The Pros and Cons of the Learned Helplessness Construct and the Battered Woman Syndrome: A Critical Analysis and Possible Reformulation

Two prevailing, contrasting theories, learned helplessness and health psychogenesis, have been proposed to describe and explain the battered woman's experience and response to domestic violence. This research examined the status (scientific adequacy) of each theory's database to determine where the preponderance of

examined the status (scientific adequacy) of each theory's database to determine where the preponderance of evidence lies.
The investigator articulated nine factors that differentiate the two theories. Trained independent raters then rated study outcomes as falling into one of three categories: pro-learned helplessness, pro-health psychogenesis, or supportive of both theories. Studies of two research samples (shelter and community) were identified and assessed (rated) along four methodological dimensions (sample, instrumentation, statistical analysis, and hypothesis testing) . Each was assigned a total design quality score, and -the studies were then divided into three groups according to the overall level of design quality (high, medium, low). A table was then created that exhibited the overall picture of design quality by outcome (the theory supported by the study).
The research relevant to each theory appeared to be roughly equal in overall design quality. The majority o f the research evidence appeared to support the pro-health psychogenesis theory. However, the difference between the percent~ge of evidence that supported psychogenesis versus learned helplessness was modest (16 studies, or 55%; 10 studies, or 34%). It may be argued that a theory that appears to be validated (correct) approximately 50 % of the time (health psychogenesis) or approximately onethird of the time (learned helplessness) does not reflect a strong or well-articulated theory. One would expect that a theory's predictions would, at least, be greater than chance.
These results suggest that rather than trying to determine which of the two theories is better, a more fruitful approach might be to develop an integrated theory that draws on the strongest elements of each. The dissertation concludes with a number of suggestions regarding how research and clinical practice might begin this process~      The study of family violence encompasses a range of subdomains, including elder, child, interpartner, and wife abuse. This rapidly growing field faces a crisis of ideas in the form of major, competing explanatory theories.

List of Tables
The current lack of consensus among family violence experts flows from debate over maj or methodological issues and multiple, simultaneous, at times competitive, conceptual lenses : psychological, sociological, feminist, and family systems. Major controversies cut across all forms of family violence with special implications for women and children.
Core dilemmas involve: • definitions, measurement, and scope of the problem; • root causes and explanatory theories; • dynamics of family violence; 2 • the relationship between types and levels of abuse, or the underlying continuity-discontinuity question: Do low levels of physical violence progress incrementally to higher levels, or are low/mild and high/severe levels of physical violence quali~atively distinct in form and type, each with different causes?; • the intergenerational transmission hypothesis, or risk and intervention focus; • treatment method: separate therapy or concurre nt treatments of perpetrator and survivor; and • the relevance, validity, and social consequences of the constructs of learned helplessness and the battered woman syndrome (BWS).
This dissertation focused on the wife abuse f arnily violence subcategory, its effects on women and children, and the strengths and competencies characteristic of the survivor-process.

Wife Abuse
The following statistics highlight the prevalence and pervasiveness of domestic violence and paint a disturbing profile of the lives and experiences of women and children living with family violence. Jones (1996), citing national statistics, underscored the general rise in rates of violent acts against women in our society.
She stated, " A few years ago the FBI reported that in the United States a man beat a woman every eighteen seconds .
By 1989 the figure was fifteen seconds. Now [in 1992) it's twelve (p . 2) ." Although these statistics reflect the general increase in men's violence toward women, the statistics on violence directed toward women and children inside the family are even more alarming. Notions of "marriage" and "family" are usually associated with intimacy and interpersonal safety. O 'Leary, Barling, Arias, Rosenbaum, Malone, and Tyree (1989), in their longitudinal analysis, reported that one out of every three marriages experienced marital violence and nearly half of all dating/marital relationships experienced some form of interpartner aggression.
According to Straus and Gelles' 1985 National Family Violence Resurvey, reported in their book Physical Violence in American Families (1990), one out of every eight women is hit by her husband each year. In addition, over the course of marriage, about one in four American couples experience husband-to-wife violence.
Hansen and Harway (1993) stated that the nature of the assaults included physical injuries such as: fractures, chronic back injuries, torn ligaments, dislocations, ruptured eardrums, broken teeth, lacerations, and stab wounds.
As Straus and Gelles (1990) demonstrated, the actual rates of assault and injury may, in fact, be double the reported rates, given the "virtual certainty" that respondents were not always completely candid in reporting violent incidents. As research evidence has suggested (e.g., Bowker, 1986;Hansen & Harway, 1993), this reticence may be partially a function of shame and humiliation. Most women seemingly would be loath to admit abuse and victimization at the hands of a spouse or partner who promised to "love, honor and cherish" them and the children of the marriage. Whereas guilt is about doing and remorse over actual transgressions, shame is about being and the broken bonds of attachment.
Crime victims have typically reported reluctance in reliving the painful details of the assault (Straus & Gelles, 1990). Battered women and children who are victims of intimate crime often feel betrayed by the intimacy associated with the crime and their close relationship with the perpetrator. · Studies conducted in Texas (Teske & Parker, 1983) and Kentucky (Schulman, 1979), have indicated that as much as 91% of spousal or intimate violence may never be reported to the police. personal, interpersonal, and system boundaries.

Family Violence and its Effects on Children
Children are also physical and emotional casualties of family violence. They are drawn into the ritualized cycles of family conflict and aggression in many direct and indirect ways. When a woman is battered and she makes the crucial decision to seek help, undertaking what is of ten the long legal and psychological road to freedom, her children automatically join in the journey. In Canada, based on an analysis of intake data, MacLeod (1987) reported that 70% of battered women seeking shelters brought children with them, with 17 % of the women bringing three or more children. Therefore, children are immersed in the crisis and abrupt transitions of the abuse experience. 8 The effects of witnessing abuse and accommodating to sudden changes in family life circumstances can affect children in many ways . MacLeod's (1987) statistics have shown that large numbers of children often witness and are embroiled in family violence. Children of family violence can become the targets not only of physical but psychological abuse. Violence does not always take the form of physical aggression and physical injury to others, although this is often the case when family abuse escalates and becomes overt.
As a number of researchers have i ndicated (e.g., Jaffe, Wolfe, and Wilson, 1990;Peled and Davis, 1995), there are other equally damaging, subtle forms of violence that take their toll on children: • a parent's inability t o s ee a child for who the child is; • a parent's blind indifference to a child's aspirations and accomplishments; • an overwrought parent's inability to be emotionally accessible or an emotional support to their child; • a parent's discouragement; • or worse, a parent's steady, deprecating criticism and emotional rejection of a child that can profoundly effect a child's mind, spirit, imagination, empathy, and expectations for his or her future. Such forms of abuse or neglect can exert a significant effect on child development (discussed in detail in Chapter 7).

The Effects of Witnessing Family Violence
Even when children escape direct physical abuse (e.g., are not hit), they often suffer, as true trauma victims do, by what they witness and experience.
Research (Arnold, 1990;Azar, 1991;Azar, Barnes, & Twentyman, 1988) has indicated that children are usually present during incidents of family violence, that is, instances of physical and verbal abuse directed at one spouse by another. Jaffe, Wolfe, and Wilson (1990) estimated that three to four million children per year witness interparental violence. Carlson (1984) estimated that 3.3 million children between the ages of three and 17 witnessed, or directly experienced, family violence.
Social learning theory, exemplified by Bandura's (1986) social cognition theory and research on modeling behavior, has strongly suggested that children absorb and acquire models of behavior through vicarious as well as observational modes of learning. Children can learn the lessons of aggression by absorbing and internalizing violent models: as they watch and witness what occurs, as they overhear the sounds of angry physical and emotional confrontation from behind closed doors, or from their beds at night.
In this paradoxical form of observed aggression, the experience can be more potent and compelling for several reasons. Most obviously , the child often feels that she or he cannot do anything about the violence; the child perceives her-or himself as a helpless and disabled witness.
Children who are involved aggressively with the abusive parent (e . g., rescuing or protecting the abused parent) are at least doing something, however destructive and inappropriate those actions might be. The child is not ready emotionally to handle this kind of power, power that the child was given by default. The child who witnesses someone hurting someone else has the added emotional burden of not perceiving him or herself as able to help, to act, or effectively intervene.
In not being able to stop people from getting hurt, the child can have his or her very notion of personal competence and selfeff icacy stripped away.
Research on childhood stress and family violence (Arnold, 1990;Carlson, 1984;Deblinger, McLeer, Atkins, Ralphe, & Foa, 1988) has suggested that being a witness to one parent's brutality toward the other, to be reduced to a helpless onlooker, can be, in itself, a traumatic and disturbing experience. To over-hear the violence, to have it take place outside of the child's plain view and real powers of cognitive integration, can be additionally destructive and troubling . When events are not seen, they can become more terrifying, as children's imaginations are left to fill in the details. When such ambiguity about potentially terrifying events exists, their impact and magnitude may be magnified. This is the poignant situation for many children immersed in family violence, dysfunction, and cycles of abuse .  (Arnold, 1990;Azar, Barnes, & Twentyman, 1988;Jaffe, Wolfe, & Wilson, 1990;Leeder, 1994;Wolfe, Jaffe, Wilson, & Zak, 1985), have included scenes of psychological and actual aggression. These involved internalized representations as well as overt scenes that showed some form of symbolic or actual violence against mind, body, relationships, and self-integrity. In effect, the child is denied important access and af f irmation--not being seen, known, heard, and found by those the child cares about and assumes care about him or her (Andreozzi, 1996) Although it was commonly believed that the learned helplessness model preceded the health psychogenesis model, they were contemporaneous theories. This common historical misimpression was partly due to the widespread popularity and publicity surrounding learned help~essness .
In the past two decades this theory overshadowed the scientific merits · of an equally plausible health psychogenesis theory. The historical development of each theory occurred as follows.

