Toward a Biopsychosocial Model of Domestic Violence Council on Family Relations is collaborating...

Toward a Biopsychosocial Model of Domestic ViolenceAuthor(s): Patrick C. McKenry, Teresa W. Julian and Stephen M. GavazziReviewed work(s):Source: Journal of Marriage and Family, Vol. 57, No. 2 (May, 1995), pp. 307-320Published by: National Council on Family RelationsStable URL: http://www.jstor.org/stable/353685 .

Accessed: 28/11/2012 13:51

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

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

http://www.jstor.org

This content downloaded by the authorized user from 192.168.82.206 on Wed, 28 Nov 2012 13:51:29 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=ncfr

http://www.jstor.org/stable/353685?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


PATRICK C. MCKENRY The Ohio State University

TERESA W. JULIAN The Ohio State University*

STEPHEN M. GAVAZZI The Ohio State University**

Toward a Biopsychosocial Model

of Domestic Violence

A sample of 102 married men were interviewed and physically assessed in an attempt to develop a biopsychosocial model of male domestic violence. Because the dependent variable, domestic violence, was censored, Tobit analysis was used to identify significant predictors. When analyzed separately, each domain was significantly related to male domestic violence. However, when all domains were considered together, only the biological and social domains yielded independent effects. Significant independent variables included alcohol, family income, and relationship quality, with testosterone approaching significance.

It is widely accepted that the etiology of family violence is a multifaceted phenomenon that can

Department of Family Relations and Human Development, Department of Black Studies, and The Ohio Agricultural Re- search and Development Center, 315 Campbell Hall, 1787 Neil Avenue, The Ohio State University, Columbus, OH 43210.

*Department of Psychiatry, Upham Hall, 473 West 12th Av- enue, The Ohio State University, Columbus, OH 43210.

**Department of Family Relations and Human Development, Marriage and Family Therapy Program, 315 Campbell Hall, 1787 Neil Avenue, The Ohio State University, Columbus, OH 43210.

Key Words: biopsychosocial, domestic violence, spouse abuse.

best be understood from a multidisciplinary perspective. No single theory or discipline has been adequate in thoroughly explaining spouse abuse (Gelles & Loeske, 1993; Hotaling & Sugarman, 1986; Howell & Pugliesi, 1988). Reviews of the literature on domestic violence typically have utilized three groupings of theories to account for the separate contributions of biological, psychological, and sociological perspectives (e.g., Stein- metz, 1987; Van Hasslet, Morrison, Bellack, & Hersen, 1988). However, to date, no studies have attempted to integrate these perspectives into what is commonly referred to as a biopsychosocial perspective.

THE BIOPSYCHOSOCIAL PERSPECTIVE

The biopsychosocial perspective is an attempt to understand health and illness through an apprecia- tion of how biological, psychological, and social elements persist in affiliation with one another. Engel's (1977, 1980) innovative work within this perspective has served to highlight the limitations of reducing explanations of dysfunction to any one of its three major components (biological considerations, psychological variables, or social context factors), and to emphasize the great bene- fits derived from their simultaneous inclusion. From the standpoint of a physician, Engel (1977) wrote:

Journal of Marriage and the Family 57 (May 1995): 307-320 307



Journal of Marriage and the Family

The boundaries between health and disease, between well and sick, are far from clear and never will be clear, for they are diffused by cultural, social, and psychological considerations.... By evaluating all of the factors contributing to both illness and patienthood, rather than giving primacy to biological factors alone, a biopsychosocial model would make it possible to explain why some individuals experience as "illness" conditions that others would regard merely as "problems of living," be they emotional reactions to life circumstances or somatic symptoms (pp. 132-133).

A more recent formulation of this perspective is presented by McDaniel, Hepworth, and Doher- ty (1992), who utilized the term biopsychosocial systems model to highlight the interactive nature of biological, psychological, and social phenomena regarding health and illness. According to this model, such phenomena are seen as not only ex- isting in an arranged hierarchical ordering, but also as having a consistent and reciprocal impact on one another. Here, biological system factors are thought to exist in and interact with psychological system factors, both of which are hypothesized to exist in and interact with family and other social system factors. Hence, descriptions that result within this framework are not simply summative but rather assert multiplicative relationships among these factors.

From this perspective, all theoretical, empirical, and clinical efforts must account for the com- plex interplay of biological, psychological, and social facets of a given intrapsychic or interpersonal dysfunction. Hence, psychiatry and associated medical models, schools of psychology, various family systems theories, and sociological models all contain their own unique limitations because of their disciplinary viewpoint on dysfunction. In addition, those who subscribe in prin- ciple to a biopsychosocial model often note how difficult the model is to carry out in practice, as exemplified in literature emanating from a variety of contexts such as pediatric illness (Wood, 1993), medical education (Engel, 1982), family medicine (Doherty, Baird, & Becker, 1986), psychiatry (Amchin, 1991), and family psychoeduca- tional approaches (Moltz, 1993).

In the area of interpersonal violence, some theoretical work has been done that attempts to integrate the biopsychosocial aspects of interpersonal violence. Dutton (1985) utilized a variety of theoretical frameworks in presenting an ecologically nested theory of interpersonal violence, including factors related to genetic predisposition, physio-

logical arousal, emotional labeling, power issues, neighborhood influences, unemployment, and the effect of cultural and societal characteristics.

Also, treatment models can be found that touch on all aspects of the biopsychosocial model. For instance, Goldner and colleagues (Goldner, Penn, Sheinberg, & Walker, 1990) have utilized a family systems orientation to conceptually link the biological, psychological, and social roots of skewed gender identities in partners presenting with domestic violence complaints. Further, the adoption of a biopsychosocial approach is consistent with newer federal funding initiatives that recognize and encourage interdisciplinary research and intervention approaches, in essence di- recting attention to biological components in addition to social science factors.

The research literature on interpersonal violence has been moving toward the integration of these three perspectives. While not fully biopsychosocial in their conceptual underpinnings, some studies within the interpersonal violence literature have generated data that directly speak to the interactive nature of certain biological, psychological, and social phenomenon. For example, Dabbs and Morris (1990) found that relationships between testosterone and antisocial tendencies in a sample of males were moderated by their socioe- conomic status. Julian and McKenry (1993) reported that men's intimate relationship quality and depression levels predicted male violence toward female partners, although alcohol usage and testosterone were not significant predictors in this study. And Leonard and Blane (1992), in a national sample of young men, found that the relationship between alcohol use and marital aggression was moderated by both the male's level of hostility and level of marital satisfaction.

The purpose of this study is to develop a model that incorporates salient predictors from the biological, psychological, and social domains that tra- ditionally have been associated with domestic violence. The relative contribution of each domain, as well as selected interactions within and across domains, is also assessed. This article focuses on husbands' abuse toward wives because this is by far the most common form of domestic violence (Stets & Straus, 1989; Van Hasslet et al., 1988).

OVERVIEW OF THE CONCEPTUAL DOMAINS

Biological Factors

Meyer-Bahlburg (1981) contended that to understand aggression, there is a need to increase our

308



Biopsychosocial Model

understanding of the role of androgens. A recent comprehensive literature review has indicated that, in a majority of studies, high testosterone levels tend to covary with high probability of aggressive behaviors, dominance status, and patho- logic forms of aggression (Archer, 1991; Meyer- Bahlburg, 1981), especially with antisocial populations (Dabbs, Frady, Carr, & Besch, 1987; Dabbs, Ruback, Frady, Hopper, & Sgoutas, 1988; Udry, 1989). For example, Rada (1978), in a study of rapists, child molesters, and normal con- trols, found significantly elevated plasma testosterone levels in subjects judged most violent.

