article leaving an abusive partner - semantic … · continuum are the theories of female...

10.1177/1524838002250769 ArticleTRAUMA, VIOLENCE, & ABUSE / April 2003Anderson, Saunders / LEAVING AN ABUSIVE PARTNER

LEAVING AN ABUSIVE PARTNERAn Empirical Review of Predictors,the Process of Leaving, and Psychological Well-Being

DEBORAH K. ANDERSONDANIEL G. SAUNDERSUniversity of Michigan

Four facets of leaving an abusive relationship are reviewed: (a) factors related toinitially leaving an abusive partner; (b) the process of leaving an abusive relation-ship; (c) the psychological well-being of survivors after leaving; and (d) the predic-tors of this well-being. The conceptual and methodological limitations of studies ineach of these areas are presented. Consistently found predictors of leaving includeboth material and psychological factors. Because battered women typically un-dergo several shifts in their thinking about the abuse before leaving permanently,research on leaving as a process is highlighted. A stress-process framework is usedto explain the seemingly paradoxical finding that some women just out of the abu-sive relationship may have greater psychological difficulties than those who arestill in it. For those experiencing the most stress, psychological health can worsenover time. Researchers and practitioners need to pay more attention to the plight ofwomen who have left abusive partners.

Key words: domestic violence, battered women, leaving, psychological well-being

STUDIES OF BATTERED WOMEN have grownexponentially in number since the discovery or,as some would say, the “rediscovery” of domes-tic violence in the mid- to late 1970s (Pleck,1987). Despite this volume of research, little isknown about the challenges faced by womenwho separate from an abusive partner. By vari-ous estimates, it appears that many or mostwomen do eventually leave abusive relation-ships (e.g., L. Okun, 1986; Strube, 1988). Most ofthe battered women in Lehnen and Skogan’s(1981) sample from the National Crime Surveyhad already divorced or separated from their as-

sailants at the time of the survey. A great deal ofresearch has focused on factors related to awoman’s decision to leave or stay and the pro-cesses involved in arriving at such a decision.However, a woman’s need for protection fromfurther abuse, practical assistance, and otherforms of intervention do not usually end whenshe leaves an abusive partner. In fact, suchneeds are likely to increase. Currently, the over-lap between research on battered women anddivorce largely remains unchartered territory.Studies on domestic violence have seldom ven-tured beyond the point of physical separation.

163

AUTHORS’ NOTE: The writing of this article was supported in part by NIMH training grant T32 MH19996.

TRAUMA, VIOLENCE, & ABUSE, Vol. 4, No. 2, April 2003 163-191DOI: 10.1177/1524838002250769© 2003 Sage Publications

At the same time, woman battering is rarelymentioned in the divorce literature, with only afew exceptions (e.g., Kurz, 1996; Molina, 1999).The present review addresses this gap in the lit-erature by focusing on battered women whohave left the assailant and extending the con-cept of leaving as a process to include the after-math of the separation.

Our review begins with a brief overview ofthe theoretical context in which research on bat-tered women’s leaving developed. It then turnsto studies that focus on factors relating to the de-cision to leave or return and the processes in-volved in terminating a relationship with anabusive partner. Finally, the challenges batteredwomen face and their subsequent psychologicalwell-being in the postseparation period are ex-amined from the available, mostly descriptiveevidence. These findings are synthesized andinterpreted from a stress-process framework toaddress the seemingly paradoxical question,Why would a substantial proportion of womenexperience low levels of well-being after leavingan abusive mate? We then identify various riskand protective factors influencing a batteredwoman’s psychological well-being afterseparation.

BACKGROUND

A wide variety of theories have been pro-posed to explain why women remain with abu-

sive partners. Prior to the1970s, Freudian notions offemale masochism pre-dominated (e.g., Snell,Rosenwald, & Robey,1964; Young & Gerson,1991). Battered womenwere believed to harbor aconscious or unconsciousneed for pain and punish-ment, which was used toexplain their “provoca-tion” leading to abuseand/or a lack of motiva-

tion for leaving. Feminist activists and social sci-entists challenged these psychodynamic viewsand helped to construct a new image of the bat-tered woman, one that emphasized gender role

conditioning, institutionalized sexism, and ex-ternal constraints on women’s ability to leave.

Thus, in the 1970s and 1980s, explanations forwhy a battered woman does or does not leavewere more likely to locate the source of thewoman’s entrapment outside of her personality.However, the emphasis on internal versus ex-ternal explanatory factors varied considerablyand many theories combined psychological dy-namics with external factors. On one end of thecontinuum are the theories of female masoch-ism mentioned earlier (Shainess, 1979). Next aretheories with interpersonal components. These

164 TRAUMA, VIOLENCE, & ABUSE / April 2003

KEY POINTS OF THERESEARCH REVIEW

• Predictors of leaving are found consistently intwo broad categories: material resources (espe-cially employment and income) and social psy-chological factors. Income variables werestronger predictors than psychological ones inmultivariate studies. A history of childhoodabuse and the nature of adult abuse were not con-sistent predictors of leaving or staying.

• Many survivors go through several phases in theprocess of leaving. They may leave and returnmultiple times, each time learning new copingskills. As with divorcing women, these phasesmay involve cognitive and emotional “leaving”before the physical leaving. The phases include(a) endurance of and managing the violencewhile disconnecting from self and others; (b) ac-knowledging the abuse, reframing it, and coun-teracting it; and (c) “breaking free,” disengaging,and focusing on one’s own needs. There appearsto be a fourth phase following separation that isnot addressed by the majority of these studies.

• Trauma effects decline in the months after separa-tion, but a substantial proportion of women con-tinue to suffer from post–traumatic stressdisorder (PTSD), depression, and other traumaticproblems. Some groups of abuse survivors expe-rience more traumatic effects and depressionright after separation than survivors still in a rela-tionship.

• After separation, continued violence and addi-tional stresses can create a negative spiral, creat-ing worsening psychological outcomes.Preliminary evidence suggests that the presenceof various coping resources such as social support,material necessities, and self-efficacy can protectagainst negative psychological outcomes. Womenoften have insufficient personal resources and re-quire additional services and support.

Battered womenwere believed toharbor a conscious orunconscious need forpain and punishment,which was used toexplain their“provocation”leading to abuseand/or a lack ofmotivation forleaving.

involve principles such as (a) intermittent rein-forcement, which Walker (1984) popularized asthe “cycle of violence”; (b) traumatic bonding,which was hypothesized to result from alternat-ing punishment and indulgences (Dutton &Painter, 1981); and (c) fear of greater harm uponseparation, which we will show later is a realis-tic fear for many women. Social learning theoryhas been applied in studies that investigatewhether battered women were likely to havewitnessed abuse or been abused in childhoodand, perhaps as a result of this abuse, developbeliefs that violence is a normal part of familylife. We also provide evidence regarding the im-pact of these childhood experiences. Learnedhelplessness theory hypothesized that motiva-tional, cognitive, and affective deficits resultwhen abused women repeatedly but unsuccess-fully attempt to get the help they need (Walker,1984). Closely related are theories of victimblaming and institutional sexism that create ap-athetic and sometimes hostile responses to bat-tered women when they seek help, leaving themisolated and without resources. On the macrolevel are patriarchal social structures that tendto keep women as a group economically de-pendent on men and patriarchal norms aboutgender roles that women tend to internalize.The above theories are described more fullyelsewhere (e.g., Barnett & LaViolette, 1993;Dobash & Dobash, 1992; Kirkwood, 1993;Strube, 1988). In sum, many potential impedi-ments to women’s leaving have been describedand investigated. As a result of theorizing sincethe 1970s, battered women are less likely to bedescribed as culpable participants in a “trou-bled relationship” than victims facing many ob-stacles restricting their alternatives to leaving anabusive relationship.

QUANTITATIVE RESEARCH PREDICTINGWOMEN’S STAY/LEAVE DECISIONS

Consistent with the background just de-scribed, studies on women’s decisions to stay orleave began to appear in the mid-1970s. Thesestudies attempted to isolate the reasons for bat-tered women’s apparent entrapment in abusive

relationships. Additionally, most of these stud-ies have tried to account both for structuralconstraints on a woman’s decision to leave orstay as well as psychological factors. This bodyof research is important for purposes of the cur-rent review inasmuch as it is among the first toshed some light on possible factors influencingthe well-being of women who leave. Whereasother recent reviews have included some stay/leave studies (e.g., Rhodes & McKenzie, 1998),the present review provides a fairly comprehen-sive assessment of those studies that directly in-volve the empirical testing of factors hypothe-sized to predict relationship status (see Table 1).1

Before presenting the results of the review, itshould be noted that making direct compari-sons between studies and resolving discrepantfindings among them is made difficult by differ-ences in the types of variables included, sampleselection (nonrandom), time of measurement,and the operationalization of independent anddependent variables. Some of the studies areretrospective whereas others are prospec-tive; some base their findings on women’sself-reports of “intentions,” and others mea-sure actual behavior. The dependent variablesin these studies, for example, range from awoman’s intentions to leave a violent partner atshelter exit, to her actual leaving/returning tothe batterer immediately after exiting a shelter,to her relationship status in the months or yearsfollowing a separation, during which she mayhave returned and left several times. Althoughthese outcomes are related (e.g., Rusbult &Martz, 1995), results could vary somewhat de-pending on which outcome measure is used.Nevertheless, findings that are consistently sig-nificant across studies with slightly differentoutcome variables can be considered more ro-bust. Keeping in mind the various operation-alizations of the dependent variables, the out-come variables will generally be referred to as“leaving” for ease of discussion. The most com-mon predictors from these studies can be cate-gorized as (a) the nature of the violence, (b) thewoman’s life history, (c) social psychologicalfactors, (d) external resources, and (e) previouscoping strategies.

Anderson, Saunders / LEAVING AN ABUSIVE PARTNER 165

(text continues on p. 170)

TABLE 1 Quantitative Studies of Battered Women’s Stay/Leave Decisions

Study Sample Size Source of Sample Design Dependent Variable Predictors of Leaving

Aguirre, 1985 312 Residents of 15 batteredwomen’s shelters

Cross-sectional Intentions to leave/return atshelter exit

2 of 8 predictors significant (multivariate): Eco-nomic independence from husband, andnumber of decisions made while at shelter

Compton, Michael,Krasavage-Hopkins,Schneiderman, &Bickman, 1989

141 Residents of a YWCA shelter Cross-sectional Intentions to live on own,return to abuser, or livewith friends/family atshelter exit

6 of (unknown number) predictors significant(multivariate): Batterer unemployed, length ofshelter stay, number of years married, yearsof abuse, number of previous separations,woman has good health

Dalto, 1983 32 Residents of a batteredwomen’s shelter

Time 1: interview at shelterTime 2: 6-week follow-up

mail survey

Relationship status atfollow-up

11 of 27 predictors significant (bivariate): Didnot seek medical attention, feminist stancetoward battering, negative attitude toward re-turning,low social pressure to return, time atshelter and looking for housing (2 variables),attributes responsibility for violence toabuser, cause of violence perceived as sta-ble, interpersonal relationships at shelter (2variables), perceived dissimilarity of rolemodel

Frisch & MacKenzie,1991

46 Former and current residents oftwo battered women’s shel-ters (currently batteredwomen were abused for atleast 4 years)

Cross-sectional Relationship status at the timeof the data collection

8 of 19 predictors significant (bivariate): Liberalattitudes toward women, self-esteem, doesnot feel controlled by outside forces, womanemployed, ever received counseling, educa-tion, incidents of serious bodily harm, attribu-tions for violence external to self

Gelles, 1976 41 Community sample of batteredwomen (obtained from policerecords, a private social ser-vice agency, and from neigh-boring families in whichwomen had sought no outsideintervention)

Cross-sectional Relationship status at the timeof data collection

3 of 6 predictors significant (bivariate): Severityof violence, frequency of violence, womanemployed

Gondolf & Fisher,1988

2,000 Residents of 50 batteredwomen’s shelters in Texas

Cross-sectional Intentions to leave/return atshelter exit

12 of 32 predictors significant (multivariate):Batterer not in counseling, own transporta-tion, child care, own income, emergencyroom care, batterer threatened to harm child,weapons used, child abuse, batterer threat-ened to kill her, not a racial minority, policenot contacted, woman does not abusealcohol

166

Herbert, Silver, &Ellard, 1991

130 (44 stillactivelyinvolved;86 notinvolved)

Community sample of batteredwomen recruited through pub-lic service announcements,television, radio, and newspa-per ads

Cross-sectional Relationship status at time ofdata collection

9 of 16 predictors significant (multivariate):Fewer positive aspects of the relationship,negative change in relationship, low family in-come, less likely to make downward compari-sons, partner blamed for abuse, manipulatoryattributions for positive behaviors, does notblame self for abuse, frequency of severeabuse, frequency of verbal abuse

Hilbert & Hilbert,1984

35 Residents of a batteredwoman’s shelter

Cross-sectional Whether woman leaves/returnsat shelter exit

6 of 20 predictors significant (multivariate):Woman’s age, low severity of violence, fre-quency of violence, shorter length of batter-ing relationship, victim income, length ofshelter stay

Hilbert, Kolia, &VanLeeuwen, 1997

216 Residents of three New Mexicobattered women’s shelters(87% living with abuser atintake)


2 of 4 predictors significant (multivariate):Length of shelter stay, verbal and physicalabuse

Jacobson, Gottman,Gortner, Berns, &Wu Shortt, 1996

45 (couples) Community sample of 45 se-verely abused marriedwomen and their batterers(recruited through advertise-ments & random phonedialing)

Time 1: Initial assessmentTime 2: 2-year follow-up

Relationship status at the 2-yearfollow-up

13 of 34 predictors significant (multivariate):Husband’s isolation of wife, husband’s degra-dation of wife, husband’s physiologicalarousal (2 variables), frequency of husbandviolence, husband negative affect (4 vari-ables), wife marital dissatisfaction, wife de-fensiveness (2 variables), wife does not usehumor to cope with abuse, wife physiologicalarousal

I. Johnson, 1992 426 Residents of a batteredwoman’s shelter


2 of 14 predictors significant (bivariate):Woman employed, low severity of violence

Lesser, 1990 58 Former residents of a batteredwoman’s shelter

Cross-sectional Length of time remained sepa-rated from abuser 1 year aftershelter stay

9 of (unknown number) predictors significant(multivariate): Social support, length of shel-ter stay, low attachment to abuser, personalfinancial independence, low family income,did not witness parental violence, youngerwomen, woman’s education, batterer abusesalcohol

Martin et al., 2000 70 Residents of a batteredwomen’s shelter

Cross-sectional Residents’ perceivedchances of leaving

0 of 3 predictors significant (bivariate).