Learned Helplessness
In the late 1970s, Lenore Walker conducted a three- year qualitative, retrospective study of battered women (N = 403). These women, primarily residents of the Rocky Mountain region, comprised a self-selected sample.
Indepth interviews with battered women were conducted to In an explore their psychological responses to violence. attempt to explain why battered women remained in intolerable relationships, Walker hypothesized and tested two theories --learned helplessness and a tensionreduction cycle theory of violence. Focusing on "leav ing the relationship" as the critical variable associated with effective coping strategies, Walker (1984) attempted to explain why battered women remained in abusive relationships .
Walker reasoned that leaving the relationship was essentially the single most positive outcome, synonymous with self-care, self-protection, and healthy, adaptive self-assertion. To explain why women stayed, she adopted and applied Seligman's (Abramson, Seligman, & Teasdale , 1978;Seligman, 1975;Seligman & Maier, 1967) reformulated learned helplessness theory to the behavioral response patterns and thought process of battered women. She concluded that battered women remained in abusive relationships because they relied on and developed ineffective coping strategies, the outgrowth of chronic abuse. Her research gene rated a great deal of public attention and her proposed theory took hold, having a far-reaching influence on prevailing treatment models, grass root women's advocacy programs , and the popularization of clinical profiles of battered women.
As a consequence of Walker's published research and learned helplessness theory, battered women were mainly seen as having cognitive, motivational, and affective deficits.
Similar to the research of cognitive attribution theorists (Abramson, Seligman, & Teasdale, 1978;Hiroto, 1974;Seligman, 1975;Seligman & Maier, 1967), Walker's theory was constructed on the hypothesis that non-contingent outcomes (e.g . , a person's perceived inability either to predict or control events) led to depression. She asserted that these deficits caused battered women to be passive, dependent, and unrealistic in their life choices and appraisal of their relationships. Descriptions of their survivor experienc e emphasized their anxiety, lack of effective coping strategies and failure to actively seek outside help.
Thus, battered women became a clinical population unable to solve problems, unwilling to seek help or follow through, and unrealistically bonded to the abuser, denying their own outrage as a means of self-pr o t e c ti o n and defense. As a result of the prevailing learned helplessness view, battered women, as a group of mental health clients and family violence survivors, we re ma inly seen as suffering from a stress response syndrome.
symptoms included anxiety, depression, agitation, denial, emotional labiality, and extreme passivity. Learned helplessness supported a psychological, internalized, developmental or genetic, and predominantly clinical model. Walker (1983) wrote, "Events which occurred in childhood, as well as in the relationship, interact with the violence and effect the woman's current state" (p.32).

Health Psychogenesis
However Similarly, the notion that unrealistic expectations of the abusive relationship guided battered women's behavior has also been challenged. The learned helplessness theory has presupposed impaired cognitive functioning and serious perceptual distortions as a function of abuse. Consequently, the phenomenon of why women remained with their abusers has been attributed to acquired psychological dysfunction . Women were thought to remain because they were unrealistic about the help that is available to them. In contrast, Gondolf 's (1988b) research has shown that battered women made numerous attempts to stop the violence and displayed a wide range of active, help-seeking behaviors. Bowker (1983aBowker ( , 1993 has also shown that battered women were effective long-term planners and that they stayed for sound, practical reasons. Her research, which involved a national sample of over 1,000 battered women, suggested that battered women remained until it was safe and advantageous for them to leave. A growing number of researchers (Bowker, 1983a(Bowker, , 1983b(Bowker, , 1993Dobash & Dobash, 1979;Okun, 1986) have shown that these women's actions were purposeful and planned, active and not passive. This viewpoint, grounded in a competency-based health psychogenesis theory, adopted from Bowker's (1993) descriptive term, presents a sociological, normative view of the battered woman's experience and behavior framed within the cultural context of broader "women's issues." The health psychogenesis theory has proposed a social, rather than a psychological, definition of and solution for the problems of wife abuse, treatment, and recovery. This dissertation critically examined both sides of the research controversy surrounding learned helplessness and a competency-based concept of treatment and recovery.

Frames of Reference for Understanding the Abuse Dynamic
Properly conceptualizing the learned helplessness- Examples of four overviews or frameworks of major explanatory theories of family violence include Hansen and Harway (1993), Browne (1989), Viano (1992), and Gelles and Straus (1979 The import of their article has sugge sted that t h e popular view of battere d women's res e arch may n ot a u ger well for theoretical models of victimization that focus on characteristics of victims. As a review of the evidence has indicated, the research literature (e.g., Bowker, 1983aBowker, , 1986Gondolf, 1988b) has strongly supported the converse: that battered women consistently demonstrated positive coping skills, ego strengths, resourcefulness, and often ingenious powers of survival.

Men Who Batter
Prior to the mid-1980s, there was a concerted effort to view the batterers' behavior as an outgrowth of normal male socialization. As Vaselle-Augenstein and Ehrlich (1992) described, ironically, the majority view that prevailed until the mid-1980s (e.g., Back et al., 1982;Bograd, 1984;Dutton, 1988;Gelles & Cornell, 1983;Walker, 1981)  A third objection has involved the following "logic": To label the batterer as dysfunctional and to include him within the psychiatric population is to plac e him in a minority o f individual s within our g e n eral population and, thus, to infer that wife battering is rare, whereas, in fact, it is not . Fourth, politically motivated feminist researchers and activists have expressed concern that to focus on the batterer's psychopathology will only divert publ ic policy makers and growing public awareness from the crucial issue, the need to stop domestic violence.
Contrary to these objections , Tolman and Bennett (1990) and Vaselle-Augenstein and Ehrlich (1992)  Based on his review of the evidence, O' Leary (1993) also concluded that the preponderance of men who batter showed evidence of personality disorders. He observed, "There seems to be little question that physically abusive men who attend programs for containment of violence are significantly different from men who are maritally discordant but not physically abusive" (p. 25) O'Leary went further, adding that, "These men are not simply products of a social system . that fosters attitudes that promulgate the domination of women" (p. 25).
In their comprehensive analysis of risk markers in husband-to-wife violence, Hotaling and Sugarman (1986) also examined the consistency and prevalence of In summary, cumulative research evidence has underscored the emergence of a psychopathology-dominated profile of the male abuser.

Children of Family Violence
Although clinical evidence has suggested that family violence damages the psychological adjustment and behavioral competence of children, these and other issues have yet to be adequately examined, tested, or clarified through formal research . There is an overall paucity of research on the effects of family violence on children .
Findings are primarily derived from shelter samples.
Beyond the general statement that abuse adversely affects child development, more refined questions have remained largely unanswered.
Clinical and research literature on battered women and family violence (Leeder, 1994;Peled & Davis, 1995;Jaffe, Wolfe, & Wilson, 1990), as well as on abusive parenting and child maltreatment (Arnold, 1990;Azar, 1991 Children who witness family violence also tend to suffer child abuse, as there is an association between wife battering and child abuse. Hughes 1982) and Straus, Gelles, and Steinmetz (1980) estimated that 30-60% of families experiencing interspousal violence also showed incidents of parental physical or sexual child abuse. In their sample of 1,000 battered women , Bowker, Arbitell, and McFerron (1988) found that approximately 70% of the wife batterers also physically abused their children.
The affects of abuse have been documented in the many disturbing behavioral symptoms and interpersonal sequelae children display (Berman, 1993;Peled & Davis, 1995 For both the batterer and battered woman, violence has been learned in the context of family patterns.

Learned Helplessness and Depression
The concept of "learned helplessness," derived from Martin Seligman's (Seligman & Maier, 1967;Seligman, 1975) extensive laboratory research, has been adopted to explain the battered woman's passivity and ostensibly nonresponsive behavior in the face of violence. Seligman originally coined this term to describe the phenomenon in which experience with uncontrollable events created passive behavior in response to a subsequent threat to well being. Using non-contingent reinforcement, he conditioned dogs to react passively (i.e., no attempt to escape) to adverse stimuli (i.e., pain/punishment), disrupting an association between behavior and expected outcome .
In one experiment, Seligman and Maier (1967) showed that dogs that had not been subjected to uncontrollable electric shocks learned quickly to avoid the subsequent shocks. However, dogs repeatedly exposed to non-contingent electric shocks failed to escape from additional electric shocks even when escape was readily apparent--crossing over a barrier into a compartment where no shocks were delivered. Similarly, Hiroto (1974) found that human subjects exposed to inescapable loud bursts of noise failed to protect themselves in a later situation where the noise could, in fact, be easily avoided .
Findings derived from these studies and replications led Seligman to conclude that both animals and humans, when exposed to an uncontrollable event, learn that control is not possible and, therefore, stopped trying t o

Learned Helplessness and the Battered Woman Syndrome
Lenore Walker (1978Walker ( , 1979Walker ( , 1984Walker ( , 1993 "Becoming angr y rather tha n depres sed and s elf-bl aming; active rather than passive ; and more reali st ic abo u t t h e likelihood of the relationship continuing on its aversive course rather than improving" (p. 87).

Violence Theory
What makes the battered woman stay in an abusive relationship? Why does she persist so long in a relationship that, in addition to its severe punitive qualities, appears depleted of any obvious reward or benefit? Several explanatory concepts, closely allied to learned helplessness, have been proposed. These have included the concept of traumatic bonding and the predictable, self-reinforcing cycle of violence. Dutton and Painter (1993) and Painter and Dutton (1985) described traumatic bonding as, "strong emotional ties that develop between two people when one person intermittently harasses, beats, threatens, abuses or intimidates the other" (p. 364) The abused woman's experience has been .likened to that of other trauma victims, and the relationship between the battered woman and her abuser has been compared to hostage and captor.
Batterers have been described as terrorists in their own homes, and the battered woman's experience as comparable to the Stockholm Syndrome (Rawlings, Allen, Graham, & Peters, 1994). The Stockholm Syndrome has been compared to the experience of prisoners of war, that is, when captives identify with and consequently surrender their will and freedom of thought to their captors.
Traumatic bonding has been described as deriving its power and tenacity from intermittent reinforcement.
However, despite the battered woman's best efforts, the tension continues to build. Eventually, the tens ion reaches its threshold and the battered woman is no longer able to contain the batterer's angry response pattern.
Next, according to Walker's (1984)  In the third phase, described by Walker (1984) as the "honeymoon" or "loving, contrite phase," the batterer may apologize and may try to win back the battered woman's affections. During this phase, it is presumably this "loving behavior" that reduces the aversive arousal and reinforces maintenance of the relationship. The loving phase stands in sharp contrast to the prior phase .
The aftermath of the acute battering incident phase is one of extreme aversive arousal coupled with dissociative feelings of disbelief and self-blame.
This last, tension-free phase provides considerable relief from, and reinforcement for; the battered woman's posttraumatic reactions . However, in contrast to Lenore Walker's conception of this phase in the abuse cycle, Dobash and Dobash's (1984) data suggested that after the first battering incident, few batterers expressed contrition. The intermittent reinforcement in this phase appears to be often simply characterized by the absence of tension and violence. This moratorium may more aptly be described as providing temporary respite for battered women and children instead of the hope and positive experiences that maintain relationships. Sociologists, exemplified by Gelles (1993) and Straus (1990), have typically approached the problems of family violence from either of two perspectives: Subcultural theory and theories of family organization. Subcultural theory has subscribed to the premise that certain cultural and ethnic subgroups tacitly approve or positively sanction wife battering.
Theories of family organization have viewed the family as a primarily conflict-ridden unit, a system characterized by high levels of potential interpersonal violence. In this light, sociological theory has often viewed the family as an incubator of potential violence, an intimate biopsychosocial system where tensions run high and conflict is inevitable. Bowker's (1993) Walker, 1988;Hansen & Harway, 1993;Leeder, 1994;and Yllo, 1993), which is closely allied to a sociological perspective, has focused on the power inequities inherent in family structure, marriage, and traditional sex role stereotyping and gender General systems theory has purported to be blame free; neutral; and, thus, unencumbered in the complexities of moral judgments. It does not "see" the individual, but rather is focused on system process.
Furthermore, it has asserted that a system is greater than the sums of its parts and that no one element or person within the system can control the remainder of the With the exception of feminist family therapy (e.g., Bograd, 1984;Hansen & Harway, 1993;Leeder, 1994;Yllo, 1993) that has attempted to combine systems theory with an activist, political stance, family therapy {e.g., pinsof & Wynn, 1995;Sprenkle, 2002) generally has focused primarily on system level views of etiology, explanation, and intervention. Minuchin (1984) has aptly characterized this viewpoint when he suggested that to separate a family, or individual behavior, from its context is akin to "studying the dynamics of swimming by examining a fish in a frying pan" {p. 34).
In the true systems spirit, he explained that an understanding and treatment of family violence "requires a frame that is semi permeable, so that the blood and tears flow in and out of the picture" {p. 34). As both quotes from Minuchin have illustrated, the fundamental issue is epistemological. Minuchin's (1974Minuchin's ( , 1984 research, combined with findings of other family therapists {e.g., Pinsof & Wynn, 1995;Sprenkle, 2002) provided evidence for the multidimensional nature of perception and the importance of system integrity. Hence, family therapy has adopted an epistemological position emanating almost exclusively from "context." In effect, the family therapist's way of knowing has derived from consideration of interpersonal connections, relationships, and interactions.
The source of this epistemological concern may be traced directly to Bateson and his concept of "epistemological error." As Dell (1986)  Feminist family therapists such as Bograd (1984Bograd ( , 1988   Procedures:

