|Home | About | Journals | Submit | Contact Us | Français|
This study examined several potential correlates of engagement and disengagement coping, including abuse-related factors, socioeconomic and social coping resources, and childhood trauma variables among a sample of battered women (N = 388). Relationship abuse frequency, particularly psychological aggression, and peritraumatic dissociation were the strongest positive predictors of the use of disengagement coping. Social coping resources, including tangible support and appraisals of social support and belonging, were associated with higher engagement coping and lower disengagement coping. A positive association was also found between interparental domestic violence and disengagement coping, and negative associations were found between both childhood physical and sexual abuse and engagement coping. Results suggest that coping strategies used by battered women are multidetermined and deserve further exploration.
The abuse of women in intimate relationships is a serious national public health problem. According to estimates derived from the National Violence Against Women Survey, approximately 1.5 million women are physically assaulted and/or raped by an intimate male partner in the United States annually, with a total of 5.3 million assaults per year (Centers for Disease Control and Prevention, 2003; Tjaden & Thoennes, 1998). In addition to fatal and nonfatal injuries, relationship abuse leads to physical health and mental health problems, decreased work productivity, and substantial legal and health care costs (Campbell, 2002; Centers for Disease Control and Prevention, 2003; Edleson & Tolman, 1992; Koss, 1990).
Because of the many deleterious impacts of relationship abuse, investigations into coping processes among battered women take on heightened importance. Effective coping behaviors and the recovery environment are critical for battered women’s positive adjustment (Carlson, 1997; Sullivan & Bybee, 1999). Most of the research in this area has focused on the impact of coping on mental health outcomes and has distinguished forms of coping that involves taking active steps to manage the abuse (engagement coping) from attempts at avoidance or tension reduction through escapist thoughts and behavior (disengagement coping). Although some inconsistencies have been reported across studies and contextual factors may limit the range and effectiveness of possible coping behaviors, evidence generally suggests that engagement coping is associated with more positive health outcomes and that disengagement coping is associated with more negative outcomes (for a review, see Waldrop & Resick, 2004). Such differences between these forms of coping are also consistent with a larger body of research within the general coping literature (Aldwin & Revenson, 1987; Compas, Malcarne, & Fondacaro, 1988; Holahan & Moos, 1990, 1991).
In addition to examining the effects of coping on health, it is also important to examine factors that may be associated with potentially effective and ineffective forms of coping since coping may differ widely across individuals and in response to various patterns of abuse (Mitchell & Hodson, 1986). Such work may ultimately assist in more fully explicating the processes of coping among this population. Therefore, the purpose of this study was to examine several potential correlates of engagement and disengagement coping. Although no published study has comprehensively examined the correlates of the coping variables of interest among battered women, some work within the general coping and domestic violence literatures suggest the potential importance of (a) abuse-related factors, (b) socioeconomic and social coping resources, and (c) childhood trauma (Holahan & Moos, 1987; Mitchell & Hodson, 1986; Waldrop & Resick, 2004).
There is evidence from samples of battered women that higher levels of abuse are positively associated with the use of both engagement (Dutton, Goodman, & Bennett, 1999; Goodman, Bennett, & Dutton, 1999; Jacobson, Gottman, Gortner, Berns, & Shortt, 1996; Marshall, 1996; Strube, 1988) and disengagement (Mitchell & Hodson, 1983) forms of coping. However, validated measures of these forms of coping have rarely been used, and few studies have assessed the impacts of psychological aggression or intimate partner sexual aggression in addition to physical assault. It is also important to consider abuse recency since battered women may be more likely to use higher levels of both forms of coping in the aftermath of an episode of abuse (Goodman, Dutton, Weinfurt, & Cook, 2003; Walker, 1989, 1991). Relatedly, it is possible that dissociative experiences during and immediately following the trauma would be associated with the subsequent use of disengagement forms of coping since both peritraumatic dissociation and disengagement coping serve to protect the individual from experiencing aspects of the trauma. The length of the abusive relationship may also be related to more disengagement coping since battered women who remain with their abuser often develop a sense of helplessness and powerlessness (Clements & Sawhney, 2000; Follingstad, Neckerman, & Vormbrock, 1988).
