Amer. J. Onhopsychiat.

63(4), October 1993

THE BATTERED WOMAN SYNDROME:

Effects of Severity and lntermittency of Abuse
Donald G. Dutton, Ph.D., and Susan Painter, Ph.D.

The concept of a battered woman syndrome was tested by assessing 50 battered

women and 25 emotionally abused women who had recently le@ their rela-
tionships. For both groups, essential features of the syndrome were present and
were signijkantly interrelated. Dynamic features of the prior abusive relation-
ship correlated signiJcantly with these sequelae. The concept of intermittency is
proposed as an alternative to the cycle of violence theory as main contributor to
the syndrome. Predictability of abuse was found to be unrelated to the intermit-
tency measure.

S ince its initial description by Walker

(1979, 1984), the battered woman syn-
drome has been widely discussed (Doug-
lence. She argued that effects of such vic-
timization were similar or identical to those
for post-traumatic stress disorder: learned
las, 1987; Schuller & Vidmar, 1992) and helplessness, re-experiencing of the trauma,
has been used as a basis for self-defense in intrusive recollections, generalized anxiety,
cases where battered women have killed lowered self-esteem, and social withdraw-
their abusers (Thyfault,Browne, & Walker, al. Finally, two apparently opposite emo-
1987). Walker (1979) initially described a tional responses are common: “psychic
three-stage battering cycle (tension build-up, numbing,” or reduced responsiveness to the
battering, contrition) that was predictable world, and generalized hyperarousal (such
to the victim. Sequelae of this repetitive as exaggerated startle responses). These re-
abuse and its anticipation included psycho- sponses are believed to be related to cu-
physiological stress, lowered self-esteem, mulative exposure to the abuse stressor.
and learned helplessness,which undercut mo- Douglas did not attempt to quantify the
tivation to leave the abusive relationship. abuse stressor per se or to specify the es-
Douglas (1987) defined the battered wom- sential dimension of spouse abuse trauma.
an syndrome as “a collection of specific Douglas also described a secondary com-
characteristics and effects of abuse on the plex of abuse sequelae that included ideal-
battered woman” @. 40), and subdivided it ization of the abuser, denial of danger, and
into three major categories: the traumatic ef- suppression of the victim’s own anger.
fects of victimization by violence, learned These responses, also common in the trau-
helplessness deficits resulting from the vio- ma literature (van der Kolk, 1987), are
lence and others’ reactions to it, and self- viewed as coping responses that occur un-
destructive coping responses to the vio- der extreme duress. Dutton and Painter

A revised version of a paper submitted to the Journal in September 1992. Authors are at: Department of
Psychology, Universiq of British Columbia, Vancouver (Dutton); and Department of Psychology, Carleton
Universily, Ottawa, Ontario (Painter).