Categorizing and Evaluating the Research Data
The research evaluation plan suggested three possible outcomes, each with its own secondary set of issues, research questions, and position on the learned helplessness/health psychogenesis controversy. Each outcome in this categorization schema offered a different perspective on the concepts and research variables of competence, survival, and recovery of battered women . The categories were defined as follows:

A pro-learned Helplessness Position
The majority of research data was found to support the learned helplessness theory and premises: that the behavior of battered women reflected cognitive, motivational, and affective distortions. Chronic abuse resulted in psychological disturbance requiring clinical intervention. Battered women developed and displayed learned helplessness, depression, and ineffective coping strategies.

A Pro-health Psychogenesis Position
The majority of data were found to support the health psychogenesis theory and premises: that the behavior of battered women reflected effective problem solving skills, accurate perceptioris, and adequate positive motivation. Battered women were psychologically healthy, proactive, competent, and emotionally strong.

Psychogenesis Position
Equally substantive data were found to support aspects of both learned helplessness and health psychogenesis theories and their underlying premises. It was proposed that such an outcome suggested two branching sub-outcomes or subcategories of experience: that battered women either displayed learned helplessness and the "battered woman syndrome," or a relatively symptom free, healthy response. This modified outcome, in turn, suggested that other kinds of hybrid outcomes were also possible. For example, the data were found to support the following outcome interpretations: (a) neither theory was well supported by its respective research databases, (b) one theory's research database was found to be more consistent, stronger, and scientifically adequate than the other, or (c) the research data appeared to support different concepts/assumptions associated with each theory, leading to the proposed integration of aspects of both theories.

Measures
This dissertation attempted to critically examine both sides of the clinical-research controversy surrounding learned helplessness/health psychogenesis.
The following four methodological criteria (research design dimensions) were used to evaluate the status (scientific adequacy) and findings of each theory's research database. These criteria were derived from a review and analysis of texts/perspectives (Campbell & Stanley, 1962;Hersen & Barlow, 1976;Huck, Cormier, & Bounds, 1974;Keppel, 1991;Kerlinger, 1964;Isaac & Michael, 1976;MacDonald, 1985;Riecken & Boruch, 1974;Tabachnick & Fidell, 1989)  The four dimensions were selected to address the issue of corroboration, at the extent to which research evidence supports core components of the theory.

A Stepwise Plan for Outcome Appraisal
The followings series of steps were implemented in order to evaluate the quality and outcome of each theory's respective research data. 4. The individual studies were also analyzed along an outcome dimension. This outcome dimension was based on the study's findings. Accordingly, study outcomes were assigned by independent raters to one of four categories: "pro-health psychogenesis," "pro-learned helplessness" , "both" (equally supportive of both theories), or "not directly applicable" (NDA), that is, findings did not pertain to either theory.

To safeguard against investigator bias, raters
were trained, and a test of interrater reliability was applied to outcome categorization and assessment. Three for shelter and community samples. They independently rated study outcomes. When disagreement between the two undergraduate raters occurred, the third graduate level rater (clinician) served as a tie-breaker. The results of these outcome ratings are discussed in the sections on "Corroboration" for shelter and community samples.
6. Because a simple outcome categorization appeared to be an insufficient measure, the decision was made to qualify each study's outcome rating by the study's overall level of design quality (high, medium, or low), using the study's assigned TDQS. Therefore, a study's outcome was considered more "significant" when it fell within the high design quality level. Tables 5.3 and 6.3, presented in Chapters 5 and 6, organized studies by level of design quality by outcome, or which theory the study's results (i.e., majority of findings) supported.
7. The initial step in the design quality by outcome analysis focused on shelter and community samples separately. Next, the design quality-by-outcome findings were integrated across shelter and community samples.

Outcome evaluation involved consideration (interaction)
of methodological adequacy and substantive outcome.
In Chapter 4, consistent with the first research objective, the results on findings derived from national probability surveys of domestic violence are analyzed and described.

CHAPTER IV
Results:

National Random Surveys of Domestic Violence
This chapter addresses the broader questions of prevalence and measurement in the area of family violence.
The focus is on national statistics regarding the experience of battered women and children. Table   4.1, which is to be discussed, summarizes major national surveys: their sample sizes, instrumentation, and major findings on partner violence and child abuse.

National Probability Surveys
But what do we actually know about battered women?
The knowledge we have of battered women, including their experience of and response to family violence, is primarily derived from three sources: randomized, national probability surveys of the general population, which are used to draw inferences about battered women's experiences; non-random shelter samples; and volunteer, self-selected community samples.  Resurvey (random sample of 6,002 families), cited in Straus and Gelles (1990); and the 1977-1978 National crime Victimization Survey (random sample of 60,000 households), cited in Gaquin (1977Gaquin ( -1978 . In 1992, the  Hansen & Harway, 1993;Leeder, 1994;Pagelow, 1981;Stark & Flitcraft, 1996) have catalogued similar injuries and documented similar violence rates.
Brown and Hendricks (1998) further corroborated this point. They stated "more than 50% of all couples experienced at least one assaultive incident during the course of their relationship, with women experiencing more than ten times the number of violent incidents than men" (p. 123). Emphasizing the serious nature of assaults, they further stated: "39% of all violent encounters involved the wife being punched with a fist, kicked, bitten, or attacked with a knife or gun" (p.

Interpreting the Reported Controversial Data on
Husband to Wife and Wife to Husband Violence 74 An overview of national statistics on the incidence of domestic violence is incomplete without a brief discussion and explication of similarly reported rates of wife-to-husband violence. These findings, derived from Straus and Gelles's (1990)  Criticisms raised by feminists have revolved around the possible psychological and political misapplication of these data. For example, feminists have expressed concern that such data will be used to "blame the victim," disparage the victim's character, discredit the merits or worthiness of the "cause" for equity between genders, and slow or ultimately forestall legal, social, and political progress for survivors of domestic violence. As many feminist researchers, and as Straus and Gelles (1990) themselves aptly pointed out, the consequences and meaning of these findings must be judged in their context. Feminists have argued that, usually, women's acts of aggression occur in response to or in defense against a husband's violent assault. Women's violence is largely reactive and self-protective. Thus, feminists have contended that violence rates must be judged in terms of directionality; whether the aggressive action is an initiating act or a reaction to or defense against assault.
In addition, considering the typical physical disparity in size and strength between men and women, women are no match for men in an aggressive interchange.
As Browne (1989) suggested, because men are physically stronger and larger than their wives, with more social and economic resources at their disposal, the physical and social consequences of marital violence are usually limited and relatively less severe when men are the victims. Accordingly, violence rates need to be assessed in terms of outcome; for example, in terms of the severity of inflicted injury . Obvi ously, women typically inflict less serious bodily injury than men. When the outcomes of husband-to-wife and wife-to-husband violence were compared, based on medical and police reports, women fared much more poorly, sustaining more severe and serious physical injury (Dobash & Dobash, 1979;Hansen & Harway, 1993).
Returning to data obtained in . National Crime surveys, Barnett and Laviolette (1993) have called attention to Schwartz's (1987)  The cruel irony of staying home because one fears violence in the streets is that the real danger of personal attack is in the home.
Offenders are not strangers, climbing through windows, but loved ones, family members.
You are more likely to be physically assaulted, beaten, and killed in your own home at the hands of a loved one than any place else, or by anyone else in our society. (p. 18) 78

Definitions, Levels, and ~es of Family Violence
The collective body of clinical and research evidence has illustrated that the concept of "family violence" is multifaceted and complex. An understanding of family violence is intimately tied to and embedded in the more general construct of and research on aggression.
The research on aggression is discussed, at this juncture, because it sheds light on a number of methodological (i.e., definition of terms and measurement) dilemmas characteristic of research on battered women. Violence toward women represents one form or variant of family violenc~ and a subset of a more general cluster of behaviors and intentions that we label as "aggression." Drawing upon research in the area of aggression, "aggression" may be broadly defined as any action or intent designed to inflict harm (physical or psychological) on another.
Herek (1992), who investigated the role and relationship of aggression in the context of hate crimes, amplified the definition of aggression to include words or actions intended to harm or intimidate an individual.
Behaviors falling within this broad definition of "aggression" ranged from obvious violent acts such as assault, rape and murder, to property crimes motivated by animosity, and more subtle forms of psychological terror and fear such as threats of violence and intimidation.
Extending this definition and the understanding of aggression further, Bandura's (1986) conceptualization set the parameters for assessing acts of aggression beyond the "act" or "consequences" of the aggressive act itself, to a consideration of the broader frame of reference or context within which the aggressive act was committed. For Bandura, the context of the aggressive act represented the by-product of and, therefore, must be evaluated in terms of the interaction between socialstructural variables and the self, including such psychological processes as subjective judgment. In emphasizing wide-spread cultural "cues" and the concept of "disinhibition", Bandura argued that predictive rules about the perception of aggressive behavior encompassed more than the behavior per se, to include social as well as self (cognitively and affectively mediated) factors.
Aggression toward women in our society is deeply embedded in the ethos of our culture. Aggressive acts are often tacitly and overtly approved or fostered by our social structure. Our tolerance of aggression is often so pervasive and its manifestations so ubiquitous, that people can become inured or desensitized to its effects.
The above discussion of varying concepts and research perspectives on "violence" was intended to highlight the definitional and, by inference, measurement dilemmas inherent in research on wife abuse. Battering encompassed the range of behaviors employed to hurt, intimidate, coerce, isolate, control, or humiliate a partner" (p. 161). As Gelles (1997), Flitcraft (1977), and Stark and Flitcraft (1996)