In a normative sample of over 4,000 male United States military veterans, Dabbs and Morris (1990) found that individuals higher in testosterone more often reported the following: having trouble with parents, teachers, and classmates; being assaultive toward other adults; going AWOL in the military; using hard drugs, marijuana, and alcohol; and having more sexual partners. Udry (1989) also found that testosterone levels were related to a variety of problem behaviors among a sample of male adolescents.

Studies of the relation between testosterone and marital relations have focused on coital frequency but little else (Booth & Dabbs, 1993). Be- cause testosterone has consistent and moderately strong links with aggression, dominance, de- pressed occupational achievement, and antisocial behavior such as fighting, nontraffic arrests, drug use, and sensation-seeking, it would appear that elevated testosterone has the potential to affect marriage adversely (Booth & Dabbs, 1993). The recurring link between testosterone and aggression and antisocial behavior could mean that men with high testosterone levels tend to carry con- tentious and hostile behavior into relationships with the opposite sex. Booth and Dabbs (1993) found in their sample of former servicemen that testosterone was positively and linearly related to every aspect of marital quality, including hitting or throwing things at spouses. Similarly, Julian and McKenry (1989) found a negative relationship between testosterone and marital happiness in a study of middle-aged males.

Another biological indicator of violent behavior is serotonin-the ubiquitous neurotransmitter that modulates the action of other brain chemi- cals. A variety of violent, impulsive behaviors has been associated with low levels of serotonin as measured through prolactin levels (Burrowes, Halles, & Arrington, 1988; Coccaro et al., 1989). Serotonin, however, does not covary with testos-

terone. Testosterone is thought to be more strong- ly correlated with outward-directed aggressive- ness and lack of socialization than it is with im-

pulsiveness, whereas serotonin is hypothesized to be related to impulsive aggression (Virkkunen & Linnoila, 1993).

Alcohol as a chemical substance has been clearly linked to aggression, yet any direct causal relationship between alcohol use and aggression has proved difficult to demonstrate (Collins, 1986; Fagan, Barnett, & Patton, 1988; Lindman, van der Pahlen, Ost, & Eriksson, 1992). In terms of domestic violence per se, alcohol use has been associated with 25% to 85% of cases (Kantor & Straus, 1987). Explanations generally have been of two overly simplistic types (Pernanen, 1991). The first is a pharmacological approach, which asserts that postdrinking violence results from alcohol acting as a catalyst for physiological changes that lead to disinhibited behavior; however, such explanations fail to identify the internal physiological mechanisms and overlook the influence of social processes on behavior. The second, a sociocultural approach, focuses on learned and shared beliefs from the cultural milieu that allow drinkers to place the responsibility for violent behavior on alcohol. What is clear from the research is that the relationship between alcohol and violence is shaped in ways as yet undetermined by a combination of individual, situational, and sociocultural factors that mediate the physiological effects of alcohol consumption (Martin, 1992).

Some recent data have attempted to link testosterone levels to alcohol in the etiology of

spouse abuse. Data indicate that men with high levels of testosterone are more apt to suffer from alcohol abuse (Dabbs & Morris, 1990). Udry (1989) concluded from his studies of testosterone among adolescent males that testosterone increases susceptibility to alcohol and that hormone levels interact with social variables in predicting a variety of problem behaviors. Studies of squirrel monkeys (Winslow & Miczek, 1985; Winslow, Ellingboe, & Miczek, 1988) indicate that testosterone may activate alcohol-sensitive brain mechanisms involved in aggressive behavior. Lindman et al. (1992) found that in abusive males high levels of testosterone often precede alcohol abuse, although the use of alcohol itself results in a sup- pression of testosterone.

A similar interaction has been found in terms of serotonin levels. Low serotonin levels are associated with a tendency to exhibit impulsive violent behavior under the influence of alcohol. This

309




is especially true for those with psychiatric diag- noses such as antisocial personality disorders (Virkkunen & Linnoila, 1993).

Social Factors

A social perspective on domestic violence places physiological and psychological variables within a wider explanatory framework that considers the impact of social institutions and social structures on social behavior (Gelles, 1993b). Empirically, much of this work has been centered in a social stress or life events paradigm.

Gelles (1987) stated that a consistent finding in domestic violence research is that violence is highly related to social stress. Life events research has indicated that negative life events, especially those threatening the status of the tradi- tional male role, are highly related to spousal abuse (Gelles, 1989; Steinmetz, 1987). According to stress theory, stress reactions such as violence are mediated by coping resources; those men with lower incomes, poorer marital quality, less social support, and alcohol abuse have been found to be most vulnerable to violent reactions (Gelles, 1994; Steinmetz, 1987).

Marital quality can serve as both a stressor and resource in the etiology of domestic violence. Gelles (1974) and Rounsaville (1978) have found that almost one-half of all cases of domestic violence are preceded by sudden transitions in inti- macy. Leonard and Blane (1992) and Pernanen (1976) noted that marital conflict precedes marital aggression and that marital conflict may interact with alcohol use in predicting domestic violence. Goode (1971) contended that as marriages decline in satisfaction, a growing sense of anger and frus- tration emerges that increases the potential for violence. Inability to communicate and negotiate conflict has been found to be highly related to physical violence between spouses (Steinmetz, 1978). Feminists and others view male abuse of female intimates as coercive control, growing out of a threat to male superior status in the marital relationship. Husbands who acknowledge and re- late positively to their wife's autonomy are least at risk for violent behaviors (Steinmetz, 1987; Yllo, 1993).

Social support, in general, is a major insulator for family violence. The ability to call on friends, family, and community for assistance appears to mediate violent reactions to stress (Gelles, 1994; Steinmetz, 1987). In general, the more a family is integrated into a community, the less likelihood

there is of violent behaviors (Milner & Chil- amkurti, 1991; Straus, Gelles, & Steinmetz, 1980).

Lower income males appear far more vulnerable to violent reactions than upper income males, although violence is by no means limited to lower income groups (Gelles & Straus, 1988; Straus et al., 1980). It is suggested that members of the middle class are socialized to mediate conflicts and are more likely to rely on verbal skills to set- tle marital disputes (Steinmetz, 1978). Lower income groups are also faced with more negative life events and fewer resources to mediate the impact of such events (Gelles, 1993a; Milner & Chilamkurti, 1991; Straus et al., 1980).

Psychological Factors

After rejecting psychological approaches to understanding domestic violence that were based on early work characterizing violent behaviors as a form of pathology, recent work has empirically and conceptually validated a psychological analysis of domestic violence (Hotaling & Sugarman, 1986; O'Leary, 1993). O'Leary (1993) concluded from his own work and that of others that personality traits or disorders play a role in the etiology of domestic violence. Although such factors have small, but statistically significant, effects at lower levels of physical violence, men who are involved in higher levels of physical aggression have much higher levels of psychological disorders than men in the general population. Psychological styles or disorders most often identified include impulsivi- ty, suspicion of others, antisocial behavior, and compulsivity (O'Leary, 1993).

