experiences (but not to a stressor video unrelated to abuse), affective benefits (lower hostility, higher positive agency) were associated with abused women's emotion-focused coping endorsements, but were not linked to problem-focused coping. It seems that in some contexts, reduced distress might limit active efforts to alter a dysfunctional situation.

• Coping
• dating abuse
• depression
• mood
• emotional stress
• young women

Introduction

Understanding what constitutes adaptive coping has been central to efforts to identify strategies for managing or diminishing the negative outcomes associated with stressful events. Coping is viewed as adaptive to the extent that it reduces immediate distress and promotes long-term well-being (Zeidner and Saklofske 1996; Snyder and Dinoff 1999). In this regard, the effectiveness of coping strategies ought to be considered within the context of specific stressor situations, taking into consideration the nature of the stressor and the individual's goals within that situation (Mattlin et al. 1990; Tennen et al. 2000). Moreover, coping efforts can vary from moment-to-moment, across domains and across time and indeed, individuals may hold multiple goals simultaneously (Zeidner and Saklofske 1996; Schwartz et al. 1999; Tennen et al. 2000; Skinner et al. 2003; DeLongis and Holtzman 2005).

Although not denying the importance of cross-situational variations, certain coping responses might be more likely to produce adaptive outcomes (e.g. problem-focused efforts), whereas others may have more variable effects across situations or across time. In this regard, avoidant coping might diminish short-term distress in a given situation and hence be considered effective (Suls and Fletcher 1985). Yet, this coping approach may work at cross-purposes with more active efforts to contend with the stressor and hence, may be linked to long-term dysfunction (Zeidner and Saklofske 1996; Snyder and Pulvers 2001). By example, women in domestic violence situations are particularly likely to use avoidant coping (Waldrop and Resick 2004), which might serve to minimize the distress associated with a specific encounter. As a result, however, rather than choosing the obviously adaptive response of leaving an unhealthy relationship, an appreciable proportion of women remain with their partners at the expense of their own safety and psychological health. To be sure, external barriers (e.g. children, financial factors) often make it difficult for women to exit domestic violence situations (Carlson 1997) and it could thus be argued that alleviation of the distress brought on by an abusive event may effectively meet the immediate needs of battered women. However, in the case of abusive dating relationships, these external barriers do not typically exist and hence coping efforts that reduce short-term distress may be counterproductive, as they sustain the dysfunctional relationship despite the relatively wide range of options that remain available.

Given that women in dating relationships often continue with their abusive partners, there appears to be a potential disjunction between the coping efforts they use to contend with their day-to-day relationship experiences and their ability to maintain or protect their longer-term well-being. Thus, the present investigation assessed the coping efforts endorsed by women who were or were not in abusive dating relationships and their implications for well-being. In an initial study, we assessed the coping propensities of women in abusive (vs. non-abusive) dating relationships and determined whether these coping efforts might be especially relevant to symptoms of depression. In addition, in a laboratory context (study 2), we evaluated whether the same coping propensities of women in abusive (vs. non-abusive) relationships that were associated with depressive affect, would bring about “apparently” positive outcomes in the context of abuse (e.g. minimizing negative mood reactions).

The stressor context: Abuse within dating relationships

Conflict within close relationships is not uncommon and when it persists, may undermine well-being (Kiecolt-Glaser and Newton 2001; Beach et al. 2003). All too often, the severity of the conflicts may evolve to become abusive and such interactions are surprisingly prevalent in dating relationships of college-aged women (White and Koss 1991; DeKeseredy and Kelly 1993; Kuffel and Katz 2002). Indeed, just under one-third reported physical abuse from their partners in the past six months and up to 80% reported a psychologically abusive incident, although these rates depended on how abuse was defined (Pipes and LeBov-Keeler 1997; Jackson 1999). Evidently, far from being unusual, dating abuse is a common stressor encountered by young women.

Physical abuse in dating relationships may engender marked psychological consequences (Arias et al. 1987; Follingstad et al. 1999; Ackard et al. 2003) and psychological abuse appears to be a key contributor to such disturbances, above and beyond the contribution of other forms of abuse (Follingstad et al. 1990; Katz and Arias 1999; O'Leary 2001; Straight et al. 2003). Psychological abuse often comprises behaviors such as name-calling, lack of affection, social isolation, humiliation and jealousy, resulting in the target of the abuse feeling threatened (Follingstad et al. 1990; Katz and Arias 1999; Straight et al. 2003). To be sure, in the absence of the more overt forms of abuse, the point at which “normal” conflicts cross the line to constitute psychological abuse may be ambiguous. In effect, dating abuse varies along multiple dimensions that may place different types of strain on the victim.

Coping adaptiveness

A wide range of coping strategies may be adopted to deal with stressors, the effectiveness of which is likely situation-specific (Tennen et al. 2000; DeLongis and Holtzman 2005). Yet, it has frequently been argued that problem-focused strategies (e.g. problem solving, cognitive restructuring or positive growth) are associated with adaptive outcomes, whereas those strategies that encourage a focus on emotions (e.g. rumination, emotional venting, self-blame) tend not to be (Lazarus and Folkman 1984; Carver et al. 1989). This said, it has been suggested that in emotionally charged situations, emotional-approach coping might facilitate individuals' ability to come to terms with their feelings and in doing so distress may be reduced (Stanton and Franz 1999). Likewise, although avoidance often works against well-being in the long-run, it may provide temporary relief from an ongoing stressor, allowing the individual to adopt more effective strategies later on (Suls and Fletcher 1985; Carver et al. 1992; Janoff-Bulman 1999). Thus, the coping typology juxtaposing the adaptiveness of problem-vs.-emotion-focused efforts is relatively simplistic. A given coping strategy can be expressed through a range of actions, have multiple functions that can vary across situations and its efficacy may also be situation-specific and may vary over time (Zeidner and Saklofske 1996; Schwartz et al. 1999; Tennen et al. 2000; Skinner et al. 2003; DeLongis and Holtzman 2005). Nonetheless, it remains tenable that broad “families” of coping (i.e. multiple dimensions of coping behaviors that tend to co-occur) might be expected to have more or less adaptive consequences when the individual is confronted with a stressor (Skinner et al. 2003).

Although the effectiveness of particular coping efforts in relation to well-being may be situation specific, it is also possible that considerations of what constitutes an adaptive outcome also ought to be context specific. For example, in response to an abusive dating relationship, experiencing distress ought to serve as an adaptive signal that encourages active coping to alter the situation. Yet, women in domestic violence situations were particularly likely to use avoidant coping (Waldrop and Resick 2004), which might have been reinforced because it temporarily alleviated distress. The use of avoidant strategies might be further reinforced among women in abusive situations, because they provide some degree of illusory control (Rosen 1996). Such perceptions of control might be adaptive if they resulted in women feeling empowered to alter their abusive situation. However, this sense of control was found not to culminate in greater problem-focused coping efforts among abused women (Pape and Arias 1995). Essentially, coping strategies may appear to be effective when they provide women with short-term “benefits” in terms of reducing distress or promoting subjective appraisals of control. Yet, because these women continue to experience abuse, the use of such strategies are likely counterproductive in that they may promote poorer long-term outcomes, such as depression.