(continued)

167

L. Okun, 1986 300 Residents of a batteredwoman’s shelter

Time 1: IntakeTime 2: 1 year following exit

(outcome form filled outby shelter worker)

Relationship status at 1-yearpostshelter follow-up (termi-nated immediately, terminatedeventually, did not terminate)

7 of 47 predictors significant (bivariate):Batterer unemployed, woman has greater in-come than batterer, woman does not takepsychoactive medications, number of previ-ous separations, length of separation,batterer’s low level of education, batterer’scriminal record of violence

Pagelow, 1981 350 Residents of various batteredwomen’s shelters and somenonshelter volunteers

Cross-sectional Length of cohabitation withbatterer after first batteringincident

2 of 2 major predictors significant (bivariate):Less severe injuries, childhood violencevictimization

Rounsaville, 1978 31 Battered women receiving ser-vices from an emergencyroom and a mental healthcenter

Cross-sectional Relationship status at the timeof data collection

5 of 15 predictors significant (bivariate): Sever-ity of abuse (2 variables), fear of being killed,contact with the police, abuse of children bypartner

Rusbult & Martz,1995


Time 1: Shelter intakeTime 2: 3 months following

exitTime 3: 6 months following

exitTime 4: 12 months following

exit

Relationship status at 3-, 6-, &12-month postshelter follow-ups

2 of 3 predictors significant (multivariate): Per-sonal resources (education, income, employ-ment, transportation), investment items (mar-ried, duration of relationship, number ofchildren)

Schutte, Malouff, &Doyle, 1988


Cross-sectional Intentions to return 4 of 7 predictors significant (multivariate): Ex-perienced childhood abuse, woman’s educa-tion, number of previous separations, numberof shelter stays

Snyder & Scheer,1981


Time 1: Shelter intakeTime 2: 6-10 weeks follow-

ing exit

Living arrangements at 6-10week postshelter follow-up

9 of 40 predictors (bivariate): Seeks more thanshort-term separation, does not seek conjointmarital counseling, unmarried, shorter lengthof marriage, number of previous separations,woman is not Roman Catholic, length of shel-ter stay, not intending to return at discharge,separated/divorced from partner at follow-up

Strube & Barbour,1983

98 Battered women seeking coun-seling with county attorney’soffice (All living w/abuser atintake)

Time 1: Initial intakeTime 2: 1-18 months follow-

ing intake

Relationship status at follow-upafter closing of case (typically2-3 months after closing)

5 of 10 predictors significant (bivariate):Woman employed, shorter length of relation-ship, no mention of economic hardship, nomention of love for the partner, partner prom-ised change at Time 1

TABLE 1 (continued)

Study Sample Size Source of Sample Design Dependent Variable Predictors of Leaving

168

169

Strube & Barbour,1984

251 Battered women seeking coun-seling w/county attorney’s of-fice (All living w/abuser atintake)

Time 1: IntakeTime 2: Roughly 2-3

months following intake

Relationship status at follow-upcontact after closing of case.

8 of 18 predictors significant: (multivariate):Woman employed, shorter length of relation-ship, no mention of economic hardship, nomention of love for partner, racial minority,somewhere else to go, number of copingstrategies used, partner did not make prom-ise to change

Truman-Schram,Cann, Calhoun, &Vanwallendael,2000a

78 Unmarried, female college un-dergraduates, all of whomhad experienced violence in adating relationship

Time 1: Initial assessmentTime 2: Roughly 1 month

after initial assessment

Relationship status at the timeof the data collection (cur-rently involved vs. left the abu-sive dating partner)

7 of (unknown number) predictors significant(multivariate): Catholic women, education ofwoman’s mother, low psychological invest-ment in the relationship, dissatisfaction withthe relationship (3 variables), shorter lengthof relationship

Williams, 2000b 100 Residents of shelters and sup-port group participants

Cross-sectional Relationship status at the timeof the data collection

1 of 5 predictors significant (multivariate): Self-efficacy for meeting personal needs

a. It should be noted that unlike the rest of the above studies, this one is included in the table even though the research focuses exclusively on “dating relationships.” It is cited in the text only withrespect to findings relating to psychological investment in the relationship, which was the primary concern of the investigators.b.Williams (2000) is much more recent than the bulk of the studies in this table and has clearly been influenced by both the stay/leave studies and the process studies described in the next sec-tion.Although this study employs a stage theory to conceptualize leaving as a process, it is categorized with the stay/leave studies because it employs a quantitative method for the purpose ofpredicting women’s likelihood of being in or out of the relationship.

The Nature of the Violence

The nature of the violence, usually assessedin terms of frequency and severity, was amongthe most likely predictors to be investigated.Whereas some perspectives such as learnedhelplessness (Walker, 1984) or theories of femalemasochism (Snell et al., 1964) would expectwomen to be more likely to remain in the rela-tionship as the abuse escalates over time, thestudies under review favor the “commonsense” hypothesis advanced by Gelles (1976),which holds that battered women will be morelikely to leave as the violence increases in its se-verity and frequency. However, the findingswere often quite inconsistent, possibly due todefinitional ambiguity of concepts such as “se-

vere” violence and differ-ences in the types of mea-surement instrumentsused. The nonsignificantfindings in numerousstudies (e.g., Gelles, 1976;L. Okun, 1986; Snyder &Scheer, 1981) are likelydue to a lack of variationwithin samples becausemost of the women experi-enced high levels of vio-

lence. Verbal and emotional abuse was some-times as good or more accurate than physicalabuse in predicting leaving (e.g., Herbert, Silver,& Ellard, 1991; Hilbert, Kolia, & VanLeeuwen,1997; Jacobson, Gottman, Gortner, Berns, & WuShortt, 1996).

Woman’s History of Other Abuse

A woman’s previous experiences with vio-lence as a child or an adult are believed by manyto play a key role in her staying with an abusiveintimate partner. In particular, factors relatingto a woman’s childhood history of family vio-lence, either as a witness or victim of abuse,were the most frequently examined—close tothree quarters of the studies in Table 1. How-ever, studies finding support for the hypothe-sized link between childhood history of vio-lence and a greater likelihood of returning werethe exceptions (Lesser, 1990). Rusbult and Martz

(1995) found a significant relationship at thebivariate level, which disappeared after otherfactors were controlled (not shown in table).These studies were outnumbered by studiesfinding that women who witnessed parental vi-olence or who were themselves abused as chil-dren, were more likely to leave than theirnonvictimized counterparts (Gondolf & Fisher,1988; Pagelow, 1981; Schutte, Malouff, & Doyle,1988). Results from Williams’s (2000) study sup-port these findings at the bivariate level only(not shown in table). Finally, the majority ofstudies that examined the relationship betweensuch childhood history of violence and stay/leave decisions reported no significant findings(e.g., Aguirre, 1985; L. Okun, 1986; Strube &Barbour, 1984). Overall, the findings are incon-sistent but violence history may sometimes bol-ster women’s determination to escape theabuse.

Social Psychological Factors

A variety of psychological and social psycho-logical factors were assessed for their predictiveability regarding the stay/leave decision. Themost frequently assessed predictor variables inthis category, however, were psychologicalcommitment to the relationship and objectiveindicators of potential commitment. Severalstudies hypothesized that women with a greatersense of commitment would be more likely toremain with the abuser compared with womenwith less commitment (Rusbult & Martz, 1995;Strube & Barbour, 1983, 1984; Truman-Schram,Cann, Calhoun, & Vanwallendael, 2000). For ex-ample, the more she has invested in terms oftime, effort, resources, legal ties, or love for herpartner, for example, the more compelled sheshould feel to justify these investments throughfurther efforts to save the relationship. Therewas a fair amount of support for these hypothe-ses across the various studies but primarily forsubjective measures of commitment. Womenwho indicated love or positive feelings for theabuser or the relationship (Strube & Barbour,1983, 1984; Truman-Schram et al., 2000) andheld traditional religious values/beliefs(Snyder & Scheer, 1981) were more likely to beinvolved in the relationship at the time of mea-


Several studieshypothesized thatwomen with agreater sense ofcommitment wouldbe more likely toremain with theabuser comparedwith women with lesscommitment.

surement. However, findings were much lesslikely to be significant when objective indicatorssuch as marital status (2 of 6 studies) and lengthof the relationship (5 of 10 studies) served as in-dicators of commitment (e.g., Compton, Mi-chael, Krasavage-Hopkins, Schneiderman, &Bickman, 1989; Martin et al., 2000; L. Okun,1986; Schutte et al., 1988).

External Resources

In a number of studies, external resources ap-peared to outweigh subjective factors in predic-tive ability. In the Strube and Barbour (1984)study, for example, a woman’s employment sta-tus was a more powerful predictor of stayingthan a woman’s positive feelings about the rela-tionship (e.g., love, partner promised tochange). Similarly, in the Lesser (1990) study,feelings of economic dependence and the needfor a place to go were mentioned as reasons forreturning by only 15% of the women, yet an ob-jective measure of women’s income was the bestsingle predictor of women’s leaving, controllingfor other factors (including life history vari-ables). Other studies report similar findings.Women who were more economically advan-taged in terms of employment status and per-sonal income were less psychologically commit-ted and significantly more likely to leave thanother women (e.g., Frisch & MacKenzie, 1991;Rusbult & Martz, 1995; Schutte et al., 1988) withfew exceptions (e.g., Dalto, 1983; Williams,2000). It may be less disturbing to a woman tobelieve that she is staying for the positive as-pects of the relationship (i.e., voluntarily) ratherthan for negative reasons, such as economicentrapment.

Income variables were not only among themost consistently related but possibly the mostpowerful predictors of the stay/leave decisionoverall, even when controlling for a variety ofpsychological and other variables. Studies thatemployed multivariate techniques and in-cluded a broad range of variables more fre-quently reported income to be the strongest pre-dictor of leaving (e.g., Aguirre, 1985; Comptonet al., 1989; Gelles, 1976; Hilbert & Hilbert, 1984;Lesser, 1990; Rusbult & Martz, 1995). Womenwho had a source of income independent of the

abuser, including welfare (Aguirre, 1985;Gondolf & Fisher, 1988; Lesser, 1990; Rusbult &Martz, 1995), or who had incomes larger thanthose of their partners (L. Okun, 1986) weremuch more likely to leave the abuser. In sum,findings for financial indicators appear quiterobust.

Unfortunately, fewer studies assessed the ef-fects of other barriers to women’s leaving.Women were more likely to leave when theyhad regular access to child care (Gondolf &Fisher, 1988), transportation (Gondolf & Fisher,1988; Rusbult & Martz, 1995), and in some cases,social support (Lesser, 1990) but not always(Strube & Barbour, 1984). Potential financial lia-bilities, such as the number, age, or presence ofchildren, as well as the woman’s age, weremostly nonsignificant (e.g., Compton et al.,1989; Gondolf & Fisher, 1988; Herbert et al.,1991; I. Johnson, 1992; L. Okun, 1986). Onestudy found women’s health problems to be asignificant predictor of returning to the abuser(Compton et al., 1989). Findings for race as apredictor variable were inconsistent (Comptonet al., 1989; Gondolf & Fisher, 1988; L. Okun,1986; Rounsaville, 1978; Strube & Barbour,1984).