0 1993 American Orthopsychiatric Association, Inc. 614

DUlTON AND PAINTER 615

(1981), for example, cited Anna Freud’s contrition phase. Schuller and Vidmar con-
( I 942) description of “identification with cluded that “not all couples go through the
the aggressor” as an explanation of how cycle of violence nor is there a universal
battered women, amongst others, cope with time frame for the cycle” (p. 280). Second,
a long-term relationship with a potentially they noted that the internal reliability of the
lethal other. Idealization of the abuser is re- symptoms comprising the “syndrome” has
lated to strength of the continued attach- not been empirically verified. In other words,
ment to him after relationship termination it is not known how internally reliable are
(Dutton & Painter, 1981). the symptoms of the battered woman syn-
Dutton and Painter argued that the per- drome, and how much they co-vary.
ceptual responses of self-derogation and The current study attempts to address
idealization of the abuser were related to these questions. It assesses three major se-
two structural features of the abusive rela- quelae of the battered woman syndrome:
tionship: a power differential and intermit- trauma symptoms, self-esteem deficits, and
tency of abuse. These features have been traumatic bonding or paradoxical attach-
found to increase attachment to an abusive ment to the former male partner (Dutton &
other in a wide variety of human and ani- Painter, 1981), to establish the extent to
mal studies (Fischer, 1955; Harlow & Har- which these factors intercorrelate and thus
low, 1971; Rajecki, Lamb, & Obsmacher, constitute a syndrome. It then relates the
1978; Scott, 1963). The concept of inter- strength of these sequelae to what we be-
mittency of abuse means that treatment oc- lieve are central features of long-term abu-
curs in negative-positive alternations, in sive relationships: power differentials and
which the onset of positive treatment is intermittency of abuse. It is hypothesized
contiguous with the offset of negative treat- that attachment, self-esteem deficits, and
ment. In abusive relationships, the negative trauma symptoms will intercorrelate signif-
treatment typically precedes the positive. icantly to form a syndrome and that this
What is essential to generating attachment syndrome will endure over a six-month pe-
is the extremity of both the good treatment riod. It is further hypothesized that the
and the maltreatment, and the temporal magnitude of these sequelae will relate sig-
juxtaposition of one extreme with the other nificantly to the intermittency of abuse and
(usually maltreatment followed immediate- to power differentials in the former rela-
ly by good treatment). While the onset of tionship.
the negative treatment may be predictable,
this predictability may be unrelated to ex- METHOD
tremity or to temporal juxtaposition. Dut- Women with a history of physical or emo-
ton and Painter maintained that intermit- tional abuse were recruited for this study
tency (or periodicity), not predictability, over a six-month period through three sour-
was the main contributor to the battered wom- ces: transition houses, a court-mandated
an syndrome and to traumatic attachment. treatment program for wife assaulters (who
Schuller and Vidmar (1992) have written were their partners), and newspaper adver-
a critical assessment of the battered woman tisements. To qualifL for the research sam-
syndrome from the perspective of its use as ple, a woman had to have left the relation-
a self-defense plea. They have raised two ship within the past six months. Sample
issues. First, they maintained that the “cy- source had no significant effect on any as-
cle of violence” cited originally by Watker sessment measure, except that transition-
(1984) as descriptive of the stressor is not house women and partners of treatment-
always present. Walker’s data revealed, for program men reported more frequent and
example, that only 65% of her cases dem- severe physical abuse than did women re-
onstrated a tension-building phase and 58% a cruited through newspaper advertisements.
616 BATTERED WOMAN SYNDROME