The present investigation evaluated the coping methods of women in abusive dating relationships and determined whether these coping efforts might be especially relevant to symptoms of depression, but at the same time be linked to affective benefits when confronted with the abuse context. Women's coping efforts in response to abuse have been considered primarily among women in long-standing domestic violence situations and hence, it is difficult to differentiate coping efforts that reflect propensities women may have brought with them to the situation relative to those that may have arisen specifically as a response to their abuse. Admittedly, to assess this distinction, a prospective analysis of women's coping over time is necessary. Yet, by assessing women in the dating phase of their relationships, it might be possible to gain insights as to whether coping efforts in general and those that are specific to their relationship conflicts, including abuse, are distinguishable from one another and from those of women whose relationships are not abusive.

To the extent that women in abusive dating relationships demonstrate patterns of coping that differ from those of women who are not in abusive relationships, it is possible that particular coping propensities evolved from such experiences because they allowed the individual to minimize distress within that situation. Thus, a second laboratory study evaluated whether affective benefits (i.e. reduced distress) were derived in response to a stressor that had features in common with the original trauma (i.e. abuse), but were less apparent in a dissimilar situation (e.g. academic pressures). Together, these studies were intended to demonstrate the importance of considering not only the situation-specific outcomes associated with particular coping efforts, but also that interpretation of adaptiveness requires consideration of the broad stressor context.

Study 1

Study 1 assessed whether women in abusive dating relationships demonstrated unique patterns of coping and evaluated the relations between coping efforts and depressive symptoms. Although the particular coping strategies individuals adopt might vary across situations (Tennen et al. 2000; DeLongis and Holtzman 2005), this does not belie the possibility that individuals may also have general response propensities stemming from either previous experiences or stable personality traits (Carver and Scheier 1994; Endler et al. 1994; Anisman and Matheson 2005). What is not clear is whether the depressive symptoms often evident among women in abusive dating relationships are linked to coping responses specific to the conflicts that arise in their relationships, or whether the depressive symptoms are related to differences in the general coping propensities women in abusive relationships employ in comparison to women who are not in such relationships. In study 1, it was hypothesized that

1. higher emotion-focused (particularly avoidant) and lower problem-focused coping would be evident among women in abusive dating relationships, relative to women who were not in abusive relationships; and

2. the presence of abuse in women's relationships would interact with whether coping was measured in response to stressors in general vs. a relationship conflict to predict depressive symptoms. Specifically, given the ongoing stressor situation abused women were experiencing, the relations between coping responses specific to a relationship conflict would be more strongly predictive of depressive symptoms than among women who were not experiencing this stressor. However, coping responses to stressors in general would be associated with depressive symptoms among both abused women and non-abused women.

Method

Participants and procedures

First year university women who were currently involved in heterosexual dating relationships for at least 1 month participated in a study concerning intimate relationships (N = 473; M age = 19.96, SD = 2.81 years). The majority of women (n = 409, 86.5%) self-identified as “going out with someone”, whereas 64 (13.5%) were living with an intimate other (but were not married). The average length of relationship was 18.72 months (SD = 13.52). Of the women reporting ethnic status, the majority was Euro-Caucasian (n = 339, 76.7%), with the remainder indicating that they were Black (n = 26, 5.9%), Asian (n = 31, 7.0%) or South Asian (n = 14, 3.2%), Hispanic (n = 12, 2.7%), Middle Eastern (n = 14, 3.2%), or Native Canadian (n = 6, 1.4%).

After providing written informed consent and background information, women completed self-report measures of depressive symptoms and the conflict tactics used in their intimate relationships. In addition, women were randomly assigned to respond to either a measure of their coping styles in response to stressors in general (n = 207) or their coping strategies in response to their most troublesome disagreement with their partner (n = 266). Women were debriefed and provided with contact information if they wished to seek help regarding problematic relationships or distress. All procedures were approved by the institutional ethics review board.

Measures

General and specific coping efforts

Although any of several multidimensional scales could have been employed, we used a 44-item scale (survey of coping profiles endorsed; SCOPE) that assessed 12 coping dimensions reflecting a broad spectrum of possible responses (Matheson and Anisman 2003). This scale differentiates between the coping strategies employed by individuals who do or do not demonstrate stress-related disorders (e.g. depression, anxiety, trauma symptoms) and between individuals who have experienced different stressor events (Matheson and Anisman 2003; Mantler et al. 2005; Matheson et al. 2005). The 12-factor structure was replicable (Mantler et al. 2005) and as reported in Matheson et al. (2005), the SCOPE subscales were highly related to those of another commonly used self-report measure (Carver et al. 1989).

Women were randomly assigned to one of two instructional sets, wherein they indicated whether they had demonstrated each of the behaviors as a way of dealing with stressors in recent weeks (general coping), or the strategies they had used in recent weeks to deal with the most troublesome disagreement they had with their partner (relationship specific coping). Each item was rated on a 5-point scale ranging from 0 (never) to 4 (almost always). Mean responses were calculated for each of the 12 subscales. As seen in Table I, as we previously reported (Matheson and Anisman 2003), the inter-item reliabilities were low for active distraction and passive resignation, but this is not unlike other coping inventories tapping into broad latent dimensions (e.g. Carver et al. 1989). In effect, these strategies subsume a range of behaviors that are valid indicators of particular strategies, but each behavior is not necessarily endorsed by the same individual (e.g. active distraction may be achieved by “shopping”, for some individuals, but by “going to the gym” for others).

Table I.  Inter-item reliabilities (Cronbach's αs) for coping responses in relation to stressors in general vs. specific to a recent relationship conflict.

Coping subscales	Example items	n of items	General	Specific
Problem-solving	Made plans to overcome your problems	4	0.72	0.69
Cognitive restructuring	Looked for how you can learn something out of your bad situation	4	0.77	0.79
Social support seeking	Asked others for help	4	0.83	0.89
Active distraction	Tried to do things which you typically enjoy	5	0.63	0.53
Humor	Told jokes about your situation	3	0.82	0.83
Rumination	Thought about your problems a lot	3	0.90	0.87
Emotional expression	Been very emotional in how you react, even to little things	4	0.80	0.77
Other-blame	Blamed others for creating your problems or making them worse	3	0.78	0.70
Self-blame	Blamed yourself for your problems	3	0.75	0.79
Emotional containment	Tried to act as if you were not upset	4	0.88	0.87
Passive resignation	Tried to just take whatever came your way	4	0.61	0.54
Cognitive avoidance	Tried to distract yourself from your troubles	3	0.78	0.83
Factor components
Problem-focused		12	0.81	0.85
Emotional-avoidant		11	0.77	0.79
Emotional-engagement		10	0.87	0.85

Abuse

Women completed the conflict tactics scale-revised (CTS-2; Straus et al. 1996), a commonly used self-report measure to assess the presence of various forms of abuse within an intimate relationship. Of interest were the tactics women's partners used toward them to resolve conflicts within the past month and in particular, the tactics involving physical assault, injury and psychological aggression. This time frame was viewed as appropriate given our focus on dating relationships. Responses were made on 6-point scales ranging from 0 (never) to 5 (more than 10 times in the past month). Internal reliabilities for physical assault (α = 0.85) and psychological aggression (α = 0.78) were high. Reliabilities were lower for injury (α = 0.64), probably due to the low incidence. Only 27 women (5.7%) reported any injuries resulting from their partners' behaviors; not surprisingly, the majority of these women (n = 21) reported that their partner had been physically abusive toward them in the past month, r = 0.72, p < 0.001. Because the response distributions for these tactics were highly skewed (sk = 5.76 and 6.66, respectively; no transformation substantially reduced this skew) and they were strongly correlated, women were categorized as in a physically abusive relationship if they reported any instance of physical assault or injury within the past month (n = 90; 19.0%).