Previous Coping Strategies

A woman’s repeated attempts to cope withthe abuse are an integral part of the overall ex-perience of violence. Studies that directly or in-directly included women’s prior efforts to copewith the violence as a predictor of the stay/leave decision serve as precursors to studies ofleaving as a process. Four (of eight) studiesfound that the greater the number of previousseparations from an abuser, the more apt awoman is to leave (Compton et al., 1989; L.Okun, 1986; Schutte et al., 1988; Snyder &Scheer, 1981). Findings from two (of three) stud-ies indicated that women who had previouslyemployed numerous other coping strategies be-sides leaving were significantly more likely toseparate from the abuser (Rounsaville, 1978;Strube & Barbour, 1984). Short-lived separa-tions and the development of new coping skillsmay enhance women’s feelings of mastery andself-efficacy, making permanent separation


more likely. However, the mixed findings onprior number of separations suggest that otherfactors may need to be considered as well, suchas the resources to which she had access and herdesire to seek long- or short-term separation.

Although many of the stay/leave studiesview women’s multiple returns as problematicand evidence of an inability or reluctance to sep-arate from an abusive mate (e.g., Lesser, 1990;Rounsaville, 1978; Strube & Barbour, 1983), oth-ers stress the multiple separations as indicatorsof persistent, strong efforts by “survivors” toleave for good (Gondolf & Fisher, 1988). In lightof women’s perseverance in the face of violence,L. Okun (1986) suggests the need to view leav-ing as a process. This positive view of leavingand returning foreshadows the process of leav-ing studies to be reviewed shortly.

Taken as a whole, the stay/leave studies havefurthered our understanding of factors affectinga woman’s leaving an abusive partner. Theypoint to an array of environmental and psycho-logical factors that influence a woman’s deci-sion to leave or stay—a marked improvementover earlier work that located the source of theproblem almost solely within the personalitiesof battered women. In sum, the current group ofstudies provides empirical backing to the claimsof advocates and theorists in the 1970s who por-trayed abused women not as masochists whoprovoked the abuse but as women facing multi-ple internal and external obstacles to leaving.

However, there are a number of conceptualproblems with this group of studies. First, des-ignating the decision to leave as the ultimateoutcome variable of interest equates leavingwith the cessation of violence. Some researchhas found that the abuse may cease while she re-mains with the abuser (e.g., Aldarondo, 1996;Bowker, 1983; J. Campbell, Miller, Cardwell, &Belknap, 1994; Horton & Johnson, 1993) or con-tinue well after the separation (e.g., Bachman &Saltzman, 1995; Wilson, Johnson, & Daly, 1995).Moreover, even though these studies consider awide variety of internal and external obstaclesas predictive factors, leaving itself is typicallyconceptualized as a single act that hinges onlyon a decision. In reality, not only can the deci-sion change several times, but leaving typicallyentails a series of other decisions and actions.

Second, because the greater emphasis isclearly on women who have not left or who arestill trying to leave, this line of research fre-quently tends to reinforce an image of the bat-tered woman as relatively passive. It can be-come difficult for researchers to conceive of heras an active agent who frequently does leave orotherwise manages to free herself from the vio-lence (Gondolf & Fisher, 1988). Furthermore, byfocusing on stay/leave outcomes, the chal-lenges faced by women in the aftermath of sepa-ration are not addressed. Despite these criti-cisms, findings from the stay/leave studies doprovide clues to some of the risk and protectivefactors faced by battered women after they sep-arate. Implicit in the why-does-she-stay ques-tion is the flip side of the coin, “What enables herto leave?”

QUALITATIVE RESEARCH THATCONCEPTUALIZES LEAVING AS A PROCESS

The 28 publications under review in this sec-tion (25 original studies, 1 theoretical article,and 2 review papers listed in Table 2) tend to bemore recent than the stay/leave decision stud-ies. Process-of-leaving researchers take issuewith the predominant focus on leaving as a sin-gle act or decision and they are also more delib-erate in their efforts to counter popular stereo-types of battered women as helpless or passive.The main research question differs primarily inits point of emphasis. Because these studies takeit for granted that many women can and doleave abusive partners, they place greater em-phasis on the flip side of the why-does-she-stayquestion and so are more likely to underscorewomen’s agency. The question that requires ex-planation is, namely, “How does she ever man-age to leave given all the strikes against her?”

Process studies typically differ qualitativelyfrom stay/leave studies in conceptualizingleaving as a complex process involving manydecisions and actions taking place over a periodof months or years. According to these studies,leaving begins with changes at the emotionaland cognitive levels well before an actual physi-cal departure. When attention is focused onlyon a single act of leaving, subtle changes in herthinking and behavior are missed and her ap-


TABLE 2 Qualitative Studies Focusing on the Process of Leaving for Battered Women

Sample Type of Consideration ofStudy Size Source of Sample Design Publication Postseparation Issues

Angless, Moconachie,& Van Zyl, 1998

21 Former residents of a battered woman’s shelter in South Africa; Womenwere located from information provided on release forms

Cross-sectional Article Approximately one page on thistopic

Brown, 1997 na na (theoretical framework for process of leaving) na Article No mention

Burke, Gielen,McDonnell,O’Campo, &Maman, 2001

78 A subset drawn from a larger women’s health study; recruited from fivesites affiliated with an urban teaching hospital, including a homelessshelter, HIV care clinic, an infant mortality prevention center, a gynecol-ogy clinic, and an outpatient drug treatment center; all were either cur-rently or formerly battered women


Campbell, Rose, Kub,& Nedd, 1998

32 Urban community sample of battered women self-identified as having aserious problem in an intimate relationship; recruited from newspaperads and bulletin board postings

Interviewed 3 timesover 2 ½ years

Article Postseparation issues and violencementioned in brief but focus is ondegree of women’s psychologicalcommitment to relationship

Carlin, 1998 5 Women who had been with an abusive partner for at least 5 years; allhad left partner within past 2 years; recruited from a counseling pro-gram for battered women

Cross-sectional Dissertation Approximately seven pages on thistopic

Dobash & Dobash,1979

109 Current or recent residents of battered women’s shelters in Scotland Cross-sectional Book Passing references topostseparation issues but no sec-tions devoted to it

Eisikovits, Buchbinder,& Mor, 1998

20 A purposive sample of Israeli women who sought help from an emer-gency hotline for survivors of domestic abuse

Cross-sectional Article No mention

Eldar-Avidan & Haj-Yahia, 2000

15 Divorced Israeli women who had been abused by their former husbands;accessed through social welfare and human service agencies

Cross-sectional Article Article devoted to this topic

Ferraro & Johnson,1983

120 Primarily residents of a battered women’s shelter (5 had received ser-vices but were not residents)

Cross-sectional Article Less than one page on this topic

Goetting, 1999 16 Convenience sample recruited from battered women’s shelters and otherorganizations and agencies sympathetic to battered women

Cross-sectional Book Some attention to postseparationissues on a case-by-case basis,but primary emphasis is onwomen’s biographies up to physi-cal exit from relationship

Hoff, 1990 9 Former residents of a battered women’s shelter; contacted through stateshelter and health networks

1-year naturalisticstudy (qualitativemethodology)

Book One chapter and brief epilogue onthis topic

Holt, 1995 10 Women who had left their abusive partners; social workers in a variety ofsettings identified and initiated contact with women for possible inclu-sion in the study

Cross-sectional Dissertation No mention

Kirkwood, 1993 30 Women who had been out of an abusive relationship for at least 1 year;recruited through newspaper and radio ads and contacts with an orga-nization for single parents

Cross-sectional Book One chapter on this topic

(continued)

173

Landenburger, 1989 30 Current or previously battered women recruited through newspaper ads,a community support group, and a battered women’s shelter


Merritt-Gray & Wuest,1995

13 Domestic violence survivors recruited by lay and professional helpers ina rural Canadian community


Mills, 1985 10 Residents from a battered women’s shelter Cross-sectional Article No mention

Molina, 1999 30 African American working women who sought legal services to obtain adivorce; 18 of these women were domestic violence survivors


Moss, Pitula,Campbell, &Halstead, 1997

30 Survivors who had terminated a relationship with an abusive partner; re-cruited through local women’s organizations & posters on collegecampuses

Cross-sectional Article Approximately two pages on thistopic

NiCarthy, 1987 33 Women who had been out of an abusive relationship for at least a year;recruited from flyers, newspaper ads, and programs servicing batteredwomen

Cross-sectional Book Approximately 9-10 pages on thistopic

A. Okun, 1998 30 Women who had left an abusive partner for a minimum of 2 years; re-cruited through newspaper ads, community television ads, and flyersplaced at local venues frequented by women

Cross-sectional Dissertation No mention

Patzel, 2001 10 Convenience sample recruited from two outreach programs and a re-source center for battered women; all had terminated their relationshipswith the batterer at least 6 months prior to the study


Rosen & Stith, 1997 22 Women had been or were currently in a violent dating relationship; re-cruited through newspaper ads, flyers reaching the general public, andreferrals from clinical colleagues

Cross-sectional Article Approximately two pages on thistopic

Sleutel, 1998 na Review of qualitative work within past 15 years na Article Less than one page on this topic

Syers-McNairy, 1990 15 Women who had terminated a relationship with an abusive partner atleast 1 year prior to the study; some recruited from women’s supportgroups at a local women’s center; others recruited from batteredwomen’s programs

Cross-sectional Dissertation Much of manuscript devoted to thistopic

Ulrich, 1991 51 Formerly battered women recruited through community newsletters, cri-sis center support groups, and therapists’ offices.


Ulrich, 1998 na Reviews some evidence from various studies that supports notion of a“process of leaving”

na Article No mention

Wilson, 1999 9 Women had left their abusive partners at least 1 year prior to study;women recruited by word of mouth and snowball sampling.

Cross-sectional Dissertation Approximately 15-20 pages de-voted to this topic

Wuest & Merritt-Gray,1999

15 Formerly battered women who had terminated the relationship at least 1year prior to the study; recruited with the help of professional and layhelpers who gave women letters explaining the study.


NOTE: na = not applicable

TABLE 2 (continued)

Sample Type of Consideration ofStudy Size Source of Sample Design Publication Postseparation Issues

174

parent staying behavior is likely to be inter-preted in a manner consistent with traditionalviews of women (e.g., she cannot leave becauseshe is helpless or dependent). To capture thecomplexity of leaving, process studies attemptto understand women’s behavior through theirown voices. Women’s accounts describe the so-cial context within which the process of leavingtakes place, thus bringing to light the courageand determination required in preparing for thefinal separation. They also highlight the insightsthat lead to growth experiences along the way,often before an act of leaving can be observed bythe outsider.

In this view, survival status is attainedthrough active and cumulative efforts aswomen gradually learn more effective strate-gies for dealing with the abuse. Most of the stud-ies that view leaving as a process describedwomen passing through a series of stages orphases leading to an eventual separation(s)from the abuser; however, periods of return toearlier phases are considered normal (J. Camp-bell, Rose, Kub, & Nedd, 1998; Patzel, 2001). In-creasingly popular among process studies is theapplication of Prochaska and DiClemente’s(1984) Transtheoretical Model of Change to bat-tered women’s leaving (e.g., J. Brown, 1997;Burke, Gielen, McDonnell, O’Campo, &Maman, 2001; Williams, 2000). The model as-sesses a person’s cognitive, emotional, and be-havioral readiness to make a needed change toher life, such as terminating the relationship.Despite differences among individual studies inthe amount of emphasis put on a distinct set ofstages, several common themes emerged. Be-cause these process studies are feminist in orien-tation, importance is placed on the role that pa-triarchy plays, particularly through traditionalfemale socialization and “family values” inwomen’s responses to their abuse (e.g., Dobash& Dobash, 1979; Hoff, 1990; Kirkwood, 1993).Under a system of patriarchy, women’s oppres-sion and devaluation seems almost natural andinevitable. Not surprisingly, most process stud-ies found that many battered women initiallydrew on traditional religious ideals and femalestereotypes to develop strategies to cope withthe violence (Mills, 1985; Wilson, 1999). For ex-ample, many adopted the role of nurturer (i.e.,

she is the only one who can help him), whileminimizing or denyingthe extent of abuse(Ferraro and Johnson,1983; Mills, 1985). Simi-larly, others stayed out ofa wifely duty to “standbehind their man.” How-ever, as Ferraro and John-son (1983) emphasize,this does not necessarilyindicate a woman’s pas-sive acceptance or resig-nation to her fate. Rather,it requires active efforts to construct a particularmeaning to her circumstances that will allowboth the continuation of the love that she feelstoward her partner and endurance of the pain ofthe abuse. Additionally, many women were iso-lated from external influences that might chal-lenge either their own or the abuser’s definitionof the situation. Moreover, African Americanwomen are socialized not to seek help fromoutside the black community to avoid bring-ing further disgrace and suffering onto theblack male by the wider society (Moss, Pitula,Campbell, & Halstead, 1997). Studies have re-ferred to this phase variously as “endurance”(Landenburger, 1989; Moss et al., 1997), “mak-ing do” (J. Campbell et al., 1998), “managing”the violence (Mills, 1985), or disconnecting fromself and others (Carlin, 1998).