A total of 75 women participated in the views were audiotaped with the subject’s
study. Their average age was 31.4, mean permission.
amount of time in the relationship was 11.5
years (range six months to 44 years), and Independent Measures
mean time separated was 20.5 weeks. On The Conflict Tactics Scale. The CTS
average, these women had initiated 2.1 (Straus, 1979) is divided into three sub-
prior separations; half of them had experi- scales: reasoning (3 items), indicating a prob-
enced some form of abuse in a previous re- lem-solving orientation; verbal aggression (7
lationship; and 22 were childless. The 75 items), indicating verbal and nonverbal
subjects were divided into a Battered group (e.g., symbolic displays of force, smashing
(N=50)and an Emotionally Abused (EA) objects) means of threatening or hurting
group (N=25). The criterion for inclusion the other; and violence (9 items), indicat-
in the EA group was fewer than two inci- ing the use of physical force as a means of
dents of physical violence during the rela- conflict resolution. Items on the violence sub-
tionship, accompanied by extreme emotional scale range in severity from “pushing” to
abuse. ‘‘using a weapon on the other.” Respon-
dents are asked to rate the type and number
Procedure of conflict tactics used by both the self and
The women were assessed for abuse ex- the other person specified in the dyad.
perience by the Conflict Tactics Scale (CTS) Psychological Maltreatment of Women
(Straus, 1979) and the Psychological Mal- Inventory. Although the CTS is useful for
treatment of Women Inventory (PMWI) studying intrafamily violence, it does not
(Tolman, 1989). The total sample reported include a broad range of nonphysical ag-
very high degrees of verbal aggression di- gression. In order to assess this, Tolman’s
rected toward them in the relationship they (1989) PMWI was included. The PMWI is
had just left. For example, the mean report composed of 58 items (rated from l=never
of verbal aggression was 55.2 on the CTS, to 5=very frequently) that delineate forms of
which places this sample beyond the 99th emotionaVverba1 abuse and dominance/iso-
percentile for population norms published lation. Dominance/isolation includes items
by Straus, Gelles, and Steinmetz (1980). related to rigid observance of traditional
Women in the Battered group reported sex roles, demands for subservience, and
physical aggression scores by their male isolation from resources. In contrast, emo-
partner of 37.5 (and severe physical aggres- tional/verbal abuse includes withholding
sion scores of 13.4), again beyond the 99th emotional resources, verbal attacks, and be-
percentile for population norms. havior that degrades women. Factor analy-
An assessment was made of intermit- ses support the inclusion of the two factors.
tency and predictability of abuse. Evalua- In the present sample, the Cronbach’s alpha
tions of both groups were made at two for the dominance/isolation subscale was
points-just after leaving the abusive rela- .82 and for the emotionalherbal subscale
tionship (Time 1) and six months later .93.
(Time 2+to ascertain whether associa- Intermittency of abuse. The measure of
tions between dependent measures had per- intermittency was designed to assess the
sisted over time. The subjects completed a juxtaposition of extreme positive and nega-
test battery, described below, of indepen- tive behavior. Respondents were asked to
dent and dependent measures. Additionally, describe the first, last, and worst incident
structured interviews were conducted at Time of abuse in detail (for emotionally abused
1 and a second set of follow-up dependent women these were incidents of conflict and
measures was collected at Time 2. All sub- emotional abuse). For each incident, a var-
jects were paid for participation. All inter- iety of behavior that included verbal and
DUTTON AND PAINTER 617

physical abuse items and threats was listed. Socially desirable responding. The Mar-
Postabuse behavior was also assessed, in- lowe-Crowne Social Desirability Scale (MC)
cluding negative behavior (threats, etc.) (Crowne & Marlowe, 1960) is a self-report
and positive “contrition” behavior (Walker, measure that contains items about every-
1978). A measure of intermittency was cre- day events that are desirable but rare. Paul-
ated by having the respondent rate (on a hus (1991) reported that it measures a sub-
scale of -5=very negative, to +5=very pos- ject’s tendency to respond in a deceptive
itive) the extremity of her partner’s behav- manner. Participants are required to check
ior before, during, and after each incident of whether each item is true or false.
abuse. The Intermittency scale was the sum
of the three positive scores (one for each of Dependent Measures
the first, worst, and last incidents) minus Dependent measures for this study were
the three negative scores. Hence, the scale collected at Time 1 and again at Time 2, six
had a theoretical range of - 15 to + 15. months later.
Predictability of abuse. Respondents were Attachment. To assess attachment, a scale
asked four questions pertaining to whether of attachment developed by Kitson (1982)
they could predict abusive outbursts from was used and was supplemented with some
their partner. As part of their description of items from a scale by NiCarthy (1982). The
the first, worst, and last incidents of abuse, Kitson scale, devised to assess attachment
they were asked if they could tell when during divorce, measures the bereavement
their partner was going to become abusive. aspect of separation and contains items
In addition, they were asked if their partner such as “I frequently find myself wonder-
went through predictable and abrupt shifts ing what he is doing” and “I spend a lot of
in mood and became suddenly angry. These time still thinking about him.” Kitson (1982)
questions were combined to form an eight- reported the psychometric qualities of the
point scale of Predictability. scale, including an alpha of .80. Supple-
Power. Two measures of the respondent’s menting the assessment of attachment were
rating of her own and her partner’s power ten items from an “idealization” measure
were taken. First, the Decision Power Index developed by NiCarthy; these included
(Blood & Wove, 1960), which assesses items such as “no one could ever under-
who has the final say on six issues (buying stand him the way I do,” “without him I
a car, having children, what apartment to have nothing to live for,” and “I love him
take, what job either partner should take, so much, I can’t think of being with anyone
whether a partner should work or not, and else.” The NiCarthy scale added an ele-
how much money to spend each week on ment of continuing obsession with the part-
food), was used. Second, a subjective mea- ner that was not included in the Kitson scale.
sure of power was used, called power dif- Since the composite scale was new, an item-
ferential, which simply asked the respon- whole correlation for each item was per-
dent to indicate on a ten-point scale how formed and only those items that had corre-
much power both she and her partner had lations over .55 were retained. Cronbach’s
I) before the violence/abuse started, 2) af- alpha for the entire 20-item scale was .92.
ter the violence/abuse started but before Self-esteem. Since self-esteem is frequent-
she left, and 3) now that she had left. The ly mentioned in the literature on the effects
definition of power on this question was of battering, it was assessed, using the
deliberately left unspecified. Finally, a var- Rosenberg (1965) Self-esteem Scale. This
iable called power shift was calculated, which 10-item self-report scale has reported al-
assessed the increase in male power differ- phas of .77 and .88 (Robinson, Shaver, &
ential (with the woman) before and after bat- Wrightsman, 1991). Responses range from
tering/abuse. “strongly disagree” to “strongly agree” on a
618 BATTERED WOMAN SYNDROME