Abram (1989) contended that antisocial disorders may trigger violent behaviors often associated with alcoholism. She noted, however, that antisocial behavior has rarely been controlled for in studies of violence. And, as noted above, such disorders may interact with both testosterone and serotonin to predict violent behaviors (Virkkunen & Linnoila, 1993). From a more normative, life events perspective, psychological functioning is viewed as a resource that mediates the stress response to stress-producing events (Steinmetz, 1987).

Based on this review of the biopsychosocial domains related to male domestic violence, we hypothesized that each domain would be indepen- dently related to male violence and that each domain would significantly add to the variance accounted for in a model containing all three do-

310




mains. Because of the strong research support for social factors as predictors of domestic violence, it was proposed that this domain would account for more variance in the dependent measure than would psychological factors after the biological base was entered. Also, the following selected interactions within and across domains were hypothesized to contribute additional variance to the

dependent measure: testosterone x alcohol use, prolactin x alcohol use, hostility x alcohol use, hostility x prolactin, testosterone x hostility, and

relationship quality x alcohol use.

METHODS

Subjects

A total of 105 males (and their marital partners) were interviewed to test the hypotheses of this study. Three subjects were eliminated from the final data analysis because of incomplete inter- views, resulting in a final sample size of 102. In order to be eligible to participate in this study, couples had to be married or considered married

by common law (a minimum of 7 years of cohabi- tation). Thirty-four of the dyads were defined as violent, and 68 were defined as nonviolent according to scores on the Conflict/Tactics Scale (Straus, 1979), indicating male physical violence toward a marital partner as reported by either spouse in the past year. The demographic characteristics of the sample are presented by violence status in Table 1. The groups were similar in terms of all demographic characteristics with the exception of family income. The mean income (M = $26,644) for the violent group was significantly (p < .001) less than the nonviolent group (M = $44,539). The mean age of the violent group was 34.76 (SD =

9.71), and the mean age of the nonviolent group was 36.97 years (SD = 9.70). The violent males were married an average of 8.15 years (SD = 7.88), whereas the nonviolent males were married an average of 9.18 years (SD = 8.42). Most study participants were Caucasian, reported being either Protestant or Catholic, and self-evaluated their health status as good or excellent. In addition, the sample reported similar numbers of children, years of education, and years lived with their marital partner.

Procedures

These dyads were purposefully recruited through mental health center and therapist referrals and

TABLE 1. MEANS AND PERCENTAGES OF SELECTED

DEMOGRAPHIC CHARACTERISTICS OF STUDY PARTICIPANTS

Violent Nonviolent Males Males

Variable (n = 34) (n = 68)

Age 34.76 (9.71) 36.9 (9.70) Income $26,644 ($15,753) $44,539 ($22,327) Education 14.03 (2.34) 14.85 (2.1) Number of

years married 8.15 (7.88) 9.18 (8.42) Number of

children 1.56 (1.24) 1.31 (1.27) Race

Caucasian 85% 88% Hispanic 0% 2% Native American 3% 0% African American 12% 10%

Religion Protestant 44% 46% Catholic 21% 19% Moslem 0% 2% Jewish 3% 6% Atheist 3% 0% Agnostic 3% 3% Other 12% 12% None 15% 13%

Health status Excellent 29% 38% Good 50% 47% Fair 15% 10% Poor 6% 4%

Note: Standard deviations are shown in parentheses following means.

through newspaper advertisements in a large, Midwestern city. The request indicated that marital couples were sought to participate in a study of physical health and marital relationships. The vast majority of the dyads were obtained through newspaper advertisements (n = 93); only nine were therapeutic referrals. In return for their par- ticipation, potential subjects were promised $60 per couple ($30 for each spouse). Utilizing data from the authors' preliminary study, a medium effect size (f-square = .15) was used to calculate the necessary sample size.

The couples were initially screened by tele- phone for marital status and relationship quality in an attempt to get a larger number of maritally distressed couples. It was predicted that many of the distressed couples would also manifest violent behaviors. After the screening, an appointment was scheduled with each dyad for an hour-long face-to-face interview, the administration of a 15- minute paper and pencil questionnaire, and the drawing of a blood sample from the male. All

311




subjects were individually interviewed by clinically trained interviewers, matched by sex of the subject. Male subjects were scheduled so that a sample of blood could be drawn between 9:00 a.m. and 10:00 a.m. The scheduling of the data collection was necessary to control for circadian influences on plasma testosterone levels. At the time of the blood draw, various other health measures were taken (i.e., blood pressure, pulse, urine, etc.). Data collection for all dyads took place at a university medical research center, and all health measures were obtained by registered nurses.

Instrumentation

Participants were asked to respond to a series of largely forced-choice interview questions and paper and pencil instrumentation. In addition, the blood serum of the males was analyzed for testosterone and prolactin levels as well as for illicit drug use. Use of prescription drugs was determined through responses to questions regarding general health status.

Violence. The Conflict/Tactics Scale (CTS; Straus, 1979) was used to indicate the extent and level of violence of males toward marital partners. The CTS has been used and refined in a number of studies of intrafamily violence (cf. Allen & Straus, 1980; Giles-Sims, 1983; Straus & Gelles, 1986). This 18-item inventory taps several categories of conflictual behavior including: (a) not physically violent, (b) indirect threats of violence, (c) direct threats of violence, and (d) severe violence. Items are organized from least to most violent, ranging from discussing the issue calmly to using a knife or gun, and yield an overall violence score or a very severe violence score, de- pending upon the severity of the items included (Straus, 1979). Statements that represent each cat- egory are measured by the length of time since the behavior last occurred (i.e., 2, 4, 8, or 12 months ago). In this study only the items from the subscale of males self-reporting physical violence toward the marital partner and from the subscale of females reporting male partner physical violence were used. The spouses' scores were summed to avoid underreporting associated with male self-report of violent behaviors (Walker, 1984). Instead of dichotomous scores (violent/ nonviolent), cumulative scores were used because of their greater sensitivity to changes in the independent variables (Makepeace, 1983).

Alcohol. The CDTect is a method for quantitative determination of carbohydrate deficient transferrin (CDT) in human serum. CDT is a biochemical marker of value in the diagnosis and management of individuals at high risk of alcohol abuse and alcoholism. Normalization of CDT occurs only after approximately 15 days of abstinence. The CDTect assay has demonstrated a clinical sensitivity of 82% and a specificity of 97%, based on a total of 2,500 individuals (Stibler, 1991). It de- tects consumption of as little as 20 g/day of alcohol. The validity of CDT as a marker of chronic alcohol consumption was tested in a racially mixed population and was found to be a highly specific marker, irrespective of ethnic background or race (Behrens, Worner, Braly, Schaffner, & Lieber, 1988). The venous blood sample was drawn between 9:00 and 10:00 a.m., then centrifuged, and plasma was withdrawn and refrigerated within 1 hour of the blood draw. The serum samples were refrigerated at -70?C. All serum samples were assayed on the same day by the same medical technologist.

Testosterone. The Coat-A-Count No Extraction Testosterone procedure was used to measure total testosterone levels (Diagnostic Products Corpora- tion, 1985; Jaffee & Behrman, 1974). Based on the findings of past investigations that have reported intrasubject consistency of testosterone levels when samples are drawn at the same time on different days, only one blood sample was drawn from each male subject (Ehrenkranz, Bliss, & Sherod, 1974; Kreuz & Rose, 1972). The venous blood sample was drawn between 9:00 and 10:00 a.m. (to control for circadian rhythms), then centrifuged, and plasma was withdrawn and refrigerated within 1 hour of the blood draw. The serum samples were refrigerated at -70?C. All serum samples were assayed on the same day by the same medical technologist. The normal range for serum testosterone is 3.6 ng/ml to 9.9 ng/ml.