Considerable coping research suggests that those with higher levels of personal and environmental resources are more likely to use engagement coping and less likely to use disengagement coping (Billings & Moos, 1981, 1982; Cronkite & Moos, 1984; Holahan & Moos, 1987; Pearlin & Schooler, 1978). In studies of battered women, greater economic resources and financial independence have similarly been associated with engagement strategies, particularly relationship termination behaviors (Rusbult & Martz, 1995; Strube & Barbour, 1983). Greater economic resources are also associated with higher levels of social support, and the size and responsiveness of battered women’s support network are other important determinants of choice of coping strategy (Mitchell & Hodson, 1983). Women in abusive relationships with a supportive and responsive social network may feel more empowered to use engagement strategies and may benefit from higher levels of resource availability (Dutton, Hohnecker, Halle, & Burghardt, 1994; Waldrop & Resick, 2004). One study found that battered women reporting higher levels of tangible support were more likely to cooperate with the criminal prosecution of the perpetrator (Goodman et al., 1999).
One previous study has shown that childhood physical abuse and witnessing inter-parental violence moderated the relationship between partner violence severity and coping behaviors, such that women with family of origin violence histories used more disengagement coping and less engagement coping in response to increasingly severe physical assault (Mitchell & Hodson, 1986). We are not aware of any previous study that has examined the impact of childhood sexual trauma on coping with relationship abuse in adulthood. The experience of uncontrollable violence in childhood is likely associated with a heavy reliance on disengagement coping behaviors to escape the trauma, and the use of these coping strategies may carry over into adulthood (Leitenberg, Gibson, & Novy, 2004). Mental health problems such as posttraumatic stress disorder (PTSD) that often accompany childhood trauma may also help account for the association between early trauma and disengagement coping (Gibson & Leitenberg, 2001).
To recapitulate, the current study examined several potential correlates of engagement and disengagement coping, with a focus on abuse-related factors, socioeconomic and social coping resources, and childhood trauma. Hypotheses included (a) that more frequent and recent abuse would be associated with higher levels of both engagement and disengagement coping, (b) that peritraumatic dissociation and longer relationship length would be associated with more disengagement coping, (c) that higher levels of socioeconomic and coping resources would be positively associated with engagement coping and negatively associated with disengagement coping, and (d) that exposure to childhood violence would be positively associated with disengagement coping and negatively associated with engagement coping.
Participants were 388 help-seeking women recruited from battered women’s shelters and nonresidential community agencies serving battered women. Inclusion criteria were as follows: (a) participants reported being involved in an intimate relationship with a male perpetrator for 3 months or longer during the previous year, (b) participants’ most recent episode of physical assault occurred more than 2 weeks but less than 6 months before study participation, and (c) participants reported experiencing at least two severe or four minor acts of physical assault on the Conflict Tactics Scale (CTS; Straus, 1979) within the previous year or a combination of at least one severe and two minor acts of physical assault. A total of 426 eligible women completed the initial screening for this investigation. Following this screening, 12 women declined to participate. An additional 12 participants were terminated from the study because of factors compromising the validity of their data (e.g., apparent drug or alcohol intoxication, the presence of psychotic symptoms), and 14 assessments were judged to be invalid after the completion of study procedures (e.g., the participant completed questionnaires quicker than possible, provided the same response option for all items, or reported the absence of physical assault following their initial screen).
Sixty-five percent of participants were African American, 29% were White, 2% were Latina, and 1% were Native American. Three percent of the participants classified their race or ethnicity as “other.” The average age of participants at the time of the study was 34.7 years old (SD = 8.3 years). During their study participation, 56% of the participants were residing in a battered women’s shelter, and 44% were seeking help from a nonresidential agency. Among the latter group, 20% continued to reside with their abusive partner.
The following measures were examined in the current investigation and drawn from a larger battery of instruments administered to study participants.