The Relationship between Physical and Emotional Abuse
Family violence takes many forms . There are overt acts of physical assault and aggression, the obvious brutality and injury that we associate with the typical image of spousal abuse and family violence. There are also subtle and insidious kinds of psychological abuse that take the form of verbal and emotional battering, degradation, ridicule, intimidation, control, and shame.
Emotional abuse pervades the lives of many women who Emotional abuse has been experienced by battered women as more detrimental than physical violence.
Despite the physical injuries that women suffered at the hands of their abusive partners, research by Walker(1984)and others (e.g., Follingstad, et al., 1990) consistently has shown that battered women e x perienced verbal humiliation and emotional abuse as the worst kind of battering. Emotional battering has been identified as the main tactic the abu s er uses t o i s olate , dominat e , demean, and confuse family members. Like physical abuse, emotional abuse forms part of the hostile traumatic bond that keeps the violence a carefully guarded secret; unites family members in a conununity of fear and silence; and keeps the "victim" mystified and enveloped by inexplicable self-recrimination and shame. The overwhelming consensus of formal researchers on the power and pain of emotional abuse is that, for abused women, physical injuries heal much more quickly than the emotional traumas and hurts they experience.

The Relationship between Verbal and Physical Aggression
A controversy central to problems of definition, measurement, and prediction in the field of family violence has revolved around the continuity-discontinuity  (Straus, 1974a(Straus, , 1974b. Studies (Straus, 1974a(Straus, , 1974bStraus, Gelles , & Steinmetz, 1980) have demonstrated that individuals who engage in verbal aggression toward the i r spouses are more likely to engage in physical violence.
As Bowker (1983aBowker ( , 1983b observed, research has shown a positive association betwee n verbal and physical aggression in direct opposition to leveling, catharsis, and ventilation theor ies of verbal aggres s i o n . Further , Murp hy' s (19 87) longitudina l study c orrobora ted the general finding that verbal aggression functions as a "precipitant" to physical aggression . Straus and Smith (1990)   Of these nine studies, Back, Post, and D'Arcy (1982), and Gellen, Hoffman, Jones, and Stone (1984) involved, respectively, psychiatric batterednonbattered group comparisons and consecutive admissions to a residential treatment program with self-selected, matched community controls. Finn (1985) and the three interrelated Follingstad and associates studies (1990,1991,1992)                  "OUT" women were expected to be more likely to recognize signs of impending violence than short term and "IN" women, 2b) short term and "OUT" women more active in their responses once they recognized physical force likely to occur , 2cl) short term and "IN" women receive more positive response, 2c2) for ST and "IN" women partners more contrite, 2dl) short term and "OUT" women would threaten batterer with consequences, 2d2) ST and "OUT" women would form an action plan (leave mate or seek outside intervention); 3) short term and "OUT" women will view reason abuse stopped as function of own actions , not outside intervention; 4a) mates of long term women predicted to be more remorseful, offer           single or multiple group studies with sample sizes of 250 or greater were assigned a value of 5. Accordingly, as  Stone, 1984;Okun, 1986;Pagelow, 1981;Schillinger, 1988) reflected a normative, racially representative dis t ribution (approximately 80 : 20 Caucasian/non-Caucasian). Three studies (Finn, 1985;Gondolf, 1988b;Snyder & Scheer, 1981) approximated an evenly divided Caucasian/non-Caucasian distr ibution , with the three remaining studies (Claerhout, Elder, and Janes, 1982;Labell, 1979;Launius and Lindquist, 1988) representing predominantly special populations, either almost exclusively Caucasian or African American.
Only four out of the 17 studi es cited in Table 5.1 reported data on educational background. Of these four studies, two (Finn, 1985;Mills, 1985) (Back, Post, & D'Arcy, 1982;Frisch & MacKenzie, 1991;Gondolf, 1988b;Labell, 1979;Launius & Lindquist, 1988;Okun, 1986;Schillinger, 1988) focused on economically disadvantaged women. Launius and Lindquist (1988) presented a special case in which the average income of their shelter sample fell between $10,000 and $20,000, or at a more economically disadvantaged level than their community sample with an average income of between $20,000 and $30,000 .
If one combines the studies involving low income, rural and low to low-middle class individuals with the seven studies involving economically disadvantaged individuals, the total percentage of economically disadvantaged samples increased to 53%. This suggested that half of the research findings reported in Table 5.1 represented economically impoverished or disadvantaged women residing in emergency shelters.
When socioeconomic status in the community sample cited in Table 6.1 is later examined, it will be seen that a similar finding holds. The principle sampling procedures across all 17 studies involved referrals, self-selected volunteers and/or non-random consecutive shelter resident or hospital admissions.  (1990,1991,1992); Frisch and MacKenzie (1991); Gellen , Hoffman, Jones, and Stone (1984); Gondolf (1988b); and Launius and Lindquist (1988) . Level four was used by Back, Post, and D'Arcy (1982); Okun (1986); and Pagelow (1981). The p rimary data gat hering methods u sed a c r os s mos t studies , whi ch f e ll into lev el one or two, involved written questionnaires and/or in-depth structured/unstructur ed face -to -face interviews with battered women.
Two studies (Back et al . , 1982;Gellen et al., 1984) us e d the MMPI. The use of multiple data sources included unobtrusive observation or participant observation in discussion groups . Unfortunately, only one study included outcome follow-up data (Snyder and Scheer, 1981) .

Design Quality and Statistical Analysis
Three criteria were applied to evaluate design quality and overall statistical analysis. These criteria refl e cted a three -level scoring sys tem. Level one included qualitative findings based on exploratory data, involving anecdotal, life-history narrative. Level two studies relied on and reported only descriptive (noninferential) analysis. Level three studies employed univariate or multivariate methods, probability assumptions, control group comparisons, tests of statistical significance, and inference. Two studies (Schillinger, 1988;Mills, 1985) were categorized as exploratory, falling into level one. Only one study (Labell, 1979) employed descriptive statistics exclusively. The remaining 14 studies (Aguirre, 1985;Back et al . , 1982;Claerhout et al. , 1982;Finn, 1985;Follingstad et al., 1990Follingstad et al., , 1991Follingstad et al., , 1992Frisch & MacKenzie, 1991;Gellen et al., 1984;Gondolf, 1988b;Launius & Lindquist, 1988;Okun, 1986;Pagelow, 1981;and Snyder & Scheer, 1981), which fell into level three, used a range of evaluation methods, including discriminate function, one-way fixed ANOVAs/MANOVAs, factor analysis, structural equation modeling, cluster analysis, and multiple regression. In effect, 82.4 % of the shelter studies used inferential methods.
Additionally, the total was 18 studies in this tally because fou r studies wer e counted twice. These studies either doubly addressed the group difference issue or combined group difference hypothesis testing with correlational analysis .
corroboration Table 5.1 summarized individual studies and described/identified major findings. Table 5.2 provided quantification (ratings) for individual studies. In Table   5.2, each study was assigned a numerical value per methodological dimension as well as a TDQS. In Figure   5.1, which defined and outlined methodological dimensions, the rationale was provided for computing the TDQS as well as a system for coding studies along a range of high, medium, and low design quality. using a third independent rater (master's level clinician) as a tie breaker, the one remaining study in question (Mills, 1985) was designated as a health psychogenesis outcome. Figures 5.1 and 5.2 provided the basis for training raters in the core concepts and assumptions of each theory.
Of the 17 studies cited in Table 5.1, two studies were categorized as "both". This left 15 studies for outcome determination. Overall, nine, or 60%, of the studies in Table 5.3 supported core concepts of the health psychogenesis theory. These studies portrayed battered women as active helpseekers (Gondolf, 1988;Labell, 1979); reality-based problem solvers whose decisions to remain with abusers involved external circumstances, including economic, occupational, and educational disadvantages (Aguirre, 1985;Gondolf, 1988;Okun, 1986;Pagelow, 1981) and psychologically competent survivors who were often the victims of "agencization" (Follingstad, et al., 1991(Follingstad, et al., , 1992Mills, 1985;Schillinger, 1988). The nine studies that supported the basic assumptions underlying the health psychogenesis theory were predominantly of high design quality (6, high quality; l, medium quality; 2, low quality).
1 _ Battered women do not develop adequate self-protection skills in childhood, resulting in adult vulnerability to physical and sexual abuse. The theory is predicated on a developmental, genetic view of childhood susceptibility or psychological risk markers that predispose the women to the violence in the abuse dynamic. 2 -Battered women adhere to traditional sex role ideations. 3 -Battered women are poor problem solvers. They demonstrate ineffective coping strategies, cognitive impairments, and behavioral and motivational deficits. 4 -Battered women employ negative self-attributions. They blame themselves and/or exonerate, excuse and forgive the abuser. 5 -Battered women are viewed as passive, anxious, and socially isolated; women who are immobilized and unable to actively seek help for themselves. 6 -Battered women demonstrate low self-esteem. Their self-esteem steadily and progressively declines as the abuse continues. 7 -The victim and abuser are both viewed as caught up in an emotionally dysfunctional relationship; both exhibit cognitive , motivational, and psychological deficits. 8 -The hallmark of a woman's mental health is defined by the single outcome criteria that she leaves her abuser. Conversely, the fact that she stays is viewed as strong evidence in favor of the learned helplessness phenomena and the woman's inadequate and impaired psychological coping strategies. 9 -This theory proposes a social -psychological model of the abuse dynamic. It incorporates concepts of social learning theory, intermittent contingency behavioral reinforcement, and a tension building/tension reduction hypothesis. The abuse is attributed to interpersonal and intrapsychic factors of the abuser and the abuse victim.

