As might be expected, no reported studies have investigated the MMPI-2 [*95] scores of battered women prior to being battered. Such a study would have to be a longitudinal study taking baseline MMPI-2 scores from a large group of subjects (perhaps high-school seniors) and then following those subjects for a substantial period of time, interviewing and administering the MMPI-2 at intervals, perhaps along with other instruments. As yet, no such study has been reported. Therefore, we do not know with any degree of certainty whether battered women, prior to being battered, were "normal" or showed certain psychological problems (and, if so, what problems). However, research on MMPI-2 scores of battered women strongly suggests that they are usually "normal" prior to the abuse.
B. MMPI and MMPI-2 Scores of Women in a Battering Relationship
In thirteen studies, the MMPI or MMPI-2 was administered to battered women either while the women were still in the battering relationship or shortly thereafter. In eight of the twelve studies of groups of battered women, all or many of the battered women were recruited from domestic violence shelters and outpatient domestic violence clinics. n30 The other four studies tested battered women in other settings. Back, Post, and D'Arcy n31 studied battered women residents of a psychiatric hospital. Gellen et al. looked at battered women who were in a residential center for "distressed women." n32 Margolin recruited only participants who were living with their spouses. n33 In a study by Charboneau, the participants were all living in their own homes, with or without their mates. n34 Finally, a case study of one battered woman was reported by Follingstad. n35
n30 Philip L. Davidson, The Development of an MMPI Profile for Battered Women: A Partial Replication of Rosewater's 1982 Study (1991) (unpublished master's thesis, University of Houston-Victoria, Houston, TX); Patricia Egan, Assessment and Diagnosis of Trauma and Psychological Problems in Abused and Battered Women (1997) (unpublished Ph.D. dissertation, The Chicago School of Professional Psychology) (U.M.I. No. 9820775); Fariha Khan et al., MMPI-2 Profiles of Battered Women in Transition, 60 J. PERSONALITY ASSESSMENT 100 (1993); Morrell & Rubin, supra note 9, Amber Rollstein & Jeffrey Kern, Correlates of Battered Women's Psychological Distress: Severity of Abuse and Duration of the Postabuse Period, 82 PSYCHOL. REP. 387 (1998); Rosewater, supra note 7; Wall, supra note 6; Judy Wilson, The Psychological Functioning, Self-Concept, and Locus-of-Control of Battered Women in a Spouse Abuse Shelter, (1990) (unpublished Ph.D. dissertation, University of Florida).
n31 Susan Back, et al., A Study of Battered Women in a Psychiatric Setting, 1 WOMEN & THERAPY 13 (1982).
n32 Murray I. Gellen et al., Abused and Nonabused Women: MMPI Profile Differences, 62 PERSONNEL & GUIDANCE J. 601 at 602 (1984).
n33 Margolin, supra note 6, at 204.
n34 Dolores Charboneau, A Comparison of Psychological Traits in Currently Battered, Formerly Battered, and Non-Battered Women (MMPI) (1986) (unpublished D. Ed. dissertation, University of South Dakota) (on file with the U. of S. Dakota Library).
n35 Diane Follingstad, A Reconceptualization of Issues in the Treatment of Abused Women: A Case Study, 17 PSYCHOTHERAPY: THEORY, RES., & PRAC. 294 (1980).
[*96] In all of the thirteen studies, except Wilson's, n36 the MMPI mean (average) clinical scale scores of the group of battered women were reported. Wilson reported the clinical scale scores of each of her sixteen participants and did not report the mean of all sixteen scores on each scale. n37 For purposes of this review, the mean scores of Wilson's sixteen participants on all clinical scales have been calculated in order to compare them with the means reported in the other studies. With the exception of the Margolin study, n38 all studies found the mean scores of the battered women were elevated above a T score of sixty-five on at least one clinical scale.
n36 Wilson, supra note 30.
n37 Id. at 185.
n38 Margolin, supra note 6.
