Patriarchal Terrorism and Common Couple Violence: Two Forms of Violence against

Women
Author(s): Michael P. Johnson
Source: Journal of Marriage and Family , May, 1995, Vol. 57, No. 2 (May, 1995), pp. 283-
294
Published by: National Council on Family Relations

Stable URL: https://www.jstor.org/stable/353683

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

National Council on Family Relations is collaborating with JSTOR to digitize, preserve and
extend access to Journal of Marriage and Family

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
 MICHAEL P. JOHNSON The Pennsylvania State University

Patriarchal Terrorism and Common Couple Violence:

Two Forms of Violence Against Women

This article argues that there are two distinct terrorized them for years. In addition, the authors
forms of couple violence taking place within fami- of the 1985 National Family Violence Survey es-
lies in the United States and other Western coun- timate that over six million women are assaulted
by their husbands each year in the United States.
tries. A review of evidence from large-sample sur-
vey research and from qualitative and quantita- But are these really the same phenomenon?
tive data gathered from women's shelters This article argues that there are, in fact, two
suggests that some families suffer from occasion-distinct forms of couple violence taking place in
al outbursts of violence from either husbands orAmerican households. Evidence from large-sam-
wives (common couple violence), while other ple survey research and from data gathered from
families are terrorized by systematic male vio-women's shelters and other public agencies sug-
lence (patriarchal terrorism). It is argued that the gests that a large number of families suffer from
distinction between common couple violence andoccasional outbursts of violence from either hus-
patriarchal terrorism is important because it hasbands or wives or both, while a significant num-
implications for the implementation of public pol-ber of other families are terrorized by systematic
icy, the development of educational programs andmale violence enacted in the service of patriarchal
intervention strategies, and the development ofcontrol.
theories of interpersonal violence.
SOCIOLOGICAL PERSPECTIVES ON VIOLENCE
AGAINST WOMEN IN THE FAMILY
You must go through a play of ebb and flow
and watch such things as make you sick at heart.
There are two major streams of sociological work
Nguyen Du (1983)
on couple violence in families, one that is gener-
We are all too familiar with stories of women ally referred to as the family violence perspective,
who are finally murdered by husbands who have and the other of which may be called the feminist
perspective (Kurz, 1989).
Work in the family violence perspective grew
out of family scholars' interest in a variety of fam-
Department of Sociology, The Pennsylvania State University,
206 Oswald Tower, University Park, PA 16802. ily conflict issues, and is generally traced to the
early work of Straus (1971) and Gelles (1974).
Key Words: domestic violence, feminism, gender, violence,
They came together in the early 1970s to develop
wife beating.

Journal of Marriage and the Family 57 (May 1995): 283-294 283

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
 284 Journal of Marriage and the Family

a research agenda based on the use of interviews VIOLENCE AGAINST WOMEN IN U.S. FAMILIES:
to elicit information regarding family violence PATRIARCHAL TERRORISM AND
from large random samples of the adult population COMMON COUPLE VIOLENCE
of the United States, conducting national surveys
in 1975 and 1985. Methodologically, work in this
The findings of the two literatures discussed
above lead to strikingly different conclusions re-
tradition has relied primarily on quantitative anal-
ysis of responses to survey questions, utilizing the garding a number of the central features of family
strengths of random sample surveys in the produc-
violence for which they both provide information
tion of estimates of prevalence, and causal analy- (gender symmetry/asymmetry, per-couple fre-
ses that rely on multivariate statistical techniques.
quency of violence, escalation of violence, and
Theoretically, the focus has been largely on com- reciprocity of violence). While these findings
monalities among the various forms of family vio-
suggest to each group of scholars that the other
lence, such as the surprising frequency of vio- misunderstands the nature of such violence, they
lence, the instigating role of stress, and public ad- suggest to me that these groups are in fact study-
herence to norms accepting the use of some ing two distinctly different phenomena.

violence within the family context. The first form of couple violence, which I will
call patriarchal terrorism, has been the focus of
In contrast, research from the feminist per-
the women's movement and of researchers work-
spective began with a narrower focus on the issue
of wife beating, developing a literature that focus- ing in the feminist perspective. Patriarchal terror-
es on factors specific to violence perpetrated ism, a product of patriarchal traditions of men's
right to control "their" women, is a form of terror-
against women by their male partners (Dobash &
Dobash, 1979; Martin, 1981; Roy, 1976; Walker, istic control of wives by their husbands that in-
1984). Methodologically, feminist analyses have volves the systematic use of not only violence,
but economic subordination, threats, isolation,
relied heavily upon data collected from battered
and other control tactics.
women, especially those who have come into
contact with law enforcement agencies, hospitals, There are a number of difficult and important
or shelters. Theoretically, the emphasis has been terminological issues here. The pattern of vio-
lence that I have just described is often referred to
upon historical traditions of the patriarchal fami-
ly, contemporary constructions of masculinity and with terms such as wife beating, wife battery, and
battered women. I have chosen to avoid these
femininity, and structural constraints that make
terms for two reasons. I avoid the restrictive term
escape difficult for women who are systematical-
ly beaten. wife in order to acknowledge recent literatures
I do not wish to give the impression that the that suggest that such a phenomenon may be in-
differences between these two literatures are abso- volved in heterosexual dating relationships (Cate,
lute, although the often-rancorous debates that Henton, Koval, Christopher, & Lloyd, 1982; Stets
have gone on between the two groups of scholars & Pirog-Good, 1987) and perhaps even in some
seem at times to suggest that there is absolutely no
lesbian relationships (Renzetti, 1992). I have cho-
overlap in methodology or theory (e.g., Dobash & sen not to switch to a simple nongendered alterna-
Dobash, 1992, pp. 251-284). The truth is that tive, such as partner, because I am convinced that
family violence researchers do acknowledge the this pattern of violence is rooted in basically pa-
role of patriarchy in wife abuse (Straus, Gelles, & triarchal ideas of male ownership of their female
partners.
Steinmetz, 1980, pp. 242-243), and do make use
of qualitative data obtained from battered wives The terminology of the battered wife is also
(Gelles, 1974). On the other side, many feminist objectionable on the grounds that it shifts the
researchers utilize quantitative data (Yllo & Bo- focus to the victim, seeming to imply that the pat-
grad, 1988) and acknowledge the role of factors tern in question adheres to the woman rather than
other than the patriarchal structure of society in to the man who is in fact behaviorally and moral-
precipitating violence against wives (Martin, ly responsible for the syndrome. The term patri-
1981). As will be seen in the next section, howev- archal terrorism has the advantage of keeping the
er, family violence researchers and feminist re- focus on the perpetrator and of keeping our atten-
tion on the systematic, intentional nature of this
searchers do clearly disagree on some very impor-
tant issues, and a case can be made that their dif- form of violence. Of course, the term also forces
us to attend routinely to the historical and cultural
ferences arise from the fact that they are, to a large
extent, analyzing different phenomena. roots of this form of family violence.