Figure 5.2: Core Concepts and Assumptions of Learned
Helplessness 1 _ A core concept of normality or prior mental health of battered women is predicated in this theory. The timeline for the abuse dynamic does not begin in a precursor childhood but in the actual first incident when a woman is hit. 2 _ As a group, battered women demonstrate fewer adherences to traditional sex role ideations than the norm. 3 -Battered women are effective problem solvers. They are cognitively realistic and devise effective behavioral, safety, and self-help plans. 4 -Battered women are less likely to blame themselves but instead blame the abuser. They do not believe the abusers contrition and are cognizant that the abuse will continue. 5 -Battered women are viewed as active help seekers in c ontrast to being compliant, passive, or help-resistant. 6 -On average, battered women show a higher sense of self-esteem and self worth than counseling controls. Their self-esteem does not portray a linear progressive self-decline, but rather is seen as curvilinear . The women's self-esteem is initially d e pressed by batteri ng but when all seems hopeless , wi t hout any exte rnal stimul us for the change, a spark of innate h e alth ignites growth in the personalities of these women. The Ushaped curve of self-esteem bottoms out and gradually battered women become stronger, accelerating their personal growth until they successfully end the violence in their lives. 7 -Battered women are healthy whereas the male batterer is more emotionally unhealthy, s ubsta nce d e p e nde nt , or p a tho l o gic al . 8 -The staying /l e a v ing crite ri on i s not vi e we d as the d e f i n ing factor determining a battered woman's health or psychological dysfunction. Battered women who stay in relationships are viewed as remaining for valid and realistic reasons, not because they are helpless and psychologically impaired. They are viewe d as practical , realistic, long-te rm pla nners who wisely a s s e ss the safet y/danger index in the a c t of l eaving the v iolent and unstable abus e r. 9 -This the ory proposes external social f~ctor s as the precipitators that create and maintain the abuse dynamic : for example, social isolation, economic dependence and disadvantage, and •agencization" or lack of appropriate responses on the part of the helping professions. This theory adopts a sociological v iew.

Psychogenes is
Of the 17 studies cited in Table 5.1, two studies were categorized as "both". This left 15 studies for outcome determination. Overall, nine, or 60%, of the studies in Table 5.3 supported core concepts of the health psychogenesis theory. These studies portrayed battered women as active helpseekers (Gondolf, 1988;Labell, 1979); reality-based probiem solvers whose decisions to remain with abusers involved external circumstances, including economic, occupational, and 139 educational disadvantages (Aguirre, 1985;Gondolf, 1988;Okun, 1986;Pagelow, 1981) and psychologically competent survivors who were often the victims of "agencization" (Follingstad, et al., 1991(Follingstad, et al., , 1992Mills, 1985;Schillinger, 1988). The nine studies that supported the basic assumptions underlying the health psychogenesis theory were predominantly of high design quality (6, high quality; 1, medium quality; 2, low quality). In contrast, six studies, or 40%, were found to support several core concepts of learned helplessness. Back, Post, and D'Arcy's (1982) study indicated that battered women, compared to nonbattered women, reported a higher incidence of childhood physical violence, thereby supporting concepts relating to childhood vulnerability.  Mills (1985) x Both x x similarly, battered women were found to have higher, clinically elevated MMPI personality profile scores than nonabused women (Gellen et al., 1984). Claerhout et al. (1982); Finn (1985); Launius and Lindquist (1988); and Snyder and Scheer (1981) provided findings that suggested battered women employed less effective coping strategies and perceived fewer alternatives, supporting the relationship between increased stress and decreased coping strategies characteristic of learned helplessness.
The six pro-learned helplessness studies reflected a different design quality breakdown. In contrast to the six high, one medium, and two low design quality tally for the pro health-psychogenesis, only one study in the learned helplessness camp was rated as high, with the five remaining studies falling in the medium design quality category.
The remaining two studies (Follingstad, et al., 1990;Frisch & MacKenzie, 1991) were more accurately categorized as "both" because they equally supported the core concepts of each theory . Chapter 6 reports the results on research findings derived from non-random community samples following the same methods outlined in this chapter.

Analysis of the Non-random Community Samples
This chapter examines the research dimensions and outcomes of non-random community samples. The same set of tables and types of research criteria described in Chapter 5 were used to evaluate community study outcomes.
Partic ipants were obtained thro ugh a v ariety o f sources--ne wspaper or television advertisements, national woman's magazine solicitations, and social service r e ferrals. Similar to the studies outlined in Table 5.1, Table 6.1 also reflects non-random samples of convenience.

sample Size, Sample Composition, Sampling Procedure, and Power
In Table 6.1, nine studies, including Campbell, et al., (1994) ;Kuhl, (1985); Walker (1984); two studies by McLeer and Anwar (1989); two studies by Flitcraft (1981, 1996) Of the twelve studies that reported racial and ethnic distribution, five of the studies (Campbell, et al., 1994;Kuhl, 1985;Rhodes, 1992;Rounsavill e, 1978;Walker , 1984) wer e raci a lly r epres enta tive. The remaining seven studies depicted a range of special populations, including four studies that were almost exclusively Caucasian, with one study repres e nting a caucasian college student sample (Launius & Jensen, 1987), and three Caucasian community samples, including two studies by Bowker (1983a) and one study by Cascardi and O'Leary (1992); two studies by McLeer and Anwar (1989) reflecting Black inner city samples; and one study (Hilberman & Munson, 1978) involving predominantly Black tenant farmers and mill families. Of interest, the shelter samples reported in Table 5.1 appeared to be more racially representative than the community samples in   women were positive (25 %), 1 8 were probable (16%), 5 were suggestive (4%), and 5 8 were negativ e (50%) (+) -for families of battered women, father was more than 3 times more likely to be child ' s abuser than of nonbattered women; 50 % of children o f at-ri s k women were abused by male batterer (+) -battered mothers were less likely to have famil y disorganization than nonbattered women (e.g. Alcoho l ism= 12 % vs . 19%, Vio l ence= 17% ....... ii::.      guilt or social impoten ce 5-Not supported 6-Not supported 7-high positive association with being slapped six or more times and Succorance (.7 1 ) : women who were slapped six or more times appeared to feel inadequate i n coping with stress and crises and appear to need to avoid confrontation and tend to retreat into fantasy positive= 5.6%, probable= 1 0.9%, suggestive= 9.2% I-' U1 w 1 : syndrome of depressive, "learned helplessness" is a critical, paralyzing factor -CES-D report  At Tl and T2 indepth interview and standardized written questionnaire BDI (Beck,197 2): enduring vs. transitory depression Tennessee Self-Concept Scale (TCSC ) (Fitts, 1972): selfesteem SCL-90 (Campbell, 1989;Derogotis, 1977) :physical symptoms of stress and grief Denyes Self-Care Agency Instrument   Bowker, 1980Bowker, -1981 Bowker, 1980, National Walker, 1978 homelessness; shame, failure, and public disgrace; loss of social identity 1)-"Ins" reached higher level of fear, anxiety, and depression at T2 (second battering incident) than "Outs". However, the curves for anger, disgust, and,  (Levenson,197 2) , Attitudes Toward Women Scale (Spence, Helmreich, 1972), MMPI-Harris-Lingoes subscales vari ables incl uded: family of origin/current demographics, relationship history and sex role socialization, se l festeem, life e v e n ts locus of contro l attribution, perceptions and beliefs of battering experience , psychological functioning, medical his t ory, psychophysiological stress responses hosti li ty rose for both "Ins " and "Outs" but the level for "Outs" is higher a t T3 (last battering incident), confirming learned helplessness in the women who remained 2)-in 65% of women report i ng three battering incidents, e v idence of tension buildi ng p h ase: in 5 8% of women , evidence of later contrition for first incident, tension bui l ding phase is 56% and 69% for later contri tion by the last incident, 71% preceded by tension building , but only 42% fol l owed by later contrition this confirms the Cyc l e of Violence hypothes i s , also indi cat ing that over t ime tension building becomes more common as contrition declines two studies (Launius & Jensen, 1987;Walker, 1984) involved middle class individuals predominantly; two studies by Bowker (1983a) represented a range of socioeconomic backgrounds; one study (Bowker, 1983b) focused on working class/middle class individuals; three studies, including the two studies by McLeer and Anwar (1989) and Rounsaville (1978), involved primarily lower socioeconomic class individuals; one study (Campbell, et al., 1994) reported one-third economically disadvantaged individuals with the remaining two-thirds representing the marginal working class; and the remaining study (Hilberman & Munson, 1978) reflected impoverishment or severe economic deprivation.
The community sample studies represented a somewhat higher proportion (60%) of economically disadvantaged women in contrast to the shelter samples (53%). The sampling procedures in the studies listed in Table 6.1 reflected similar methodologies identified in As an adjunct to Table 6.1, Table 6.2 was constructed to present the methodological ratings assigned to the studies and also provide a TDQS. These numerical ratings provided the underpinnings for the subsequent text and basis for discussion throughout the following sections.

Instrumentation
In contrast to shelter samples, the community samples reflected a 21:79 split or slightly higher representation of multidimensional, multiple data source methods of measurement and instrumentation. One study (Hilberman & Munson, 1978), coded at level one, (using original instrumentation/scoring systems) and two studies (McLeer and Anwar, 1989), coded at level two, adopted a primarily medical (data source) access route to the exploration of the imp~ct and effects of domestic violence.
For example, in one study, McLeer and Anwar (1989) replicated Stark and Flitcraft's (1981)  In a third study that investigated the psychological impact of physical abuse on symptom formation using informal narratives and treatment data, Hilberman and Munson (1978)  In focus and intent, the more methodologically sophisticated level three and four studie s f e ll into three categories, addressing three broad issues . Of the eleven (level three and four) studies cited in Table 6.1, the two Stark and Flitcra ft studies (1981, 1 996), (c oded at level four), represented a medical perspective on abuse identification. Their first investigation (1981) underscored the discrepancy between institutional versus interview-focused identification of prevalence rates.
Their second study (1996), which will be discussed in the Alienation. Kuhl (1985) Leary (1992) tested basic assumptions regarding the relationship of depression, loss of self-esteem, and self-blame attribution to the frequency, severity, and consequences of abuse.
Building on their prior research, Campbell et al.