Additional effort and knowledge is usuallyrequired for her to begin questioning her situa-tion, let alone give up the dream of a happy lifewith a loving partner. Nevertheless, processstudies almost invariably found that most of thewomen reported another shift in their perspec-tive that sometimes occurred suddenly butmore often developed gradually as women ex-perienced fleeting insights about themselvesand the relationship (e.g. , Eisikovits ,Buchbinder, & Mor, 1998; Holt, 1995; Mills, 1985;A. Okun, 1998; Rosen & Stith, 1997). In this pe-riod of reframing (Moss et al., 1997; Patzel,2001), acknowledgement/contemplation (J.Brown, 1997; Burke et al., 2001), shrinking/lossof the self (Landenburger, 1989; Mills, 1985), orcounteracting the abuse (Merritt-Gray & Wuest,1995), women began to redefine the relationship


Women’s accountsdescribe the social

context within whichthe process of

leaving takes place,thus bringing to light

the courage anddetermination

required in preparingfor the finalseparation.

as abusive and label themselves as victims. Cat-alysts that helped to bring about this shift inthinking included an increase in the level of theviolence or fewer periods of love and affection;loss of hope that the relationship will get better;witnessing the effect of the abuse on the chil-dren; or external influences such as friends,family, and helping professionals who offersupport and alternative perspectives (e.g.,Dobash & Dobash, 1979; Eisikovits et al., 1998;Ferraro & Johnson, 1983; Kirkwood, 1993;Ulrich, 1991). Importantly, Ferraro and Johnson(1983) were among the few to report how somewomen began reinterpreting the violence whenfinancial and material resources necessary forescape became available. As women recognizedthe negative impact of the violence on theirwell-being, suppressed feelings of loss, failure,anger, and/or fear commonly began to surface.However, such feelings provided the fuel manywomen needed to begin searching for a way out(Carlin, 1998; Eisikovits et al., 1998; Hoff, 1990;Kirkwood, 1993; A. Okun, 1998).

Finally, in the disengagement (Landenburger,1989) or breaking free stage (Merritt-Gray &Wuest, 1995), women’s priorities seemed tochange from focusing only on the needs of part-ners to protecting her own needs and reconnect-ing with the self (Carlin, 1998; Wilson, 1999).Most of the process studies noted how a“woman’s agenda to maintain her relationship”was slowly “replaced with an agenda to leavethe relationship” (Rosen & Stith, 1997, p. 177).At this stage, women began to engage in activi-ties they believed would help them leave, suchas finding a safe place to think, focusing onother areas of life, finding social support, turn-ing to alcohol, making safety plans, enrolling inself-defense classes, setting some limits in therelationship, and making small decisions thathelped increase self-efficacy and self-worth(e.g., Goetting, 1999; Merritt-Gray & Wuest,1995; NiCarthy, 1987). Although manywomen’s newfound feelings about themselveswere mixed with pain and anguish, they beganleaving in many ways that were not always visi-ble to the casual observer, such as arriving homelater than usual, withdrawing from their part-ners emotionally, accepting jobs that took themaway from home, separating their belongings

from the abuser’s while living under the sameroof, and seeking temporary separations at ashelter (e.g., Landenburger, 1989; Merritt-Gray& Wuest, 1995). These temporary separationswere not a sign of weakness; rather, they seemedto give many women enough autonomy andself-confidence to make the final break, al-though sometimes not before a “last straw” epi-sode of abuse occurred (e.g. , Angless,Maconachie, & Van Zyl, 1998; Goetting, 1999;Holt, 1995; Rosen & Stith, 1997).

Clearly, process studies highlight the cour-age, determination, and persistence involved inleaving an abusive partner. By underscoring theagency of battered women, the stereotype of thewoman as a passive victim is replaced with anew and more complex psychology of woman-as-survivor, in which battered women slowlyregain control over their own lives. From thisstandpoint, the expectation that women willhave the strength and resolve to maintain thisindependence after the separation seems to fol-low more easily, unlike most of the work on thestay/leave decision. However, several limita-tions of this research need to be addressed. Theretrospective nature of the data, with one excep-tion (e.g., J. Campbell et al., 1998), andnonrandom sampling procedures should firstbe noted. The women in these studies were fromselect samples (e.g., recruited from shelters, bat-tered women’s groups, and community adver-tisements), and most had already made the finalbreak. Furthermore, it is not known to what ex-tent the data might be subject to recall errors asmost of the interviews took place quite a fewmonths or years after the separation or shelterstay. Additionally, very little attention is givento ways in which the leaving process may differby race or ethnicity. One study found thatwomen of color (primarily African American)were much more similar than different fromtheir White counterparts, despite the multipleoppressions faced by minority women (J.Campbell et al., 1998). However, possibly due toracial oppression, African American women inWilson’s (1999) sample appeared to be moreconscious of the sociopolitical context that sup-ports men’s domination of women.

Most important, however, in trying to coun-ter older notions of female masochism and pas-


sivity, the process studies (e.g., Hoff, 1990;Kirkwood, 1993) face conceptual limitations.The parameters for the study of batteredwomen’s leaving behavior are similar in at leasttwo ways to earlier research emphasizingintrapsychic processes. First, whereas the no-tion that battered women can be survivors aswell as victims was a necessary and valuablecontribution to the field, many of the processstudies, with a few exceptions (Dobash &Dobash, 1979; A. Okun, 1998; Syers-McNairy,1990), come close to the psychologicalreductionism for which theories of female mas-ochism and learned helplessness have been crit-icized (see Dobash & Dobash, 1988, for a critiqueof learned helplessness). Internal explanationsfor women’s decisions need also to be viewed inlight of major structural constraints faced bybattered women, as many of the stay/leavestudies have effectively demonstrated. How-ever, many of the process studies, including thenewer ones employing the TranstheoreticalModel of Change, seem to suggest that leavingan abusive relationship is primarily contingenton changes in the subjective meaning of the situ-ation to the women (e.g., J. Brown, 1997; Burkeet al . , 2001; J . Campbell et al . , 1998;Landenburger, 1989; Mills, 1985; Ulrich, 1998).For example, Mills (1985) characterizes the pro-cess of leaving as one in which battered womenshift from being “compliant zombies to reflec-tive actors who decide to leave their husbands”(p. 115). At the same time, Landenburger (1989)emphasizes that battered women need to learnmore about abusive relationships. However, inthe policy and practice implications section ofthis study, no mention is made of the need forthe provision of external resources, nor thelarger social forces that contribute to a woman’scontinued victimization or harassment shouldshe decide to separate.

Second, the primary goal of most processstudies has been to demonstrate the ways inwhich women still technically “in” the relation-ship progressively undergo significant growthexperiences that are part of the leaving processand are otherwise overlooked by traditional re-search. Although this has been an importantcontribution, there seems to be an assumption,with a few exceptions (e.g., J. Campbell et al.,

1998), that the only way a woman can regain heragency is to leave and that staying is always un-healthy. This view is perhaps most apparentamong process studies employing theTranstheoretical Model, which is often appliedto populations with addictive and problem be-haviors such as smoking, drug use, and delin-quency (Prochaska, Norcross, & DiClemente,1994). Many well-intentioned practitioners andbattered women’s activists who adopt this vieware likely to be of little help to women who expe-rience less severe abuse and may wish toachieve nonviolence from within the relation-ship (Peled, Eisikovits, Enosh, & Winstok, 2000).

Moreover, given the intensity of effort goinginto showing that these women living with abu-sive partners are not helpless victims, perhaps itis not surprising that most of the process studiesalso neglect or give minimal attention to thepostseparation period. In most of the studies,the process of leaving seems to abruptly endwith the woman’s physical departure from therelationship, or else the possibility of ongoingviolence or other obstacles following the sepa-ration is mentioned in passing (e.g., Anglesset al., 1998; J. Brown, 1997; J. Campbell et al.,1998; Eisikovits et al., 1998; Holt, 1995;Landenburger, 1989). As shown in Table 2, only7 of the 28 qualitative studies devote a some-what considerable proportion of space to the af-termath of separation (Eldar-Avidan & Haj-Yahia, 2000; Hoff, 1990; Kirkwood, 1993;Molina, 1999; NiCarthy, 1987; Syers-McNairy,1990; Wuest & Merritt-Gray, 1999), and this pe-riod is the primary concern of only four of thestudies (Eldar-Avidan & Haj-Yahia, 2000;Molina, 1999; Syers-McNairy, 1990; Wuest &Merritt-Gray, 1999) .However, the challengesto maintaining a life inde-pendent from the abuserwould seem to constitutean equally important partof the process. Notably, aparallel trend cannot befound in the general di-vorce literature. When di-vorce is treated as a multistage process, the pro-cess frequently begins with the initial sepa-


There seems to be anassumption, with a

few exceptions . . . ,that the only way awoman can regain

her agency is toleave and that

staying is alwaysunhealthy.

ration (e.g., Bursik, 1991; Gray, Koopman, &Hunt, 1991; Pledge, 1992), although there aresome exceptions (e.g., Guerin, Fay, Burden, &Kautto, 1987). Thus, the majority of the processstudies on battered women either ignore ordownplay the importance of the postseparationperiod, including the violence and other stress-ors that often occur then.

POSTSEPARATION PSYCHOLOGICALWELL-BEING

Psychological Outcomes of Abuse

The preceding sections identified some keyfactors predicting who is most likely to leave anabusive partner as well as the processlike natureof this decision. The emphasis shifted from“What keeps her there?” to “How does she evenmanage to leave?” For the next set of studies, in-quiry centers on the question of “What happensafter she leaves?” As the focus now turns to bat-tered women in the postseparation stage of theprocess, psychological well-being will replacerelationship status as the outcome of interest in-asmuch as leaving does not necessarily mean asafer, happier life. Relationship status may berelated less to well-being but seems to deter-mine the types of environmental circumstancesand stressors to which a woman is exposed.Given that battered women who separate willencounter a unique set of circumstances andstressors, it is important to determine how thepsychological well-being of these women maydiffer from that of women who are still livingwith the batterer and from nonabused divorc-ing women.

Although many studies include both womenwho are currently involved with the abuser andwomen who have since left, very few havemade comparisons between these groups whenassessing psychological well-being. Among theexceptions, several studies have found that av-erage levels of depression, PTSD, or othertrauma symptoms among battered women whohave left can equal (Herbert et al., 1991;Rounsaville, 1978) or exceed (Kemp, Green,Hovanitz, & Rawlings, 1995; Lerner & Kennedy,

2000; Walker, 1984) those of women still in therelationship. Recency of separation in one of thecross-sectional studies was related to lower lev-els of psychological well-being among womenwho had left (Lerner & Kennedy, 2000). Thesefindings are confirmed by several longitudinalstudies that found a significant reduction in thelevel of trauma symptoms for recently sepa-rated battered women over a 6-month (Dutton& Painter, 1993) and 1-year interval (Mertin &Mohr, 2001). At the same time, Anderson (2001)found that positive aspects of well-being, spe-cifically perceived quality of life, increased overmultiple time points for women who remainedseparated from their abusers at least 2 years af-ter shelter exit.

Although the psychological well-being ofmany separated battered women appears to in-crease with time, mean rates of well-being pro-vide limited information. Anderson, Saunders,Yoshihama, Bybee, and Sullivan (in press) as-sessed the course of recovery at multiple timepoints over a 2-year period for 94 batteredwomen who had left their abusive partners.Findings from the study indicate the need to dis-tinguish among various subgroups of womenwho have left, as the depression levels for onethird of the sample remained moderate to highor actually increased significantly with time.Prevalence rates of depression and anxiety werereported in another study of separated batteredwomen. At the time of shelter residence, 42% ofthe 59 women in the Mertin and Mohr (2001)study met the full DSM-IV (Diagnostic and Statis-tical Manual of Mental Disorders, fourth edition)(American Psychiatric Association, 1994) diag-nostic criteria for PTSD, whereas 14% still metthe criteria 1 year later. Similar percentages forclinical levels of anxiety were reported at bothtime points. The prevalence rates at the 1-yearfollow-up, however, are likely underestimatedbecause 41 women who were unavailable forthe follow-up interview were significantly morelikely to believe they would be killed by theirpartner at Time 1. If substantial numbers ofwomen who separate are highly distressed,then more research concerning the relevant riskand protective factors affecting their psycholog-ical well-being is needed.


Predictors of Postseparation Well-Being:A Stress Process Model

Drawing from the work of Pearl in,Lieberman, Menaghan, and Mullan (1981), theremainder of the review employs a stress pro-cess conceptual framework from which thefindings on predictors of women’s postsepara-tion well-being can be organized, interpreted,and integrated. The model may not be superiorto others but can provide a valuable, heuristicframework. This approach seems applicable tothe study of battered women in the separationprocess for at least several reasons. First, thestress experience is viewed not as an isolatedevent but an ongoing process. Similarly, “leav-ing” for the battered woman is the continuationof a process that begins at the emotional andcognitive level while she is still in the relation-ship and extends well beyond her physical de-parture. Second, such models place a heavy em-phasis on the combined impact of variousenvironmental factors accompanying a majorstressor such as domestic violence and subse-quent separation. Asevere stressor rarely occursas an isolated event but is likely to trigger a se-ries of other stressors affecting multiple life do-mains. Leaving an abusive partner sets in mo-tion numerous, sudden, and often dramaticchanges in women’s life situations. The multi-ple stressors facing women potentially involvevarious emotional and financial losses; new oraltered responsibilities as a single parent or soleprovider; repeated exposures to violence; andlimited access to key social, material, and inter-nal coping resources. Thus, it is theoreticallypossible for the negative effects of stress to becumulative.