four-point scale; the higher the score, the based on emotional and verbal abuse. For
greater the self-esteem. this group, the Intermittency score was
Trauma symptoms. The Trauma Symp- 15.8 ( S B 5 . 8 ) and the Predictability score
tom Checklist (TSC-33) (Briere & Runtz, was 5.1 out of a possible 8 (SD=1.2). So-
1989) is a brief (33-item), reliable instru- cial desirability measures (the Marlowe-
ment showing predictive and construct va- Crowne Scale) did not correlate signifi-
lidity. It has been shown to discriminate fe- cantly with reports of partners' physical or
male victims of childhood sexual abuse from emotional abuse, leading to the conclusion
nonvictimized women. The TSC-33 contains that these reports were uncontaminated by
five subscales: dissociation, anxiety, depres- the respondents' motive to "image man-
sion, hypothesized post-sexual abuse trau- age" by suppressing or inflating reported
ma (PSAT), and sleep disturbance. The abuse rates.
PSAT-hypothesized includes those symp-
toms thought to be most characteristic of Intercorrelations of Dependent Measures
sexual abuse experiences but that may also TABLE1 shows intercorrelations of the
occur as a result of other types of trauma. dependent variables at Time 1 and Time 2,
Analysis of the internal consistency of the broken down separately for the Battered
five subscales indicated reasonable relia- and EA groups. For the Battered group,
bility, with an average subscale alpha of significant correlations were found between
.71 and a total alpha for the TSC-33 of .89 all pairs of the three dependent measure
(Briere & Runtz, 1989). scales at both Time 1 and Time 2.
For the EA group, TSC-33 scores corre-
RESULTS lated significantly with attachment (+.44,
Violence Profiles p<.Ol) and self-esteem (-.40,p<.Ol). How-
Respondents reported CTS annual physi- ever, unlike the Battered group, EA women
cal aggression scores by their former part- showed no significant correlation between
ner as follows: Battered group M=37.5 attachment and self-esteem (-.2 1, NS). This
(SB17.6), EA group M=1.2 (SB2.0), in- pattern was repeated at Time 2. For bat-
dicating frequent physical abuse for the tered women, the three sequelae of batter-
Battered group. On the Tolman PMWI, the ing measured in this study were signifi-
Battered group's scores were as follows: cantly intercorrelated. This intercorrelation
domination/isolation=79.1 (SB25.9) and exhibits durability up to six months.
emotional abuse=95.5 (SD=15.9), indicat- TABLE 2 shows the intercorrelations of
ing that frequent emotional abuse accom- Time 1 with Time 2 measures taken six
panied physical abuse for these women. months later for the entire sample. Despite
Corresponding scores for the EA group the lengthy interval between Time 1 and
were: domination/isolation=43.1 (SB27.5) Time 2, each dependent measure taken at
and emotional abuse=69.4 (SB20.1). The
Battered group indicated a mean Intermit- Table 1
tency score for their partners of 19.0 ( S B INTERCORRELATIONSWITHIN TIME PERIOD
6.7) on the 30-point Intermittency scale as- TIME1 TIME 2
sessing alternating good-bad treatment and GROUP TRMA ATTACH TRMA ATTACH
a mean Predictability score of 5.7 out of a Battered
possible 8 ( S B 1 . 5 ) .Interestingly, the EA Attachment +.39*** + .55'*'
Self-esteem +.22' -.28 -.67"' -.48"
group also reported intermittency of emo- Emotionally
tional abuse, with similar crests (first, Abused
Attachment +.44** +.41"
worst, and last incident) and troughs. The Self-esteem -.40" -.21 -.56" -.I7
subjective ratings of how good or bad treat- Note: TRMA=Trauma. AlTACH=Attachment.
ment was during abusive episodes was ~ x . 0 5 **DC.OI. . ***~.001.
DUTTON AND PAINTER 619