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
Two Forms of Violence Against Women 285

The second form of couple violence, which I of men who were battered by their wives or
will call common couple violence, is less a prod- lovers. Results from the Conflict Tactics Scale
uct of patriarchy, and more a product of the less- (CTS) used in the National Family Violence Sur-
gendered causal processes discussed at length by veys (NFVS)-both the 1975 study upon which
Straus and his colleagues working in the family Steinmetz relied and the 1985 replication-do in-
violence tradition (Straus & Smith, 1990). The dicate almost perfect symmetry in the use of vio-
dynamic is one in which conflict occasionally lence by men and women against their partners.
gets "out of hand," leading usually to "minor" (For a thorough methodological critique of the
forms of violence, and more rarely escalating into CTS, see Dobash & Dobash, 1992, and Dobash et
serious, sometimes even life-threatening, forms of al., 1992. For earlier responses to many of those
violence. criticisms, see Straus, 1990a, 1990b. Although I
am in essential agreement with many of the criti-
Gender Symmetry/Asymmetry
cisms of the CTS, data presented below indicate
that the patterns of violence discovered in shelter
The importance of the distinction between com- samples and national samples differ dramatically
mon couple violence and patriarchal terrorism even is when violence is assessed with the CTS in
most forcefully illustrated in the heated debate both settings. This provides strong evidence that
over the extent to which women are perpetrators the differences are not due merely to the deficien-
of couple violence. One of the surprising findings cies of the CTS.) For any use of violence, the
of Straus and his colleagues' national surveys was 1975 national figures for men and women were
that women were evidently as likely to utilize vio- 12.1% and 11.6%, respectively; in 1985 the com-
lence in response to couple conflict as were men. parable figures were 11.3% and 12.1%. For seri-
One family violence researcher unfortunately ous violence (a subset of the figures for any use of
chose to refer to these women's use of violence violence, including only acts judged to have a
against their partners as "the battered husband high probability of producing serious injury, such
syndrome" (Steinmetz, 1978a), suggesting that as hitting with a fist), the 1975 figures were 3.8%
women's violence against men represented the for men, 4.6% for women; in 1985 the comparable
same sort of phenomenon as the male violence figures were 3.0% and 4.4% (Straus & Gelles,
that was being reported to women's shelters 1990, p. 118). In all cases, the gender differences
across the country. The feminist scholars strongly are less than 2%.
disagreed (Adams, Jackson, & Lauby, 1988; These findings contrast dramatically with
Berk, Loseke, Berk, & Rauma, 1983; Dobash & those from shelter populations, from hospitals,
Dobash, 1992, pp. 251-284; Dobash, Dobash, and from the courts. For example, Gaquin (1978)
Wilson, & Daly, 1992; Fields & Kirchner, 1978; reported that National Crime Survey data (United
Pleck, Pleck, Grossman, & Bart, 1978; Wardell, States) for the period 1973-75 indicate that 97%
Gillespie, & Leffler, 1983). Unfortunately this de- of assaults on adults in the family were assaults
bate has been structured as an argument about theon wives. Analyses of police files in the U.S. and
nature of family violence, with both sets of schol-Britain show similar patterns (Dobash & Dobash,
ars overlooking the possibility that there may be 1992, p. 265; Martin, 1981, pp. 13-14). Kincaid's
two distinct forms of partner violence, one rela-(1982, p. 91) analysis of family court files in On-
tively gender balanced (and tapped by the surveytario, Canada, found 17 times as many female as
research methodology of the family violence tra- male victims, and Levinger's (1966) study of di-
dition), the other involving men's terroristic at-vorce actions in Cleveland, Ohio, found 12 times
tacks on their female partners (and tapped by themore wives than husbands mentioning physical
research with shelter populations and criminalabuse (37% vs. 3%). Fields and Kirchner (1978,
justice and divorce court data that dominates thep. 218) reported that Crisis Centers in the New
work in the feminist tradition). York City public hospitals counseled 490 battered
The Steinmetz (1978) article that introducedwives and only two battered husbands during the
the term battered husband to the literature relied last half of 1977.
primarily on data from large-scale survey research The most likely explanation for these dramatic
to make a case for the position that women are just differences in the gender patterns of violence in the
as violent as men in intimate relationships, and national surveys and in statistics collected by pub-
that there was therefore a need for the develop- lic agencies is not that one or the other methodolo-
ment of public policy that would address the needs gy misrepresents the "true" nature of family vio-