Finally, process models conceptually allowfor diverse outcomes depending on the number,strength, and combinations of environmentalfactors present. In other words, some womenwill be more or less traumatized due to differ-ences in the level of overall stress as well as inthe amount of social support, self-efficacy, andfinancial and material resources. Below we re-view evidence for the impact of three types offactors believed to influence women’spostseparation psychological well-being: pri-mary stressors that set the stress process in mo-

tion (the violence that caused or contributed tothe separation); secondary stressors, or thosethat accompany the primary stressor chrono-logically; and coping resources that can serve tocushion the impact of stressors or prevent theirfuture reoccurrence.

Violence

A source of stress that sets battered womenapart from other divorcing and separatingwomen is the experience of abuse. Besideshaving to cope with thetraumatic effects of pre-separation violence,many battered womenare subject to ongoing orescalated violence afterleaving. According to theNational ViolenceAgainst Women Survey,over one half millionwomen are stalked by(typically an ex-) intimatepartner annually in theUnited States (Tjaden & Thoennes, 2000). Datafrom national crime surveys in the UnitedStates and Canada estimate that comparedwith married women, separated women areabout 25 times more likely to be assaulted by ex-mates (Bachman & Saltzman, 1995) and 5 timesmore likely to be murdered (Wilson & Daly,1993). Furthermore, 24%-35% of those whowere assaulted before separation experienceeven more severe violence after leaving (H.Johnson, 1995; Statistics Canada, 2001). At leastone recent study found that the length of timeout of a physically abusive relationship (from 4days to 4 years) was associated with a greaterfrequency of stalking victimizations by the for-mer partner in the past month, even after con-trolling for previous levels of physical and emo-tional abuse (Mechanic, Weaver, & Resick,2000).

Unfortunately, domestic violence research istypically focused on the violence taking placeprior to separating from the abuser. Only onequantitative study directly examined the rela-tionship between postseparation violence andwell-being. Mertin and Mohr (2001) found that


“Leaving” for thebattered woman is

the continuation of aprocess that begins

at the emotional andcognitive level while

she is still in therelationship and

extends well beyondher physical

departure.

continued abuse after separation was signifi-cantly and positively correlated with anxiety,depression, and PTSD. A number of existingqualitative studies, however, provide morevivid accounts of the stress that postseparationabuse can have on a woman and the direct ef-fects it can have on her well-being. Whereasonly 14% of B. Brown and Foye’s (1982) di-vorced homemakers (n = 134) described interac-tions with their estranged husbands as “worri-some,” fear of retaliation and continuedvictimization was a common theme in the sto-ries of most of the battered women who left theirabusers (Angless et al., 1998; Hoff, 1990;Kirkwood, 1993; Kurz, 1996; Landenburger,1989; Moss et al., 1997; NiCarthy, 1987; Tutty,1998; Wuest & Merritt-Gray, 1999). These stud-ies make note of other forms of violence oftenmissed by quantitative measures. Harassmentand intimidation by batterers ranged from beg-ging, sending flowers, and threatening suicideto pressuring the children for information, mak-ing menacing phone calls, trumping up falsechild abuse charges against her, and threateningto “blow” her head off (Raphael, 1999; Tutty,1998; Wuest & Merritt-Gray, 1999). The presenceof restraining orders and mandatory arrest lawsseemed to provide little relief for most women(Kirkwood, 1993). Having to live like a fugitiveand lack of protection from others were ratedamong the biggest problems by women fromSouth Africa (Angless et al., 1998). Likewise,about half of the Canadian women in the Tutty(1998) study continued to feel threatened bytheir partners at least 4 to 6 months after leavinga shelter for battered women. Some womenwere still having to endure the abuser’s contin-ued manipulation and control several years af-ter leaving (Moss et al., 1997). Those who finallyfound safety expressed how much the violencehad been a drain on their emotional resources.One third of Kirkwood’s (1993) sample ex-pressed relief as the “dominant feeling” in theirlives. “Relief was in response to the lessening ofthe emotional or physical danger, of the need tobe constantly vigilant of their partners’ moodsand behavior, and of the expenditure of energyin worry, fear, and survival” (p. 107). These

women seemed to be enjoying a higher per-ceived quality of life and less depression.Similar sentiments were echoed in Eldar-Avidan and Haj-Yahia’s (2000) sample of di-vorced Israeli women.

Secondary Stressors

Secondary stressors, brought on by the sepa-ration itself, simultaneously involve multiplelosses on one hand and new or altered responsi-bilities on the other. In this respect, divorce issometimes conceptualized as a dual transition(B. Brown & Foye, 1982) and is typically consid-ered one of the most stressful life experiences(see Pledge, 1992, for a review). Separatedwomen experience significantly more negativelife changes (psychological, physical, and so-cial) on average within the 1st year than do mar-ried women (Nelson, 1989). Researchers havebegun to delineate some of the stressful life dis-ruptions that are specific to the divorce process,three of which are discussed below.

Attachment and loss. At the emotional level,one key issue facing many women who separateis the presence of lingering feelings of attach-ment and loss around the former intimate part-ner. One small body of research begins with theassumption that feelings of attachment and lossin battered women are more situational and notqualitatively different from those of otherwomen leaving troubled nonviolent relation-ships (e.g., J. Campbell, 1989; Eldar-Avidan &Haj-Yahia, 2000; Molina, 1999; Turner &Shapiro, 1986; Varvaro, 1991). This research—for example, J. Campbell (1989)—borrowsheavily from the bereavement literature in sug-gesting that attachment issues in batteredwomen might be more appropriately viewed asa normal and expected grieving response tohaving lost a major attachment figure. Whenwomen are in the process of leaving, they tendto miss the positive qualities of their partnersand the relationship (Landenburger, 1989).However, Dutton and Painter’s (1993) longitu-dinal data indicated that the attachment of re-cently separated women in their sample


decreased by 27% over a brief period of 6months, which suggests the situational natureof the attachment.

Additionally, attachment in divorcingwomen must be understood in terms ofwomen’s socialization, specifically in terms ofidealized, culturally prescribed gender roles aswell as their access to power and resources rela-tive to men. In qualitative interviews, batteredwomen who had left commonly reported an on-going attachment that was sometimes less re-lated to feelings about the partner himself thanto the loss of a relationship with the highly val-ued wife-mother role (Carlin, 1998; Hoff, 1990;Mills, 1985; Moss et al., 1997; NiCarthy, 1987).They felt great sadness in having to let go ofthese childhood dreams. In some battered andnonbattered women, the identity disruptionwas so keenly felt, they reported feeling a com-plete loss of their core selves (Kirkwood, 1993;Mills, 1985; Moss et al., 1997) requiring them toundergo a process of “identity transition” to“reconstruct a new sense of self” (DeGarmo &Kitson, 1996, p. 983). In the Degarmo and Kitson(1996) longitudinal study of divorcing and wid-owed women, the greater the centrality of the“coupled identity” to a woman, the greater wasthe psychological distress both shortly after thepartner’s death/filing for divorce and 1 year af-ter. J. Campbell and associates (1989; J. Camp-bell, Kub, Belknap, & Templin, 1997) found onlyweak or qualified support for the relationshipbetween the valuing or defining of oneself interms of the traditional wife/mother role andgreater depression among battered women.However, it is important to note that batteredand nonbattered control women were recruitedon the basis of their currently having serious dif-ficulties within an intimate relationship, so feel-ings of loss may not yet have been as great tothese women. In sum, these findings reflect thepowerful role of socialization in shapingwomen’s reactions to such deeply felt loss.

Other losses. Finally, many women undergo asudden major loss of economic resources as a re-sult of the separation. According to data fromnational samples, divorcing women experi-enced declines in financial well-being averagingabout 30% to 36% when there were young chil-

dren in the household (e.g., Bianchi, Subaiya, &Kahn, 1999; Duncan & Hoffman, 1985). Formore than one quarter of single mothers, the de-clines in income relative to needs were greaterthan 50%. Many battered women arrive at shel-ters with no money or possessions except for theclothes they are wearing (Tutty, 1998). Domesticviolence has been linked to homelessness andother economic problems in samples of poorwomen (e.g., Browne & Bassuk, 1997; Tolman &Rosen, 2001). Nevertheless, only two quantita-tive studies under review have directly exam-ined the impact of the actual loss on women’spostseparation psychological well-being. Satoand Heiby (1992) link the loss of household in-come and other primarily material losses (e.g.,home, job, security) directly to greater depres-sion in battered women after controlling for theeffects of physical abuse, prior depression, andother variables. Similarly, Berman (1988) founda strong trend in his multivariate analysis to-ward higher rates of depression in divorcedwomen who experienced the greatest losses inincome.

Changes in family responsibilities. Adding tothe complexity of divorce, other stressors ap-pear in the form of new and often conflicting re-sponsibilities, especially those associated withthe single parent/economic provider role. Atthe same time, many women have to find em-ployment or increase their income to compen-sate at least in part for the loss of their partner’sfinancial contributions. Family demands mayincrease largely due to numerous changes infamily composition, roles, and relationships aswell as children’s difficulties in understandingand adjusting to these changes (e.g., Buehler,Hogan, Robinson, & Levy, 1985-1986; Dixon &Rettig, 1994; Gerstel, Riessman, & Rosenfield,1985). Clearly, these challenges also loom largein the minds of battered women who separate.Children must be settled in the new home,schools, day care, and the neighborhood. A sub-stantial number of battered women mentiontheir children’s difficulties in sorting out whathappened (Wuest & Merritt-Gray, 1999). In theHumphreys (1995) study, every woman sponta-neously mentioned concerns around her chil-dren, especially in trying to adjust her work


schedule so that children would need a baby-sitter as little as possible. This frequently provesextremely challenging given the continual closesupervision required for preschool children andthe time involved in transporting school-agedchildren to and from school and other activities(B. Brown & Foye, 1982). Issues concerning jobsand children were among the most importantand frequently mentioned 4 to 6 months afterhaving left the shelter, for Tutty’s (1998) sample.Parenting responsibilities and getting or keep-ing a job were also among the top three worriesof divorcing homemakers in another study (B.Brown & Foye, 1982). Competing goals andtime demands seem to be more the rule than theexception for divorcing women and this hasdefinite implications for their psychologicalwell-being.

Studies that focused on the number of recentchanges across multiple life domains, andwomen’s feelings about the changes, seem toshow fairly consistent results with respect topsychological well-being. For example,Voydanoff and Kelly’s (1984) study of 468 mar-ried and divorced parents indicated that par-ents in households experiencing three or moremajor changes in family life during the year pre-ceding the interview (e.g., divorce, increased ex-penses) had higher levels of perceived timeshortages. This was especially characteristic ofhouseholds with single custodial mothers.Moreover, feelings about single parenthood andparenting skills were significantly correlatedwith mothers’ general sense of psychologicalwell-being (Thiriot & Buckner, 1992). Buehleret al. (1985-1986) found that the more divorce-related stressors experienced by single mothersin various areas, including those relating toparent-child relationships, the former spouse,or financial, housing, and legal issues, the worsewas the women’s level of parenting satisfaction.In short, there may be a cumulative effect ofmultiple stressors that negatively affects divorc-ing women’s well-being in the form of timepressures, job tensions, and low levels of satis-faction with family life simultaneously.

Cumulative effects of stressors. One study longi-tudinally tested the long-term effects of multi-ple stressors, including violence and secondary

stressors (i.e., attachment/loss, changes infamily responsibilities, financial losses), on de-pression in a sample of 94 women who had re-mained separated from their abusers for at least2 years (Anderson et al., in press). The samplewas divided into thirds by overall level of expo-sure to stress. Depression for those with thehighest stress levels near the time of shelter exitremained moderately high or actually increasedfor some over the 2-year period. The gap in psy-chological well-being between the highest andlowest stress groups widened with time. Thesefindings are consistent with a stress process con-ceptual framework in at least several ways: (a)Mental health outcomes varied among womenexposed to a similar traumatic stressor (i.e., se-vere violence), which suggests that women wholeave their abusive partners are not a single pop-ulation; and (b) downward spirals in well-beingover time is one form the stress process can take.It may be that initially high levels of stress cantrigger a series of subsequent negative events ormore chronic strains that place an individual atan ever greater risk for negative mental healthconsequences if coping resources are not pres-ent in sufficient quantities (e.g., Pearlin, 1999;Wheaton, 1999).

Coping Resources

Although a link between the violence and/orsubsequent separation from the abusive partnerand psychological distress seems clear, not allwomen will develop clinical depression, PTSD,or other trauma symptoms. Reactions to stress-ors will vary even after differences in the sever-ity of the violence and secondary stressors havebeen considered. A major explanatory factor forsuch differences lies in women’s access to vari-ous coping resources that provide them with themeans to exert some control over their lives ingeneral and to respond to particular life stress-ors when they arise. Coping resources include(but are not limited to) various external and in-ternal necessities such as material goods andservices, income, social support, and self-effi-cacy. Coping resources are believed to be espe-cially relevant for battered women and otherwomen who separate from their intimatepartners.