1,4 1% of attachment scores were account-

Table 2
INTERCORRELATIONS OF TIME 1 ed for by a composite variable composed of
AND TIME 2 MEASURES power shift (8=-.36), dominance/isolation
TIME 2
(/3=-.33), and length of relationship (B=-
TRAUMA ATTACH- SELF-
.18).
TIME 1 SYMPTOMS MENT ESTEEM Self-esteem scores at Time 1 were 29% ac-
Trauma counted for by relationship variables: length
symptoms +.40**' +.27" -.36" of relationship (8=.33), power differential
Attachment +.34" +.68"' -.17
Self-esteem -.16 -.07 +.27' (8=.24), physical abuse by the partner (B=-
.58), and intermittency of abuse (p=.60).
Relationship variables did best at ac-
Time 1 correlated significantly with its coun- counting for attachment at Time 2 (55% of
terpart measure taken at Time 2. variance). This was a composite variable
composed of dominance/isolation (8=.23),
Relationship Variables power differential (6=-.2 l), and intermit-
and Battered Woman Syndrome tency @=.3 I). Trauma symptoms at Time
TABLE3 shows Pearson correlations be- 2 had 47% of their variance accounted for
tween predictor variables from the prior re- by relationship variables, suggesting a de-
lationship and battered woman syndrome layed effect of relationship trauma on symp-
measures. The strongest associations be- tom onset. Both dominance/isolation ( B
tween individual predictor and dependent =.47) and total physical abuse (8=.21) were
variables in this study were intermittency instrumental in this regression.
correlated with attachment to the abuser at Finally, a discriminant function analysis
Time 1 (+.62) and Time 2 (+.60). Both were was run on composite distress by assigning
significant (p < .001). women to high and low battered woman syn-
To estimate better the overall effect of drome groups on the basis of self-esteem,
relationship variables on postrelationship experienced trauma symptoms, and contin-
measures, composite measures were con- uing attachment at Time 2. Using all avail-
structed and entered into a multiple re- able predictor variables, a five-variable
gression on the various dependent mea- composite explained 81% of the variance
sures of the study. Relationship variables in composite distress and correctly classified
included intermittency, power shift (change 9 1.3% of the women according to distress
in victim-partner power differential pre-post group assignment. The main contributors to
battering), total physical abuse, dominance, this composite variable (with beta weights)
emotional abuse, and length of relationship. were dominance/isolation (8=.8 l), inter-
In this analysis, relationship variables ac- mittency (/3=.77), total physical abuse by
counted for a substantial amount of the partner (B=.59), emotional abuse (8=.49),
postrelationship variables' variance. At Time and power shift (/3=.45).