Two studies that reported individual scores on the MMPI clinical scales reported that some battered women did not show high elevations on any MMPI clinical scales: three out of thirty-one participants in the study by Khan et al. n39 and five out of sixteen participants in Wilson's study n40 showed no significant elevations. The fact that some battered women do not show MMPI elevations would shed some doubt on the "trait" hypothesis that battered women are battered because they are "crazy" and provoke abuse. n41
n39 Khan et al., supra note 30, at 104.
n40 Wilson, supra note 30, at 136.
n41 Assuming that battered women are "normal" prior to the battering, it would be important to investigate what protects some battered women from developing the symptoms of psychological impairment measured by the MMPI-2. For example, can it be accounted for by a short length of abuse, low severity of abuse, supportive relatives and friends, some inner strength of those women, or some other factor(s)? There is an enormous literature on the subject of psychological resilience, but a discussion of that issue is beyond the scope of this article.
The code patterns for the mean clinical scale scores found in the thirteen studies are as follows:
Back et al. (1982)
Gellen et al. (1984)
Khan et al. (1993)
463 (but none at sixty-five or above)
Morrell & Rubin (2001)
681 (PTSD) & 468 (non-PTSD) n42
Rollstin & Kern (1998)
6824 (the scores for the 2 & 4 scales were
n42 Morrell and Rubin administered the MMPI-2 and the Posttraumatic Stress Diagnostic Scale (PDS) to 93 battered women. The PDS identified 58 as meeting DSM-IV (1994) criteria for PTSD. The non-PTSD participants commonly reported PTSD symptoms but did not meet full criteria for a PTSD diagnosis. Morrell & Rubin, supra note 9, at 153. See discussion of PTSD at text accompanying notes 81-99 infra.
[*97] The similarities in the code patterns are striking. Six out of the thirteen studies reported a three-point code with a combination of 4, 6, and 8.
Egan's study provides a good illustration. A comparison between the mean scores of the thirty-eight battered women and the mean scores of the thirty-four women in the control group reveals the following: n43
Elevations on the 6 (paranoid) scale for battered women are not surprising and, in fact, might be expected, because fearfulness and suspicion of others might be viewed as a logical outcome of having been abused by one with whom an individual had an intimate relationship. Intimate relationships are supposed to be based on trust, and abuse in an intimate relationship would destroy that trust.
Elevations on the 8 (schizophrenia) scale for battered women are also not surprising, because the eight scale measures confusion in thought processes and feelings of being overwhelmed. It is not necessary to be schizophrenic in order to receive an elevated score on the 8 scale. Being abused by an intimate partner could certainly cause confusion in thought processes and feelings of being overwhelmed.
Elevations on the 4 (psychopathic deviate) scale, on the other hand, are more difficult to interpret. Follingstad, a single case study, was the only study that reported a three-point code that did not include an elevated four scale. n44 With the exception of Morrell and Rubin, all the studies of groups of battered women reported three-point code scores for mean MMPI scale scores that included a 4. Morrell and Rubin reported a 681 three-point code for their battered women who met the criteria for PTSD and a 468 code for their non-PTSD battered women. n45
n44 Diane R. Follingstad, A Reconceptualization of Issues in the Treatment of Abused Women: A Case Study, 17 PSYCHOTHERAPY: THEORY & PRAC. 294 (1980). In this case, the woman and man were dating, not married or living together, and their relationship was not a lengthy one.
n45 The fact that the PTSD battered women had a 681 3-point score does not mean their 4 scale was not elevated--in fact, the mean 4-scale elevation was actually higher for the PTSD battered women (70.32) than for the non-PTSD women (66.83). Morrell & Rubin, supra note 9, at 154.
[*98] As a noted text on the MMPI indicates, "Scale 4 was developed to measure the 'personality characteristics of the amoral and asocial subgroup of persons with psychopathic personality disorders' . . ." n46 "Psychopathic" is an extremely strong term. No one would feel safe recommending custody of a child to a psychopath.
n46 FRIEDMAN ET AL., supra note 11, at 101, quoting W.G. DAHLSTROM ET AL., AN MMPI HANDBOOK: VOL. 1. CLINICAL INTERPRETATION 195 (rev'd ed. 1972).