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
 286 Journal of Marriage and the Family

lence, but that the two information sources deal of escalation. The evidence from the NFVS sug-
with nearly nonoverlapping phenomena. The com- gests that so-called minor violence against
mon couple violence that is assessed by the large- women does not escalate into more serious forms
scale random survey methodology is in fact gender of violence. Feld and Straus (1990) reported data
balanced, and is a product of a violence-prone cul- relevant to this question based on a 1-year follow-
ture and the privatized setting of most U.S. house- up survey of 420 respondents from the 1985
holds. The patriarchal terrorism that is tapped in re- NFVS. My own reanalysis of their published data
search with the families encountered by public shows almost no tendency to escalation. For ex-
agencies is a pattern perpetrated almost exclusively ample, among husbands who had perpetrated no
by men, and rooted deeply in the patriarchal tradi- acts of minor or severe violence in Year 1 (the
tions of the Western family. year prior to the 1985 interview), 2.6% had
moved to severe violence in Year 2. Among those
who had committed at least one act of only minor
Per Couple Frequency
violence, only 5.8% had moved to severe vio-
With regard to the frequency of couple violence lence; among those who had used severe violence
in "violent" families, we are fortunate to have in Year 1, only 30.4% had been that violent in
data using the same data collection instrument Year 2. Thus, these data indicate that not only is
(the CTS) with survey samples and shelter sam- there virtually no tendency to escalation (fully
ples. According to Straus (1990b), among women 94% of perpetrators of minor violence do not go
who report to NFVS researchers that they have on to severe violence), but that in most (70%) of
been assaulted by their husbands in the previous the cases of severe violence there is, in fact, a de-
year, the average number of such assaults per escalation. Data on frequency show much the
woman was six (n = 622); for those in the sample same pattern.
who had used the services of a shelter, the aver- A very different pattern is observed in re-
age was 15.3 (n = 13). search with shelter populations. According to
In dramatic contrast, Straus cited studies of Pagelow (1981), "one of the few things about
shelter populations in Maine (Giles-Sims, 1983) which almost all researchers agree is that the bat-
and Michigan (Okun, 1986), utilizing the same terings escalate in frequency and intensity over
series of survey questions, that find an average time" (p. 45).
annual number of incidents per woman in the 65 Why does patriarchal terrorism escalate while
to 68 range! Although Straus argued that the common couple violence does not? Common
NFVS probably "underrepresents" certain types couple violence is an intermittent response to the
of violence against women (among the 622 as- occasional conflicts of everyday life, motivated
saulted women in the sample, only four had been by a need to control in the specific situation (Mi-
assaulted as many as 65 times), he evidently con- lardo & Klein, 1992), but not a more general need
tinued to think of this as just another point on a to be in charge of the relationship. In contrast, the
continuum of violence, referring to the missed causal dynamic of patriarchal terrorism is rooted
cases as "cases of extreme violence" (Straus, in patriarchal traditions, adopted with a
1990b, p. 85). Although Straus recognized and vengeance by men who feel that they must con-
discussed the possibilities raised by this "under- trol "their" women by any means necessary. As
representation" for resolving differences between one husband responded to his wife's protests re-
the conclusions of shelter research and survey re- garding a violent episode during their honey-
search, and even referred to the possibility of a moon, "I married you so I own you" (Dobash &
"qualitatively different experience," he does not Dobash, 1979, p. 94). Escalation in such cases
seem to have taken the next step, to suggest that may be prompted by either of two dynamics.
perhaps we are dealing with decidedly different First, if his partner resists his control, he may es-
phenomena and should adopt a terminology that calate the level of violence until she is subdued.
would mitigate against the mistaken assumption Second, even if she submits, he may be motivated
that common couple violence is merely less se- not only by a need to control, but by a need to
vere or less frequent than patriarchal terrorism. display that control, yielding a pattern observed
by Dobash and Dobash (1979, p. 137), in which
Escalation no amount of compliance can assure a wife that
she will not be beaten:
The two literatures also appear to uncover dra-
For a woman simply to live her daily life she is
matically different patterns of behavior in termsalways in a position in which almost anything

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
Two Forms of Violence Against Women 287