Design Quality and Statistical Analysis
The same three-level system for assessing methodological adequacy (design quality and appropriate statistics) was applied to the community studies cited in  Flitcraft, 1981Flitcraft, , 1996 and theory evaluation (psychological versus sociological) on the basis of a multifactorial descriptive statistic method (Campbell, 1990;Rounsaville, 1978).
The remaining eight studies reflected more complex, multifaceted designs. Community level three studies employed an array of statistical procedures including ANOVAs/ANCOVAs/MANOVAs, multiple regression, factorial analysis, scale development and item correlation, and path analysis. Out of the total eight, level three studies, three studies used correlational techniques either exclusively or in conjunction with group differences methods . For example, Kuhl (1985) investigated the degree of association between domestic abuse items and scale scores on the Gough Adjective Check List, which measures the personality needs/structures of battered women. Cascardi and O'Leary (1992) examined the intercorrelation among measures of depression, selfesteem and self-blame attribution, for severely battered women. As part of her comprehensive analysis, in addition to emphasis on group differences, Walker (1984) used correctional methods extensively for scale and instrument development as well as to measure dimensions of learned helplessness (childhood, current relationship, etc.) .
Curiously, in contrast to 13, or 72%, of the studies in Table 5.1 on shelter samples that examined group differences ("Ins" versus "Outs" or battered versus nonbattered), of the community samples reported in Table   6.1 only five, or 33%, of the studies represented group differences designs. Campbell et al. (1994), Launius and Jensen (1987), and Rhodes (1992) Walker (1984) and Bowker (1983b) investigated the perceptions, beliefs, and actions that presumably distinguished "Ins" and "Outs". Although the shelter sample appeared to reflect a more frequent use of the group differences model, the two group differences studies that launched the debate and controversy between the two competing models arose from community research.
These studies are the landmark research on health psychogenesis (Bowker, 1983a(Bowker, , 1983b and learned helplessness (Walker, 1984). Of note, the overall tally in design quality analysis is 15 rather than 14, the actual number of samples cited in Table 6.1: One study (Walker, 1984) was counted twice because it employed both group differences and correlational models. As previously described (see the section on corroboration in Chapter 5), Tables 6.1, 6.2, and 6.3 build logically and cumulatively from qualitative to quantitative findings. Table 6.3 formed the basis for discussion in this section on the analysis of studies and also the upcoming outcome evaluation. Table 6.3 summarized TDQS and outcome results.
In this section, results reported in Table 5.3 on shelter samples and Table 6.3 on community samples are compared.
In the shelter samples, nine studies, or 60%, supported the basic concepts of the health psychogenesis theory; six studies, or 40%, supported the learned helplessness theory; and two studies were labeled "not directly applicable" because they did not directly address core assumptions or constructs of either theory; and one study was labeled "both" because it supported each theory equally. The NDA studies were excluded from the analysis of outcome.
In the community sample analysis, the results were quite similar. Seven studies, or 64%, supported health psychogenesis; four studies, or 36%, supported learned helplessness; two studies were labeled "not directly applicable" because they did not address any core assumptions of either theory; and one I\ study was labeled "both" because it examined both theories and found equal evidence to support each theory . studies labeled NDA or "both" were excluded from the outcome analysis. Overall, seven studies supported the health psychogenesis theory. Flitcraft's (1981, 1996)  behavior, the capacity to persevere and pursue long-term goals in favor of less effective compulsive ones, and reality based reasons for their "waiting" behavior in their efforts to find the right time to leave.
Cascardi and O' Leary's (1992) study was also rated as pro-health psychogenesis. However, this study raises an interesting dilemma and will be used to illustrate how the decision was made to consider it supportive of the health psychogenesis theory. Although the findings were mixed, the weight of the evidence pointed more to a proheal th psychogenesis viewpoint. / Campbell (1989) designed an initial study to test the power of the theories of depression and learned helplessness and found relatively equal evidence supporting both theories. However, she noted problems with some aspects of the learned helplessness theory. In a more extensive, longitudinal study, using a larger sample of abused and nonabused women, Campbell et al. (1994) found little support for core concepts of learned helplessness theory, such as loss of self-esteem, depression, lack of self-care agency, and relationship control. Her research failed to find a difference between abused and nonabused women on these variables.
Four studies provided evidence in favor of learned helplessness.
In investigating the relation between domestic violence assessment items and personality needs, as measured by Gough's Adjective Check List, Kuhl (1985) identified personality needs that closely approximated character\ stics of women experiencing learned helplessness as a function of battering. Kuhl found these women to be cautious, conflict avoidant, unable to cope with traumatic stressors and, being dissatisfied with their current status, taking refuge in fantasy . Hilberman and Munson (1978), who studied chronically abused and impoverished women, reported a wide range of psychological distress concurrent with the abuse. The psychological sequelae included depression, manicdepression, schizophrenia, character disorder, and alcoholism. This is not to suggest that the abuse "caused" the psychological disorder , but rather to state these conditions were ide ntified along with the presence of abuse in these women's lives (i.e., comorbid factors) Walker's (1984)  In addition, the two McLeer and Anwar (1989) studies may be aptly characterized as "not directly applicable" because, while they touched on domestic violence issues and the battered woman's experience, they did not directly test any of the hypotheses or core constructs of either theory. The remaining study (Rounsaville, 1978) was categorized as "both" because it provided approximately equal evidence for key aspects of each theory .
In effect, Rounsaville's research findings appeared to support assumptions underlying both theories and argued for the integration of salient features of each to yield a systems view of the battering experience. Walker's (1984) and Bowker's (1983aBowker's ( , 1983b  a ccompanying child abus e; g end e r , a ge and rac e of the child; maternal stress (combination of maternal health, negative life events, and family disadvantage); child rearing practices; length of time exposed to the violence; and impact of shelter residence.
These studies of moderator variables (Hansen & Harway, 1993;Peled & Davis, 1995;Straus & Gelles, 1990) have yielded several preliminary findings. Children who witnessed woman battering and who were also physically or sexually abused showed more behavior problems. In addition, the more chronic and severe the abuse, the greater was the likelihood of long-lasting and pervasive effects. The more violent the home, the more likely children were to become involved in criminal behavior and the court system . The influence of abuse was also found to be more profound when the child was younger and developmentally unequipped or less prepared, on cognitive and emotional grounds, to process and integrate the traumatic events.
Research on gender as a mediating variable has yielded mixed results and appeared to reflect interaction between the age of the child and the onset and duration of the violence (Hinchey & Gavelek, 1982;Levine, 1975;Stark & Flitcraft, 1996). However, the handful of studies, to date, has suggested a preliminary sketch or profile. The profile appeared to follow cultural norms associated with gender-based differences in socialization, a point that, in part, has confounded the capacity to draw clear inferences from research findings.
This qualification noted, research has suggested that boys tend to act out their sadness and rage and are Interestingly, they also found that of the 209 children studied, one-third displayed somatic, psychological, and behavioral problems .
A childhood immersed in emotional combat and chronic, unrelenting physical aggression appears to compromise the child's emotional outlook and interpersonal future. In exploring the intergenerational link and relationship between childhood violence and adult violence, Gelles and Straus (1988) repor ted that between 18 and 70% of children who grew up in violent homes will "recreate" that violence in their adult families. The broad range in this statistical estimation requires further clarification. Zigler (1987, 1993) reviewed three sources of research data, providing evidence on the intergenerational transmission hypothesis. These data included case studies on emergency room visits, human service agency records, and self-report studies. After a careful review of research, Zigler (1987, 1993) concluded that the most accurate estimate within this wide range is about a 30% rate of intergenerational transmission. Although a history of childhood family violence clearly does not necessarily make a child become an adult abuser, there is a definite relationship between the experience of abus~ and its future effects on the child (Egeland, 1993). Gelles (1997)

Protective Factors and Resilience
What protective factors can safeguard the child from a potential future as an abuser? What factors appear to account for those resilient children, the 70% who do not go on to abuse? What positive aspects within the violent, conflict-ridden family environment protect the child? In examining factors that contributed to the resiliency of children, Garmezy (1983) identified three major categories of protective influences: dispositional attitudes of the child (e.g., ability to cope with and adjust to new situations), support within the family system (e.g., good, stable relationship with one parent), and support figures outside the family system (e.g., available resources and positive role models, such as peers and relatives) .
Research on the resiliency of childhood trauma survivors has shown that the presence of a nurturing, emotionally available and responsive parent--one who is sensitive to the child's age-and stage-appropriate developmental needs, with whom the child forms a genuinely close emotional bond, and with whom the child can identify--is a crucial factor determining the wellbeing and predicted resilience of children who witness family abuse. Pett (1982) d e fined a good relationship with one parent as a relationship characterized by the absence of severe criticism and the presence of high warmth . In most instances, the nurturant and stabilizing parent in abusive homes is the mother. Battered women have frequently reported (Hansen & harway, 1993;Leeder, 1994) placing the safety and well-being of their children before their own welfare and personal needs.
This fourth category of studies that focused on intervention programs more closely approximated efforts to achieve a systems viewpoint. However, the kinds of data collected (i.e., described in the research literature) did not capture the transactional dynamics of A salient factor mediating the impact of family violence on the identified child was level of maternal stress (Arnold, 1990). Hughes and Barad (1983) studied self-concept, level of anxiety, and psychological problems in 65 children (ages 2-12) residing with their mothers in a shelter. The children were rated by mothers, teachers, and staff, and were found to display a below average self-concept score for the preschool age group, more aggressive behavior in school age shelter boys than girls, and a general tendency for mothers to rate their children more negatively than other observers. Hughes et al. (1989)  In examining interparental violence and the ways children are at risk, Levine (1975) found that children experiencing family violence displayed an array of symptoms, including insomnia, anxiety, tics, aggressive behavior, truancy, and enuresis.
In an urban hospital setting, Stark and Flitcraft (1996) matched and identified battered women and their abused children.
They concluded that child abuse and wife battering often occurred simultaneously in a family violence context. In a similar developmentally focused study, Hinchey and Gavelek (1982)  Peled and Edleson (1992) constructed a group program that drew upon interview data from child witnesses, battered women, perpetrators and group leaders.
One other study attempted a type of systems level analysis. Holden and Ritchie (1991) examined the relationship between marital discord, parental behavior, and child behavior. Based on maternal self-report and mother-child observations of 37 battered women and 37 comparison mothers and their children (ages 2-8), the only robust self-report differences found between the two groups of mothers were the level of stress and reports of inconsistency in parenting.
In addition, the expected difference between the mothers' reports of the two groups of fathers was supported: abusive fathers were perceived as more violent and irritable. In the violent family, maternal stress and paternal irritability were the two significant predictors of child behavior problems, whereas in comparison families, only maternal stress was a reliable predictor.
However, none of these 13 studies provided a phenomenological or genuinely transactional view of the mother-children unit as a newly emerging family system.
It appeared that the primary method of analysis involved studying battered women and children separately.
Although they were undergoing the realities of an immediately felt and shared family crisis, their p e rceptions, b e liefs, and experi ences were treated as individual and unrelated. This certainly points to a dearth of information on battered mothers and children as an evolving system in crisis. Based on a review of the literature, it appears that the study of family violence would benefit from transactional research models, given that mothers and children often experience the battering as a system or unit.
In addition, among the national studies on family violence repor ted in Table 4 . 1, only two of the three studies (Straus, Gelles, and Steinmetz, 1980;Straus & Gelles, 1990) gathered data on child abuse as a concomitant to wife-battering. In reviewing and analyzing the shelter sample studies, with the exception of one study (Labell, 1979) that simply reported an N for shelter children or data in the form of demogr aphic statistics, no studies provided insight or information into children's shelter experience and immediate response to family violence, upheaval, and crisis.
A similar condition held when community samples were examined. Only two studies (Hilberman & Munson, 1978;Stark & Flitcraft, 1996) included data on children.  Hastings, , 1991Murphy, Meyer, & O'Leary, 1992;Vaselle-Augenstein & Ehrlich, 1992)  Research has corroborated these clinical conjectures. For example, often the initial battering incident has been reported to occur around a dependency issue, evoked by events such as the honeymoon, the wife's pregnancy, and birth of the first child (Hamberger & Hastings, 1991;Okun, 1986;Vaselle-Augenstein & Ehrlich, 1992;Walker, 1984). Research has suggested that relationship events (i.e. pregnancy, birth of first child) evoke the conflicting fears of engulfment and abandonment or alienation. Pregnancy has been found to raise issues of dependency and responsibility. Parenthood signals a shift and rearrangement in interpersonal boundaries in order to include a new family member.
Dependency is a two-sided coin. Batterers not only reported fearing the threat of emotional fusion but also fearing separation. For example, the batterer's behavior and violent outbursts (Gondolf, 1985(Gondolf, , 1988a were often motivated by fear of abandonment, accounting for the increased risk and threat of violence that women experienced when they leave the batterer. Rounsaville (1978) found that 71% of the battered wives he studied reported that their husbands had threatened to commit suicide if they left. As Rounsaville's research showed, even five years after the divorce or separation, many of the male batterers were still following or harassing their former spouses/partners.