Prolactin. Human prolactin is a single-chain polypeptide of 199 amino acids. Prolactin is pro- duced by the anterior pituitary and its secretion is regulated physiologically by inhibitory and re- leasing factors of the hypothalamus (Bowers, Friesen, Hwang, Guyda, & Folkers, 1971; Tal- walker, Ratner, & Meites, 1963). It is a promising index of overall, central 5-HT serotonin activity (Coccaro et al., 1989). Its existence as a distinct chemical entity, separate from growth hormone,

312




was established through a series of studies (Frantz, Kleinberg, & Noel, 1972; Niall, 1972). Quantitative measurement of serum prolactin levels were assayed by the IMx Prolactin assay, a microparticle enzyme immunoassay (Abbott Lab- oratories, 1991). Sensitivity of the IMx Prolactin assay was calculated to be better than or equal to 0.6 ng/ml. Specificity was determined by study- ing the interference of triglycerides (up to 1,000 mg/dl), hemoglobin (up to 750 mg/dl), and biliru- bin (up to 50 mg/dl). No interference with the determination of prolactin using this assay was ob- served. Expected values range for males (n = 111) between 1.58 and 23.12 (M = 6.75) (Abbott Labo- ratories, 1991). The venous blood sample was drawn between 9:00 and 10:00 a.m., then cen-

trifuged, and plasma was withdrawn and refrigerated within 1 hour of the blood draw. The serum samples were refrigerated at -70?C. All serum samples were assayed on the same day by the same medical technologist.

Negative life events. The Life Experiences Survey (LES) is a 57-item, self-report measure of events that have occurred in the past year; 47 of the items are designed for use with the general population (Sarason, Johnson, & Siegel, 1978). Events include marriage, change in residence, major per- sonal illness, loss of job, and detention in jail. For each of the 47 events, respondents may indicate whether or not they experienced the event in the past year and, if they experienced it, the extent to which they viewed the event as positive or negative on a 7-point rating scale, ranging from "extremely negative" (-3) to "extremely positive" (3). Summing the ratings provided a negative change score, a positive change score, and a total change score. Only the negative change scores were used in this study because this subscale is theoretically most highly related to extreme stress responses. Reliability data collected in several past studies ranged from .56 to .88 (Sarason et al., 1978).

Relationship quality. The Autonomy and Related- ness Inventory (ARI) assessed the quality of the male's relationship with his female partner. Specifically, the scale measures relatedness versus detachment/rejection and autonomy versus control. This scale is thought to be particularly appropriate for the study of marital quality in terms of violent communication because the con- ceptualization and measurement focuses on perceived autonomy and control by an intimate,

which is thought to be a major component of the relationship dynamics in domestic violence. The ARI is a short version of the Marital Autonomy and Relatedness Inventory (MARI). Thirty-two short statements about interpersonal behavior of the intimate partner are asked, with five possible responses that range from "not at all like her" (1) to "very much like her" (5) (Schaefer & Edger- ton, 1982). Instrument items include such descrip- tors as "talks over her problems with me," "is there when I need her." Scale reliabilities are appropriate for research, and validity of the ARI is supported by moderate to strong correlations with the Spanier Dyadic Adjustment Scale (Schaefer &

Edgerton, 1982). The Cronbach alpha reliability in this study was .71.

Family income. Family income was measured by male self-report of the annual income provided by his employment as well as by his marital partner's employment. Some researchers contend that violent behavior may be best explained by structural factors such as resource inequality (Vold, 1986). The males' reports and their marital partners' reports of family income were very similar (violent males and female partner: r = .87, p < .001; nonviolent males and female partner: r = .89, p < .001).

Social support. The Inventory of Social Support- ive Behaviors (Barrera, Sandler, & Ramsay, 1981) was used to assess objective social support or the frequency with which male subjects reported receiving specific types of support. This instrument consists of 40 items with a Likert-response pattern that ranges from "not at all" (1) to "about every day" (5). High scores indicate more frequent receipt of such support. Types of support include the following: "did some activity together to help you get your mind off things," "loaned you over $25," "provided you with some transportation." This study utilized 28 of the original 40 items. The Cronbach alpha reliability in this study was .94.

Psychological symptoms. The Psychiatric Symp- tom Checklist 90/Brief Symptom Inventory (SCL-90, BSI) is a multidimensional paper and pencil inventory designed to assess psychopathology in psychiatric and medical outpatients (Dero- gatis, Lipman, & Covi, 1973). Respondents are asked to rate the extent to which they have been bothered by each symptom in the last few months. Likert-type responses range from "not at all" (1) to "extremely" (5). Symptoms assessed

313




include such items as "nervousness or shakiness inside," "feeling blue," and "never feeling close to another person." The checklist is scored on 10 primary symptom dimensions plus three global indices that reflect distinct aspects of psychological disorder. The primary symptom dimensions are: (a) somatization, (b) obsessive-compulsive, (c) interpersonal sensitivity, (d) paranoid ideation, (e) depression, (f) anxiety, (g) hostility, (h) pho- bia, (i) psychoticism, and (j) addictive. The Glob- al Severity Index (GSI) combines information on the number of symptoms. Only three subscales were used in this study: anxiety, hostility, and paranoia. The SCL-90 has been normed on a number of populations including psychiatric inpa- tients, alcoholics, and drug abusers. Validity studies have indicated high concurrent validity with the Minnesota Multiphasic Personality Inventory (MMPI) (Derogatis, 1992; Derogatis, Rickels, & Rock, 1976). Cronbach alpha reliabilities in this study were .73 for paranoid ideation, .72 for anxiety, and .85 for hostility.

RESULTS

The means, standard deviations, and ranges of the independent variables are presented in Table 2, and the univariate correlations (Pearson's r) are presented in Table 3 with a log score used for the dependent measure for violence because of its skewed distribution (Moore & McCabe, 1989). As mentioned above, 34 of the couples were clas- sified as maritally violent, that is, having experienced at least one male-initiated violent incident in the past year. Husbands who were violent reported 3.7 incidents of violence toward their wives, and their wives reported a mean of 3.3 incidents; the correlation between these spousal reports was statistically significant (r = .75, p < .001). These mean scores were lower than the

TABLE 2. MEANS AND STANDARD DEVIATIONS

OF INDEPENDENT VARIABLES

Variable Mean Range SD

Alcohol 14.09 5.2-56.5 6.95 Testosterone 4.4 1.6-10.6 1.62 Prolactin 11.1 3.7-43.5 6.49 Relationship quality 122.34 79-154 18.05 Negative life events 5.59 0-22 5.27 Social support 64.43 28-125 19.57 Family income 38,575 5,000-99,000 12,292 Anxiety .61 0-2.16 .48 Hostility .79 0-3.67 .71 Paranoia .74 0-3.2 .60

scores of 9.7 (Browning & Dutton, 1986) and 7.8 (Claes & Rosenthal, 1990) found in two recent studies utilizing more clinical samples.