Material resources. Material necessities arecrucial for a woman to establish a life independ-ent from her former partner. For example, af-fordable housing and transportation areespecially important for a battered woman asthese resources literally make her less availableto be beaten (Ellis, 1992). Food, clothing, childcare, and a personal source of income are essen-tials in maintaining her independence. How-ever, of 11 possible areas of unmet needs, thebattered women in Sullivan and associates’(Sullivan, Basta, Tan, & Davidson, 1992;Sullivan & Davidson, 1991) shelter sample mostoften reported needing help with obtaining ma-terial goods and services such as furniture andclothing (83%-84%). Not surprisingly, com-pared with battered women who returned to theabuser, women who were ending their relation-ships were even more likely to need help withaccessing resources such as finances, education,legal assistance, and transportation.

The need for housing and economic resourceswas documented in several studies as the mostpressing concern among battered women whohad recently left (Kirkwood, 1993) and divorc-ing women from the general population (Amato& Partridge, 1987; Berman, 1988; B. Brown &Foye, 1982; Miller, Smerglia, Gaudet, & Kitson,1998). Even among the employed AfricanAmerican women in Molina’s (1999) sample(40% of whom had been battered), divorce pre-sented substantial financial difficulties for a ma-jority (57%) due to low salaries and lack of childsupport. Many battered women in Hoff’s (1990)and Kirkwood’s (1993) qualitative studies re-marked that their energy during their shelterstay, or shortly after, was almost totally ab-sorbed in practical concerns such as securingpermanent housing or a fixed address for them-selves and their children, a necessary prerequi-site for receiving government financial aid(Ellis, 1992). They frequently reported gettinglittle cooperation from housing authorities andother institutions. When they did locate hous-ing, it was often too expensive, too small for awoman with children, or located in a high-crimearea where they felt unsafe. These issues werecited as a major source of stress by manywomen. Only a single empirical study specifi-cally tested for the effects of housing stressors

for divorcing women and men (e.g., residentialmoves) and found partial support for its nega-tive impact on single parents’ adjustment(Buehler et al., 1985-1986). Other studies of bat-tered women (e.g., J. Campbell et al., 1997;Dutton & Painter, 1993; Mitchell & Hodson,1983) and other divorcing women (e.g., Duffy,1989; Lindsay & Scherman, 1987; Miller et al.,1998; Shapiro, 1996; Thabes, 1997; Thiriot &Buckner, 1992; Wilcox, 1986) are fairly consis-tent in finding that various measures of incomeand to a lesser extent, education, predict betterpostseparation psychological well-being.

Social Support. Of all thecoping resources, socialsupport—most broadlydefined as “the availablesocial relationships thatobjectively may be calledupon for help in times ofneed” (Hobfoll & Vaux,1993, p. 687)—has beenthe most widely re-searched in the domesticviolence and generalstress fields. Most quanti-tative studies of batteredwomen assessing the im-pact of social support on various measures ofpsychological well-being have found a signifi-cant, positive relationship (e.g., Anderson et al.,in press; Astin, Lawrence, & Foy, 1993; R. Camp-bell, Sullivan, & Davidson, 1995; Kemp et al.,1995; Mertin & Mohr, 2001; Mitchell & Hodson,1983; Tan, Basta, Sullivan, & Davidson, 1995).However, only two of the studies under reviewspecifically tested for the efficacy of social sup-port for women who have since left the shelteror otherwise remained separated from theabuser (Mertin & Mohr, 2001; Anderson et al., inpress). In the Anderson et al. (in press) study,separated battered women with superior sup-port systems at the time of shelter exit had sig-nificantly lower levels of depression than thosewith less support more than 2 years later evenwhen controlling for other coping resources,stressors, and baseline levels of depression.These findings fit with those of studies withqualitative data on the importance of postsepa-


The need for housingand economicresources was

documented inseveral studies as the

most pressingconcern among

battered women whohad recently left . . .

and divorcing womenfrom the general

population.

ration social support. Friends, family, supportgroups, and new romantic partners providedsupport in the form of advice and information,practical assistance, companionship, and emo-tional support (e.g., Hoff, 1990; Molina, 1999;Syers-McNairy, 1990; Tutty, 1998; Wuest &Merritt-Gray, 1999).

Internal resources. Also very important for thebattered woman are internal resources such ashaving confidence in her ability to exert somemeasure of control over her immediate environ-ment, variously referred to as self-care agency(e.g., J. Campbell, 1989), locus of control (e.g., R.Campbell et al., 1995; Garvin, Kalter, & Hansell,1993; Orava, McLeod, & Sharpe, 1996), mastery(Mitchell & Hodson, 1983; Nurius, Furrey, &Berliner, 1992), and self-efficacy (e.g., Benight,Swift, Sanger, Smith, & Zeppelin, 1999; Sullivan,Campbell, Angelique, Eby, & Davidson, 1994;Sullivan, Tan, Basta, Rumptz, & Davidson,1992). Numerous studies, including some well-controlled prospective analyses, have foundone or more of these closely related internal re-sources to be directly and significantly relatedto subsequent positive adjustment in batteredwomen (e.g., J. Campbell et al., 1997; R. Camp-bell et al., 1995; Dutton & Painter, 1993), withgreater feelings of self-worth related to higherlevels of well-being in terms of perceived qual-ity of life, less depression, and other psychologi-cal improvements. Although separatingbattered women typically exit the relationshipwith an eroded sense of self, their final resolveto leave and maintain their independence alsoprovides them with a sense of accomplishment(e.g., Carlin, 1998; Wilson, 1999). A number ofother investigators (J. Campbell, 1989; J. Camp-bell et al., 1997; Sullivan & Bybee, 1999) have be-gun to call attention to these existing innerresources so that service providers can helpwomen enhance their capacity for resilienceand growth at a time when they are quite vul-nerable.

Institutional resources. From the literature re-viewed thus far, it seems apparent that manyseparating battered women are in need of a vari-ety of resources and that their personal re-sources as well as those of their informal social

support networks can be overwhelmed by themultitude of stressors. The threat of ongoing orescalated violence is frequently accompaniedby housing problems and a lack of transporta-tion, job opportunities, and affordable child care(Correia, 2001). For many women who are ableto call on the support of close friends and familymembers, both emotional and financial help isusually temporary, particularly when the threatof violence continues after separation (e.g.,Angless et al., 1998).

Institutional support, on the other hand, canbe more substantial and provide women withresources that may be more lasting than per-sonal resources. Quantitative studies assessingthe impact of institutional resources on sepa-rated battered women’s well-being are rare andmost of them include women in their sampleswho are still with their abusers. Nevertheless,Sullivan and associates (e.g., Sullivan & Bybee,1999; Sullivan et al., 1994) assessed the efficacyof a postshelter advocacy intervention programusing an experimental design. At shelter exit,278 battered women (some who were returningto the assailant) indicated what issues theywould like to work on with trained para-professionals should they be randomly selectedto receive a 10-week advocacy intervention, 4 to6 hours per week. Controlling for baseline per-ceptions of life quality and other variables, theinvestigators found that battered women whoreceived the support of advocates reported sig-nificantly greater satisfaction in attainingneeded resources, fewer incidences of abuse bytheir assailants, and higher levels of quality oflife at subsequent follow-up interviews, com-pared with women who were not randomly se-lected to participate in the intervention.

Another follow-up program for batteredwomen who had exited shelters employedtrained professionals who assumed more of ageneral, supportive counseling or case managerrole on many issues, including self-esteem andeffective ways of accessing needed services andresources (Tutty, 1998). Results of this study arealso very encouraging, with a greater propor-tion of follow-up clients’ seeking further educa-tion or job training than those in the non-follow-up group. Quantitative measures showed sig-nificant improvements in self-esteem and in


emotional support for participants 4 to 6months into the program. The qualitative com-ponent of the study indicated that women wereparticularly pleased with the emotional supportthey received but also felt they benefited fromthe help with accessing community resources tosatisfy legal, financial, or other needs. However,it is difficult to isolate the specific componentsof the programs that contributed the most tothese positive outcomes as many of the womenwere simultaneously involved in batteredwomen’s support groups as part of the follow-up program.

In at least two evaluative studies of profes-sionally led support groups for batteredwomen, there were pretest to posttest improve-ments in self-esteem, belonging support, locusof control, and perceived stress (Tutty, Bidgood,& Rothery, 1993). Participants were also beingsubjected to less violence at the time of theposttest. Moreover, group participants showedgreater decreases in depression, anxiety, andother negative outcome variables comparedwith controls who were not part of the interven-tion. No differences between participants andnonparticipants emerged in another study, pos-sibly because of the small sample size (Cox &Stoltenberg, 1991).

Despite the small sample sizes and explor-atory nature of many existing evaluative stud-ies, both the qualitative and quantitative evi-dence suggest the promise of postseparationand postshelter services for battered women.Although these kinds of piecemeal services can-not adequately address the many significantneeds of this population, findings from thesestudies provide evidence that women benefitfrom such resources.

SUMMARY AND CONCLUSIONS

The findings on women’s stay/leave deci-sions consistently revealed two broad catego-ries of predictors of leaving: material resources(especially employment and income) and socialpsychological factors such as negative feelingstoward the partner. Income variables werestronger predictors than psychological ones inmultivariate studies. A history of childhood

abuse and the nature of adult abuse were notconsistent predictors of leaving or staying.These findings imply that counselors need tomaintain a dual consideration of material re-sources as well as the psychological/emotionallives of survivors. Policies that bolster women’sincome are especially needed, along with jobsupports such as child care and transportation.Research in this area needs to focus more onwomen who are more representative of the pop-ulation and not part of help-seeking samples.

Survivors typically go through severalphases or shifts in their thinking during the pro-cess of leaving. They may leave and return mul-tiple times, each time learning new copingskills. As with divorcing women, these phasesmay involve cognitive and emotional “leaving”before the physical leaving. The phases include(a) endurance of and managing the violencewhile disconnecting from self and others; (b) ac-knowledging the abuse, reframing it, and coun-teracting it; (c) breaking free, disengaging, andfocusing on one’s own needs. There appears tobe a fourth, postseparation phase that is not ad-dressed by the majority of these studies. Thefindings of studies on the process of leaving canhelp advocates and counselors understand thatthe leave/return cycle is an opportunity to learnbetter coping skills and is not a sign of failure.Counselors can use the process model to edu-cate women and validate their experiences. Thistype of research would benefit from larger, morerepresentative samples. The research also needsstandardized measures and longitudinaldesigns.

Fewer research efforts have focused on thepsychological well-being of survivors who haveleft their abusers. In the months after separa-tion, well-being seems to improve for a majorityof women over time despite the difficulties theyface. Many survivors experience more trau-matic effects and depression right after separa-tion than survivors still in a relationship. How-ever, a substantial minority who are exposed tothe highest levels of stressors and are most lack-ing in needed resources are at risk for continuednegative or worsening psychological outcomes.The stressors they face have been largely under-estimated or overlooked. Preliminary evidence


suggests that the presence of various coping re-sources such as social support, material necessi-ties, and self-efficacy can protect against nega-tive psychological outcomes. Women oftenhave insufficient personal resources and requireadditional services and support.

Research on the psychological well-being ofseparated survivors highlights again the needfor policies that increase resources for employ-ment, employment training, child care, andtransportation. Domestic violence programsand other social service agencies need to domore to reach out to separated battered womenand assess their needs. More studies are neededwith specific measures of social support, andmore research is needed on identifying factors

early in the separation process that signal nega-tive outcomes. The increased focus in researchand practice on victims who are struggling toleave or who have left highlights the multiplechallenges they face in leaving an abusive part-ner. The research also points to some promisingsteps that can be taken to help these victims alle-viate the traumatic effects of battering.

NOTE1. Table 1 represents a fairly comprehensive inclusion of such

studies. We focused on the most rigorous ones with a sample sizeof at least 30. We also included reviews of these articles. The firstphase of the search process involved locating articles via thePsycINFO database, using the keywords battered women, domesticviolence, family violence, leaving, and staying. We then located nu-merous studies from the reference lists of the above studies.

IMPLICATIONS FOR PRACTICE, POLICY, AND RESEARCH

Practice and Policy

Predictors of staying/leaving• Advocates and counselors need to keep dual focus

on material and emotional factors related to leaving.• Policies need to provide a means out of poverty.Leaving as a process• Counselors can use the stages to educate survivors.• All practitioners need to view leave/return cycles as

opportunities for new coping skills.Psychological well-being• There needs to be more focus on outreach to women

who have left violent relationships.• There need to be policies to support more resources

such as job training, child care, and transitionalhousing.

Research

Predictors of staying/leaving• There is a need for studies with a greater range of vi-

olence severity and that include victims not seekinghelp.