The more common form of abuse was psychological aggression, with 39.5% (n = 187) of women scoring 3 (i.e. at least 3–5 instances within the past month) or greater (M = 3.45, SD = 0.58). The extent of psychological aggression was positively related to reports of physical assault, r = 0.53, p < 0.001. Although the distribution was mildly skewed (sk = 1.96), the sample size was sufficiently large that the effects of this skewness would be negligible (Tabachnick and Fidell 2007). Thus, psychological aggression was analyzed in its raw form as a continuous variable.

Depressive symptoms

The 21-item Beck Depression Inventory (Beck et al. 1961) was used to assess depressive symptoms. Respondents selected from sets of response choices that reflected increasing degrees of problem severity. Responses were summed to provide an index of depressive affect (α = 0.87).

Results

Factor structure of coping styles and strategies

In order to facilitate data analyses and minimize issues arising from multicollinearity, coping measures are often reduced to two or three superordinate dimensions that capture the stable variations of individuals' coping efforts (Endler and Parker 1994). To determine the reduced factor structure, a principle components analysis was conducted on women's responses to the 12 subscales assessing their general coping endorsements. Some have suggested that because the factor structure may vary, it ought to be re-evaluated within each context (Carver et al. 1989); yet it was desirable to assess the viability of a common factor structure in relation to general vs. specific coping efforts in order to provide a basis for comparison. Thus, the factor structure derived from the exploratory analysis of coping responses to general stressors was used as the basis for a subsequent confirmatory factor analysis conducted on the coping strategies women used in relation to a specific relationship conflict.

A parallel analysis (comparing derived eigenvalues to average eigenvalues generated from random data sets) (Zwick and Velicer 1986) indicated that the 12 coping responses to stressors in general could be reduced to three dimensions. The principle components analysis indicated that these three dimensions accounted for 47.5% of the total variance. As an oblique rotation resulted in components that were only mildly related (rs = − 0.02–0.23), a varimax rotation was used to identify the subscales comprising each component. Based on loadings of 0.45 or higher, the first component comprised 5 subscales reflecting problem-focused coping (problem-solving, cognitive restructuring, active distraction, social support seeking and humor). The second component reflected emotional-avoidant coping efforts, including self-blame, emotional containment, and passive resignation. Finally, the third factor comprised emotional-engagement coping behaviors, namely rumination, emotional expression and other-blame. The only subscale that did not load onto any of the three components was cognitive distraction and thus was not considered further.

A confirmatory factor analysis assessed whether this three-factor structure fit the coping strategies women indicated using in response to their most troublesome relationship conflict. Goodness of fit criteria for confirmatory models has been the subject of much discussion in recent years, with the cutoffs for fit criteria becoming increasingly strict to minimize concerns regarding model misspecification. Further, multiple fit indices are commonly reported, as it is possible that any given model may meet the criteria of some fit indices but not others. In the present investigation, goodness of fit was evaluated using four criteria: a nonsignificant chi-square goodness of fit (χ²) and the ratio of the observed chi-square to degrees of freedom (χ²/df) ≤ 3; Bentler's comparative fit index (CFI) ≥ 0.95; and the standardized root mean residual of approximation (RMSEA) ≤ 0.06 (Hu and Bentler 1999). Although the full model did not fit the data well, χ²(41) = 120.50, p < 0.001, χ²/df = 2.94, CFI = 0.88, RMSEA = 0.08, model-trimming involving the elimination of active distraction and humor from the problem-focused coping factor resulted in better model fit, χ²(24) = 60.9, p < 0.01 (a result typically found with large (>200) sample sizes), χ²/df = 2.53, CFI = 0.93, RMSEA = 0.07. In fact, these fit estimates were in line with those reported in psychometric evaluations of other empirically based measures of coping (e.g. Endler and Parker 1994). Although the best fit was achieved if the coping subscale reflecting other-blame was also removed from the emotional engagement factor, χ²(17) = 37.5, p < 0.01, χ²/df = 2.21, CFI = 0.96, RMSEA = 0.06, the decision was made to retain this subscale as blaming someone else for one's difficulties might be relevant to psychological well-being in the context of a relationship conflict, including abuse. Thus, the factor structure derived on the basis of women's coping responses to stressors in general was trimmed to exclude active distraction and humor and applied to coping strategies in response to an intimate relationship conflict for subsequent analyses.

Variations of coping as a function of abuse

It was expected that women's coping efforts would vary as a function of whether or not they were in an abusive relationship. To this end, a series of hierarchical regressions was conducted wherein physical abuse (present vs. absent), psychological aggression (continuous) and a dummy variable reflecting the specificity of the coping context (general vs. specific), as well as the interactions between each form of abuse and coping specificity were the predictor variables. The three coping dimensions, namely problem-focused, emotional-engagement and avoidant coping, served as the outcome variables. The specificity of the context was a significant predictor of problem-focused coping, R²_cha = 0.082, F(1,468) = 41.92, p < 0.001, emotional engagement, R²_cha = 0.045, F(1,468) = 22.74, p < 0.001 and avoidant coping, R²_cha = 0.014, F(1,468) = 5.84, p < 0.01. As seen in Table II, women reported greater use of all three coping methods when they were asked about stressors in general, relative to a specific relationship conflict. In addition, relative to non-abused women, those women in either physically or increasingly psychologically abusive relationships were more likely to endorse emotional-engagement (β = 0.12, 0.10, ps < 0.05, respectively), R²_cha = 0.037, F(2,469) = 8.92, p < 0.001 and avoidant coping efforts (β = 0.10, 0.15, ps < 0.05, respectively), R²_cha = 0.044, F(1,469) = 10.90, p < 0.001, but not problem-focused coping, F < 1. None of the interactions between abuse and the specificity of coping was significant, Fs < 1.

Table II.  Means (SDs) of coping responses to general stressors or specific relationship conflicts and depressive symptoms as a function of abuse in women's dating relationships in study 1.

	Physical assault		Psychological aggression
	Absent	Present	< 3 events	>3 events
Coping with general stressors	(n = 170)	(n = 36)	(n = 129)	(n = 77)
Problem-focused	2.62 (0.85)	2.58 (0.73)	2.64 (0.84)	2.56 (0.80)
Emotional-engagement	1.91 (0.80)	2.28 (0.81)	1.88 (0.79)	2.14 (0.82)
Emotional-avoidant	1.86 (0.88)	2.37 (0.79)	1.80 (0.92)	2.19 (0.78)
Coping with relationship conflict	(n = 212)	(n = 54)	(n = 156)	(n = 110)
Problem-focused	2.12 (0.71)	2.32 (0.64)	2.14 (0.70)	2.19 (0.70)
Emotional-engagement	1.57 (0.75)	1.91 (0.75)	1.51 (0.73)	1.83 (0.77)
Emotional-avoidant	1.72 (0.65)	2.01 (0.63)	1.69 (0.64)	1.90 (0.65)
Depressive symptoms	8.92 (7.08)	12.30 (8.10)	8.05 (6.27)	11.94 (8.32)

Predicting psychological well-being

It was hypothesized that women in abusive relationships would report greater depressive symptoms, but that certain coping methods might be associated with the diminution of such outcomes. Hierarchical regressions were conducted wherein involvement in an abusive relationship (psychological (standardized) or physical) was entered on the first step, followed by the dummy variable reflecting specificity of the coping context and standardized scores for the three coping dimensions (in separate analyses). The two-way interactions among these variables were entered in subsequent steps, followed by the three-way interaction terms.