Table 3
SIGNIFICANT CORRELATIONS OF PREDICTOR TO DEPENDENT MEASURES

AlTACHMENT TRAUMA SYMPTOMS SELF-ESTEEM

MEASURE TIME 1 TIME2 TIME 1 TIME2 TIME 1 TIME2
lntermittency of abuse +.62"* +.60"'
Power differential +.27" +.31*'
Physical violence (CTS) -.58"'
Dominancelisolation -.33' +.44" -.27'
Emotional abuse +.29'

p . 0 5 , **p~.01,
--*p<.ooi.
620 BATTERED WOMAN SYNDROME

DISCUSSION (91.3% correct classification into distress

Both hypotheses of the current study categories) lends support to this link. In ef-
were confirmed. Battered women experi- fect, the totality of symptom scores com-
enced three aspects of the battered woman prising the battered woman syndrome is af-
syndrome: high rates of trauma symptoms, fected by the severity of physical abuse, the
lowered self-esteem, and heightened “para- intermittency of abuse-good treatment, dyad-
doxical attachment” to the batterer. These ic power losses to the woman, and emo-
effects were all significantly intercorrelat- tional abuse. It is suggested that each of
ed, forming a syndrome, or complex, that these contributors be thoroughly assessed
persisted for at least six months. Further- in court cases involving battered women.
more, these effects were significantly re-
lated to the intermittency of positive-nega- Intermittency vs. Cycle of Violence
tive treatment, to power differentials in the Criticisms of the battered woman syn-
former relationship, and to the extremity of drome (Schuller & Vidmar, 1992) have fo-
the battering. Examination of the symptom cused on the “cycle of violence” aspect of
scores for the Emotionally Abused group battering described by Walker (1979, 1984),
revealed that patterns at Time 1 changed whereby abusive males go through three
somewhat when physical abuse occurred distinct phases: tension build-up, acute
only once. For this group, the significant battering, and contrition or loving respite.
negative correlation of attachment to self- The present data suggest that intermittency
esteem did not occur, but other intercorre- of abuse, not a battering cycle per se, is a
lations remained significant(as with the Bat- major determinant of postseparation dis-
tered group). This different pattern for the tress and the battered woman syndrome.
Emotionally Abused women was repeated Whereas a “cycle of violence” perspective
at Time 2. describes violence as going through pre-
The trauma symptoms experienced by dictable (and mood-driven) cycles, the term
women in the current sample included height- “intermittency,”as defined in this research,
ened anxiety, dissociation, depression, and simply indicates that extreme positive be-
sleep disturbance. Almost half the variance havior and extreme negative behavior oc-
in these symptoms was attributable to rela- cur with temporal contiguity; they need not
tionship variables (that is, the severity and be cyclical or predictable. While women in
intermittency of psychological abuse, dom- our sample did find their partner’s behavior
ination, and battering in the relationship predictable (generating mean scores on pre-
just ended) as much as six months after rela- dictability of 5.7 out of a possible 8), pre-
tionship dissolution. Composite distress, or dictability and intermittency were not sig-
battered woman syndrome, scores (composed nificantly correlated (Pearson’s r=. 1 1, NS).
of persistent attachment to the abuser, trau- Indeed, predictability did not correlate sig-
ma symptoms, and low self-esteem) for these nificantly with any extremity measures of
women were largely accounted for by domin- negative or positive behavior, and it was
anceholation, intermittency of abuse, total intermittency, not predictability, that fore-
physical abuse, emotional abuse, and pow- cast postrelationship distress. It appears
er shift (losses in power to the woman and that it is the extremity and juxtaposition of
increases in power to the man following positive and negative behavior that con-
abuse). Hence, a rather direct link is estab- tribute to battered woman syndrome, rather
lished between these abuse factors and post- than the predictability of abuse per se.
relationship distress as assessed by a com- Intermittency can be cyclical, as in the case
posite battered woman syndrome score. The of abusive men with borderline personality
power of these relationship variables in cor- organization (Dutton, in press; Dutton &
rectly predicting postrelationship distress Starzomski, in press), but it can influence at-
DUTTON AND PAINTER 62 1