If a custody evaluator did not look more deeply into the possible reasons for an elevation on the 4 scale, the evaluator might infer that the individual had serious characterological traits similar to the "typical" high scale 4 persons, who
show impulsiveness, poor interpersonal judgment, unpredictability, social alienation, and a reduced sense of responsibility and morals. They may evidence poor work and marital adjustment. They tend to sacrifice long-term goals for short-term desires and seem limited in their capacity to anticipate consequences. Social relations typically are shallow, and strong loyalties rarely are developed. Although these persons sometimes make good first impressions, their unreliability, self-orientation, manipulativeness, and resentment soon become apparent. n47
n47 FRIEDMAN, supra note 11, at 296.
This is a very damning description to apply to a parent in a custody evaluation. To make matters worse, the most common three-point code types for battered women in the thirteen studies are various combinations of 4, 6, and 8. A 46 code has been described as follows: "Key features are anger, resentment, distrust, sullenness, irritability, hypersensitivity to criticism or to demands by others, and projection of blame onto others." n48 A 48 code has been described as follows:
Adults with this code almost always show severe problems, usually as a major personality disorder or a psychotic process. Distrust is a central characteristic for this group and is characterized by a sense of profound alienation and disconnection from others. n49
n48 Id. at 300.
n49 Id. at 307.
A custody evaluator seeing such an MMPI profile for a parent would be loathe to suggest custody to such an individual.
A custody evaluator who looked more deeply into the possible meanings of Scale 4 elevations, however, would find much cautionary language about such elevations that would be relevant to battered women. For example,
Duckworth and Anderson (1995) viewed "fighting something" (p. 164) as a cardinal feature of an elevated Scale 4. They suggested that the individual may be in conflict with his or her parents, friends, spouse, society, or school and that it is essential to examine the context in which the person is being assessed. n50
n50 Id. at 105 (emphasis added), quoting from J.C. DUCKWORTH & W. ANDERSON, MMPI INTERPRETATION MANUAL FOR COUNSELORS AND CLINICIANS (1995).
[*99] The context of custody litigation and domestic violence would certainly be essential to examine, but often this caution is not heeded.
Another important issue regarding Scale 4 is its source. In a 1992 article, Rhodes described her study of MMPI 4 scale scores of battered women and nonbattered women. She pointed out that it is not surprising that battered women have elevated scores on the 4 scale when the criterion group used to create the 4 scale probably contained many victims of physical and sexual abuse. n51 Thus, it is logical that abused women would score high on the four scale. However, that does not mean that they would have scored high on the 4 scale if they had not been abused.
n51 Nancy Rhodes, Comparison of MMPI Psychopathic Deviate Scores of Battered and Nonbattered Women, 7 J. FAM. VIOLENCE 297 (1992). Rhodes states (298-99):
The criterion group used by Dahlstrom et al., [AN MMPI HANDBOOK] (1972) in developing the scale was made up of predominantly female delinquent individuals who were placed in a psychiatric setting by court order. Delinquency was defined as stealing, lying, truancy, sexual promiscuity, alcohol abuse, and forgery (but did not include capital offenses) . . .
It is particularly relevant to note the similarities between the criterion group used in the development of scale 4 and the profile of the typical victim of sexual abuse. Behavioral indicators of sexual abuse include running away from home, substance abuse, sexual promiscuity, truancy, and shoplifting. Such antisocial behaviors are also associated with physical abuse and neglect (Edwards and Gil, [BREAKING THE CYCLE: THE ASSESSMENT AND TREATMENT OF CHILD ABUSE AND NEGLECT] 1986) and apply to both males and females. This implies that it is quite possible that many members of the criterion group used by Dahlstrom et al. (1972) in the development of the Psychopathic Deviate scale were themselves victims of physical or sexual abuse.