she does may be deemed a violation of her wife- occur on average more than once a week, and es-
ly duties or a challenge to her husband's authori- calate in seriousness over time. The violence is
ty and thus defined as the cause of the violence almost exclusively initiated by the husband, most
she continues to experience. (p. 137)
wives never attempt to fight back, and, among
those who do, about one-third quickly desist,
Reciprocity and Initiation of Violence leaving only a small minority of cases in which
the women respond even with self-defensive vio-
On the issue of reciprocity, the NFVS analysts re-
lence. These patterns have led researchers in the
port a pattern in which two-thirds of the families feminist tradition to conclude that violence
in which the husband has been violent also in-
against women in the family has its roots in the
volve a violent wife, and in which "women initi-
patriarchal structure of the U.S. family. The cen-
ate violence about as often as men" (Stets &
tral motivating factor behind the violence is a
Straus, 1990, p. 161).
man's desire to exercise general control over
Research with shelter populations provides "his" woman.
quite a different picture. Pagelow, for example, re-
It is important not to make the mistake of as-
ported that only 26% of her respondents say they
suming that this pattern of general control can be
fight back; another 16% indicate that they had
indexed simply by high rates of violence. Al-
once tried, but stopped when it made things worse
though the average frequency of violence among
(Pagelow, 1981, p. 66). She also suggested, al-
cases of patriarchal terrorism may be high, there
though she is not entirely clear (Pagelow, 1981,
may well be cases in which the perpetrator does
pp. 65-66), that none of her respondents had initi-
not need to use violence often in order to terrorize
ated the violence in the incidents on which they
reported. Giles-Sims's (1983, pp. 49-50) data for his partner. Feminist theorists and shelter activists
a shelter population show dramatic lack of re- argue that since patriarchal terrorism has its roots
ciprocity in the use of violence, as reported in re- in a motive to exercise general control over one's
sponse to the CTS. The five most severe forms of partner, it is characterized by the use of multiple
violence were roughly twice as likely to have been control tactics (Dobash & Dobash, 1979). The
used by the men as the women, and in some cases Duluth Domestic Abuse Intervention Project
the differences are even more dramatic (e.g., 84% (Pence & Paymar, 1993) has developed a useful
of the men had beat up their spouse, as compared graphic representation of this pattern that captures
with 13% of the women), and this in spite of the the importance of not becoming overly focused
fact that "the men had almost all abused the on the violent control tactics that are only part of
women seriously enough to cause injury. In many an overall pattern (see Figure 1). The patriarchal
cases the beatings had been life threatening" terrorist will use any combination of these tactics
(Giles-Sims, 1983, p. 50). The feminist scholars that will successfully (a) control his partner and
(b) satisfy
also point out that when women murder they are 7 his need to display that control.
times more likely than men to have acted in self-Researchers in the family violence perspective
defense (Martin, 1981, p. 14). We may sum up describe
the a dramatically different pattern of vio-
feminist research with testimony to the United lence, one in which the complexities of family
States Commission on Civil Rights to the effect life produce conflicts that occasionally get "out of
that "most women who have been violent towards hand" in some families, incidents occurring in
their husbands have done so only as a last resort, those families an average of once every 2 months.
in self-defense against longstanding terror and The violence is no more likely to be enacted by
abuse from their husbands" (United States Com- men than by women, and violent incidents are ini-
mission on Civil Rights, 1978, pp. 450-453, cited tiated as often by women as by men. In this com-
in Dobash & Dobash, 1992, p. 257). mon couple violence, there appears to be little
likelihood of escalation of the level of violence
Patriarchal Terrorism and over time. I would argue that this type of violence
Common Couple Violence is usually not part of a pattern in which one part-
ner is trying to exert general control over his or
The interpersonal dynamic of violence againsther partner. Although it is possible that a relative-
women uncovered by the researchers working in ly infrequent, nonescalating use of violence is in
the feminist tradition is one in which men system-some cases part of a generally successful use of
atically terrorize their wives, thus the term patri-
other control tactics (the "success" precluding the
archal terrorism. In these families the beatingsneed to use frequent or extreme violence), I will

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
 28 Journal of Marriage and the Family

FIGURE 1. POWER AND CONTROL WHEEL

Note: Reprinted from Pence and Paymar (1993).

argue next that it is more likely that the national validity of two radically different descriptions of
surveys that uncover this pattern reach only popu- the nature of couple violence in the United States.
lations in which violence is a relatively isolated The feminists have argued that the description of
reaction to conflict (common couple violence), violence against women that is derived from fam-
while studies using data from shelters and other ily violence research is seriously flawed and sim-
public agencies reach primarily victims of vio- ply cannot be reconciled with the results of femi-
lent, but multifaceted, strategies of control (patri- nist research.
archal terrorism). I disagree, arguing that such apparent incon-
sistencies would be expected if the two literatures
SURVEY SAMPLES AND SHELTER SAMPLES are dealing with different phenomena. I propose
that the dramatic differences in the patterns of vi-
The debate that has arisen between the feminist olence described by these two research traditions
researchers and the family violence researchersarise because the sampling decisions of the two
continues to be framed as a contention over the traditions have given them access to different,

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
Two Forms of Violence Against Women 289

largely nonoverlapping populations, experiencing mittent, an unusual response to family conflict, do

different forms of violence. not need or want such services. The woman or
man who is struck or pushed by his or her partner
a few times a year will not in most cases report
The Sampling Biases of Surveys and Shelters
the incident to the police, or go to a shelter, or file
One of Straus's (1990b) responses to feminist cri- for divorce or need to seek medical treatment.
tiques of the NFVS was focused around what he Such sources of data are therefore heavily biased
called "the clinical fallacy," which arises "be- in the direction of providing access only to cases
cause women whose partners stopped assaulting of patriarchal terrorism, and, even among those
them are unlikely to seek help from a shelter" (p. cases, biased in the direction of the most egre-
86). The next, perfectly reasonable step in his line gious cases.
of thinking was to call into question generaliza- The biases of shelter samples, although hard to
tion from shelter populations to "the general pop- document, are perhaps obvious. The biases of
ulation." The problem is that he did not at this random sample surveys, however, may require a
point acknowledge that survey research also miss- bit of documentation.
es a significant segment of the general population.
Instead, his discussion of the "representative sam- Do the Survey Numbers Seem to
ple fallacy" fell into the trap of assuming not only Include Patriarchal Terrorism?
that the shelter samples are not representative, but
that a random sample is. He assumed that random Straus and his colleagues (Gelles & Straus, 1988;
sample surveys provide information regarding Straus & Gelles, 1990) have continued to make
"the characteristics and experiences of the total use of terminology that implies that their survey
population who manifest a certain problem" data include the phenomenon that is encountered
(Straus, 1990b, p. 86). I will argue that they do in battered women's shelters, referring to severe
not. violence against wives as "wife beating." Howev-
The sampling bias in survey research comes in er, there are data available from both the NFVS
large part from the fact that even the best de- and from women's shelters regarding the number
signed survey projects are unable to gather infor- of women who have experienced patriarchal ter-
mation from the total target sample, and nonre- rorism, data that produce estimates that are so di-
spondents may differ in important ways from re- vergent that we must conclude that the NFVS
spondents. For example, men who systematically simply does not give access to this phenomenon.
terrorize their wives would hardly be likely to
agree to participate in such a survey, and the Violence data. Straus (1990b) reported that, in the
women whom they beat would probably be terri- 1985 NFVS, four women reported a frequency of
fied at the possibility that their husband might assaults equal to or above the average for shelter
find out that they had answered such questions. populations. Since this is the number of women
Support for the argument that such families are above the average for shelter populations, if we
not represented in the survey data may be found assume a symmetrical distribution of frequency of
in the fact that among the 182 victims of so-called violence for shelter populations, the total popula-
"wife beating" in the 1985 survey research sam- tion should be projected from double that figure,
ple, only four had been assaulted 65 times or or eight. The projection to the total U.S. popula-
more (the average for shelter populations). Intion yields an estimate of about 80,000 women
contrast, if Straus and his colleagues are correct, whose beatings fall into a frequency range com-
and occasional family violence is normative in parable to that of shelter populations, and who
the sense of being expected and tolerated, if not might therefore seriously consider the possibility
accepted, then many, if not most, families in- of moving to shelter housing.
volved in common couple violence may well We can compare this figure with an estimate
agree to participate in a survey on family life. of the number of women actually requesting
What about the data sources for most of the housing in shelters in the United States. Although
feminist research-shelters, hospital emergency the National Coalition on Domestic Violence can-
rooms, and the criminal and divorce courts? Cer-not provide such statistics, this source suggested
tainly there are equally serious biases in thesethat I contact the Pennsylvania Coalition on Do-
mestic Violence for the best available statistical
sources of data. It is likely, for example, that most
families in which couple violence is only inter-data; another source suggested Minnesota. In