Need £or Control, Lack 0£ Verbal Assertiveness
Batterers appear to need to be more in control in their intimate relationships than nonbatterers.
According to research (e.g., Allen, Calsyn, Fehrenbach, & Benton, 1989;Vaselle-Augenstein & Ehrlich, 1992), the need for control is also related to problems with dependency, adequacy, and self-perception of the skills  Allen et al. (1989), who argued that batterers attempt to control "not out of a need to dominate, but rather because they feel inadequate in close relationships" (p . 142) .

The research reviewed by Vaselle-Augenstein and
Ehrlich (1992) strongly suggested that men who batter do not show a significantly greater need for power in comparison to nonbatterers, but that they are significantly less verbally assertive. Men who batter were found to be signific antly less verbally assertive than either happily married or unhappily married nonbatterers. Maiuro, Cahn, and Vitaliano (1986) further examined the "verbal assertion deficit" hypothesis.
These researchers administered their Assertiveness and Aggression Inventory, an instrument that provides separate scales designed to assess the ability to refuse requests and the ability to initiate requests, to a clinical sample of male batterers (N=78) and a comparison group of nonviolent men (N=2 9).
Interestingly, Maiuro et al. (1986) found that batterers and nonbatterers did not differ significantly on the measure of refusal assertiveness; however, batterers were found to be significantly lower on the measure of request assertiveness. In addition, significant differences in anger and hostility were also found between groups, with a positive correlation between refusal behavior and overt anger/hostility. Conversely, a negative correlation was found between initiating/request behavior and covert anger/hostility, suggesting the presence of social skill deficits in male batterers. These authors concluded that , because batterers were also found to be more hostile than nonbatterers, batterers appeared to be able to defend their rights, whereas they had difficulty with the positive expression of needs. Accordingly, growing research evidence (Gondolf, 1985(Gondolf, , 1988aOkun, 1986) has suggested that the batterer's excessive need for control is associated with related emotional difficulties in intimate relationships . Thus, batterers often perceive their wives' assertiveness as personally damaging to them.

Problems of Intimacy and Dissociative Reactions
Research {Hastings & Okun, 1986) supported the well-documented clinical observation and seeming consensus that batterers have extreme difficulties with intimacy. Batterers have often reported simultaneously feeling both emotionally isolated and extremely dependent on their wives, to whom they tacitly assign the role of emotional arbiter (Dutton & Strachan, 1987;Gondolf, 1988a ;Gondolf, 1988b). It is typical to find a relationship dynamic in which the batterer perceives himself as assigned the role of "reason" and perceives his wife as assigned the role of "emotional interpreter." The batterer, who is afraid of his feelings and must keep them at a distance, dissociates from affect. For him, his wife "holds" the feeling dimension of the relationship.
Further research {Dutton & Strachan, 1987;Goldstein & Rosenbaum, 1985;Neidig, Freidman, & Collins, 1986;Rosenbaum & O'Leary, 1981), comparing batterers to nonbatterers or normative population data, showed that men who batter displayed fewer ego strengths, lower selfesteem, and inadequate verbal assertiveness skills in the context of request behavior. Using spouse-specific a ss ertivene ss score s d erived from r esponses on the Thematic Apperception Test in relation to need for power , Dutton and Strachan (1987)  Multiaxial Inventory (MCMI), Tolman and Bennett (1990) raised an important methodological point. They contended that the failure to observe elevated mean scores on these instruments may be a function of the way the data are organized and interpreted. Reinforcing Hastings and Hamberger's (1986) position, Tolman and Bennett (1990) suggested that male batterers' psychopathology was undetected because its manifestation ( highly selective about whom they permit to get close. Murphy et al. (1992) found that physically abusive men displayed more autonomous and expressive personality disturbance--patterns of affective dysregulation and antisocial tendencies --than maritally distressed nonabusive men. When compared with controls, these abusive men displayed significantly different and more aberrant personality profile patterns.

Impulse Control, Hostility, and Suspicion
Growing research evidence has qualified the relatively rudimentary clinical understanding of impulse control and suggested two different interpretations of the dynamics of impulse control among men who batter.
The impulse control of men who batter was found to be selective. Batterers were not uniformly disinhibited.
As research indicated (Bernard & Bernard, 1983;Sonkin, Martin, & Walker, 1985), batterers often have awakened their victims from sleep in order to beat them; aimed their blows where the injuries or bruises will not show; and stopped before they kill their victims. Batterers also have shown impulse control when they choose not to assault superiors or coworkers who frustrate or oppose them.
Clinicians (e.g., Bograd, 1984;Leeder, 1994;Hansen & Harway, 1993) have contended that these findings suggest that the batterer may store up aggression until it can be discharged safely and selectively against weaker individuals. Battered women have frequently reported that assaults or attacks occurred around ostensibly trivial incidents (e.g., an over-cooked or late meal, a misplaced or un-ironed shirt). Research has strongly suggested that batterers may deliberately create "crises"--"mountain out of molehill" confrontations--as an intentional or unwitting means of discharging pent up aggression (Gondolf, 1985(Gondolf, , 1988a(Gondolf, , 1988bOkun, 1986;Stark & Flitcraft, 1996).  Carlson (1990), Rounsaville (1978), and Sonkin et al. (1985) reported that between 44%-67% of batterers who were seen in treatment programs were found to have had previous Similarly, Maiuro et al. (1988) found that batterers obtained significantly higher scores (that approximated the scores of men who were generally assaultive) than non-batterers on a measure of suspicion, derived from the

MMPI.
Typically, the batterer's suspicions revolved around his wife's sexuality and sexual behavior. As studies of battered women have documented, batterers were often convinced that their wives were having affairs or that their wives used sex to gain an advantage. For example, Gayford found (1975Gayford found ( , 1994  used the mechanism of projection to justify the battering and abuse, and ultimately blamed the victim. The most common observation among psychotherapists (e.g., Coleman, 1980;Gondolf, 1985) has been that batterers feel little guilt or concern other than concern generated by selfinterest (e.g., possible arrest, trial, or permanent loss of and separation from spouse). Whether batterers, as a group, are capable of genuine remorse has strong implications for treatment and intervention. Murphy, Meyer, and O'Leary (1992)  Hamberger ) administered the MCMI, the Beck Depression Inventory, and the Novaco Anger Scale to batterers and non-batterers and found significant differences on most of the scales. The results indicated greater overall pathology in the batterers, particularly in regard to comfort in intimate relationships.
After reviewing the research on men who batter, vaselle-Augenstein and Ehrlich (1992) concluded that remorse was rarely manifested in the lives and personality dynamics of this group of men. They observed that subsequent data contradict Walker's (1979Walker's ( , 1984 earlier theoretical assertion that the batterer expressed authentic contrition for his behavior. Dobash and Dobash (1984) found that batterers rarely, if ever, experienced regret even after a severe attack and, if contrition was expressed, it was usually short-lived, occurring only after the first attack.
Similarly, Kelly, Berscheid, Christensen, Harvey, Huston, Levinger, et al. (1983) found that battered women were split relatively evenly in their perceptions regarding their husband's remorse over the attack. In fact, almost half of the sample did not subscribe to the conviction that the batterer was regretful or contrite at all. Hastings and Hamberger (1986) supported these observations on the basis of observed differences in test profiles between batterers and nonbatterers on the MCMI, which showed that batterers, in comparison to nonbatterers, were generally lacking in empathy.
These findings have seriously challenged the credence attached to several core concepts underlying learned helplessness as an explanatory mechanism accounting for battered women's behavior. In contrast to learned helplessness, the research has suggested the following description : Many battered women seemed aware of, and objective about, the limitations of their abusers and were not gullible in their belief of contrition or spontaneous positive change. Many women did not remain in the violent relationship due to the assumption that the batterer is all powerful and they were helpless; that he was right in his perceptions and they were wrong in their beliefs; that he was the injured party and it was "all their faultn; and that he, rather than they, defined their identity. The fol l owing provides a summary and overview of the personality and characterological deficits prevalent in male batterers' profiles . Bernard and Bernard (1983)

Research Plan
The analysis and evaluation of research findings supporting learned helplessness and health psychogenesis proceeded along nine steps. These steps are summarized below.
1. At the single study level, the literature on shelter and community samples was surveyed and studies were identified.
2. Studies were reviewed and salient study characteristics and major research findings were summarized (by this investigator) and were reported (described) in Tables 5.1 and 6.1.

3.
In  When a finding in a study was not directly applicable ("NDA"), it was excluded from the outcome analysis.
Similarly, when an entire study showed no direct relationship to a test of core concepts of either theory, it was also eliminated f r om the overall analysis .
Accordingly, study outcomes were assigned to one of four Inter-rater reliability was considerably high (94% agreement, shelter studies; 91% agreement, conununity studies). When the two independent raters did not agree, the judgment of a third independent rater was used to break the tie.
8. A single outcome measure did not appear to be sufficient, and thus the decision was made to qualify each study's outcome rating by the study's overall level of design quality (high , medium, low) . Therefore, a study's outcome was conside red more significant as design quality increased. Tables 5.3 and 6.3 (in Chapters 5 and 6) organized studies by level of design quality and outcome, or by which theory the research supported.
9. Next, outcome findings were integrated across shelter and conununity samples.