Leaving as a process• Larger, more representative samples are needed.• Longitudinal studies with standardized measures

can further test the information gathered from quali-tative studies.

Psychological well-being• There is a need for more longitudinal studies that in-

clude subgroups of women under different levels ofstress.

• There is a need for studies with specific measures ofsocial support.

REFERENCESAmerican Psychiatric Association. (1994) Diagnostic and

statistical manual of mental disorders (4th ed.). Washing-ton, DC: Author.

Aguirre, B. (1985). Why do they return? Abused wives inshelters. Social Work, 30, 350-354.

Aldarondo, E. (1996). Cessation and persistence of wifeassault: A longitudinal analysis. American Journal ofOrthopsychiatry, 66(1), 141-151.

Amato, P., & Partridge, S. (1987). Widows and divorceeswith dependent children: Material, personal, family,and social well-being. Family Relations, 36, 316-320.

Anderson, D. (2001). Predicting long-term psychological well-being in women who have left their abusive partners.

Unpublished doctoral dissertation, University ofMichigan.

Anderson, D., Saunders, D., Yoshihama, M., Bybee, D., &Sullivan, C. (in press). Long-term trends in psychologi-cal well-being among women separated from abusivepartners. Violence Against Women.

Angless, T., Maconachie, M., & Van Zyl, M. (1998). Batteredwomen seeking solutions: A South African study. Vio-lence Against Women, 4(6), 637-657.

Astin, M., Lawrence, K., & Foy, D. (1993). Posttraumaticstress disorder among battered women: Risk and resil-iency factors. Violence and Victims, 8(1), 17-28.

Bachman, R., & Saltzman, L. (1995). Violence against women:Estimates from the redesigned survey. Washington, DC:U.S. Department of Justice.


Barnett, O., & LaViolette, A. (1993). It could happen to any-one: Why battered women stay. Newbury Park, CA: Sage.

Benight, C., Swift, E., Sanger, J. Smith, A., & Zeppelin, D.(1999). Coping self-efficacy as a mediator of distress fol-lowing a natural disaster. Journal of Applied Social Psy-chology, 29(12), 2443-2464.

Berman, W. (1988). The relationship of ex-spouse attach-ment to adjustment following divorce. Journal of FamilyPsychology, 1(3), 312-328.

Bianchi, S., Subaiya, L., & Kahn, J. (1999). The gender gap inthe economic well-being of nonresident fathers andcustodial mothers. Demography, 36(2), 195-203.

Bowker, L. (1983). Beating wife beating. Lexington, MA:Lexington Books.

Brown, B., & Foye, B. (1982). Divorce as a dual transition:Interpersonal loss and role restructuring. In V. Allen &E. van de Vliert (Eds.), Role transitions: Explorations andexplanations (pp. 315-329). New York: Plenum.

Brown, J. (1997). Working toward freedom from violence:The process of change in battered women. ViolenceAgainst Women, 3(1), 5-26.

Browne, A., & Bassuk, S. (1997). Intimate violence in thelives of homeless and poor housed women: Prevalenceand patterns in an ethnically diverse sample. AmericanJournal of Orthopsychiatry, 67(2), 261-278.

Buehler, C., Hogan, M., Robinson, B., & Levy, R. (1985-1986). The parental transition: Divorce-related stressorsand well-being. Journal of Divorce, 9(2), 61-79.

Burke, J., Gielen, A., McDonnell, K., O’Campo, P., &Maman, S. (2001). The process of ending abuse in inti-mate relationships: A qualitative exploration of theTranstheoretical Model. Violence Against Women, 7(10),1144-1163.

Bursik, K. (1991). Correlates of women’s adjustment dur-ing the separation and divorce process [Special issue].Journal of Divorce & Remarriage 14(3/4), 137-162.

Campbell, J. (1989). A test of two explanatory models ofwomen’s responses to battering. Nursing Research,38(1), 18-24.

Campbell, J., Kub, J., Belknap, R. A., & Templin, T. (1997).Predictors of depression in battered women. ViolenceAgainst Women, 3(3), 271-293.

Campbell, J., Miller, P., Cardwell, M., & Belknap, R. (1994).Relationship status of battered women over time. Jour-nal of Family Violence, 9(2), 99-111.

Campbell, J., Rose, L., Kub, J., & Nedd, D. (1998). Voices ofstrength and resistance: A contextual and longitudinalanalysis of women’s responses to battering. Journal ofInterpersonal Violence, 13(6), 743-762.

Campbell, R., Sullivan, C., & Davidson, W. (1995). Womenwho use domestic violence shelters: Changes in depres-sion over time. Psychology of Women Quarterly, 19, 237-255.

Carlin, R. (1998). Developing self: Women’s process in end-ing an abusive relationship. (Doctoral dissertation, Uni-versity of Toronto, 1998). Dissertation Abstracts Interna-tional, 59(6-A), 1918.

Compton, W., Michael, J., Krasavage-Hopkins, E.,Schneiderman, L., & Bickman, L. (1989). Intentions for

postshelter living in battered women. Journal of Commu-nity Psychology, 17, 126-128.

Correia, A. (2001). Housing and battered women: A case studyof domestic violence programs in Iowa. Washington, DC:U.S. Department of Justice. Retrieved May 20, 2001,from http://www.vaw.umn.edu/FinalDocuments/housing.asp

Cox, J., & Stoltenberg, C. (1991). Evaluation of a treatmentprogram for battered wives. Journal of Family Violence,6(4), 395-413.

Dalto, C. (1983). Battered women: Factors influencingwhether or not former shelter residents return to theabusive situation. (Doctoral dissertation, University ofMassachusetts, 1983). Dissertation Abstracts Interna-tional (4-B), 1277.

DeGarmo, D. S., & Kitson, G. C. (1996). Identity relevanceand disruption as predictors of psychological distressfor widowed and divorced women. Journal of Marriage& the Family, 58(4), 983-997.

Dixon, C., & Rettig, K. (1994). An examination of incomeadequacy for single women two years after divorce.Journal of Divorce and Remarriage, 22(1/2), 55-71.

Dobash, R., & Dobash, R. (1979). Violence against wives: Acase against the patriarchy. New York: Free Press.

Dobash, R., & Dobash, R. (1988). Research as social action:The struggle for battered women. In K. Yllo & M.Bograd (Eds.). Feminist perspectives on wife abuse (pp. 51-74). Newbury Park, CA: Sage.

Dobash, R., & Dobash, R. (1992). Women, violence, and socialchange. London: Routledge.

Duffy, M. (1989). Mental well-being and primary preven-tion practices in women heads of one-parent families.Journal of Divorce, 13(1), 45-64.

Duncan, G., & Hoffman, S. (1985). A reconsideration of theeconomic consequences of marital dissolution. Demog-raphy, 22, 485-497.

Dutton, D., & Painter, S. (1981). Traumatic bonding: Thedevelopment of emotional attachments in batteredwomen and other relationships of intermittent abuse.Victimology: An International Journal, 1(4), 139-155.

Dutton, D., & Painter, S. (1993). Emotional attachments inabusive relationships: A test of traumatic bonding the-ory. Violence and Victims, 8(2), 105-120.

Eisikovits, Z., Buchbinder, E., & Mor, M. (1998). What itwas won’t be anymore: Reaching the turning point incoping with intimate violence. Affilia, 13(4), 411-434.

Eldar-Avidan, D., & Haj-Yahia, M. (2000). The experienceof formerly battered women with divorce: A qualita-tive, descriptive study. Journal of Divorce and Remarriage,32(3/4), 19-40.

Ellis, D. (1992). Woman abuse among separated anddivorced women: The relevance of social support. In E.Viano (Ed.), Intimate violence: Interdisciplinary perspec-tives (pp. 177-189). Washington, DC: Hemisphere.

Ferraro, K., & Johnson, J. (1983). How women experiencebattering: The process of victimization. Social Problems,30(3), 325-338.

Frisch, M., & MacKenzie, C. (1991). A comparison of for-merly and chronically battered women on cognitive


and situational dimensions. Psychotherapy, 28(2), 339-344.

Garvin, V., Kalter, N., & Hansell, J. (1993). Divorcedwomen: Factors contributing to resiliency and vulnera-bility. Journal of Divorce and Remarriage, 21(1/2), 21-39.

Gelles, R. (1976). Abused wives: Why do they stay? Journalof Marriage and the Family, 38, 659-668.

Gerstel, N., Riessman, C., & Rosenfield, S. (1985).Explaining the symptomology of separated anddivorced women and men: The role of material condi-tions and social networks. Social Forces, 64(1), 84-101.

Goetting, A. (1999). Getting out: Life stories of women who leftabusive men. New York: Columbia University Press.

Gondolf, E., & Fisher, E. (1988). Battered women as survivors.Lexington, MA: Lexington Books.

Gray, C., Koopman, E., & Hunt, J. (1991). The emotionalphases of marital separation: An empirical Investiga-tion. American Journal of Orthopsychiatry, 61(1), 138-143.

Guerin, P., Fay, L., Burden, S., & Kautto, J. (1987). The evalu-ation of treatment of marital conflict: A four-stage approach.New York: Basic Books.

Herbert, T., Silver, R., & Ellard, J. (1991, May). Coping withan abusive relationship: I. How and why do womenstay? Journal of Marriage and the Family, 53, 211-325.

Hilbert, J., & Hilbert, H. (1984). Battered women leavingshelter: Which way do they go? Adiscriminant functionanalysis. The Journal of Applied Social Sciences, 8(2), 291-297.

Hilbert, J., Kolia, R., & VanLeeuwen, D. (1997). Abusedwomen in New Mexican shelters: Factors that influenceindependence on discharge. Affilia, 12(4), 391-407.

Hobfoll, S., & Vaux, A. (1993). Social support: Socialresources and social context. In L. Goldberger & S.Breznitz (Eds.), Handbook of stress: Theoretical and clinicalaspects (pp. 686-705). New York: Free Press.

Hoff, L. (1990). Battered women as survivors. New York:Routledge.

Holt, D. (1995). An exploration of the decision to leave anabusive partner. (Doctoral dissertation, University ofAlabama, 1995). Dissertation Abstracts International,56(6-A), 2417.

Horton, A., & Johnson, B. (1993). Profile and strategies ofwomen who have ended abuse. Families in Society: TheJournal of Contemporary Human Services, 74(8), 481-492.

Humphreys, J. (1995). The work of worrying: Batteredwomen and their children. Scholarly Inquiry for NursingPractice, 9(2), 127-145.

Jacobson, N., Gottman, J., Gortner, E., Berns, S., & WuShortt, J. (1996). Psychological factors in the longitudi-nal course of battering: When do the couples split up?When does the abuse decrease? Violence and Victims,11(4), 371-392.

Johnson, H. (1995). Risk factors associated with non-lethalviolence against women by marital partners. In C. Block& R. Block (Eds.), Questions and answers in lethal andnonlethal violence (Vol. 3). Washington, DC: NationalInstitute of Justice.

Johnson, I. (1992). Economic, situational, and psychologi-cal correlates of the decision-making process of bat-

tered women. Families in Society: The Journal ofContemporary Human Services, 73, 168-176.

Kemp, A., Green, B. L., Hovanitz, C., & Rawlings, E. I.(1995, March). Incidence and correlates ofposttraumatic stress disorder in battered women: Shel-ter and community samples. Journal of Interpersonal Vio-lence, 10(1), 43-55.

Kirkwood, C. (1993). Leaving abusive partners: From the scarsof survival to the wisdom for change. Newbury Park, CA:Sage.

Kurz, D. (1996). Separation, divorce, and woman abuse.Violence Against Women, 2(1), 63-81.

Landenburger, K. (1989). A process of entrapment in andrecovery from an abusive relationship. Issues in MentalHealth Nursing, 10, 209-227.

Lehnen, R., & Skogan, W. (1981). The National Crime Survey:Working papers, Vol. 1, current and historical perspectives.Washington, DC: U.S. Department of Justice.

Lerner, C. F., & Kennedy, L. T. (2000.) Stay-leave decisionmaking in battered women: Trauma, coping and self-efficacy. Cognitive Therapy and Research, 24(2), 215-232.

Lesser, B. (1990). Attachment and situational factors influ-encing battered women’s return to their mates follow-ing a shelter program. In K. Pottharst (Ed.), Researchexplorations in adult attachment (pp. 81-128). New York:Peter Lang.

Lindsay, G., & Scherman, A. (1987). The impact of childrenand time lapse since divorce on women’s postdivorceadjustment. Journal of Mental Health Counseling, 9(3),174-183.

Martin, A., Berenson, K., Griffing, S., Sage, R., Madry, L.,Bingham, L., et al. (2000). The process of leaving an abu-sive relationship: The role of risk assessments and deci-sion-certainty. Journal of Family Violence, 15(2), 109-122.

Mechanic, M., Weaver, T., & Resick, P. (2000). Intimate part-ner violence and stalking behavior: Exploration of pat-terns and correlates in a sample of acutely batteredwomen. Violence and Victims, 15(1), 55-72.

Merritt-Gray, M., & Wuest, J. (1995). Counteracting abuseand breaking free: The process of leaving revealedthrough women’s voices. Health Care for Women Interna-tional, 16, 399-412.