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
 290 Journal of Marriage and the Family

1985-86, Pennsylvania shelters housed or turned rough comparability of "domestic violence con-
away 6,262 different women (Pennsylvania tacts" in the shelter data and "used the services of
Coalition Against Domestic Violence, 1995). Ex- a women's shelter" in the NFVS data, we must
trapolating that figure to the total U.S. population conclude that the NFVS successfully interviewed
(i.e., multiplying by 19.20), we get an estimate of about one-sixth of the users of shelter services in
120,230 women who actually tried to use shelter its target sample. The Minnesota Department of
housing, a number roughly 1.5 times the 80,000 Corrections reported shelter contacts with 36,189
that the NFVS suggests might even consider such women in 1985, which extrapolates to about 2.1
housing. The Minnesota data provide an even million women nationwide, 16 times the extrapo-
more dramatic contrast. According to the Min- lation from the NFVS.
nesota Department of Corrections (1987), in Certainly, there are a great many problems
1985, 8,518 women were housed or turned away with the statistical manipulations presented
from shelters. Extrapolating to the U.S. popula- above. Pennsylvania and Minnesota are certainly
tion (i.e., multiplying by 57.72), we get an esti- unusual states, having been in the forefront of the
mate of 491,659 women to compare with the shelter movement. In addition, the meaning of
NFVS estimate of 80,000. "using the services of a shelter" and "shelter con-
If we take the shelter data as representing the tacts" may be different in the two data sources,
absolute minimum number of women who con- and there may be hidden problems of distinguish-
sider using shelter services each year (they in- ing multiple contacts with the same woman from
clude, after all, only those women who not only contacts with different women. The extrapola-
considered such action, but took it), we would es- tions from the NFVS and shelter statistics are so
timate that the NFVS reaches one-sixth to two- divergent, however, that it is unlikely that any of
these problems would alter the conclusion that the
thirds of the victims of patriarchal terrorism in its
NFVS simply does not provide valid information
target sample. However, given the difficulty most
women find making the decision to seek help regarding the prevalence or nature of patriarchal
(Kirkwood, 1993), most shelter activists assume
terrorism.
that there are at least five terrorized women in the
community for every one that seeks shelter, sug- Response rates. This conclusion implies, of
gesting that the NFVS may collect data from only course, that a large number of the cases of patriar-
1/13 to 1/7 of such couples in its target sample. chal terrorism in the target sample fell into the
category of nonrespondents. Were there enough
Data on use of shelter services. There is another nonrespondents in the NFVS to make this argu-
potential source of data on patriarchal terrorism in ment tenable? There are two ways to look at the
the NFVS. Straus (1990b) reported that there number of nonrespondents in the NFVS. First, we
were 13 women in the NFVS who had used shel- can ask how many of the respondents who were
ter services. That figure extrapolates to about screened as eligible did not complete the inter-
128,600 women nationwide. Unfortunately, the view. According to Gelles and Straus (1991, p.
survey wording, referring to the use of the ser- 25), that number is 1,149, representing about 16%
vices of a women's shelter, is ambiguous. Most of those screened eligible, and providing the basis
so-called women's shelters in the U.S. actually for the commonly reported 84% response rate in
function as comprehensive resources for women the NFVS. If we assume that roughly half of the
who have been victimized by patriarchal terror- nonrespondents are women, we get a figure of
ism (many also address issues of sexual assault 574 female nonrespondents.
and child sexual abuse). Most of the women who However, there is a second way to look at re-
use the services of such organizations do not actu- fusals, including as nonrespondents some portion
ally move into a shelter facility. I will, therefore,of the 6,166 people whom Gelles and Straus list-
compare the figure of 128,600 derived from the ed as "unable to screen for eligibility." I presume
NFVS with an estimate from shelters of the num- that these are people who refused to answer even
ber of women who contact them annually regard- the screening questions, since people who were
ing domestic violence. not reached after multiple attempts were listed
The Pennsylvania Coalition Against Domestic separately in the discussion of the sampling
Violence reported 41,425 domestic violence con- methodology. Although it therefore makes sense
tacts with different women in 1985-86, which ex- to label these people as refusals, it is likely that
trapolates to 795,360 nationwide. If we assume only some of them were eligible for the interview