Learned Helplessness
In an attempt to explain the battered woman's behavior, Lenore Walker (1979Walker ( , 1984 proposed the "battered women syndrome." Although Walker rejected the notion of a victim-prone personality that predisposed the battered woman to violence, she did purport to identify characteristics of the battered woman in relation to the abuse experience. These characteristics encompassed socialization, gender, and psychological factors.  (Walker 1979(Walker , 1984 attributed this susceptibility factor to a range of experiences. These women reached a high point of fear, anxiety, and depression; then became less fearful, anxious, and depressed as their anger, disgust, and hostility toward the abuse rose and as their "resigned acceptance" decreased. Acknowledging the inf luence and interaction of gender in conjunction with personal history as a precipitating factor in "abuse vulnerability," Walker and Browne (1985) added to the data that supported the consensus that no particular personality pattern or trait leads to a person becoming a victim. Interestingly , a core concept of Walker's application of the learned helplessness theory to battered women, was predicated on a genetic view of childhood susceptibility. In the learned helplessness theory , women were viewed as vulnerable and, therefore, easy prey to their abusers.
Although the timeline for the onset of learned helplessness behavior, symptoms, and characteristics was designated , according to Walker's application, as beginning at the first abuse incident, her analysis of the progress i on and proclivi ty toward learned helplessness was described as of ten beginning in the woman's early childhood . The core concepts and basic assumptions of the learne d helplessness theory were summarized in Chapter 5 ( Figure 5.2) .

Psychogenesis
Despite the popularity of Walker's views, no substantive or consistent body of research has yet emerged that confirms a victim-prone personality in relation to battered women. Instead, research has shown that battered women conside r and try a broad range of solutions to resolve their re lations hip problems and stop the abuse. Studies have indicated that battered women employ considerably more effective problem solving strategies than comparison groups of women in conflicted/distressed but man-battering relationships (Campbell, 1989); actively seek help in contrast to being compliant, passive, or help-resistant (Gondolf, 1988b); and show higher self-esteem than controls, a greater sense of self-worth, and less adherence to traditional sex role ideations (Hotaling & Sugarman, 1986). This research evidence contradicted the view that battered women displayed low self-esteem, denied or avoided help seeking behavior, were passive, and conformed more typically to traditional sex roles.
According to Bowker's (1983aBowker's ( , 1993 Bowker's research (1983aBowker's research ( ,1992described the battered woman's experience as more active and rational than the concepts of learned helplessness and the "battered woman syndrome" would lead one to believe. In contrast to Walker , Bowker (1983a)  Similar to other researchers (e.g., Gondolf, 1988b ;Stark & Flitcraft, 1996)  For Bowker and socially-minded theorists, the timeline for the abuse dynamic did not begin in a precursor childhood but in the actual first incident when a woman was hit.
A number of studies have corroborated the prior mental health, well-being, and normalcy of battered women . They also · have yielded n egat i ve evidence regarding self-blame and a cognitively deficient state.
In this research light, battered women emerged as "heroes" and survivors who made informe d decisions that, at first glance, appeared counter-intuitive to a nonabused social culture; but, on closer examination, made s ound, practical sense .

Learned Helplessness versus Health Psychogenesis
The main objective of this dissertation was to systematically evaluate the strength of the research evidence on both sides of the learned helplessness-health psychogenesis controversy and debate. The research procedure involved a two -st ep process: research analysis on the shelter and community sample level, followed by an overall evaluation of the shelter-community sample outcomes by de s ign quality . The goal was to shift from the evaluation of design quality and outcome at the level of individual studies to a broader level of assessment.
In Chapter 3 on Methodology predictive outcomes were identified for classifying a study ' s findings: a prolearned helplessness position, a pro-health psychogenesis position, and a modified pro -lear n e d helplessness/prohealth psychogenesis position. In a pro-learned helplessness outcome , the majority of data would be found to support the learned h elplessnes s theory, concepts, and assumptions: that the behavior of battered women reflected cognitive, mo t i vational, and affective distortions . In a pro -h ealth psychoge nesis outcome, the majority of data would be found to support the health psychogenesis theory, concepts, and assumptions: that the behavior of battered women reflected effective problemsolving skills, accurate perceptions, and adequate positive motivation. Battered women were psychologically healthy, proactive, competent, and emotionally strong.
In a modified pro-learned helplessness/pro-health psychogenesis outcome, study results would be found to equally support aspects of both theories . Furthermore, outcomes found to equally support both theories, or aspects of each theory, might suggest reconsideration of both theories, or the importance of heterogeneity: that different women displayed contrasting characteristics, that is, battered women either displayed learned helplessness and the "batte red woman syndrome," symptomfree, healthy responses, or a combination of both.
Tables 5.2 and 6.2 summarized the results of the individual design quality ratings per dimension per study. These tables also presented a total design quality score (TDQS) for each study. The tables were organized by sample type: Table 5.2 represented shelter samples and Table 6.2 depicted community samples. This provided the first step in the analysis: a numerical value of design quality t h at reflected high, me dium, or low methodological adequacy. In addition, each study was evaluated in terms of outcome: or which theory it generally supported .
Therefore, in the first step of this analysis, a study was classified into one of four categories: prolearned helplessness, pro-health psychogenesis, "both" in which data equally supported learned helplessness and health psychogenesis, or as "not directly applicable" (NDA) . NDA studies focused on general issues of abuse and battering (e.g., prevalence, demographics) but did not directly test core assumptions or related hypotheses of either theory.
Outcomes were analyzed in relation to design quality (TDQS). Thus, both the type of outcome and level of design quality were considered. Tables 5.3 and 6.3 organized studies into high, medium, or low design quality and reported their outcome: whether they supported learned helples sness, health psychogenesis, both theories equally, or were not directly applicable. Table 5.3 represented outc omes on shelter samples and Table 6.3 represented outcomes on community samples.
To provide an additional and somewhat different perspective on the data, dropped from this second step of the analysis, given the lack of relevance and applicability to the research questions.) In essence, studies that fell within the category of "bothu reflected the modified pro-learned helplessness/pro-health psychogenesis outcome predicted in this research.

Results
As Table 5.3 on shelter samples indicated, in the high design category, three, or 10.3%, of the studies supported lear n e d helples s ness, twelve, or 41.3 %, of the studies, supported health psychogenesis and one, or 3.4% of the studies supported both learned helplessness and health psychogene sis. At the medium or average design quality level, six, or 20.6%, of the studies supported learned helpl essnes s a nd two studie s , or 6.8 %, s upported health psychogene sis, and two studies, or 6.8 %, supported both theories. In the low design category, one, or 3.4% of the studies supported l e arned helplessness and two , or 6.8%, supported health p s y c hogenesis concepts.  12, or 41.3 %, of the studi es supported health psychogenesis; and one, or 3.4%, supported both theories.
Of the medium design quality studies, six, or 20.6%, favored learned helplessnes s; two, or 6 . 8%, favor e d health psychogenesis; and two, or 6.8%, supported both.
In sum, the majority of res e arch (particularly that of high design quality) f a v ore d heal t h psychoge nesis theory, and its basic concepts and assumptions. These It was also observed that, to date, there is very little research evidence of a systems perspective on the mother-children relationshi p minus the male batterer .
Although the importance of accrued data on. a systems perspective has been underscored in the treatment, research , and theoretical l iterature (Bograd , 1984 ;Hansen & Harway, 1993;Gondolf, 1988a;Pinsof & Wynn , 1995;Sprenkle, 2002), very little research on the ways battered women and childre n interac t, develop, cope, and change as a family unit after the abuse has ended is available . This lack of a systems perspective may, in part , be a r e fl ec t i on of t h e practi cal, methodological, and treatment complex ities inherent in research on the social problem of domestic vi o lence .

Discussion and Recommendations
Although the majority of the re s earch eviden ce in this investigation appeared t o support the pro-health Another limitation of this research involves the potential influence/confound of investigator bias.
Although independent raters were used, the initial data (study findings) which formed the basis for the independent ratings, were identified (organized) by this investigator. However, the process of identifying findings and study characteristics was relatively straight-forward and require d minimal influence. These Where is she in her one-t o-s even on average attempts to leave? Is she staying for v alid pra ctical reasons?
Research consensus (Gelles, 1997;Leeder, 1994;Straus, Gelles, & Steinmetz, 1990; has suggested that a battered woman is at greatest risk to her physical safety when she does leave the abuser. Therefore, it may be argued that, in contrast, a battered woman's decision not to leav e impulsively (without a safe plan in place) may be a more accurate index of problem solving. It may prove more informative for future research to devise multidi mensional outcome criteria and to more closely examine the underlying cognitive, motivational, and emotional processes informing the battered woman's complex s urvival behaviors.
In One other limitation of this research pertained to the method of obtaining ratings. Individual study findings were identified by this investigator; this list of findings per study provided the initial data pool for interrater reliability. Three raters (two undergraduate students in psychology and one graduate level mental health clinician, serving as a tie breaker) were trained (by this investigator) in the core concepts and assumptions of each theory and asked to independently rate the findings as providing evidence for learned helplessness, health psychogenesis, or equally supporting both theories . This may present a possible confound.
Additionally , the identification and listing of findings was relatively straightforward and, therefore, less susceptible to investigator bias.
Based on the above discussion, the following recommendations are proposed: 1. In the context of clinical practice with battered women, learned helplessness has almost, at times, functioned at the level of an assumptive given. Learned helplessness and the battered woman syndrome, concepts well known in popular culture, have also been widely embraced by the clinical community. However, in light of the preliminary evidence obtained in this investigation, it would seem advantageous to rethink these concepts: i.e., consider both theoretical orientations in devising assessment/treatment protocols .
2. Adopt an integrated meta-theoretical perspective that draws the on aspects of both learned helplessness and health psychogenesi s .
3. Based on this meta-theoretical approach, develop a multi-dimensional decision branching tree of outcome, assessment, diagnosis, and treatment in clinical work and research on battered women and children. View battered women and children as a heterogeneous rather than homogeneous popu l ation wi t h a wide range of socialservice needs.
4 . Based on the pauc ity of research that addresses mothers and children a s a n e wly evo l v ing family unit, design research that simultaneously studies and combines data on both mother's and children's experiences. Create a transactional view of the ir family experience and recovery.
5 . Conduct further re search on children, per se, that is developme ntally -f ocused and gender-specific.
6. Paralleling the recommendation for transactional research on the mother -child family unit, create educational treat ment programs that unite and integrate personal narratives and different perspectives of mothers and children as survivors of abuse.