Multivariate data analysis was conducted to assess the degree of relationship between the independent variables and the dependent variable. The independent variables were analyzed by structuring them into three groups of predictors (i.e., biological, social, and psychological). These groups first were introduced separately into the equation with the dependent measure, husband violence toward his wife. Although the dependent variable is continuous, it is censored. That is, there is a "piling up" of cases at its lower limit of zero. Thus, ordinary least squares regression, a common statistical analysis in much of the violence literature, is not an appropriate statistical technique in this study because it does not consid- er the concentration of cases at the lower limit of zero (Stets, 1991). Tobit analysis was used instead because it provides maximum likelihood parameter estimates for equations in which the dependent variable is censored (Amemiya, 1974). A log likelihood ratio is computed to test for statistical significance, and it has a chi-square distribution. Further, the Tobit model has a test analo-

TABLE 3. CORRELATION (PEARSON'S r) MATRIX OF STUDY VARIABLES

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)

1. Alcohol 2. Testosterone -.07 3. Prolactin -.10 -.02 4. Relationship quality -.03 .04 .05 5. Negative life events .01 .04 .09 -.20* 6. Social support .12 .18 .12 .32** .08 7. Family income .02 .12 -.18 .12 -.25* -.18 8. Anxiety -.03 -.07 -.02 -.33** .36** .06 -.14 9. Hostility -.06 .05 -.02 -.36** .35** .06 -.21* .68** -

10. Paranoia -.16 .00 .01 -.40** .28** .02 -.20* .53** .59** 11. Violence .31** .14 -.01 -.31** .30** .10 -.38** .19 .36** .29**

*p<.05. **p<.01.

314




gous to the R2 goodness-of-fit statistic reported in regression analysis. This statistic is called the likelihood ratio index or rho-square, and is a measure of how well the model approximates the ob- served data. Likelihood ratios have an upper bound of about .3 (Hensher & Johnson, 1981; Tardiff, 1976).

Tobit analysis of male violence toward a female intimate using the group of biological variables resulted in a significant equation. Alcohol use and testosterone were significant individual predictors in the equation, indicating that males with greater alcohol use and higher levels of testosterone express violence against their female partner more frequently. The Tobit analysis of male violence toward a female intimate using the

group of psychological variables showed that only hostility was significantly related to the dependent measure, indicating that males with higher levels of hostility are more inclined toward violence against their female partner. The Tobit analysis of male violence toward a female intimate using the group of social variables showed that family income and relationship quality were significant independent predictors in the model. Males with lower family incomes and lower relationship quality appear to express violence toward their female partner more frequently (see Table 4).

TABLE 4. TOBIT ANALYSIS FOR BIOLOGICAL, SOCIAL, AND PSYCHOLOGICAL MODELS

Parameter Domain Estimate Chi-square

Social Marital life events -.14 6.10** Negative life events .18 1.06 Social support .05 1.28 Family income -.00 11.49***

Log likelihood 143.01 Log likelihood ratio 45.4*** Rho-square .14

Physiological Alcohol .05 15.49** Prolactin .00 .01 Testosterone .19 4.2*

Log likelihood 148.05 Log likelihood ratio 35.32**** Rho-square .11

Psychological Anxiety -2.17 1.23 Hostility 2.83 3.81* Paranoia 1.39 .94

Log likelihood 154.15 Log likelihood ratio 24.43*** Rho-square .07

*p< .05. **p<.01. ***p< .001. ****p< .0001.

The complete Tobit model indicated that only alcohol use, family income, and relationship quality were significantly related to domestic violence, with testosterone remaining marginally significant (See Table 5). The three groups then were entered into Tobit equations in hierarchical fashion to assess the unique or independent variance accounted for by each cluster. Three different equations were created with one domain omitted from each equation. A log likelihood ratio statistic was computed to compare the explanatory power of these reduced models with the complete model to assess the independent strength of the omitted cluster. The null hypothesis is that the omitted set of variables has no impact on violence probabilities. Results of the Tobit analysis indicated that the physiological set of variables had a significant effect on violence (log likelihood ratio = 28.58, p < .0001), as did the social domain (log likelihood ratio = 24.46, p < .0001). Interestingly, as a group the psychological variables did not have a statistically significant independent effect on violence probability.

As noted above, previous research has suggested interactions between certain variables; based on this research literature, several interaction terms were created, and the dependent variable was regressed on these: testosterone x alcohol use, prolactin x alcohol use, testosterone x

prolactin, hostility x alcohol use, hostility x prolactin, testosterone x hostility, and relationship quality x alcohol use. These interaction terms were tested individually and entered following their corresponding main effects. However, none

TABLE 5. TOBIT ANALYSIS OF COMBINED MODEL

Parameter Domain Estimate Chi-Square

Social Marital quality -.11 4.68** Negative life events .14 .91 Social support .02 .34 Family income -.00 12.40***

Physiological Alcohol .05 23.67*** Prolactin -.01 .50 Testosterone .07 2.46*

Psychological Anxiety -1.57 1.14 Hostility .82 .43 Paranoia 1.55 2.01

Log likelihood 127.69 Log likelihood ratio 75.62**** Rho-square .23

*p<.10. **p<.05. ***p<.01. ****pp<.0001.

315




of these interaction terms had a significant impact on the level of male violence expressed toward female partners.

DISCUSSION

The findings from these analyses provide some support for a biopsychosocial approach to the understanding of domestic violence. When examined separately, each conceptual domain was significantly related to the dependent measure. In the full model, however, only the biological and social domains were statistically significant. Fur- ther, no significant interactions were found between any of the biological, psychological, and social context variables.

It is interesting to note the relative indepen- dence of measures representing the three conceptual domains in this study. The biological variables were not significantly related (Pearson's r) to any of the social or psychological variables, nor were any of the hypothesized interactions significant. Similarly, while the social and the psychological variables were interrelated within their respective domains, there were few significant relationships between variables across their domains, and again no significant interaction terms. These findings would suggest an independent effect of the various domains on male aggression toward wives in this study. Also, such findings would support a linear relationship between the independent variables and domestic violence in contrast to those preliminary studies that have found interaction effects across conceptual domains. It should be noted here that the lack of interaction effects may have been due to a number of factors, including the use of a relatively small sample, as well as the appearance of restricted variance in some of the measures utilized in this study. Selectivity cannot be ruled out given that causation cannot be determined.

The data indicate that the social cluster constituted the best set of predictors, providing support for a life events perspective on domestic violence. In particular, family income and relationship quality were found to constitute a combined set of stressors and resources that predicted a sizable effect on the dependent measure. Additionally, in the zero-order correlations, the life events measure was significantly related to domestic violence. Given the strong relationship between family income and violence probability found in this study, an alternative hypothesis was tested. This hypothesis was based on research that suggests

that family income is highly related to domestic violence and thus might overshadow other variables. Hence, a stepwise Tobit analysis control- ling for family income was conducted. This analysis did not result in a change in the previously identified significant predictors in the Tobit model with all variables entered.

Additionally, the strong association between relationship quality and violence probability prompted further analysis based on research that suggests an "extent of interaction" hypothesis (cf. Murphy & Cascardi, 1993). Relationships that are of longer duration and/or include more frequent contact have been shown to contain higher rates of aggressive interaction (Mason & Blankenship, 1987; Stets, 1991). However, Pearson's r correlations between length of marriage and both relationship quality and violent behaviors proved to be nonsignificant (r = .16 and .06, respectively).