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
Two Forms of Violence Against Women 291

in any case, and we probably should not count all violence likely to lead them to consider seeking
of them as refusals. Since 47% of the respondents shelter. This select group thus includes only cases
who were screened for eligibility were deemed el- in which the women are not generally afraid of
igible, it seems reasonable to add 47% of the their partner-because they have not experienced
6,166-or 2,898-to our pool of nonrespondents, a general pattern of control-that is, women who
yielding 4,047 nonrespondents and an alternative are victims of common couple violence.
response rate of 60%, not the 84% usually report-
ed. If we assume that roughly half of the nonre- SUMMARY
spondents are women, we have 2,024 female non-
respondents. Certainly, the case for two forms of violence, one
If we assume the worst, that the eight most relatively nongendered, the other clearly patriar-
severely abused women in the NFVS represent chal, is not ironclad. However, I am not the first
only 1/30 of such women in the target sample, the scholar to suggest the possibility that there are
other 29/3-or 232-women would represent only multiple forms of couple violence. In fact, at
40% of the 574 female nonrespondents or 11% of about the time she was developing her case for
the 2,024 female nonrespondents (the base de- the "battered husband syndrome," Steinmetz
pending upon your choice of definition of nonre- (1978b) published an excellent article making a
sponse). Similarly, if as suggested in the worst distinction that is quite similar to the distinction
case scenario above, the 13 shelter clients who re- between patriarchal terrorism and common cou-
sponded represent only 1/16 of such people in the ple violence. More recently, Lloyd and Emery
target sample, the other 15/16-or 195-women (1994) have emphasized variability in their re-
(shelter clients who presumably refused to partici- view of the literature on couple violence. They
pate in the survey) would represent only 34% of present nine "tenets" that explicate the interper-
the 574 female nonrespondents, or 10% of the sonal and contextual dynamics of aggression in
2,024 female nonrespondents. Although these intimate relationships. Two of those nine tenets
percentages are not small by any means, there are focus on the likelihood of multiple forms of cou-
clearly enough nonrespondents in the NFVS to ple violence, and in both cases the authors are
cover even the worst-case estimate of underrepre- able to cite relevant data to support their position
sentation of patriarchal terrorism. (Lloyd & Emery, 1994, pp. 37-40).
Thus, I would argue that the sampling biases Nevertheless, since the heart of the distinction
of shelter research and "random" sample research between common couple violence and patriarchal
put them in touch with distinct, virtually nonover- terrorism is one of motivation, the evidence pre-
lapping populations of violent families. On the sented above can only be suggestive. What is re-
one hand, shelter samples include only a small quired is research that can provide insight into
portion of the women who are assaulted at least motivation. One way to get at motivation would
once by their partner in any particular year. (For be to gather information concerning a range of
1985, the NFVS estimate is six million such conflict and control tactics from each couple. Pa-
women, while the shelter extrapolations suggest triarchal terrorism is presumed to involve acts of
that at most two million women contacted shel- violence that are embedded in a larger context of
ters, many fewer seeking services that would control tactics. Common couple violence is pre-
make them likely to show up in a shelter research sumed to show a less purposive pattern, erupting
sample.) Of course, this select group is likelyastoit does from particular conflicts rather than
include only women who feel they must enlist from a general intent to control one's partner. A
help to escape from a man who has entrapped second approach to motivation is in-depth inter-
them in a general pattern of violence and control, viewing of couples who are involved in violence,
that is, victims of patriarchal terrorism. eliciting interpretations of the psychological and
On the other hand, the extrapolations from the interpersonal causes of specific incidents or pat-
NFVS and the Minnesota and Pennsylvania shel- terns of control. The goal is to go beyond the be-
ter data indicate that survey research reaches onlyhavioral description of particular acts to develop a
a small fraction of the women who experience narrative
se- of each incident's development, as pre-
vere violence or who make use of the services of
sented and interpreted by perpetrators and targets
shelters. The vast majority of NFVS respondentsof violence. Both of these sorts of data are com-
who experience couple violence have not contact-
monly collected in the work with shelter samples.
ed shelters and have not experienced the level of
However, we also need this kind of data from

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
 292 Journal of Marriage and the Family

samples that target populations that are more like- common couple violence, which does not gener-
ly to include examples of both forms of violence. ally escalate. Thus, advice that is based on a mis-
Let me conclude with a partial list of the rea- taken assumption of impending terrorism may do
sons for my belief that the distinction between some women a great disservice. One can also
common couple violence and patriarchal terror- imagine similar scenarios of misinterpretation and
ism is important. The first, and most important, misplaced advice in family counseling or other
has to do with the role of scientific understanding therapeutic relationships. As in most areas of in-
in the shaping of social policy. The issue is per- tervention, family practitioners will be most ef-
haps best illustrated in the debate regarding the fective if they work with a set of alternative inter-
gender symmetry/asymmetry of couple violence. pretive frameworks rather than with a single-
The failure to make a distinction between patriar- minded assumption that every case of violence
chal terrorism and common couple violence has fits the same pattern.
led some analysts to make the logical error of The third area in which problems may be cre-
leaping from (a) the description of a few case ated by the conflation of different forms of vio-
studies of terrorism perpetrated against men and lence is in theoretical interpretation. If the two
(b) frequency estimates of common couple vio- forms of violence have different psychological
lence against men from survey research to (z) the and interpersonal roots, then theory development
conclusion that there is a widespread "battered will either have to proceed along different lines
husband" syndrome. This erroneous conclusion for each, or move in the direction of synergistic
may be used in campaigns against funding for theories that explicate the conditions under which
women's shelters (Pleck et al., 1978), opponents particular combinations of the same causal factors
arguing that shelters should not be funded unless might produce qualitatively different patterns of
they devote equal resources to male and female violent behavior. For example, we are beginning
victims. Although it is indisputable that some to try to develop an understanding of the dynam-
men are terrorized by their female partners (I ics of lesbian couple violence, a phenomenon that
have worked with some at my local shelter), the must seem somewhat mysterious if we assume
presentation of survey data that tap only common that all violence within couples follows the pat-
couple violence as evidence that men are terror- tern found in patriarchal terrorism. If we were to
ized as frequently as women produces a danger- assume a unitary phenomenon, we would develop
ous distortion of reality. a theory of lesbian violence that focused heavily
A similar distortion occurs when stories of pa- on the conditions under which some lesbians
triarchal terrorism against women are used to de- might fall into patriarchal family forms. It may be
scribe the nature of family violence, while num- more reasonable to assume that the bulk of vio-
bers that probably apply only to common couple lence in lesbian relationships is of the common
violence (survey extrapolations) are used to de- couple variety and involves causal processes that
scribe its prevalence. If the arguments presented are very similar to those involved in nonlesbian
above are correct, random sample surveys cannot common couple violence, having little to do with
produce estimates of the prevalence of patriarchal the taking on of patriarchal family values.
terrorism. We must develop methods of collecting Alternatively, using the synergistic approach
and extrapolating effectively from shelter, hospi- to theory development, we might note that (a)
tal, police, and court data. some, if not all, of the causal factors involved in
A second major problem arises when educa- patriarchal terrorism may also be involved in
tional and therapeutic efforts targeted at preven- common couple violence and vice versa, (b)
tion and intervention are governed by the assump- many of these factors are best conceptualized as
tion of one form of couple violence. For example, continuous variables, and (c) although some of
in women's studies texts and in training manuals them are sex-linked, there is probably consider-
at women's centers, one often finds the statement able overlap in the gender distributions (Taylor,
that couple violence always escalates. Unaccept- 1993). The following partial list of causal factors
able as any one incident of violence in a relation- may be used to illustrate these three points: (a)
ship may be, if the arguments above are correct, it motivation to control, (b) normative acceptability
is certainly not the case that escalation is an in- of control, (c) inclination to use violence for con-
evitable part of male violence, let alone an in- trol, (d) physical strength differences that make
evitable part of the violence in lesbian relation- violence effective, (e) inclination to expressive
ships, which is almost certainly more likely to be violence, (f) victim deference, and (g) structural