tachment, trauma, and self-esteem without little is known about the experiences shap-
cyclicity. ing these apparently opposite emotional re-
Dutton and Painter (1981) have de- sponses. Herman (1992) described trauma
scribed the theoretical basis for the de- effects as a dialectic process characterized
velopment of strong and continuing attach- by extreme affective states: “floods of in-
ment in abusive relationships because of tense, overwhelming feeling and arid states
intermittency, and have reviewed research of no feeling at all” (p. 47). With time and
on animal learning in which intermittent healing, the balance and integration of these
good-bad treatment was shown to increase states eventually results in a stabilizing of
attachment to 230% of the level of that affective lability.
from consistent good treatment (Rajecki, The present study also did not examine
Lamb, & Obsmacher, 1978). Such attach- learned helplessness (Walker, 1979), an ac-
ment appears to constitute a cornerstone of quired motivational deficit thought to un-
the battered woman syndrome, and may dercut attempts to leave the batterer. How-
contribute to the apparently precipitous de- ever, since our sample was contacted after
cision some battered women make to return leaving, they may not, ips0 facto, have
to their partner. demonstrated this effect as strongly as
lntermittency has a stronger effect on at- would battered women still in an abusive
tachment at Time 2 than at Time 1, con- relationship. It is sampling problems such
firming another prediction of traumatic as these that continue to make the empiri-
bonding theory as originally proposed by cal study of battering sequelae problem-
Dutton and Painter (1981). The metaphor atic. However, for the variables studied,
they proposed for traumatic bonding was self-esteem, attachment, and trauma symp-
that of an abuse victim attached to the per- toms demonstrate a similarity in how they
petrator by an elastic band that begins to “behave” in response to features of prior
stretch with her initial leaving. With time abuse such as intermittency and power dif-
away from the relationship, the pressure on ferentials. These features, along with the
the stretching band continues to build until extremity of physical and emotional abuse,
she returns, apparently impulsively, to the should constitute part of the assessment of
abuser. The delayed effect of dynamic rela- battering dynamics.
tionship variables on postrelationship at-
tachment and the experience of trauma
REFERENCES
symptoms is a noteworthy phenomenon for Blood, R. Jr., & Wolfe, D. (1960). Husbands and
professionals working with battered women. wives: The dynamics of married living. Glencoe,
Dutton and Painter (1981) also suggested IL: Free Press.
Briere, J., & Runtz, M. (1989). The trauma symptom
that microsystem (dyad) power imbalances checklist (TSC-33): Early data on a new scale,
contribute to low self-esteem and attach- Journal of Interpersonal Violence, 4, 151-163.
ment in the person with less power. The Crowne, D.P., & Mulowe, D.A. (1960). A new scale
of social desirability independent of psycho-
contributions Of the power to pathology. Journal of Consulting Psychoiogy, 24,
attachment at Time 1 and the power differ- 349-354.
ential measure to attachment at Time 2 are Douglas, M A . (1987). The battered woman syn-
drome. In D.J. Sonkin (Ed.), Domestic violence on
consistent with this suggestion. trial: Psychological and legal dimensions o f fam-
ily violence ( p i 39-54). New York: Springer:
Implicationsfor Research Dutton, D.G. (in press). Behavioral and affective cor-
relates of borderline personality organization in
C1early, more empirical to be wife assaulters; International Journal of Law and
done on the battered woman syndrome. Psychiatry.
The present study did not, for example, at- Dutton, D.G., & Painter, S.L. (1981). Traumatic
bonding: The development of emotional attach-
tempt to assess affective reactions such as ments in battered and other relationships
psychic numbing or hyperreactivity, and of intermittent abuse. Victimology, 6, 139-155.
 BATTERED WOMAN SYNDROME