As seen in Table III, being involved in physically or psychologically abusive dating relationships was associated with higher depressive symptoms relative to women who were not in such relationships, although only psychological aggression accounted for unique variance. Women's coping endorsements were further associated with depressive symptoms, in that lower problem-focused coping, greater emotional-engagement and avoidant coping were associated with higher depressive symptoms. Finally, as seen in Table III, depressive symptoms varied as a function of the interaction between problem-focused coping and involvement in an abusive relationship and particularly a psychologically aggressive one. Simple slope analyses (Aiken and West 1991) assessed the relation between problem-focused coping and depressive symptoms at one SD below (minimal) and above the mean (severe) of psychological aggression. Although problem-focused coping was mildly related to reduced depressive symptoms when psychological aggression was minimal, β = − 0.18, p < 0.05, this relation strengthened as the level of psychological aggression became more intense, β = − 0.53, p < 0.001. None of the relations between coping and depressive symptoms was moderated by whether coping was assessed in relation to stressors in general vs. a relationship conflict.

Table III.  Regression analyses of moderating effects of abuse on the relations between coping propensities (each dimension in a separate analysis) and depressive symptoms (outcome) in study 1.

	Coping dimension
	Problem-focused			Emotional-engagement			Emotional-avoidant
	r	β	R^2	r	β	R^2	r	β	R^2
Step 1: abuse			0.083***
Physical	0.18**	0.07		–	–	–	–	–	–
Psychological	0.28***	0.25***		–	–	–	–	–	–
Step 2:			0.002
Measure specificity	− 0.04	− 0.04		–	–	–	–	–	–
Step 3:			0.080***			0.140***			0.133***
Coping dimension	− 0.29***	− 0.30***		0.40***	0.39***		0.40***	0.38***
Step 4:
Two-way interactions with measure			0.009			0.007			0.004
Step 5: coping * abuse			0.023***			0.004			0.008
* Physical	0.15*	0.24
* Psychological	− 0.10*	− 0.17***
Step 6:
Three-way interactions			0.009			0.002			0.008

* p < 0.05; ** p < 0.01; *** p < 0.001. Note: Steps 1 and 2 were identical for all three regression analyses. Standardized regression coefficients were from the step of entry.

Discussion

In the present investigation, women's coping responses were found to reflect three broad factors, problem-focused efforts and two dimensions reflecting emotion-focused coping, namely engagement and avoidant strategies. Moreover, as in previous research (Carver and Scheier 1994; Endler et al. 1994), there appeared to be a congruence in the structures reflecting general coping propensities and those that were specific to relationship conflicts. It appears that our factor structure was relatively stable and was reminiscent of the three-factor structure commonly identified in previous research (e.g. Endler and Parker 1994). This said, the conceptual meaning of the three factors may not be entirely clear-cut and to some degree, the labeling (and subsequent interpretation) of the factors reflected an attempt to achieve consistency with prior research. For example, although the problem-focused dimension included problem-solving and social support seeking, it also subsumed cognitive restructuring (e.g. finding the silver lining), as well as humor and active distraction (e.g. taking time out to go to a movie). This combination of strategies typically falls under the rubric of problem-focused coping (e.g. Carver et al. 1989). However, in actuality, the latter efforts are not necessarily directly in the service of resolving the problem, but arguably might reflect emotion-focused efforts that allow the individual to “calm down”, which, in turn, might facilitate their ability to problem solve. Similarly, the labels distinguishing between emotional engagement vs. avoidant coping might perhaps be misleading (e.g. self-blame might not necessarily facilitate an ability to avoid confronting one's emotions or the situation). In this instance, once again, we opted for maximizing consistency in terms of our ability to interpret factors relative to other research and additionally, relied on our subjective understanding of the contexts in which these strategies were being assessed (i.e. in the context of abuse, self-blame might result in women not holding their partners responsible for their abusive behaviors). Indeed, labeling and defining core dimensions of coping, along with finding consensus regarding the behaviors that constitute particular coping efforts is a problem that is endemic to the field. This has led some researchers to call for a reconceptualization of the organizational structures and processes that group various coping behaviors in a manner that allows for flexibility and multiplicity in the functions and adaptive outcomes that various coping actions might reflect (Skinner et al. 2003).

As expected, depressive symptoms were more pronounced among women in abusive dating relationships. Just as in previous research pointing to the insidiousness of psychological abuse (Follingstad et al. 1990; Katz and Arias 1999; O'Leary 2001), it seemed that although physical and psychological aggression were both associated with greater symptomatology, depressive symptoms were primarily linked to the psychological aspect. In particular, when controlling for one another, only psychological aggression was related to depressive symptoms. Although it was possible that this finding reflected the low base rates of physical assault within this sample of women, there is ample reason to believe that psychological abuse may be a key contributor to psychological disturbances above and beyond the contribution of other forms of abuse. Indeed, much like the present investigation, once the effects of verbal assault were controlled, physically abused women were similar to non-abused women with respect to depressive symptoms and self-esteem (Orava et al. 1996). In contrast, when the effects of physical abuse were controlled for, psychological abuse remained significantly correlated with psychological symptoms, suggesting that it was, at least, as detrimental to well-being as physical abuse (Arias and Pape 1999; Katz et al. 2000; O'Leary 2001).

Consistent with findings that depressive symptoms were greater among women who employed emotion-focused coping efforts (Carver et al. 1989; Matheson and Anisman 2003), women in abusive dating relationships were more likely to endorse emotion-focused coping (engagement and avoidant). Although women were equally likely to use problem-focused coping, the negative relation between problem-focused coping and depressive affect was especially pronounced among psychologically abused women. These data are in line with the view that problem-focused coping efforts are typically adaptive, protecting against negative psychological outcomes and this would be particularly notable under conditions of a relatively severe ongoing stressor, such as abuse. What was surprising, however, was that this outcome was marked among women experiencing psychological, but not physical, abuse. Although speculative, it is possible that in the case of physically abusive relationships, attempts to directly negotiate the situation increase the threat of violent confrontation as it challenges the perpetrator's desire for power and control (Follingstad et al. 1999; Banister et al. 2003), thereby limiting its relation to diminished distress.

It will be recalled that although coping strategies may vary across situations, individuals may also bring with them particular coping propensities. In the present investigation, women indicated having used each of the three coping methods more often when their point of reference was stressors in general, relative to specific relationship conflicts. However, the predictive utility of the three coping dimensions in relation to depressive symptoms did not differ depending on whether coping was evaluated in response to stressors in general or in response to a relationship conflict. Because a between groups design was employed, wherein some women indicated their coping responses to stressors in general and others reported the strategies used to deal with a relationship conflict, the observed similarities were not due to a common response bias. Of course, it could be argued that when asked about coping in response to stressors in recent weeks, women in abusive relationships were primarily responding to their ongoing relationship conflicts. Hence, the coping efforts reported in response to stressors in general would necessarily have been similar to the strategies indicated in response to specific conflicts. However, given that the same pattern was found for women who were not in abusive relationships, it seems unlikely that finding similarities between the general and specific contexts was simply an artifact of abused women reporting on frequent relationship stressors. This said, it is possible that the commonalities between self-reported coping efforts in response to general vs. relationship specific stressors might have stemmed from women's perceptions concerning how they believed they would cope with a stressor and might not reflect their behaviors when a stressor was actually encountered (Schwartz et al. 1999). Ultimately, however, it will be necessary to assess coping responses across stressors within the same individuals in order to determine whether response tendencies across stressors are related to the coping propensities that individuals bring with them.