Based on the hierarchical ordering, the biological domain also constituted an independent effect, primarily as a result of the strength of the alcohol use variable. This finding is consistent with the vast literature supporting the relationship between alcohol and abuse. However, this study is unique in its use of the CDTect, a measure only recently introduced into this country and offering the most reliable and valid means of testing for recent alcohol use available. Compared with the normative data, the subjects in this study fell below the cut- off score of 20 for excessive use of alcohol in the last 14 days for both violent males (M = 16.93, SD = 8.92) and nonviolent males (M = 12.71, SD = 5.29).

Testosterone did approach significance and might have been significant with greater variability within the measure. In general, the mean scores were very low-a mean of 4.18 compared with the normal range of 3.6 ng/ml to 9.9 ng/ml. Only two men scored above the normal range. Howev- er, testosterone was significant when the physiological cluster was assessed separately. Prolactin, as a measure of serotonin, was likewise not related to male violence, perhaps also due to its low variability, well within normal range. In addition, prolactin may not have been the best proxy for serotonin because it does not vary as much among males; however, far more intrusive measures are required to obtain other indices of serotonin. The hypothesized interactions between and among these and other variables were also limited by the prolactin and testosterone scores of this particular sample, as noted above. Perhaps the hypothesized interaction effects are dependent on acute inges-

316




tion of alcohol along with higher levels of testosterone and/or prolactin. It should be noted that the subjects were also tested for marijuana and cocaine use. Urinalysis indicated that only negligi- ble levels of marijuana were present in two nonviolent subjects, and no marijuana or cocaine use was indicated in the violent subjects. Prescription drug use was minimal in each group. Reported comorbidity rates regarding polydrug use and violent acts (Hayes & Emshoff, 1993) underscore how the lack of illegal drug ingestion becomes a further indication of the restricted range of individual and interpersonal functioning levels in the present sample.

When examined separately, the psychological domain also accounted for a significant effect on the dependent measure, with hostility as the only variable significantly related to domestic violence. Hostility could be perceived either as a direct effect or as a response to stressors (negative life events, relationship quality, and family income) with which they were correlated. However, none of the variables within the psychological domain were significant in the combined model. Perhaps this domain would have been a stronger predictor if the level of violence had been greater; psychological factors have a greater impact on males exhibiting higher levels of aggression (Ho- taling & Sugarman, 1986; O'Leary 1993). The low level of violence and the lack of variablity within this measure might also have limited the power of the other independent variables in predicting relationship violence. The scores of the subjects on the three SCL-90/BSI subscales also were well within the middle band of normative scores, although in each case the violent men's scores were greater. The means on each of the three SCL-90/BSI subscales (anxiety, hostility, paranoia) for the nonviolent men (.57, .68, .19, respectively) placed this group in the 64th, 61st, and 50th percentiles of the normative group scores, respectively, as reported by Derogatis (1992). A similar comparison of the violent men's scores (.68, 1.01, .32) placed this group in the 64th, 65th, and 53rd percentiles of the normative group scores, respectively.

Although only exploratory, the findings of this study indicate the potential of a biopsychosocial approach to the understanding of domestic violence. It is probable that the strength of many of the predictors, and thus the power of the model, could have been enhanced by the inclusion of males with scores beyond the normal range on such variables as alcohol use, testosterone, pro-

lactin, paranoia, and anxiety. Although clinical samples have been overused in family violence research, such a sample, including more patholog- ical and/or severely distressed subjects, may have been more appropriate in developing the biopsychosocial model specified in this study. Also, a larger number of violent males in general would have lessened the risk of a Type II error.

For some researchers and practitioners alike, there is often concern when biological factors are examined in relation to dysfunctional behavior that such findings might be used to absolve individual responsibility. Similarly, the examination of psychopathology as an underlying mechanism of family violence has been criticized for concep- tualizing violent behaviors within the family as only extreme responses and abnormal behaviors. It is hoped that such concern has been minimized by the study's intent of integrating both psychological and biological perspectives with the more thoroughly studied social factors.

Indeed, the findings of the study indicate that the best estimates of violence probability are the social variables, given a sample where scores on the variables within both biological and psychological domains were largely within normal range. These findings are consistent with many other studies cited above that have generated evi- dence regarding the primacy of social context variables, and in a sense they argue against the necessity for a biopsychosocial approach to domestic violence. Also, the biopsychosocial perspective holds that it is not simply the summative relationships but also the interactive relationships among biological, psychological, and social context variables that fully explain phenomena related to health and pathology (Engel, 1977; Mc- Daniel et al., 1992). Hence, further disputation of this approach is evidenced through the lack of significant interactions found between variables representing the three domains of the model. In certain ways, then, the influence of biological and psychological factors on violence probability estimates may be seen as secondary to and perhaps dependent on social context variables.

Further advancements in empirical efforts will also need to occur if researchers are to remain consistent with the holistic biopsychosocial model. For instance, it would have been prefer- able to have studied other members of the male's family or community system. Data were available from the wives on many measures and were indeed used in the compilation of the violence score. However, due to sample size restrictions

317




and the interest in individual factors that previous literature has related to domestic violence, a more expansive analysis was not conducted.

This study was also limited by its cross-sec- tional design, which did not allow for causal infer- ences. Researchers have generally neglected to examine the temporal relationships between wife abuse and characteristics correlated with it; thus it cannot be established whether correlates are risk factors or merely consequences of abuse (Sedlak, 1988). A longitudinal analysis also would have al- lowed for better detection of possible interactions of study variables. For example, it was not possible to examine Lindman et al.'s (1992) conclusion that high levels of testosterone are thought to precede alcohol abuse, with chronic alcohol use then resulting in lower testosterone levels.

Self-report is always a limitation in studies of sensitive topics. However, this study made some attempts to compensate for this limitation by using couple scores on the CTS and by using biological indicators, including the CDTect alcohol screening device, which do not rely on self-report. It is hoped that the use of clinically trained interviewers limited the influence of social desir- ability in the administration of the SCL-90/Brief Symptom Inventory.

In some respects, the ability to translate this study's findings into treatment and policy considerations is a difficult task. This is especially true in light of the antagonism that has existed histori- cally between domestic violence researchers and victim advocates (Gelles, 1994). However, it is believed that the biopsychosocial model offered here may at the very least be a small but significant step toward the utilization of the "contextual approach" advanced by Jacobson (1994a, 1994b), a perspective that advances an inclusive and eco- logical viewpoint regarding the study of relationships that contain interactions of violence.

NOTE

Salaries and research support for this study were provided, in part, by state and federal funds appropriated to the Ohio Agricultural Research and Development Cen- ter, The Ohio State University (H-075).

REFERENCES

Abbott Laboratories. (1991). Prolactin. Abbott Park, IL: Abbott Laboratories.

Abram, K. M. (1989). The effect of co-occurring disorders on criminal careers: Interaction of antisocial personality, alcoholism, and drug disorders. Internation- al Journal of Law and Psychiatry, 12, 133-148.

Allen, C., & Straus, M. (1980). Resources, power, and husband-wife violence. In M. Straus & G. Hotaling (Eds.), The social causes of husband-wife violence. Min- neapolis: University of Minnesota Press.

Amchin, J. (1991). Psychiatric diagnosis: A biopsychosocial approach using DSM-III-R. Washington, DC: American Psychiatric Press.

Amemiya, T. (1974). Multivariate regression and simultaneous equation models when the dependent variables are truncated normal. Econometrica, 42, 999- 1011.

Archer, J. (1991). The influence of testosterone on human aggression. British Journal of Psychology, 82, 1- 28.