In a 1990 dissertation, Wilson expressed her suspicions that the psychological deficits ascribed to her battered women sample by the MMPI results might be caused by (or enhanced by) the abuse they had suffered. n52 She called for more research to address that issue:
There are few conclusive studies concerning battered women and their psychological characteristics. Earlier studies blamed the victim for her abuse. More recent theories postulate that battered women have psychological characteristics similar to prisoners of war or victims of brainwashing techniques. Long-term follow up studies are needed to assess the changes which occur in these characteristics as a battered woman escapes her violent home and rebuilds a healthy, nonviolent life for herself and her children. n53
n52 Wilson, supra note 30.
n53 Id. at 154-55.
Wilson called for longitudinal studies to answer the "characterological versus reactive" question:
Do battered women have dysfunctional psychological areas as a result of their childhood experiences that influence their choices of mates and choices of behavior once in a spouse abuse marriage or is there a post-battering personality which is a result of the battering?" n54
n54 Id. at 154.
[*100] C. MMPI Scores Tend to Normalize After the Abuse Ends
A good way to investigate whether a battered woman's elevated MMPI scores are characterological or reactive would be to compare those scores with the MMPI scores of the same woman after the battering relationship has been terminated for some period of time. If the elevations decreased, that would seem to support the reactive theory. Follingstad's pioneering study in 1980 made such a comparison, with results that suggest the validity of the reactive theory. n55 Follingstad found that the three highest MMPI scores of her battered patient, "Barbara," while the abuse was going on were a T-score of 80 on the 3 (Hysteria) scale, of 72 on the 7 (Psychasthenia) scale, and of 70 on the 6 (Paranoia) scale. n56 Nine months after Barbara terminated therapy, which was a few months after she successfully broke off the abusive relationship, Barbara's MMPI scores were substantially reduced: 55 on the 3 scale and 55 on the 7 scale. The 6 scale remained elevated (67), but Follingstad noted that such an indication of continuing anger and distrust was "at least partially understandable due to the negative experiences she had undergone." n57
n55 Follingstad, supra note 43.
n56 Id. at 297.
n57 Id. at 301-02.
A less ideal, but still somewhat persuasive, method to investigate whether a battered woman's elevated MMPI scores were characterological or reactive would be to compare the MMPI scores of groups of currently (or very recently) battered women with scores of different groups of formerly battered women. Several studies have made such comparisons. n58 These studies tend to show that MMPI elevations decrease after the battered women has been out of the abusive relationship for a period of time and therefore lend support to the view that such elevations are reactive rather than characterological.
n58 Back et al., supra note 31; Charboneau, supra note 34; Gellen et al., supra note 32; Margolin, supra note 6; Rosewater, supra note 7.
Charboneau, another 1980s pioneer in this field, compared MMPI scores of twelve currently battered women, twelve formerly battered women, and twelve never battered women. n59 Charboneau had hypothesized that her study might find "traits that may be influential in keeping currently battered women in abusive situations and unable to overcome their victimization." n60 The data, however, seem to point more in the direction of a reactive state theory. n61
n59 Charboneau, supra note 34.
n60 Id. at 6 (emphasis added).
n61 In different parts of her dissertation, Charboneau seems to take different positions. On the one hand, when discussing the elevations of the three groups on Scale 1, Charboneau notes that "Some of the characteristics which Graham attributed to people with elevated scores on Scale 1 would seem appropriate for battered women. These characteristics include excessive bodily concern (possibly concern over pain from beatings). . . ." Id. at 96. Thus, she seems to characterize the elevations on Scale 1 as reactive.
On the other hand, in the "Implications and Discussion" section of her dissertation, Charboneau states:
The findings that battered women score high on Scale 4 with consistency may suggest not only that battered women have feelings of non-conformity, rebellion, anger, and disregard for social custom but also may suggest something more. In what may be the major finding of this study, this researcher hypothesizes that these women's feelings may cause them to be attracted to men who also have [the same] feelings. Furthermore, these men's feelings of anger appear to be vented through acts of violence. Id. at 92.
The logical leaps in this paragraph are not sufficiently justified by the data. From elevations on scale 4 Charboneau leaps to hypothesizing that because battered women have such elevations they have certain feelings, that because of those feelings they are probably attracted to men with similar feelings, and that men with those feelings vent their anger "through acts of violence." In other words, certain traits in certain women cause them to be attracted to men who subsequently batter them.