Study 2

Consistent with prior research examining the coping strategies adopted by women in violent domestic relationships (Waldrop and Resick 2004; Arriaga and Capezza 2005), in study 1, women in abusive dating relationships reported greater emotional-engagement and avoidant coping, both of which were associated with greater depressive symptoms. The question remains whether abused women simply have generally poorer coping skills (and hence are inclined to use these strategies across various stressor types), or whether the use of such apparently maladaptive coping efforts was relied upon because they actually provide some benefit, albeit transient, in the context of abuse.

Study 2 assessed whether the coping propensities of women in abusive dating relationships might be rooted in prior experiences, particularly as previous assault trauma has frequently been reported among women who experience abuse (Banyard et al. 2000; Humphrey and White 2000). In effect, general coping propensities may be relevant to understanding women's reactions to abuse because such methods were established as an effective response to previous experiences of a similar nature. If women's coping propensities evolved because they evoked some benefit in relation to prior experiences, it would seem reasonable to expect that the use of these coping efforts would evoke positive affective benefits in response to stressors with features reminiscent of these events (e.g. abuse), but not necessarily in response to other stressor situations. In the present study, participants were shown a video depicting a young woman discussing her abusive relationship, after which mood reactions were measured. To evaluate the contextual specificity of the benefits associated with particular coping propensities, mood reactions were also determined in response to a second video in which a young woman described her academic worries. It was hypothesized that

1. women in abusive dating relationships would be more likely to report prior assaultive experiences relative to non-abused women and that these experiences would account for the variations of coping between women in abusive and non-abusive dating relationships; and

2. women's responses to each of the two videos would vary as a function of their coping propensities and involvement in an abusive dating relationship. (a) In response to abuse-related reminder stimuli, there would be interactive effects between the presence of abuse in women's relationships and their coping orientation on mood reactivity. Specifically, women in abusive dating relationships who more strongly endorsed emotion-focused coping behaviors would report lower negative mood. For women in non-abusive relationships, coping was not expected to be related to mood states, as the abuse reminder cues would not represent a stressor relevant to their own experiences. (b) In response to stimuli unrelated to abuse (academic stressor), coping endorsements would demonstrate relations with mood states, irrespective of the presence of abuse in women's relationships. In this instance, affective benefits linked to greater emotion-focused coping would not be evident. Instead, greater emotion-focused and lower problem-focused coping would be associated with more negative mood.

Method

Participants and procedures

First year university women who indicated in a screening measure that they were currently in a heterosexual relationship were invited to participate in a study concerning intimate relationships (N = 80; M age = 19.3, SD = 1.47). Women identified themselves as going out with someone (n = 72, 90.0%), or living with an intimate other (n = 8, 10.0%). The average length of relationship was 16.5 months (SD = 11.3). The majority of women were Euro-Caucasian (n = 62, 81.6%), with the remainder indicating that they were Black (n = 3, 3.9%), Asian (n = 5, 6.5%), Hispanic (n = 3, 3.9%), Middle Eastern (n = 6, 2.6%), or Native Canadian (n = 1).

In an initial session, women completed questionnaires assessing coping responses to stressors in general, depressive symptoms, and the abusive nature of their relationships, with follow-up sessions occurring three and five months later (to attenuate possible carry-over effects associated with questions regarding abuse and the influence of the videos). In all sessions, women provided written informed consent, which included, in the first session, permission to recontact them for follow-up. As incentives, women were eligible to receive course credit (if they were in Introductory Psychology) and were included in a draw for $100 that was held at the end of each week of data collection. At the end of each session, women were debriefed regarding relationship conflicts and well-being, and provided contact information in the event that they wished to seek help associated with distress symptoms. A full debriefing was given following the final session. All procedures were approved by the institutional ethics review board.

Approximately 13.7% of participants withdrew by the final session of the study, leaving a total of 69 women. There were no significant differences between women who withdrew vs. those who continued in the study with respect to initial levels of satisfaction with the amount of time spent with their partners or the quality of time spent together. By the final session, 12 women had ended their relationships and of these, five had been in abusive relationships. All women were included in the analyses, as none had commenced a new relationship.

Stress reactivity task

In each of the two follow-up sessions, women watched a 7-min stressor video, counter-balanced to control for possible order effects. In fact, there were no differences in women's mood reactions to the two videos as a function of the order in which they were presented (ps > 0.05). One video presented cues associated with abuse. In increasing severity through the video, a woman described her boyfriend's impatience with her, including his explosive temper and his belittlement and humiliation of her at social events. She does not feel that she can talk to anyone because his behavior probably reflects something wrong with her. Finally, she admits to being frightened because her boyfriend has become physically rough with her, including hitting, shoving and kicking with minor injuries. The second video portrayed a stressor that was relevant to these women's day-to-day lives, but did not entail abuse. The young woman conveyed unexpected difficulty in meeting course requirements, juggling the demands of school, part-time work and social/family life, conflicts with others and concerns about her future. In both videos, the woman is sufficiently distressed to be seeking help.

Measures

In a preliminary session, women completed the CTS-2 (Straus et al. 1996) to assess the presence of various forms of abuse within women's dating relationships within the past month, along with the SCOPE to assess coping styles in recent months (Matheson and Anisman 2003). Using the same criteria employed in study 1 for defining an abusive relationship, 12 women (15.0%) reported experiencing at least one incident of physical assault or injury within the past month, whereas psychological aggression was reported by 27.5% (n = 22) of women (i.e. at least 3–5 instances within the past month). The extent of psychological aggression was positively related to reports of physical assault (r = 0.47, p < 0.001).

Trauma history

The Traumatic Life Events Questionnaire (Kubany et al. 2000) is a 23-item self-report measure of exposure to a broad spectrum of potentially traumatic events. Events are described in behaviorally descriptive terms (consistent with the DSM-IV stressor criterion A1). The frequency of occurrence of each event was indicated using a 7-point scale, ranging from 0 (never) to 6 (more than five times). When events were endorsed, respondents indicated whether they experienced intense fear, helplessness, or horror (PTSD stressor criterion A2 in the DSM-IV). Trauma history scores were determined in relation to four discrete categories (Breslau et al. 1999), including shocking event (e.g. car accident), death of a loved one, trauma to others (e.g. witnessing assault) and assault. The frequencies associated with events that women reported as causing fear, helplessness, or horror were summed. Of particular interest in the present study were events involving assault, including childhood physical or sexual abuse, physical assault by a stranger, rape, being stalked, or threats of violence from someone other than their intimate partner.

Affective reactions to stressor videos

The 20-item positive and negative affect schedule (PANAS; Watson et al. 1988) was used to measure affect following the stressor videos. Each adjective was rated on an 11-point scale ranging from 0 (not at all) to 10 (extremely) in terms of how descriptive it was of “the way you presently feel”. The PANAS has been critiqued due to the confounding of valence (e.g. happy vs. sad) and activation (e.g. alert vs. indifferent, respectively) properties (Russell and Carroll 1999). Given the potential importance of this distinction within an abusive context (e.g. women's negative mood state and inactivity might not co-occur), instead of forming the usual positive and negative subscales, the 20 items were subjected to a principle components analysis to permit the disentangling of valence and activation, if appropriate. Based on a parallel analysis, three components were extracted, accounting for 55.1% of the variance. As an oblique rotation indicated that the factors were not correlated (rs from 0.05 to − 0.16), a varimax rotation was conducted. Given the small sample size, each factor comprised only items loading greater than 0.60 (Tabachnick and Fidell 2007). The first factor comprised eight adjectives reflecting anxiety (afraid, distressed, guilty, jittery, upset, nervous, scared, ashamed) (α = 0.90), whereas the second factor comprised five items reflecting a sense of positive agency (active, determined, enthusiastic, proud, strong) (α = 0.78). Finally, two adjectives, hostile and irritable, formed their own factor (r = 0.50).