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
Two Forms of Violence Against Women 293

commitment to the relationship. All could con- REFERENCES

ceivably be involved in the generation of particu-

Adams, D., Jackson, J., & Lauby, M. (1988). Family
lar cases of either patriarchal terrorism or com- violence research: Aid or obstacle to the battered
mon couple violence, each can be conceived as a women's movement. Response, 11, 14-16.
continuous variable, and all are likely to be at Berk, R. A., Loseke, D. R., Berk, S. F., & Rauma, D.
(1983). Mutual combat and other family violence myths
least imperfectly linked to gender. The behavior
In D. Finkelhor, R. Gelles, G. Hotaling, & M. A. Strau
described in this paper as patriarchal terrorism, (Eds.), The dark side of families: Current family vio
however, may develop only from the co-occur- lence research (pp. 197-212). Newbury Park, CA: Sage
rence of high values on some particular subset of Cate, R. M., Henton, J. M., Koval, J., Christopher
the causal variables. If all other combinations of F. S., & Lloyd, S. (1982). Premarital abuse: A social
psychological perspective. Journal of Family Issues, 3
the same variables produce either no violence79-90.
at
all or a pattern recognizable as common coupleDobash, R. E., & Dobash, R. P. (1979). Violence
violence, this complex combination of weakly against wives. New York: Free Press.
gender-linked, continuous variables would pro- Dobash, R. E., & Dobash, R. P. (1992). Women, vio-
duce a strongly gender-linked pattern of two lence and social change. New York: Routledge.
Dobash, R. P., Dobash, R. E., Wilson, M., & Daly,
types of couple violence. Under such conditions,
M. (1992). The myth of sexual symmetry in marital vio-
even relatively weak links of the various factors
lence. Social Problems, 39, 71-91.
to gender might produce empirical patterns of pa-Feld, S. L., & Straus, M. A. (1990). Escalation and
desistance from wife assault in marriage. In M. A.
triarchal terrorism that occur almost exclusively
Straus & R. J. Gelles (Eds.), Physical violence in Ameri-
among men in heterosexual relationships, accom-
can families (pp. 489-505). New Brunswick, NJ: Trans-
panied by the occasional occurrence of a similaraction Publishers.
pattern among women-even in lesbian relation-Fields, M. D., & Kirchner, R. M. (1978). Battered
ships-and in gay male couples. women are still in need: A reply to Steinmetz. Victimol-
Finally, we have to ask, "How on earth could ogy, 3, 216-226.
Gaquin, D. A. (1978). Spouse abuse: Data from the
two groups of social scientists come to such dif-
National Crime Survey. Victimology, 2, 632-643.
ferent conclusions about something as unsubtle asGelles, R. J. (1974). The violent home: A study of
family violence?" We owe it to the families thatphysical aggression between husbands and wives. New-
are the focus of our work not to get so caughtburyup Park, CA: Sage.
Gelles, R. J., & Straus, M. A. (1988). New York:
in the defense of our initial positions that we Simon
fail & Schuster.
to see important insights that can be gained fromGelles, R. J., & Straus, M. A. (1991). Second Na-
our disagreements. The social policy, educational,
tional Family Violence Survey: Survey methodology
and therapeutic implications of what we do are (3rd rev.). In R. J. Gelles & M. A. Straus (Eds.), Physi-
cal violence in American families, 1985 (2nd release)
too important for us to allow our deep moral aver-
[Computer file]. Durham, NH: University of New
sion to violence to blind us to important distinc-
Hampshire Family Research Laboratory [Producer].
tions. Yes, all family violence is abhorrent, but
Ann Arbor, MI: Inter-University Consortium for Politi-
not all family violence is the same. If there cal
areand Social Research [Distributor].
Giles-Sims, J. (1983). Wife-battering: A systems the-
different patterns that arise from different societal
ory approach. New York: Guilford.
roots and interpersonal dynamics, we must make
Kincaid, P. J. (1982). The omitted reality: Husband-
distinctions in order to maximize our effective-
wife violence in Ontario and policy implications for ed-
ness in moving toward the goal of peace in our
ucation. Maple, Ontario: Learners Press.
private lives. Kirkwood, C. (1993). Leaving abusive partners.
Newbury Park, CA: Sage.
Kurz, D. (1989). Social science perspectives on wife
NOTE abuse: Current debates and future directions. Gender
and Society, 3, 489-505.
I owe thanks to Kathleen Barry and our colleagues in Levinger, G. (1966). Sources of marital dissatisfac-
Vietnam, who started me off in this direction. Kathleen tion among applicants for divorce. American Journal of
Barry, Donna Hastings, Stephen Marks, Bob Milardo, Orthopsychiatry, 36, 803-807.
Marylee Taylor, and anonymous JMF reviewers gave Lloyd, S. A., & Emery, B. C. (1994). Physically ag-
me invaluable comments on early drafts. Versions of gressive conflict in romantic relationships. In Dudley
this article have been presented at the 1993 Conference Cahn (Ed.), Conflict in close relationships (pp. 27-46).
on the Family and the Status of Women in Ho Chi Hillsdale, NJ: Lawrence Erlbaum.
Minh City, the 1993 National Council on Family Rela- Martin, D. (1981). Battered wives. Volcano, CA:
tions annual meetings, and the 1994 International Con- Volcano Press.
ference on Personal Relationships. Milardo, R. M., & Klein, R. (1992). Dominance
norms and domestic violence: The justification of ag-
gression in close relationships. Paper presented at the

