Fathers for Life
Fatherlessness, the lack of natural fathers in children's lives
| Home | In The News | Our Blog | Contact Us | Share

Fathers for Life Site-Search

Site Map (very large file)
Table of Contents
Children—Our most valued assets?
Educating Our Children for the Global Gynarchia
Child Support
Civil Rights & Social Issues
Family Law
Destruction of Families
Divorce Issues
Domestic Violence
Gay Issues
Hate, Hoaxes and Propaganda
Help Lines for Men
Law, Justice and The Judiciary
Mail to F4L
Men's Issues
The Politics of "Sex"
Our Most Popular Pages
Email List
References - Bibliography

You are visitor

since June 19, 2001


Male and Female Perpetrated Partner Abuse


Table of Contents


Chapter 1


Chapter 2 Part 1


Chapter 2 Part 2


Chapter 2 Part 3


Chapter  3 Part 1


Chapter 3 Part 2


Chapter 3 Part 3


Chapter 3 Part 4


Chapter 4


Chapter 5 Part 1


Chapter 5 Part 2


Chapter 5 Part 3


Chapter 5 Part 4


Chapter 5 Part 5


Chapter 5 Part 6


Chapter 6 Part 1


Chapter 6 Part 2


Appendix A


Appendix B


Appendix C



Male and Female Perpetrated Partner Abuse: Testing a Diathesis-Stress Model 

by Reena Sommer

Chapter 2; Part 1

CHAPTER TWO (part 1)


This section is a review of the partner abuse literature and focuses on the issues examined in the research.  Included is a discussion of the prevalence of male and female perpetrated violence.
    As well, an overview of theoretical perspectives that have been applied in previous research, the risk factors included in this research, and a detailed discussion of the model employed in this study are provided.

Partner Abuse Methodology

Family violence research is relatively new. The focus of much of the family violence research from the 1970's onward has been on determining the prevalence and incidence of violence within intimate relationships. Later investigations have been on the correlates of domestic violence.  Family violence, like many other new research areas, is challenged by the need to develop effective methodologies.

In a review of literature on wife assault, Dutton (1988) pointed to the difficulties associated with gathering data on its occurrence.
    The challenge associated with breaking through the wall of privacy surrounding events within the family has been a major impediment to accurate data collection.  According to Hotaling and Straus (1980), the normative kinship and household structures appear to insulate the family from social controls and social assistance in coping with intrafamilial conflict.

Dutton (1988) cited two widely used methods of estimating wife assault. The first involves an actual estimate of the frequency and incidence of abuse within the general population drawn from a representative sample.  Examples of such surveys are the National Crime Survey (U.S. Department of Justice, 1980), the National Survey on Family Violence (Straus et al., 1980; Straus & Gelles, 1986) and the Canadian Urban Victimization Survey (Solicitor General of Canada, 1985). Whereas the above studies have focused on the occurrence of abuse nationally, other family violence studies have had a more limited focus targeting cities and regions (Brinkerhoff & Lupri, 1988; Schulman, 1981; Smith, 1987).

The second method of estimating partner abuse has been through the examination of the clinically based data (Dutton, 1988) to determine what proportion of this population engages in domestic assault.  Studies of this nature target the clientele of social services, hospitals, and the criminal justice system. Although one would expect that the occurrence of abuse is higher among respondents in these groups, the medical community in particular has been criticized for failing to recognize the role of domestic abuse in patients presenting physical injuries (Johnson, 1988).

The use of convenience samples in gathering of data on partner abuse represents, yet, a third popular method of estimating the occurrence of violence within intimate relationships.  The use of convenience samples has been useful in targeting certain groups such as students (Makepeace, 1981), and provides a basis for comparison with other populations (Johnson, 1988).  Added benefits of this method are time and cost effectiveness.

For the purposes of the discussion that follows, the above typology of data collection methods will serve as a useful framework to aid in the understanding of the occurrence of violent behaviour between intimate partners.  To facilitate appropriate comparisons, only studies utilizing the CTS (Straus, 1979) to investigate the perpetration of abuse will be presented in the following section.
    Finally, an attempt will be made to differentiate between prevalence (i.e., abuse which has "ever" occurred) and incidence (i.e., abuse which has occurred within a specific time-frame) rates of partner abuse.

Prevalence of Partner Abuse

General Population Based Surveys

Probably the most cited research estimating the occurrence of several forms of family violence have been the surveys conducted by Straus et al. (1980) and Straus and Gelles (1986).  The former found that nearly half (49.5%) of all married or cohabiting couples surveyed had experienced some form of violence during the course of their relationship.  In the 1975 survey (Straus et al., 1980), the overall annual rate of husband-to-wife abuse was found to be 12.1 percent, while the 1985 resurvey demonstrated a nonsignificant decline in the rate of the same form of abuse (11.3%).  Severe abuse perpetrated annually by males was 3.8 percent in the 1975 survey and 3.0 percent in the 1985 resurvey.