[*101] The mean MMPI scores of Charboneau's three groups on the five highest scales are as follows:
Currently Battered Women
Formerly Battered Women
The fact that the MMPI elevations are lower for formerly battered women than for currently battered women could be interpreted as evidence that the elevations decreased when the women were no longer being abused, indicating that the abuse had caused the elevations. Similarly, the MMPI clinical scale scores for never battered women are (with the one exception of the 6 scale) even lower than the clinical scale scores for the formerly battered women.
Lynne Bravo Rosewater was the first researcher to address head-on the question of whether the elevated MMPI clinical scale scores for battered women were a cause or a result of their abuse:
Repeatedly I have seen professionals fail to distinguish the symptoms of victims of violence from the symptoms of the sufferers of mental illness or to understand their interplay. . . . In blatant victim blaming fashion, two common errors are made: the extreme fearfulness (paranoia) and confusion created by repeatedly experiencing violence are misdiagnosed as psychiatric symptoms, and/or the woman is diagnosed as having a character disorder, which is seen as a predisposition for the violence that occurs. Thus the victimized woman is viewed either as "crazy," with her tales dismissed as ravings, or as inadequate and provoking the violence in her life. n62
n62 Rosewater, supra note 7, at 200.
Rosewater administered the MMPI to three groups of currently battered women: Group I (fifty women) in a battered woman's shelter; Group II [*102] (twenty-nine women) from an early intervention program; and Group III (twenty-seven women) from a witness victim service center. n63 She compared the mean MMPI clinical scale scores for these three groups to the mean scores of twelve formerly battered women who had been clients of the three previously mentioned programs and had suffered no abuse for at least one year. n64
n63 Id. at 204.
n64 Id. at 204-05.
Rosewater also collected information from each participant concerning the length of time in the relationship, the length of time battered, the level of violence of the batterer's behavior, the level of physical damage to the woman, and the frequency of battering on a scale of one to six from "less than once every 2 years" to "more than once a week." n65
n65 Id. at 205-06.
One of the goals of Rosewater's research was to determine if a composite "MMPI profile" exists for battered women and, if so, what it would appear to signify in terms of possible diagnoses. n66 Another goal was to determine whether any of the other abuse variables she measured, such as the length of time in the relationship, correlated with the elevations on the MMPI profiles. n67
n66 Id. at 204.
n67 Id. at 212.
The MMPI pattern that emerged for the currently battered women was a 486 profile for Groups I and II and a 468 profile for Group III. For interpretation purposes, those code types are the same. n68 The elevated Scale 4 (T=71) of the currently battered women measures anger, the elevated Scale 8 (T=70) measures confusion, and the elevated Scale 6 (T=70) measures fear.
n68 Id. at 207.
If a mother going through a custody evaluation were to have a 486/468 pattern, the custody evaluator could conclude, on the basis of commonly used MMPI interpretation manuals, that the mother had a personality disorder or paranoid schizophrenia. Rosewater pointed out, however, that if the MMPI profile was placed within the context of the woman's life, it would indicate "a reactive behavior set to being a victim of violence, which includes anger, confusion, fearfulness, weakness, and a sense of pessimism." n69 Thus, taken out of context, the MMPI scores of a battered woman might lead a mental health evaluator to misdiagnose the woman as severely mentally ill--even psychotic--while she was actually suffering from the effects of the abuse. Rosewater thought the battered woman's condition could be most accurately described as Posttraumatic Stress Disorder (PTSD). n70
n69 Id. at 211.
n70 Lynne B. Rosewater, Feminist Therapy: Implications for Practitioners, in WOMEN AND MENTAL HEALTH POLICY 197-206 (Lenore Walker, ed. 1984). See discussion of PTSD at text accompanying notes 81-99 infra.
[*103] The formerly battered women had lower elevations than the currently battered women. Their MMPI mean profile was 849. That profile sets them apart from the currently battered women in terms of "less alienation and less feeling of inferiority." n71 In other words, the formerly battered women appeared to be recovering from the effects of having been abused.
n71 Rosewater, supra note 7, at 211.