Intimate partner physical assault and sexual aggression were measured using the Conflict Tactics Scale–2 (CTS2; Straus, Hamby, Boney-McCoy, & Sugarman, 1996). The present investigation examined the 12-item Physical Assault subscale and two items based on the CTS2 sexual coercion subscale designed to assess (a) the use of threats or force to coerce the victim to engage in oral or anal sex and (b) the use of threats or force to coerce the victim to engage in vaginal intercourse. Participants indicated the frequency that each abusive behavior occurred during the previous 12 months on a scale ranging from 0 (never) to 6 (more than 20 times). Each item was recoded to reflect the actual frequency of the behavior (i.e., three to five times equals a score of 4), and total abuse scores were computed by summing the recoded frequency scores for all items in each subscale (see Straus, 1990). Several studies have documented the construct validity and internal consistency reliability of the CTS2 (Connelly, Newton, & Aarons, 2005; Newton, Connelly, & Landsverk, 2001).
Intimate partner psychological aggression was assessed with the 14-item version of the Psychological Maltreatment of Women Inventory (PMWI; Tolman, 1989, 1999). Participants responded to each item based on the frequency that each abusive behavior occurred in the past year on a scale from 1 (never) to 5 (very frequently). The PMWI consists of two subscales capturing Emotional/Verbal abuse and Dominance/Isolation behaviors. Items in both of these subscales were summed to arrive at a total score. Several researchers have demonstrated the reliability and validity of the PMWI and its abbreviated version and have upheld the two-factor structure of the measure (Dutton & Hemphill, 1992; Tolman, 1999).
Abusive relationship characteristics were assessed via interview and included two factors pertaining to participants’ most recent abusive relationship: the length of the abusive relationship (how long have you and [partner’s name] been involved in a romantic relationship?) and the recency of the abuse (when was the last episode of physical force or threat of physical force by [partner’s name]?).
Peritraumatic dissociation was assessed with an eight-item questionnaire adapted from the Peritraumatic Dissociative Experiences Questionnaire (PDEQ; Marmar et al., 1994). The PDEQ retrospectively assesses dissociative experiences at the time of the trauma, such as amnesia, altered time perceptions, and depersonalization. Respondents indicated how often they engaged in each dissociative behavior at the time of the worst episode of physical assault on a scale from 0 (none of the time) to 4 (all of the time), and these scores were summed. Several investigations attest to the measure’s reliability and construct validity (Marmar, Weiss, & Metzler, 1997; Marmar et al, 1994; Marshall, Orlando, Jaycox, Foy, & Belzberg, 2002). The internal consistency reliability estimate for the PDEQ in the present study was.80.
Socioeconomic resources were examined with two interview questions about personal income and years of formal education. For personal income, respondents reported which of six income ranges best applied to their earnings. Responses were then recoded to reflect estimated income (i.e., $20,001 to $30,000 equals a score of $25,000).
Perceived social support was assessed using the three subscales of the brief version of the Interpersonal Support Evaluation List (ISEL; Cohen & Hoberman, 1983). The Tangible Support subscale measures the perceived availability of material resources. Appraisal of Support reflects respondents’ perceptions of the availability of people to talk with about personal problems. The Belonging subscale is designed to measure the perceived availability of others with whom the respondent can participate in social activities. Responses to the ISEL are given on a scale from 1 (definitely false) to 4 (definitely, true), and subscale items are summed. The ISEL has strong construct validity and internal consistency reliability (Cohen & Hoberman, 1983). Internal consistency reliability estimates for the ISEL subscales ranged from.74 to.83 in this sample.
Childhood physical abuse was examined with the Physical Punishment scale of the Assessing Environments-III (AE-III; Berger, Knutson, Mehm, & Perkins, 1988). The AE-III assesses punitive experiences and characteristics associated with abusive family environments during childhood and adolescence. The Physical Punishment scale (AE-III-PP) consists of 12 true/false items and focuses on the experience of physical discipline during childhood (before age 16). Severity of the behaviors captured in the AE-III-PP range from mild (e.g., spanking) to severe (e.g., choking). Positively endorsed items were summed to arrive at a total score, with higher scores indicative of more types of physically abusive childhood experiences. Berger et al. (1988) demonstrated the validity and test-retest reliability of the AE-III and the AE-III-PP.