The annual rates for wife-to-husband abuse were similar to those of males, although their patterns of occurrence were in the opposite direction.  However, no tests of significance were performed to establish sex differences in the occurrence of partner abuse.  For overall abuse, the 1975 findings (Straus et al., 1980) indicated that 11.6 percent of females abused their male partners, followed by a slight, but nonsignificant increase (12.1%) reported in the 1985 resurvey. Rates of severe annual abuse perpetrated by females were somewhat higher than for males in both surveys (4.6% and 4.4%, respectively).  Finally, estimates of the occurrence of couple violence were also found to be similar during the two sampling periods; showing a nonsignificant decline in both the annual rates of overall (16.0% and 15.8%, respectively) and severe abuse (6.1% and 5.8%, respectively).

Several regional and urban surveys have also been conducted in both the U.S. and Canada.  Brinkerhoff and Lupri (1988) conducted face-to-face interviews with 562 randomly selected couples in Calgary, Alberta during 1981.  Their findings indicated that 37.8 percent of the couples reported that they engaged in some form of abuse tactics during the course of their relationship.  For husband-to-wife abuse, overall abuse during the past year was reported to occur among 10.3 percent of the couples and severe abuse occurring during the past year was reported to occur among 4.8 percent of the couples.  Alternatively, for wife-to-husband abuse, overall and severe abuse during the past year was reported to occur among 13.2 percent and 10.7 percent of the couples, respectively.
    Finally, overall and severe mutual abuse occurring during the past year was reported by 14.3 percent and 6.0 percent of couples, respectively.

The rates of both male and female perpetrated severe violence reported by Brinkerhoff and Lupri (1988) are twice as high as those reported by Straus et al. (1980) and Straus and Gelles (1986).
    Furthermore, while the higher rates of female perpetrated violence are consistent with the results of the National Survey on Family Violence, the rates of overall and severe abuse reported in Brinkerhoff and Lupri (1988) are again higher.  The authors attributed this disparity in the reporting of abuse to the fact that both partners were interviewed; lessening the likelihood of under-reporting.  The authors cited the higher crude divorce rates of Calgary residents (i.e., twice as high as the national Canadian average), as well as the city's experience of an economic boom that was followed by its crash, as factors possibly related to the rates of abuse reported.

In 1987, Kennedy and Dutton (1989) investigated the occurrence of wife assault in a random sampling of two Canadian cities (Calgary and Edmonton, Alberta) and rural residents in the Province of Alberta.  Weighting procedures were applied to ensure the proportionate reporting of results.  The final weighted male and female sample was as follows: Edmonton (n=254), Calgary (n=281) and rest of the province (n=510).

Unfortunately, the authors did not report a breakdown of abuse by region. When the annual rates for abuse for overall (11.2%) and severe (2.3%) wife assault are considered, it is noted that the former is almost identical to the rate reported by Straus and Gelles (1986), whereas the latter is somewhat lower (i.e., 11.3% and 3.0%, respectively).  Rates for overall (15.5%) and severe (5.5%) couple violence were also found to be similar to those of Straus and Gelles' (1986) survey (i.e., 15.8% and 5.8%, respectively).

It is possible that the high rates of abuse reported in Brinkerhoff and Lupri's (1988) study of Calgary residents were influenced by the reports of abuse within the rural Alberta sub-sample. Another possibility may be that the dynamics associated with the slump in that city's economy during 1981 were no longer relevant during the time of data collection for this study seven years later.  However, because the authors failed to report the incidence of abuse by region, the above assertions are strictly speculative.

In summary, the data from general population based studies investigating the incidence of partner abuse seem to indicate that approximately one in nine married or cohabiting men and women abuse their intimate partners annually each year.  The studies reviewed also indicate that mutual or couple violence is reported to occur more frequently than either male or female perpetrated violence alone.

Clinically Based Data

Most of the partner abuse investigations have focused almost exclusively on wife abuse and have been based upon data drawn from the reports of victims of wife assault (Gondolf, 1988), women's shelter workers (Ontario Association of Professional Social Workers, 1987), police and court records (Caputo, 1988) reports of hospital staff (Stark, Flitcraft & Frazier, 1979) and data gathered from individuals in treatment programs (Telch & Lindquist, 1984).Gondolf, Mulvey and Lidz (1990) investigated the prevalence of wife assault among 389 males who visited the emergency room in a psychiatric training and research hospital. Results indicated that 36 percent of the cases reviewed were identified as perpetrators of various forms of assault occurring across time. Of those, 35 percent were identified as perpetrators of family assault only, 12 percent as perpetrators of family and non-family assault, and the remaining 53 percent as perpetrators of non-family assault only.