Dutton, D.G., & Starzomski, A. (in press). Psy- Rosenberg, M. (1965). Society and the adolescent
chological differences in court-referred and self- sel/'-image. Princeton, NJ: Princeton University
referred wife assaulters. Criminal Justice and Be- Press.
havior. Schuller, R.A., & Vidmar, N. (1992). Battered wom-
Fischer, A.E. (1955). The effects of dflerential early an syndrome evidence in the courtroom. Law and
treatment on the social and exploratory behavior Human Behavior, 16, 273-291.
ofpuppies. Unpublished doctoral dissertation, De- Scott, J.P. (1963). The process of primary socializa-
partment of Psychology, Pennsylvania State tion in canine and human infants. Monographs of
University. the Society for Research in Human Development,
Freud, A. (1942). The ego and the mechanisms of de- 28, 1-34.
fense. Madison, a: International Universities Press. Straus, M.A. (1979). Measuring family conflict and
Harlow, H., & Harlow. M. (1971). Psychopathology violence: The Conflict Tactics Scale. Journal of
in monkeys. In H.D. Kimmel (Ed.), Experimental Marriage and the Family, 41, 75-88.
psychopathology (pp. 112-136). New York: Aca- Straus, M.A., Gelles, R.J., & Steinmetz, S. (1980). Be-
demic Press. hind closed doors: Violence in the American family.
Herman, J. (1992). Trauma and recovery: The aJer- Garden City, NY:Anchor Presfiubleday.
math of violence--from domestic abuse to politi- Thyfault, R.K., Browne, A., & Walker, L.E. (1987).
cal terror. New York: Basic Books. When battered women kill: Evaluation and expert
Kitson, G.C. (1982). Attachment to the spouse in di- witness testimony techniques. In D.J. Sonkin
vorce: A scale and its application. Journal of Mar- (Ed.), Domestic violence on trial: Psychological
riage and the Family, 44, 379-393. and legal dimensions of family violence (pp.
NiCarthy, G. (1982). Getting free: A handbook for 71-85). New York: Springer.
women in abusive relationships. Seattle: Seal Tolman, R.M. (1989). The development of a mea-
Press. sure of psychological maltreatment of women
Paulhus, D.L. (1991). Measurement and control of re- by their male partners. Violence and Victims, 4,
sponse bias. In J.P. Robinson, P.R. Shaver, & L.S. 159-1 77.
Wrightsman (Eds.), Measures ofpersonality and van der Kolk, B. (1987). Psychological trauma.
social psychological attitudes (pp. 93-1 04). San Washington, DC: American Psychiatric Press.
Diego: Academic Press. Walker, L. (1978, August). Psychotherapy and coun-
Rajecki, P.M., Lamb, M., & Obsmacher, P. (1978). seling with battered women. Paper presented at the
Toward a general theory of infantile attachment: A annual meeting of the American Psychological
comparative review of aspects of the social bond. Association, Washington, DC.
Brain and Behavioral Sciences, 3, 417-464. Walker, L. E. (1979). The battered woman. New
Robinson, J.P., Shaver, P.R., & Wrightsman, L.S. York: Harper & Row.
(1991). Measures of personality and social psy- Walker., L.E. (1984).
\ , The battered woman svndrome.
chological attirudes. San Diego: Academic Press. New York: Springer.

For rcprints: Donald G. Dutton, Ph.D., The University of British Columbia, Department of Psychology, 2509-2136 W. Mall,
University Campus, Vancouver, BC V6T 124, Canada