Results

Coping styles and previous trauma

It was expected that women in abusive relationships would be more likely to have encountered previous assaultive trauma experiences and that these would give rise to distinctive coping orientations. Logistic regression analysis with the four types of previous trauma events entered simultaneously as predictor variables, indicated that only traumatic assault experiences were uniquely associated with a greater likelihood of physical abuse in women's current relationships, β = 0.56, p < 0.05. A comparable multiple regression analysis predicting psychological aggression also indicated a unique relation with women's assault experiences, β = 0.24, p < 0.05. Approximately two-thirds of women in physically abusive relationships (n = 8/12; 66.7%) reported previous assault (including threats of violence from someone other than their partner, childhood physical abuse and stalking), as did almost half of the women who reported three or more psychologically aggressive actions from their partners in the past month (n = 10/22; 45.5%). In contrast, less than a quarter of women who were not in physically (16/68; 23.5%) or psychologically aggressive relationships (14/58; 24.1%) indicated such experiences. Women did not differentially report trauma involving experiences of a shocking or unexpected event (n = 36, 45.0%), the unexpected death of a loved one (n = 40, 50.0%), or witnessing something traumatic happen to someone close to them (n = 33, 41.3%).

Zero-order correlations indicated that previous traumatic assault experiences were uniquely related to higher levels of endorsing avoidant coping, r = 0.24, p < 0.05 and marginally related to emotional engagement coping, r = 0.20, p = 0.07, but not problem-focused coping, r = 0.03, ns. Likewise, physical abuse in women's current dating relationships was related to greater endorsements of avoidant coping, r = 0.28, p < 0.01, whereas psychological aggression was not, r = 0.11, ns; neither form of abuse was associated with problem-focused or emotional-engagement coping behaviors. Importantly, when the relations between avoidant coping and each of traumatic assault history, pr = 0.16, ns and current physical abuse, pr = 0.21, p = 0.07, were assessed controlling for one another, neither remained significant. Evidently, the avoidant coping behaviors endorsed by women in physically abusive dating relationships were at least partially reflective of their prior assault experiences.

Finally, both physical assault, r = 0.46, p < 0.001, and psychological aggression, r = 0.41, p < 0.001, were associated with greater depressive symptoms, as were prior assault experiences, r = 0.42, p < 0.001. Notably, the relations between current dating abuse and symptoms were not accounted for by prior assault experiences, as these relations remained significant (rs = 0.29, 0.33, p < 0.01, respectively) even after controlling for such experiences.

Mood reactions to stressor cues

It was argued that emotion-focused coping might be associated with reduced distress when women in abusive relationships were exposed to reminders of abuse. Hierarchical regressions were employed to assess the interactive effects of abuse and coping on each dimension of self-reported mood (anxiety, hostility and positive agency) in response to the video stressors depicting abuse or academic pressures. In each analysis, the extent of physical and psychological (standardized) abuse was entered on the first step, coping propensities (standardized) on the second (in three separate regressions, given sample size) and the two-way interactions between each form of abuse and coping on the third. Omnibus analyses were conducted to assess whether effects were moderated by the video being watched (i.e. abuse vs. academic pressures). For the sake of reporting simplicity, and given a priori hypotheses regarding reactions to the two videos, only the simple effects analyses for each of the two videos are reported in full. Where different patterns were implicated in response to the two videos, the omnibus findings confirming the moderating effects of the video are provided in a summary at the end of the Results section.

Reactions to abuse stressor video

As seen in Table IV, women's feelings of anxiety in response to the abuse video were greater among those who experienced physical abuse, but were unrelated to levels of psychological aggression. Further, levels of anxiety were not predicted by any of the three coping dimensions (over and above abuse), nor did coping interact with abuse to predict women's self-reported anxiety in response to this video.

Table IV.  Regression analyses assessing moderating effects of coping propensities on the relations between abuse and affective reactivity to abuse stressor video in study 2.

	Anxiety			Hostility			Agency
	r	β	R^2	r	β	R^2	r	β	R^2
Step 1: abuse			0.111*			0.027			0.015
Physical	0.25**	0.40**		0.01	− 0.10		0.10	0.06
Psychological	− 0.04	− 0.26*		0.14	0.20		0.11	0.08
Model I
Step 2: coping			0.015			0.020			0.002
Problem-focused	0.12	0.12		0.13	0.14		0.07	0.04
Step 3: coping * abuse			0.015			0.043			0.105*
* Physical							− 0.11	0.13
* Psychological							− 0.31**	− 0.41**
Model II
Step 2: coping			0.006			0.005			0.000
Engagement	0.15	0.08		− 0.09	− 0.07		0.03	0.02
Step 3: coping * abuse			0.020			0.042			0.129**
* Physical							0.21	− 0.00
* Psychological							0.36*	0.36*
Model III
Step 2: coping			0.057*			0.021			0.004
Avoidant	0.32**	0.25*		0.12	0.15		− 0.04	− 0.06
Step 3: coping * abuse			0.043			0.091*			0.114**
* Physical				− 0.22*	− 0.39*		0.22*	0.23
* Psychological				− 0.18	− 0.05		0.33**	0.23^+

⁺p < 0.10; * p < 0.05; ** p < 0.01. Note: Standardized regression coefficients were from the step of entry.

Neither problem-focused nor emotional-engagement coping influenced levels of hostility, but the interaction between physical abuse and avoidant coping was significant (Table IV). Follow-up simple slope analyses indicated that among women not in physically abusive relationships, avoidant coping was mildly associated with greater feelings of hostility following the abuse video, β = 0.26, p < 0.05, whereas women who were in abusive relationships and favored an avoidant coping style were remarkably less likely to report such hostility, β = − 0.79, p < 0.001.

Women's self-reported feelings of positive agency following the abuse video varied as a function of the interaction between their abusive experiences and coping. As seen in Table IV, the relations between all three of the coping dimensions and levels of positive agency were moderated by the extent of psychological aggression women experienced. Follow-up simple slope analyses (at one SD below (minimal) and above (severe) the mean of psychological aggression) indicated that when minimal psychological aggression was reported, greater problem-focused coping was associated with higher levels of positive agency following the abusive video (Figure 1, top; β = 0.32, p < 0.05); this relation was attenuated as women experienced more severe psychological aggression, β = − 0.23, ns. In contrast, emotional-engagement (Figure 1, middle; β = − 0.34, p < 0.05) and avoidant coping (Figure 1, bottom; β = − 0.35, p < 0.05) were related to lower feelings of agency when psychological aggression was minimal. However, when women experienced higher levels of psychological aggression, emotional-engagement, β = 0.33, p < 0.05 and avoidant coping, β = 0.38, p < 0.05, were associated with a stronger sense of positive agency.

Figure 1 Relations between problem-focused (top), emotional-engagement (middle) and avoidant (bottom) coping styles and self-reported feelings of positive agency in response to the abuse stressor video among women reporting varying degrees of psychological aggression in their dating relationships, ranging from minimal (one SD below the mean) to severe (one SD above the mean).