Childhood sexual abuse was measured with two items from the Sexual Abuse Exposure Questionnaire (SAEQ; Rowan, Foy, Rodriguez, & Ryan, 1994). Items assessed the number of times that the participant reported being (1) sexually coerced and (2) raped before the age of 16. Responses to these items were summed to arrive at a total score.
Interparental domestic violence was assessed via the AE-III interview (Berger et al., 1988). Participants provided a true/false response to the following item: My parent(s) repeatedly used to use physical force against one another. Responses to this item could reflect any level of interparental physical violence, from low-level physical assault (i.e., slapping, shoving) up to more serious uses of force (i.e., beating up, choking).
Coping strategies were examined with the 72-item Coping Strategies Inventory (CSI; Tobin, Holroyd, Reynolds, & Wigal, 1989). The CSI consists of eight subscales that assess coping strategies employed in response to stressful events. In a factor analysis of the CSI, Tobin et al. (1989) showed the measure to contain two tertiary factors. One factor represented engagement coping and included the Problem Solving, Cognitive Restructuring, Express Emotions, and Social Support CSI subscales. The other factor represented disengagement coping and included the Problem Avoidance, Wishful Thinking, Self-Criticism, and Social Withdrawal subscales. Responses to the CSI are provided on a 5-point Likert scale ranging from 0 to 4, and summed scores indicate a greater likelihood of using the method of coping in question. Previous investigations have demonstrated the test-retest reliability, internal consistency, and criterion and construct validity of the CSI (Cook & Heppner, 1997; Tobin et al., 1989). Internal consistency reliability estimates for engagement and disengagement coping were.92 and.91, respectively, and these two subscales were positively correlated with one another, r =.15, p <.05.
All potential participants completed an initial screen via telephone to determine study eligibility. Eligible participants who elected to participate were scheduled for two baseline appointments approximately 2 weeks after the initial screening appointment, usually within several days of each other. During these sessions, participants completed several questionnaires on a laptop computer and interviews with trained clinicians who had extensive experience with victims of interpersonal trauma. Debriefings were conducted with participants following completion of all instruments. Interviewers inquired about participants’ safety, and safety planning information and clinical referrals were provided. To further ensure safety, no phone messages were left for participants, no mail was sent to the residences of participants, and security equipment was installed at the research site to protect a woman if her abuser followed her to the site. Each woman had an opportunity to discuss the impact of her study participation with her interviewer and completed an instrument designed to assess her experience with the study (see Griffin, Resick, Waldrop, & Mechanic, 2003).
First, descriptive statistics were computed for all study variables. Next, zero-order correlations were computed to examine the bivariate associations between the predictor and outcome variables. A series of multiple regression analyses then examined associations between the predictor variables of interest and the coping outcome variables. Separate regressions were conducted for the three categories of correlates (abuse-related factors, socioeconomic and social coping resources, childhood trauma) and the engagement coping and disengagement coping outcomes. Effect sizes were interpreted in terms of suggestions made by Cohen (1988).
Table 1 presents descriptive statistics for all of the study variables. On the CTS2, participants reported experiencing an estimated average of 85 physical assaults and 13 incidents of sexual aggression. Average scores on the PMWI measure similarly reflected high levels of abuse victimization. Participants’ average relationship length was approximately 7 years, and their most recent episode of abuse occurred about 42 days before their initial study session. The mean AE-III-PP score for the current study was slightly higher than 4, which is the score that has been shown to differentiate respondents that were physically abused as children from those who were not (Zaidi, Knutson, & Mehm, 1989), suggesting considerable childhood abuse among the current sample. On the CSI, participants’ mean scores reflect more usage of both types of coping relative to one previous study of women experiencing childhood sexual abuse (Coffey, Leitenberg, Henning, Bennett, & Jankowski, 1996) and higher disengagement coping than a sample of women experiencing dating violence (Coffey, Leitenberg, Henning, Turner, & Bennett, 1996). Forty percent of participants reported exposure to interparental violence during childhood.
Bivariate associations between the abuse-related factors and engagement and disengagement coping outcomes, as well as results from the multiple regression analyses examining associations between these predictor and outcome variables, are presented on Table 2. None of the abuse-related factors were significantly associated with engagement coping at the bivariate level, and the abuse-related variables accounted for only 2% of the variance in engagement coping.