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms
 294 Journal of Marriage and the Family

Pre-Conference Theory Construction and Research Straus, M. A. (1971). Some social antecedents of
Methodology Workshop, National Council on Family physical punishment: A linkage theory interpretation.
Relations annual meeting, Orlando, FL. Journal of Marriage and the Family, 33, 658-663.
Minnesota Department of Corrections. (1987). Min- Straus, M. A. (1990a). The Conflict Tactics Scales
nesota program for battered women: Advocacy pro- [sic] and its critics: An evaluation and new data on va-
gram: Data summary report (through 1986). St. Paul, lidity and reliability. In M. A. Straus & R. J. Gelles
MN: Author. (Eds.), Physical violence in American families (pp. 49-
Nguyen Du. (1983). The tale of Kieu (Translated and73). New Brunswick, NJ: Transaction Publishers.
annotated by Huynh Sanh Thong). New Haven: Yale Straus, M. A. (1990b). Injury and frequency of as-
University Press. sault and the "representative sample fallacy" in measur-
Okun, L. (1986). Woman abuse: Facts replacing ing wife beating and child abuse. In M. A. Straus & R. J.
myths. Albany: State University of New York Press. Gelles (Eds.), Physical violence in American families
Pagelow, M. (1981). Woman-battering: Victims and(pp. 75-91). New Brunswick, NJ: Transaction Publish-
their experience. Newbury Park, CA: Sage. ers.

Pence, E., & Paymar, M. (1993). Education groups Straus, M. A., & Gelles, R. J. (1990). Societal
for men who batter: The Duluth model. New York: change and change in family violence from 1975-1985
Springer. as revealed by two national surveys. In M. A. Straus &
Pennsylvania Coalition Against Domestic Violence. R. J. Gelles (Eds.), Physical violence in American fami-
(1995). Personal correspondence. lies (pp. 113-131). New Brunswick, NJ: Transaction
Pleck, E., Pleck, J. H., Grossman, M., & Bart, P. B. Publishers.
(1978). The battered data syndrome: A comment on Straus, M. A., Gelles, R. J., & Steinmetz, S. K.
Steinmetz' article. Victimology, 2, 680-683. (1980). Behind closed doors. Newbury Park, CA: Sage.
Renzetti, C. M. (1992). Violent betrayal: Partner Straus, M. A., & Smith, C. (1990). Family patterns
abuse in lesbian relationships. Newbury Park, CA: and primary prevention of family violence. In M. A.
Sage. Straus & R. J. Gelles (Eds.), Physical violence in Ameri-
Roy, M. (Ed.). (1976). Battered women: A psychoso- can families (pp. 507-526). New Brunswick, NJ: Trans-
cial study of domestic violence. New York: Van Nos- action Publishers.
trand Reinhold. Taylor, M. C. (1993). Personal correspondence.
Steinmetz, S. K. (1978a). The battered husband syn- United States Commission on Civil Rights. (1978).
drome. Victimology, 2, 499-509. Battered women: Issues of public policy. Washington,
Steinmetz, S. K. (1978b). Wife-beating: A critique DC: Author.
and reformulation of existing theory. Bulletin of the Walker, L. E. (1984). The battered woman
American Academy of Psychiatry and the Law, 6, 322- syndrome. New York: Springer.
334. Wardell, L., Gillespie, D. L., & Leffler, A. (1983).
Stets, J. E., & Pirog-Good, M. A. (1987). Violence Science and violence against wives. In D. Finkelhor,
in dating relationships. Social Psychology Quarterly, 50, R. J. Gelles, G. T. Hoteling, & M. A. Straus (Eds.), The
237-246. dark side of families: Current family violence research
Stets, J. E., & Straus, M. A. (1990). Gender differ- (pp. 69-84). Beverly Hills, CA: Sage.
ences in reporting marital violence and its medical and Yllo, K., & Bograd, M. (Eds.). (1988). Feminist per-
psychological consequences. In M. A. Straus & R. J.spectives on wife abuse. Newbury Park, CA: Sage.
Gelles (Eds.), Physical violence in American families
(pp. 151-180). New Brunswick, NJ: Transaction Pub-
lishers.

This content downloaded from

158.144.113.36 on Thu, 10 Oct 2024 14:23:52 UTC
All use subject to https://about.jstor.org/terms