Reactions to academic stressor video

Levels of anxiety in response to the academic video, as in the case of the abuse video, were related to the presence of physical abuse in women's relationships (Table V). In addition, feelings of anxiety in response to the academic stressor video were predicted by the interactions between abuse, and particularly psychological aggression, and all three of the coping dimensions. Follow-up simple slope analyses indicated that none of the coping dimensions was related to anxiety when psychological aggression was minimal (Figure 2). However, as the intensity of psychological aggression increased, anxiety was greater with lower problem-focused (Figure 2, top; β = − 0.49, p < 0.01), and higher emotional-engagement (Figure 2, middle; β = 0.57, p < 0.001) and avoidant coping styles (Figure 2, bottom; β = 0.79, p < 0.001).

Table V.  Regression analyses assessing moderating effects of coping propensities on the relations between abuse and affective reactivity to an academic stressor video in study 2.

	Anxiety			Hostility			Agency
	R	β	R^2	r	β	R^2	r	β	R^2
Step 1: abuse			0.293***			0.177**			0.036
Physical	0.54***	0.57***		0.41***	0.48***		0.09	0.20
Psychological	0.27*	− 0.06		0.14	− 0.13		− 0.09	− 0.21
Model I
Step 2: coping			0.069**			0.027			0.037
Problem-focused	− 0.28**	− 0.26**		− 0.17	− 0.16		0.20*	0.19
Step 3: coping * abuse			0.140***			0.063			0.017
* Physical	− 0.45***	− 0.18
* Psychological	− 0.29**	− 0.26*
Model II
Step 2: coping			0.043*			0.022			0.001
Engagement	0.32**	0.21*		0.24*	0.15		0.07	0.03
Step 3: coping * abuse			0.107**			0.061			0.005
* Physical	0.47***	0.02
* Psychological	0.33*	0.32*
Model III
Step 2: coping			0.041*			0.048*			0.022
Avoidant	0.36***	0.22*		0.35***	0.23*		− 0.10	− 0.16
Step 3: coping * abuse			0.128***			0.037			0.009
* Physical	0.52***	0.02
* Psychological	0.32**	0.36***

*p < 0.05; ** p < 0.01; *** p < 0.001. Note: Standardized regression coefficients were from the step of entry.

Figure 2 Relations between problem-focused (top), emotional-engagement (middle) and avoidant (bottom) coping styles and self-reported feelings of anxiety in response to the academic stressor video among women reporting varying degrees of psychological aggression in their dating relationships, ranging from minimal (one SD below the mean) to severe (one SD above the mean).

Self-reported hostility following the academic stressor video was related to the presence of physical abuse in women's relationships, but not with psychological aggression (Table V). Only avoidant coping was directly related to higher levels of hostility in response to the academic stressor, but this form of coping did not interact with abuse. Finally, comparable regression analyses conducted to predict levels of positive agency following the academic stressor video revealed no significant relations with either abuse or coping.

Summary

Irrespective of their coping propensities, women in physically abusive relationships reported greater anxiety in response to both videos than did women not in such relationships. The only instance in which coping influenced physically abused women's mood was in regard to greater avoidant coping being associated with lower feelings of hostility following the abuse cues. Indeed, the omnibus analysis indicated that this effect was unique to the abuse video, as the three-way interaction between video (abuse vs. academic pressures; within subjects), physical abuse (present vs. absent) and avoidant coping was significant, R²_cha = 0.158, F(1,65) = 5.17, p < 0.001.

Although women in psychologically aggressive relationships did not report overall differences of mood in relation to either video, the more severe the psychological aggression women experienced, the more their coping propensities were related to mood reactions. Specifically, consistent with our hypotheses, with greater levels of psychological aggression in their dating relationships, women were more likely to indicate a sense of positive agency if they employed either emotion-focused engagement or avoidant coping. The increased agency was not linked to a greater inclination toward problem solving. This pattern of results was unique to the abuse video, in that omnibus analyses indicated that the three-way interactions between video, psychological aggression and each of emotional-engagement, R²_cha = 0.218, F(1,65) = 8.09, p < 0.001, avoidant coping, R²_cha = 0.139, F(1,65) = 10.48, p < 0.01, and problem-focused coping, R²_cha = 0.120, F(1,65) = 8.85, p < 0.01, were significant in relation to positive agency.

In contrast to abuse cues, in response to the academic video, women in psychologically abusive relationships who endorsed greater problem-focused and lower emotion-focused coping reported lower anxiety. Once again, the uniqueness of this pattern of relations to the academic video context was supported by omnibus analyses indicating significant three-way interactions between video, psychological aggression and each of emotional engagement, R²_cha = 0.159, F(1,65) = 12.30, p < 0.001, avoidant coping, R²_cha = 0.179, F(1,65) = 14.15, p < 0.001, and problem-focused coping, R²_cha = 0.087, F(1,65) = 6.23, p < 0.05.

Discussion

As previously reported (Banyard et al. 2000; Humphrey and White 2000; Wolfe et al. 2004), women in abusive dating relationships more frequently reported previous assaultive trauma (e.g. childhood abuse, threat of violence, being stalked). It is unlikely that this relation reflected a reporting bias, as they did not indicate experiencing other forms of trauma more frequently (i.e. shock, loss of a loved one, witnessing trauma to others). Of course, the possibility cannot be dismissed that women who had previously encountered abuse might be more sensitive or reactive to later abuse, and hence more likely to report such events when they occurred.

Previous assault trauma may set in motion a cascade of changes related to belief systems, perceptions of self and others, and coping orientations, thereby increasing the likelihood of subsequently encountering or not leaving an abusive situation (VanderKolk and Fisler 1994; Banyard et al. 2000). In fact, early life experiences and prior trauma have been found to influence how individuals coped with subsequent stressors (VanderKolk and Fisler 1994; Runtz and Schallow 1997; Leitenberg et al. 2004; Goldenberg and Matheson 2005; Matheson et al. 2005). In the present study, women who reported previous assault were more likely to endorse avoidant coping, but not problem-focused or emotional-engagement coping behaviors. As the relations between current abuse and avoidant coping were nonsignificant when previous assault experiences were controlled, the avoidant coping of women currently in abusive relationships may stem, at least in part, from prior trauma.

Given that women's coping propensities were linked to their stressor history, it was expected that the potential benefits derived from their coping efforts might be evident in response to reminder cues of current stressors that had commonalities with such events. Indeed, when shown a video conveying abuse, women currently in physically abusive relationships who endorsed avoidant coping reported lower levels of hostility. Women in psychologically abusive relationships also demonstrated affective benefits, in that those who endorsed emotion-focused coping (either avoidant or engagement) expressed higher levels of positive agency in response to this video. As this outcome was not evident in relation to problem-focused coping, it is possible that an emotion-focused orientation may have served to diminish negative or facilitate positive affect among women in abusive relationships, without actually being linked to strategies to alter their situations (see also Pape and Arias 1995).

When shown a video regarding academic pressures, women who experienced abuse reported greater anxiety than did non-abused women, possibly emanating from the fact that their academic performance was indeed suffering. Importantly, in response to this stressor, the coping propensities of abused women followed a typical pattern, in that higher problem-focused and lower emotion-focused coping were associated with lower anxiety. Evidently, in response to stressor cues that were not directly linked to their abusive situation or prior trauma, coping predicted affective reactions precisely in the manner that would normally be viewed as adaptive.