Behrens, V., Worer, T., Braly, L., Schaffner, F., & Lieber, C. (1988). Carbohydrate-deficient transferrin (CDT), a marker for chronic alcohol consumption in different ethnic populations. Alcohol Clinical Experimental Research, 12, 427-432.

Booth, A., & Dabbs, J. (1993). Testosterone and men's marriages. Social Forces, 47, 334-355.

Bowers, C. Y., Friesen, H. G., Hwang, P., Guyda, H. J., & Folkers, K. (1971). Prolactin and thyrotropen release in man by synthetic pyroglutamylhistidylpromi- namide. Biochemical Biophysical Research Communi- cations, 45, 1033-1041.

Browning, J. J., & Dutton, D. G. (1986). Assessment of wife assault with the Conflict Tactics Scale: Using couple data to quantify the differential in reporting effect. Journal of Marriage and the Family, 48, 375-79.

Burrowes, K., Halles, R., & Arrington, E. (1988). Research on the biologic aspects of violence. Psychi- atric Clinics of North America, 11, 499-507.

Claes, J. A., & Rosenthal, D. M. (1990). Men who batter women: A study in power. Journal of Family Vio- lence, 5, 215-224.

Coccaro, E. F., Siever, L. J., Kalr, H. M., Maurer, G., Cochrane, K., Cooper, T. B., Mohs, R. C., & Davis, K. (1989). Serotonergic studies in patients with affective and personality disorders. Archives of General Psychia- try, 46, 587-598.

Collins, J. (1986). The relationship of problem drinking to individual offending sequences. In A. Blum- stein, J. Cohen, J. Roth, & C. Wisher (Eds.), Criminal careers and career criminals (pp. 125-176). Washing- ton, DC: National Academy Press.

Dabbs, J. M., Jr., Frady, R. L., Carr, T. S., & Besch, N. F. (1987). Saliva testosterone and criminal violence in young adult prison inmates. Psychomatic Medicine, 49, 174-182.

Dabbs, J. M., Jr., & Morris, R. (1990). Testosterone, social class, and antisocial behavior in a sample of 4,462 men. Psychological Science, 1, 209-211.

Dabbs, J. M., Jr., Ruback, R. B., Frady, R. L., Hop- per, C. H., & Sgoutas, D. S. (1988). Saliva testosterone and criminal violence among women. Personality and Individual Differences, 9, 269-275.

Derogatis, L. R. (1992). SCL-90R: Administration, scoring, and procedures manual. Towson, MD: Clinical Psychometric Research.

Derogatis, L. R., Lipman, R. S., & Covi, L. (1973). The SCL-90: An outpatient psychiatric rating scale: Pre- liminary report. Psychopharmacology Bulletin, 9, 13.

Derogatis, L. R., Rickels, K., & Rock, A. (1976). The SCL-90 and the MMPI: A step in the validation of a

318




new self-report scale. British Journal of Psychiatry, 128, 280-289.

Diagnostic Products Corporation. (1985). Coat-A- Count: Testosterone. Los Angeles: Author.

Doherty, W. J., Baird, M. A., & Becker, L. A. (1986). Family medicine and the biopsychosocial model: The road toward integration. Advances, 3, 17-28.

Dutton, D. G. (1985). An ecologically nested theory of male violence toward intimates. International Journal of Women's Studies, 8, 404-413.

Ehrenkranz, J., Bliss, E., & Sherod, M. H. (1974). Plasma testosterone: Correlation with aggressive behavior and social dominance in man. Psychometric Medicine, 36, 469-475.

Engel, G. L. (1977). The need for a new medical model: A challenge for biomedicine. Science, 196, 129- 136.

Engel, G. L. (1980). The clinical application of the biopsychosocial model. American Journal of Psychia- try, 137, 535-544.

Engel, G. L. (1982). The biopsychosocial model and medical education: Who are to be the teachers? New England Journal of Medicine, 306, 802-805.

Fagan, R., Barnett, O., & Patton, J. (1988). Reasons for alcohol use in maritally violent men. American Jour- nal of Drug and Alcohol Abuse, 14, 371-392.

Frantz, A. G., Kleinberg, D. D., & Noel, G. L. (1972). Studies on prolactin in men. Recent Progress in Hormonal Research, 28, 527-590.

Gelles, R. J. (1974). Child abuse as psychopathology: A sociological critique and reformation. In S. Stein- metz & M. Straus (Eds.), Violence in the family (pp. 190-204). New York: Dodd, Mead.

Gelles, R. J. (1987). Family violence. Newbury Park, CA: Sage.

Gelles, R. J. (1989). Child abuse and violence in single parent families: Parent absence and economic depri- vation. American Journal of Orthopsychiatry, 59, 492- 501.

Gelles, R. J. (1993a). Alcohol and other drugs are associated with violence: They are not its cause. In R. J. Gelles & D. R. Loseke (Eds.), Current controversies on family violence (pp. 182-196). Newbury Park, CA: Sage.

Gelles, R. J. (1993b). Through a sociological lens: Social structure and family violence. In R. J. Gelles & D. R. Loseke (Eds.), Current controversies on family violence. Newbury Park, CA: Sage.

Gelles, R. J. (1994). Research and advocacy: Can one wear two hats? Family Process, 33, 93-96.

Gelles, R. J., & Loseke, D. R. (1993). Introduction: Examining and evaluating controversies on family violence. In R. J. Gelles & D. R. Loseke (Eds.), Current controversies on family violence (pp. ix-xvii). Newbury Park, CA: Sage.

Gelles, R. J., & Strauss, M. A. (1988). Intimate violence: The causes and consequences of abuse in the American family. New York: Simon & Schuster.

Giles-Sims, J. (1983). Wife battering: A systems theory approach. New York: Guilford.

Goldner, V., Penn, P., Sheinberg, M., & Walker, G. (1990). Love and violence: Gender paradoxes in volatile attachments. Family Process, 29, 343-364.

Goode, W. (1971). Force and violence in the family. Journal of Marriage and the Family, 33, 624-636.

Hayes, H. R., & Emshoff, J. G. (1993). Substance abuse and family violence. In R. L. Hampton, T. P. Gul- lotta, G. R. Adams, E. H. Potter, & R. P. Weissberg (Eds.), Family violence: Prevention and treatment (pp. 281-310). Newbury Park, CA: Sage.

Hensher, D. A., & Johnson, L. W. (1981). Applied discrete-choice modeling. London: Croom Helm.

Hotaling, G. T., & Sugarman, D. B. (1986). An analysis of risk markers in husband to wife violence: The current state of knowledge. Violence and Victims, 1, 101-124.

Howell, M. J., & Pugliesi, K. L. (1988). Husbands who harm: Predicting spousal violence by men. Journal of Family Violence, 3, 15-27.

Jacobson, N. S. (1994a). Contextualism is dead: Long live contextualism. Family Process, 33, 97-100.

Jacobson, N. S. (1994b). Rewards and dangers in re- searching domestic violence. Family Process, 33, 81-86.

Jaffee, B., & Behrman, N. (1974). Methods of hormone radioimmunoassay. New York: Academic Press.

Julian, T., & McKenry, P. (1989). Relationship of testosterone to men's family functioning at mid-life: A research note. Aggressive Behavior, 15, 281-289.

Julian, T. W., & McKenry, P. C. (1993). Mediators of male violence toward female intimates. Journal of Family Violence, 8, 39-56.