All three of the abuse measures and PDEQ scores were significantly positively correlated with disengagement coping, with effect sizes falling within the small to medium range (Table 2). When the abuse-related factors were entered together into a regression equation predicting disengagement coping, only PMWI and PDEQ scores remained significantly predictive of this outcome, such that higher psychological aggression frequency and peritraumatic dissociation were associated with more disengagement coping. The abuse-related factors accounted for 15% of the variance in disengagement coping.
The three perceived social support variables were significantly positively associated with engagement coping, with medium effect sizes obtained (Table 3). When considered together, only the ISEL Belonging subscale remained associated with the outcome. Contrary to expectations, in this regression analysis, personal income was significantly negatively associated with engagement coping, such that higher income was associated with less engagement coping. The effect size for this association was in the small range of magnitude. The coping resource variables accounted for 16% of the variance in engagement coping.
Each of the perceived social support variables was significantly negatively associated with disengagement coping, with effect sizes ranging from small to medium (Table 3). When entered together into a multiple regression equation, the effects of the ISEL Tangible Support subscale was reduced to nonsignificance, and the ISEL Appraisal of Support and ISEL Belonging subscales remained associated with disengagement coping. The coping resource variables accounted for 11% of the variance in the disengagement coping outcome variable.
Both the AEIII Physical Punishment Scale and the SAEQ child coercion and rape variable were negatively associated with engagement coping, with small effect sizes obtained (Table 4). When the childhood trauma variables were considered together as predictors of engagement coping in a multiple regression analysis, only the SAEQ variable remained a significant predictor. This set of variables accounted for 3% of the variance in engagement coping.
The only significant childhood trauma correlate of disengagement coping was the AEIII interparental domestic violence variable (Table 4). Positive associations were found between interparental violence and disengagement coping both at the bivariate level and when accounting for the other childhood trauma variables in a multiple regression analysis. Three percent of the variance in disengagement coping was accounted for by the childhood trauma variables.
Findings revealed significant associations between abuse-related factors, social coping resources, childhood trauma variables, and engagement and disengagement coping. Consistent with one previous study (Mitchell & Hodson, 1983), relationship abuse frequency was more strongly associated with disengagement coping than engagement coping. Higher levels of each form of abuse were associated with disengagement coping behaviors, and psychological aggression was the strongest unique predictor of disengagement coping A growing literature shows psychological aggression to be a particularly robust predictor of PTSD symptoms among battered women (Arias & Pape, 1999; Dutton et al., 1999; Taft, Murphy, King, DeDeyn, & Musser, 2005). The stronger relationship between psychological aggression and disengagement coping in this study may reflect the particularly damaging effects of this form of abuse on mental health.
Peritraumatic dissociation represented the only other abuse-related factor that was associated with coping. This variable was positively associated with disengagement coping across analyses, suggesting that dissociative experiences during and immediately following incidents of relationship abuse were related to a pattern of escapist coping thoughts and behaviors. These findings are consistent with a prior study by Marmar, Weiss, Metzler, and Delucchi (1996), who found a strong positive association between peritraumatic dissociation and disengagement forms of coping among a sample of rescue workers. Further, prior research across a range of trauma groups has shown peritraumatic dissociation to be strongly associated with higher levels of PTSD (see Ozer, Best, Lipsey, & Weiss, 2003). It is possible that PTSD, which is typically characterized by behavioral avoidance and emotional numbing symptoms, serves as a mechanism whereby peritraumatic dissociation leads to disengagement coping.
Coping resource variables were the strongest predictors of engagement coping. Tangible support, appraisals of support, and a greater sense of belonging were all associated with this outcome at the bivariate level, and belonging remained associated with engagement coping when accounting for the other coping resource variables. Prior work suggests that battered women with a responsive support network display more active, engagement forms of coping (Goodman et al, 1999). Such social networks may foster a sense of empowerment and greater access to resources that assist with engagement coping (Dutton et al., 1994). Interestingly, the tangible and emotional support that these women received from others were more important determinants of their use of engagement coping strategies than were personal socioeconomic resource variables.