It should be noted that the sample size of women who reported abuse in study 2 was relatively small, particularly with respect to the prevalence and severity of physical abuse. The primary issue regarding small sample sizes is the power to detect meaningful differences as significant; in the present study, meaningful effect sizes (i.e. R² = 0.13 based on Cohen 1992) were found to be significant and indeed, relatively small effects were also significant. A second issue with small sample sizes may be the generalizability of the findings. In this regard, the rates of abuse evident in the present study were consistent with prevalence rates of dating abuse in women's relationships reported in previous research (e.g. Kasian and Painter 1993; Katz and Arias 1999), suggesting that, at least in this regard, the sample was representative of the relationships of college-aged women.

General discussion

As previously observed, a chronic stressor, in this case an abusive dating relationship, was associated with greater depressive symptoms (Ackard et al. 2003). These symptoms were generally at subsyndromal levels and only some women would have met the criteria for clinically relevant pathology. Nevertheless, the limited symptom severity should not be misconstrued to suggest that the psychological problems experienced were not significant, as early signs of distress may be harbingers of more profound disturbances.

Ordinarily, depressive symptoms have been associated with the endorsement of emotion-focused coping strategies at the expense of problem-focused coping and this pattern of relations was also apparent in the present investigation. Although women in abusive relationships reported higher depressive symptoms, the relations between symptoms and coping orientations did not vary as a function of abuse. Further, despite the context-specific nature of coping (Tennen et al. 2000; DeLongis and Holtzman 2005), the relations between coping and depressive symptoms did not vary as a function of whether coping was assessed in relation to stressors in general vs. a specific relationship conflict. Taken together, these data are consistent with the view that coping orientations may have broad effects across contexts (Carver et al. 1989). However, it ought to be considered that depressive affect may have increased reporting of negative experiences and that negative affect was also the basis for women's higher endorsements of emotion-focused coping, irrespective of the stressor context.

The relations between experiencing abuse in women's relationships and their coping orientations were, in actuality, relatively modest (i.e. reflected by small to medium effect sizes, Cohen 1992). In this regard, it is possible that the coping options available to women in abusive dating relationships were not especially limited relative to women who were not in such relationships. In line with this, among women in abusive relationships, the use of problem-focused coping was more strongly linked to lower depressive affect, as well as to reduced anxiety in response to reminder cues of a stressor unrelated to their abusive situation (i.e. academic pressures). Thus, it appears that abused women did not lack the capacity to cope effectively and indeed demonstrated an adaptive pattern in response to stressors unrelated to their abusive situation. However, in considering abuse, it is important to distinguish between psychological and physical components and in fact, in the present investigation coping appeared to have particularly marked consequences among women experiencing psychological, rather than physical abuse. It is possible that the role of coping is influenced by the ambiguity of a situation, being more closely aligned with affective outcomes when the event is open to interpretation.

It will be recalled that affective outcomes might appear to be beneficial in a given situation, but may undermine well-being in a broader context. In this regard, among women in abusive relationships, avoidant coping has been associated with positive affective benefits (Pape and Arias 1995; Rosen 1996). Likewise, in the present study, when shown a video portraying abuse, women in physically abusive relationships who endorsed avoidant coping reported lower levels of hostility and women in psychologically aggressive relationships reported higher positive agency. It is unlikely that these latter mood benefits reflected a form of denial, as the positive affect reported by women in psychologically abusive relationships was also associated with emotional-engagement coping behaviors that included other-blame. Moreover, the feelings of hostility reported by women in psychologically aggressive relationships were not diminished by their coping propensities. Thus, rather than minimizing the perceived severity of the abuse, the ability of psychologically abused women to derive positive agency might be indicative of active efforts to manage an emotionally charged situation (Stanton and Franz 1999). In this sense, it is tempting to conclude that the emotion-focused coping of women in psychologically abusive relationships provided emotional benefits and hence ought to be regarded as adaptive.

This conclusion must be tempered, however, as the positive agency expressed by women in such relationships occurred in the absence of a problem-solving orientation. Rather than confronting their abusers, the positive affect experienced by these women might be linked to reinterpreting their partners' abusive behaviors in positive terms (Herbert et al. 1991; Arriaga 2002); only when this positive framing was absent was the stability of the relationship undermined (Katz et al. 2000). Thus, although the capacity to derive positive affect or meaning within a stressful situation is typically viewed as adaptive (Davis et al. 1998), interpreting abusive relationships as positive is clearly dysfunctional.

As noted earlier, the specific mood benefits derived by women in physically or psychologically abusive relationships differed from one another. Albeit speculative, these differential responses may be related to the nature of the abuse experience conveyed in the video. Although the goal of these videos was to present reminders to women of their own abusive situations, they might also have served a social comparison function. Specifically, relative to the woman in the video who reported intense psychological abuse, coupled with physical abuse, the psychologically abused participants in the present study might have perceived their own situations as relatively benign and under their own control (“if he was as bad as the guy in the video, I'd leave”). In this regard, information that provides a basis for social comparison might be an important determinant of women's evaluations of their own relationships (Pipes and LeBov-Keeler 1997) and hence might dictate coping outcomes.

Taken together, the present findings suggest that although the emotion-focused coping orientations of women in abusive dating relationships were linked to depressive symptoms, they also facilitated women's ability to derive a “positive spin”. However, despite enabling women to diminish distress and enhance a sense of positive agency, avoidant coping responses in relation to abuse could be viewed as maladaptive, as these affective outcomes were not associated with greater problem-focused coping, hence leaving women vulnerable to further abuse. As well, the coping pattern that was associated with enhanced mood in response to the abuse video was the same coping pattern that was related to poorer affective responses to other stressors (i.e. academic pressures) and greater depressive symptoms. Thus, although coping adaptiveness ought to be evaluated in a situation-specific manner, a lack of consideration of the broader stressor context may obviate the limits to the apparent adaptiveness of these strategies.

In summary, women may enter their abusive relationships in an already vulnerable state both in terms of depressive symptoms, as well as less functional strategies for coping. Yet, in a context unrelated to abuse, these women appeared to employ coping efforts in a manner that appropriately reduced distress. Thus, their responses in relation to abuse may be uniquely problematic and might reflect defensive biases that enable women to experience a sense of control and justify maintaining the relationship. Of course, this does not imply that women should be blamed for their continued involvement in abusive dating relationships. Rather, these findings suggest that these women are using their coping skills to adapt to and endure highly aversive interpersonal situations. Given that this pattern of responses is adopted in the dating stages of the relationship, these women's worldviews may limit their ability to appraise and recognize their situation for what it is. Thus, attempts to establish interventions for protecting young women from such relationships might be best served if they targeted the defensive biases they rely on to tolerate or justify their situations.

Acknowledgements

We thank Irina Goldenberg and Alia Offman for their role in the data collection. This research was funded by the Canadian Institutes of Health Research. Hymie Anisman holds a Canada Research Chair in Behavioral Neuroscience.

Notes

^† Given this finding, it was noteworthy that participants' grade point averages at the end of the year were accessed (with permission) and even after controlling for their self-reported incoming grades (which did not vary as a function of abuse, F < 1), relative to women who were not in abusive relationships (M = 7.00 “B-”, SD = 2.38), those in physically abusive relationships (M = 4.95 “C”, SD = 3.04) had significantly lower grades at the end of their first year at university, F(1,71) = 7.45, p < 0.01, η² = 0.095.