Kantor, G. K., & Straus, M. A. (1987). The "drunken bum" theory of wife beating. Social Problems, 34, 213- 230.

Kreuz, L. E., & Rose, R. M. (1972). Assessment of aggressive behavior and plasma testosterone in a young criminal population. Psychosomatic Medicine, 34, 321- 332.

Leonard, K. E., & Blane, H. T. (1992). Alcohol and marital aggression in a national sample of young men. Journal of Interpersonal Violence, 7, 19-30.

Lindman, R., von der Pahlen, B., Ost, B., & Eriks- son, C. J. (1992). Serum testosterone, cortisol, glucose, and ethanol in males arrested for spouse abuse. Aggres- sive Behavior, 18, 393-400.

Makepeace, J. M. (1983). Life events stress and courtship violence. Family Relations, 32, 101-109.

Martin, S. E. (1992). The epidemiology of alcohol- related interpersonal violence. Alcohol Health and Re- search World, 16, 230-237.

Mason, A., & Blankenship, V. (1987). Power and affiliation motivation, stress, and abuse in intimate relationships. Journal of Personality and Social Psychology, 52, 203-210.

McDaniel, S. H., Hepworth, J., & Doherty, W. J. (1992). Medical family therapy: A biopsychosocial approach to families with health problems. New York: Basic Books.

Meyer-Bahlburg, H. F. L. (1981). Androgens and human aggression. In P. Brian & D. Benton (Eds.), The biology of aggression (pp. 263-290). Alphen ann den Rijn, The Netherlands: Sijhoff & Noordhoff.

Milner, J. S., & Chilamkurti, C. (1991). Physical child abuse perpetrator characteristics: A review of the literature. Journal of Interpersonal Violence, 6, 345- 366.

Moltz, D. A. (1993). Bipolar disorder and the family: An integrative model. Family Process, 32, 409-423.

Moore, D., & McCabe, G. (1989). Introduction to the practice of statistics. New York: Freeman.

319




Murphy, C. M., & Cascardi, M. (1993). Psychologi- cal aggression and abuse in marriage. In R. L. Hampton, T. P. Gullotta, G. R. Adams, E. H. Potter, & R. P. Weissberg (Eds.), Family violence: Prevention and treatment (pp. 86-112). Newbury Park, CA: Sage.

Niall, H. D. (1972). The chemistry of the human lac- togenic hormones. In A. R. Brynes & K. Griffiths (Eds.), Prolactin and carcinogenesis: Proceedings of the Fourth Tenovus Workshop (pp. 116-123). Cardiff, Wales: Alpha Omega Alpha.

O'Leary, K. D. (1993). Through a psychological lens: Personality traits, personality disorders, and levels of violence. In R. J. Gelles & D. R. Loseke (Eds.), Cur- rent controversies on family violence (pp. 7-30). New- bury Park, CA: Sage.

Pernanen K. (1976). Alcohol and crimes of violence. In B. Kissin & H. Bagleitor (Eds.), The biology of alcoholism: Vol. 4. Social aspects of alcoholism (pp. 351- 444). New York: Plenum.

Rada, R. (1978). Clinical aspects of the rapist. New York: Grune & Stratton.

Rounsaville, B. J. (1978). Battered wives: Barriers to identification and treatment. American Journal of Or- thopsychiatry, 48, 487-494.

Sarason, I., Johnson, J., & Siegel, J. (1978). Assess- ing the impact of life changes: Development of the life experiences survey. Journal of Consulting and Clinical Psychology, 64, 932-946.

Schaefer, E. S., & Edgerton, M. (1982). Autonomy and relatedness inventory (ARI). Unpublished manuscript, University of North Carolina at Chapel Hill, School of Public Health.

Sedlak, A. J. (1988). Prevention of wife abuse. In V. B. Van Hasselt, R. L. Morrison, A. S. Bellack, & M. Hersen (Eds.), Handbook of family violence (pp. 223-256). New York: Plenum.

Steinmetz, S. K. (1978). The battered husband syn- drome. Victimology, 2, 499-509.

Steinmetz, S. K. (1987). Family violence: Past, present, and future. In M. B. Sussman & S. K. Steinmetz (Eds.), Handbook of marriage and the family (pp. 725- 766). New York: Plenum.

Stets, J. E. (1991). Psychological aggression in dating relationships: The role of interpersonal control. Journal of Family Violence, 6, 97-114.

Stets, J. E., & Straus, M. A. (1989). The marriage license as a hitting license: A comparison of assaults in dating, cohabiting, and married couples. In M. Pirog- Good & J. E. Stets (Eds.), Violence in dating relationships: Emerging issues (pp. 33-52). New York: Praeger.

Stibler, H. (1991). Carbohydrate-deficient transferrin in serum: A new marker of potentially harmful alcohol consumption reviewed. Clinical Chemistry, 37, 2029- 2037.

Straus, M. A. (1979). Measuring intrafamily conflict and violence: The conflict tactics (CT) scale. Journal of Marriage and the Family, 41, 75-88.

Straus, M., & Gelles, R. (1986). Societal change and change in family violence from 1975 to 1985 as re- vealed by two national surveys. Journal of Marriage and the Family, 48, 1-15.

Straus, M. A., Gelles, R. J., & Steinmetz, S. K. (1980). Behind closed doors: Violence in the American family. Garden City, NY: Anchor/Doubleday.

Talwalker, P. K., Ratner, A., & Meites, J. (1963). In- hibition of pituitary prolactin synthesis and release by hypothalamic extract. American Journal of Physiology, 205, 213-218.

Tardiff, T. (1976). A note of the goodness of fit statistics for Probit and Logit models. Transportation, 5, 377-388.

Udry, J. R. (1989). Biosocial models of adolescent behavior problems. Unpublished manuscript, University of North Carolina at Chapel Hill.

Udry, J. R. (1991). Predicting alcohol use by adolescent males. Journal of Biosocial Science, 23, 381-386.

Van Hasselt, V. B., Morrison, R. L., Bellack, A. S., & Hersen, M. (1988). (Eds.), Handbook of family violence (pp. 334-356). New York: Plenum.

Virkkunen, M., & Linnoila, M. (1993). Brain serotonin, type II alcoholism and impulsive violence. Jour- nal of Studies on Alcohol, 11, 163-169.

Vold, G. B. (1986). Theoretical criminology. New York: Oxford University Press.

Walker, L. E. (1984). The battered woman syn- drome. New York: Springer.

Winslow, J., Ellingboe, J., & Miczek, K. (1988). Ef- fects of alcohol on aggressive behavior in squirrel monkeys: Influence of testosterone and social context. Psy- chopharmacology, 95, 356-363.

Winslow, J., & Miczek, K. (1985). Social status as determinant of alcohol effects on aggressive behavior in squirrel monkeys. Psychopharmacology, 85, 167-172.

Wood, B. L. (1993). Beyond the "psychosomatic family": A biobehavioral family model of pediatric illness. Family Process, 32, 261-278.

Yllo, K. A. (1993). Through a feminist lens: Gender, power, and violence. In R. J. Gelles & D. R. Loseke (Eds.), Current controversies on family violence (pp. 47- 62). Newbury Park, CA: Sage.

320



Toward a Biopsychosocial Model of Domestic Violence Council on Family Relations is collaborating...

Documents

Transcript of Toward a Biopsychosocial Model of Domestic Violence Council on Family Relations is collaborating...