Childhood physical abuse and sexual abuse were negatively associated with engagement coping at the bivariate level, and childhood physical abuse was also predictive of less engagement coping in a regression analysis. Interparental domestic violence was positively associated with disengagement coping at the bivariate level and in light of the other childhood trauma variables. Disengagement strategies may be particularly adaptive during childhood to escape from the violence but may be ineffective in responding to trauma in adulthood (Leitenberg et al., 2004). It is also possible that PTSD and other mental health problems resulting from childhood trauma increase the use of disengagement coping (Gibson & Leitenberg, 2001).
Kraemer et al.’s (1997) risk factor typology that distinguishes fixed markers from variable risk factors may be particularly useful in considering the potential clinical implications of current study findings. Fixed markers are factors that do not vary over time among individuals and assist in identifying populations at risk. In this study, interparental domestic violence and peritraumatic dissociation represented fixed marker variables that may serve to identify those who are more likely to use disengagement coping methods in response to abuse. Variable risk factors change over time, either naturally or through manipulation, and represent potential points of intervention. For example, findings involving the social coping resource variables suggest efforts to enhance social and tangible support may be associated with increased engagement coping and less use of disengagement coping. Since engagement coping is associated with more positive psychological outcomes and disengagement coping is associated with negative psychological outcomes (Waldrop & Resick, 2004), results highlight the importance of interventions that target these factors. Findings also suggest that eliminating the relationship abuse may serve to decrease disengagement coping in particular.
Future studies should investigate a number of other potentially relevant correlates of coping. Cognitive appraisal variables, such as perceptions of threat and the controllability of the stressor, have been highlighted in general coping research and theory (Folkman, Lazarus, Dunkel-Schetter, DeLongis, & Gruen, 1986; Lazarus & Folkman, 1984) and are likely to be relevant for battered women’s choice of coping strategies. Mental and physical health functioning and other life stressors may play also play important roles with respect to coping. Problems in these areas may diminish both social and economic resources necessary to use engagement coping strategies (Aldwin & Revenson, 1987). It is also important to consider prior coping efforts and societal factors such as criminal and legal system responsiveness since more positive institutional responses may lead to feelings of empowerment and increased use of engagement coping (Mitchell & Hodson, 1986; Waldrop & Resick, 2004).
The use of retrospective reports mandates caution when interpreting study results because encoding and memory alterations have been linked with the presence of dissociation and trauma symptoms (Bremner, Vermetten, Southwick, Krystal, & Charney, 1998; Zoellner, Sacks, & Foa, 2003). The emotional or psychological state of participants may also have influenced retrospective reports. Further, childhood sexual abuse was assessed using a two-item composite, and interparental domestic violence was assessed with a one-item indicator. It is possible that these measures did not fully capture important aspects of these forms of childhood trauma. Future investigations should more comprehensively assess these trauma variables to better understand their relationships with different forms of coping. Finally, the coping behavior of the women in this study may not be representative of the larger population of battered women because participants were selected in part on the basis of their help-seeking behavior (e.g., seeking shelter). However, it is important to note that study participants reported higher levels of both engagement and disengagement forms of coping relative to prior research in this area. Regardless, current study findings should be replicated among community samples of battered women to minimize this threat to external validity.
With these cautions in mind, this study represents an initial attempt to document factors associated with the use of engagement and disengagement coping strategies among a sample of battered women. Results suggest that battered women’s use of different coping strategies is multidetermined and associated with a variety of proximal and distal factors. Additional research is needed to more fully explicate the coping process among battered women to inform prevention and intervention efforts for this population.
The study reported here was supported by National Institute of Mental Health Grant MH55542, awarded to Patricia Resick.
Casey T. Taft, National Center for PTSD, VA Boston Healthcare System, and Boston University, School of Medicine.
Patricia A. Resick, National Center for PTSD, VA Boston Healthcare System, and Boston University, and, University of Missouri, St. Louis.
Jillian Panuzio, National Center for PTSD, VA Boston Healthcare System.
Dawne S. Vogt, National Center for PTSD, VA Boston Healthcare System, and Boston University, School of Medicine.
Mindy B. Mechannic, California State University, Fullerton.