Mertin, P., & Mohr, P. (2001.) A follow-up study ofposttraumatic stress disorder, anxiety, and depressionin Australian victims of domestic violence. Violence andVictims, 16, 645-654.

Miller, N., Smerglia, V., Gaudet, D., & Kitson, G. (1998).Stressful life events, social support, and the distress ofwidowed and divorced women. Journal of Family Issues,19(2), 181-203.

Mills, T. (1985). The assault on the self: Stages in copingwith battering husbands. Qualitative Sociology, 8(2), 103-123.

Mitchell, R., & Hodson, C. (1983). Coping with domesticviolence: Social support and psychological healthamong battered women. American Journal of CommunityPsychology, 11(6), 629-654.


Molina, O. (1999). The effect of divorce on African Ameri-can working women. Journal of Divorce and Remarriage,32(1/2), 1-15.

Moss, V., Pitula, C., Campbell, J., & Halstead, L. (1997). Theexperience of terminating an abusive relationship froman Anglo and African American perspective: A qualita-tive descriptive study. Issues in Mental Health Nursing,18, 433-454.

Nelson, G. (1989). Life strains, coping, and emotional well-being: A longitudinal study of recently separated andmarried women. American Journal of Community Psy-chology, 17(4), 459-483.

NiCarthy, G. (1987). The ones who got away: Women who leftabusive partners. Seattle, WA: Seal.

Nurius, P., Furrey, J., & Berliner, L. (1992). Coping capacityamong women with abusive partners. Violence and Vic-tims, 7(3), 229-243.

Okun, A. (1998). Battered women who leave: An explor-atory model of factors involved in the separation pro-cess. (Doctoral dissertation, University of Michigan,1998). Dissertation Abstracts International, 59(2-B), 882.

Okun, L. (1986). Woman abuse: Facts replacing myths. NewYork: State University of New York Press.

Orava, T., McLeod, P., & Sharpe, D. (1996). Perceptions ofcontrol, depressive symptomotology, and self-esteemof women in transition from abusive relationships. Jour-nal of Family Violence, 11(2), 167-186.

Pagelow, M. (1981). Factors affecting women’s decisions toleave violent relationships. Journal of Family Issues, 2(4),391-414.

Patzel, B. (2001.) Women’s use of resources in leaving abu-sive relationships: A naturalistic inquiry. Issues in Men-tal Health Nursing, 22, 729-747.

Pearlin, L. (1999). The stress process revisited: Relectionson concepts and their interrelationships. In C.Aneshensel & J. Phelan (Eds.(, Handbook of the sociologyof mental health (pp. 395-415). New York: Kluwer Aca-demic/Plenum.

Pearlin, L., Lieberman, M., Menaghan, E., & Mullan, J.(1981). The stress process. Journal of Health and SocialBehavior, 22, 337-356.

Peled, E., Eisikovits, Z., Enosh, G., & Winstok, Z. (2000).Choice and empowerment for battered women whostay: Toward a constructivist model. Social Work, 45(1),9-25.

Pleck, E. (1987). Domestic tyranny: The making of social policyagainst family violence from colonial times to the present.New York: Oxford University Press.

Pledge, D. S. (1992). Marital separation/divorce: A reviewof individual responses to a major life stressor. Journal ofDivorce & Remarriage, 17(3/4), 151-181.

Prochaska, J., & DiClemente, C. (1984). The transtheoreticalapproach: Crossing traditional boundaries of change. Home-wood: Dow Jones/Irwin.

Prochaska, J., Norcross, J., & DiClemente, C. (1994).Changing for good. New York: William Morrow.

Raphael, J. (1999). Keeping women poor: How domesticviolence prevents women from leaving welfare andentering the world of work. In R. Brandwein (Ed.), Bat-

tered women, children, and welfare reform: The ties that bind(pp. 31-43). Thousand Oaks, CA: Sage.

Rhodes, N., & McKenzie, E. (1998). Why do batteredwomen stay? Three decades of research. Aggression andViolent Behavior, 3(4), 391-406.

Rosen, K., & Stith, S. (1997). Surviving abusive dating rela-tionships: Processes of leaving, healing, and movingon. In G. K. Kantor & J. L. Jasinski (Eds.), Out of the dark-ness: Contemporary perspectives on family violence (pp.170-182). Thousand Oaks, CA: Sage.

Rounsaville, B. (1978). Theories in marital violence: Evi-dence from a study of battered women. Victimology,3(1/2), 11-31.

Rusbult, C., & Martz, J. (1995). Remaining in an abusiverelationship: An investment analysis of nonvoluntarydependence. Personality and Social Psychology Bulletin,21(6), 558-571.

Sato, R., & Heiby, E. (1992). Correlates of depressive symp-toms among battered women. Journal of Family Violence,7(3), 229-245.

Schutte, N., Malouff, J., & Doyle, J. (1988). The relationshipbetween characteristics of the victim, persuasive tech-niques of the batterer, and returning to a battering rela-tionship. Journal of Social Psychology, 128(5), 605-610.

Shainess, N. (1979). Vulnerability to violence: Masochismas process. American Journal of Psychotherapy, 33, 174-189.

Shapiro, A. D. (1996). Explaining psychological distress ina sample of remarried and divorced persons: The influ-ence of economic distress. Journal of Family Issues, 17(2),186-203.

Sleutel, M. 1998. Women’s experiences of abuse: A reviewof qualitative research. Issues in Mental Health Nursing,19, 525-539.

Snell, J., Rosenwald, R., & Robey, A. (1964). Thewifebeater ’s wife: A study of family interaction.Archives of General Psychiatry, 11, 107-112.

Snyder, D., & Scheer, N. (1981). Predicting disposition fol-lowing brief residence at a shelter for battered women.American Journal of Community Psychology, 9(5), 559-566.

Statistics Canada. (2001). Family violence in Canada: A statis-tical profile. Ottawa, Ontario: Canadian Centre for Jus-tice. Retrieved November 3, 2002, from http://www.statcan.ca

Strube, M. (1988). The decision to leave an abusive rela-tionship: Empirical evidence and theoretical issues.Psychological Bulletin, 104(2), 236-250.

Strube, M., & Barbour, L. (1983). The decision to leave anabusive relationship: Economic dependence and psy-chological commitment. Journal of Marriage and the Fam-ily, 45, 785-793.

Strube, M., & Barbour, L. (1984). Factors related to the deci-sion to leave an abusive relationship. Journal of Marriageand the Family, 46, 837-844.

Sullivan, C., Basta, J., Tan, C., & Davidson, W. (1992). Afterthe crisis: A needs assessment of women leaving adomestic violence shelter. Violence and Victims, 7(3), 267-274.


Sullivan, C., & Bybee, D. (1999). Reducing violence usingcommunity based advocacy for women with abusivepartners. Journal of Consulting and Clinical Psychology,67(1), 43-53.

Sullivan, C., Campbell, R., Angelique, H., Eby, K., &Davidson, W. (1994). An advocacy intervention pro-gram for women with abusive partners: Six-month fol-low-up. American Journal of Community Psychology,22(1), 101-122.

Sullivan, C., & Davidson, W. (1991). The provision of advo-cacy services to women leaving abusive partners: Anexamination of short-term effects. American Journal ofCommunity Psychology, 19(6), 953-960.

Sullivan, C., Tan, C., Basta, J., Rumptz, M., & Davidson, W.(1992). An advocacy intervention program for womenwith abusive partners: Initial evaluation. American Jour-nal of Community Psychology, 20(3), 309-332.

Syers-McNairy, M. (1990). Women who leave violent rela-tionships: Getting on with life: I and II. (Doctoral disser-tation, University of Minnesota, 1990). DissertationAbstracts International, 51(6-A), 2159.

Tan, C., Basta, J., Sullivan, C., & Davidson, W. (1995). Therole of social support in the lives of women exitingdomestic violence shelters: An experimental study.Journal of Interpersonal Violence, 10(4), 437-451.

Thabes, V. (1997). Survey analysis of women’s long-term,postdivorce adjustment. Journal of Divorce & Remarriage,27(3/4), 163-175.

Thiriot, T., & Buckner, E. (1992). Multiple predictors of sat-isfactory post-divorce adjustment of single custodialparents. Journal of Divorce and Remarriage, 17(1/2), 27-48.

Tjaden, P., & Thoennes, N. (2000). Extent, nature, and conse-quences of intimate partner violence: Findings from theNational Violence Against Women Survey (Report spon-sored by the National Institute of Justice). Washington,DC: National Institute of Justice.

Tolman, R., & Rosen, D. (2001). Domestic violence in thelives of women receiving welfare. Violence AgainstWomen, 7(2), 141-158.

Truman-Schram, D., Cann, A., Calhoun, L. , &Vanwallendael, L. (2000). Leaving an abusive datingrelationship: An investment model comparison ofwomen who stay versus women who leave. Journal ofSocial and Clinical Psychology, 19(2), 161-183.

Turner, S., & Shapiro, C. (1986). Battered women:Mourning the death of a relationship. Social Work, 31,372-376.

Tutty, L. (1998, July). The importance of follow-up programs forabused women. Paper presented at Program Evaluationand Family Violence Research: An international confer-ence at the University of New Hampshire, Durham.

Tutty, L., Bidgood, B., & Rothery, M. (1993). Supportgroups for battered women: Research on their efficacy.Journal of Family Violence, 8(4), 325-343.

Ulrich, Y. (1991). Women’s reasons for leaving abusivespouses. Health Care for Women International, 12, 465-473.

Ulrich, Y. (1998). What helped most in leaving spousalabuse: Implications for Interventions. In J. Campbell(Ed.), Empowering survivors of abuse: Health care for bat-tered women and their children. Thousand Oaks, CA: Sage.

Varvaro, E. (1991). Using a grief response assessment ques-tionnaire in a support group to assist battered women intheir recovery. Response, 13(4), 17-20.

Voydanoff, P., & Kelly, R. (1984). Determinants of work-related family problems among employed parents.Journal of Marriage and the Family, 46, 881-892.

Walker, L. (1984). The battered woman syndrome. New York:Springer.

Wheaton, B. (1999). The nature of stressors. In A. Horwitz& L. Scheid (Eds.), A handbook for the study of mentalhealth: Social contexts, theories, and systems (pp. 176-197).New York: Cambridge University Press.

Wilcox, B. (1986). Stress, coping, and the social milieu ofdivorced women. In S. Hobfoll (Ed.), Stress, social sup-port, and women (pp. 115-132). New York: Hemisphere.

Williams, J. (2000). Factors related to women’s stages of ter-minating violent intimate relationships. (Doctoral dis-sertation, Indiana University, 2000). DissertationAbstracts International, 61(3-B), 1333.

Wilson, J. 1999. Surviving well: Decision-making and resil-iency of African American and Euro-American womenwho have successfully left loved and dangerous part-ners. (Doctoral dissertation, University of Oklahoma,1999). Dissertation Abstracts International, 60(5-A), 1796.

Wilson, M., & Daly, M. (1993). Spousal homicide risk andestrangement. Violence and Victims, 8, 3-15.

Wilson, M., Johnson, H., & Daly, M. (1995). Lethal and non-lethal violence against wives. Canadian Journal of Crimi-nology, 37, 331-362.

Wuest, J., & Merritt-Gray, M. (1999). Not going back: Sus-taining the separation in the process of leaving abusiverelationships. Violence Against Women, 5(2), 110-133.

Young, G., & Gerson, S. (1991) New psychoanalytic per-spectives on masochism and spouse abuse. Psychother-apy, 28(1), 30-38.

SUGGESTED FUTURE READINGSRhodes, N., & McKenzie, E. (1998). Why do battered

women stay? Three decades of research. Aggression andViolent Behavior, 3(4), 391-406.

Sleutel, M. (1998). Women’s experiences of abuse: Areviewof qualitative research. Issues in Mental Health Nursing,19, 525-539.

Anderson, D., Saunders, D., Yoshihama, M., Bybee, D., &Sullivan, C. (in press). Long-term trends in psychologi-cal well-being among women separated from abusivepartners. Violence Against Women.

Deborah K. Anderson is a National Insti-tute of Mental Health postdoctoral researchtrainee at the University of Michigan’s Insti-tute for Research on Women and Gender. Herresearch focuses primarily on violence against


women, women’s psychological well-being, and the pro-cess of leaving abusive partners. Other research interestsinclude evaluation of domestic violence training pro-grams. She recently coauthored an article assessing theeffectiveness of such a training program for child protec-tion workers and their supervisors.

Daniel G. Saunders, Ph.D., is an associateprofessor at the University of MichiganSchool of Social Work where he teaches

courses on direct practice and domestic violence. Hisresearch on domestic violence focuses on abuser types andtreatment, the traumatic aftermath of violence victimiza-tion, and the attitudes and responses of professionals. Hewas a postdoctoral research fellow at the Family ViolenceResearch Program of the University of New Hampshireand a postdoctoral research fellow and associate scientistat the Department of Psychiatry at the University of Wis-consin–Madison. His professional publications haveappeared in numerous journals and books.