Jaffe, Wolfe, Telford and Austin's (1986) study reviewed 2003 assault occurrences causing bodily harm during a one year period. Of those,443 (22.1%) were identified as assaultive incidents where there was an established male-female relationship (married,/cohabiting) and the female was the victim of a physical attack or was threatened with violence by a male (as determined by the CTS). Reports of assaults perpetrated against males by females were not provided by their investigation.

While the first study reviewed demonstrates that types of abuse other than wife assault are evident within clinical samples, the focus of research has been primarily on wife abuse.

Data Gathered from Convenience Samples

The use of convenience samples enables the investigation of abusive behaviour within a segment of a "normal" population (i.e., non-clinical) without the need for rigorous sampling techniques.
    Group designs utilizing comparative analyses provide researchers with an opportunity to determine the extent to which the characteristics of a clinical group (i.e., abusers in treatment) resemble those of a "normal" group derived from convenience sampling.

Convenience sampling methodology in family violence research has been variously applied.  For example, convenience samples have been utilized in studies that focus on courtship violence (eg., Barnes, Greenwood & Sommer, 1991; Makepeace, 1986), as well as those that investigate methodological techniques (e.g., Szinovasz, 1983) such as longitudinal designs (eg., Malone et al., 1989) or group comparisons (eg., Jouriles & O'Leary, 1985; Lloyd, 1990).

Research by Jouriles and O'Leary (1985) attests to the versatility of convenience sampling because it allows researchers to compare the effects of a particular dependent measure within two populations.  They investigated the interspousal reliability of retrospective reports of marital violence in a community sample (n=37 couples) and a sample of couples involved in marital therapy (n=65 couples).  The results demonstrated that 27 percent of males and 6 percent of females within their community sample reported having perpetrated abuse against their partners, while their clinical sample reported abuse by 37 percent of males and 23 percent of females.
    Yet, when the reports of perpetrated and sustained abuse by males and females were further examined,  correlations derived were only low to moderate.  The authors concluded that clinic husbands tended to under-report their own perpetration of abuse while the clinic wives tended to over-report their own victimization.

This latter finding brings to mind another important function of convenience sampling; the validity of self reports of marital violence.  Arias and Beach's (1987) study investigated this issue in a sample of 82 married men and 90 married women recruited through advertisements placed in local fliers. Descriptive analyses conducted on the demographic data indicated that the sample could be characterized as typically middle class (i.e., three years post secondary education, two children and a median individual income for males of $25,000 per year and $7,000 per year for females).

Further analyses indicated that 32 percent of males and 39 percent of females reported perpetrating violence against their partner while 34 percent of both males and females reported sustaining the same.  Like the previous study, this research also provides evidence for the differential reporting of husbands' and wives' reports of enacted and sustained abuse. In explaining their findings, Arias and Beach (1987) implicated the role of social desirability as a factor affecting the accuracy with which the prevalence of abuse is estimated. Responses given may be guided by the notion that the perpetration of abuse against women is less acceptable than abuse directed toward men.  The authors suggested the use of corroborative sources (i.e., police blotters and eye witness reports) would be useful in achieving more precise estimates of abuse reports.


The discussion to this point has focused on the different sampling strategies employed to estimate the prevalence and incidence of domestic violence. Regardless of the approach employed, it is clear that the experience of abuse between intimate partners is not a "rare occurrence".  This review also indicates that being a victim of violence within an intimate relationship is not limited to women alone.  Women and men across relationship types engage in comparable rates of domestic violence.  This review provides support for Steinmetz's (1978) paper describing the "battered husband syndrome" and challenges the remarks of her critics (Pleck, Pleck, Grossman & Bart, 1978).

There also appears to be some indication that the abuse reported is likely to be bidirectional; that both partners participate actively in its occurrence. Finally, this review indicates a need for methods that address issues such as the context of abuse, its initiation and its resolution could also be addressed and be better understood.

Risk Factors in Partner Abuse

Socio-demographic Risk Factors

Sociologically based family violence research investigates the occurrence of relationship violence within the context of our social fabric.  In an effort to develop a socio-demographic profile of the abuser, researchers have therefore focused on the interrelationship between age, marital status, education, employment status, religion, race, and abuse within an intimate relationship. Sociologists have extended their study of socio-demographics to include the investigation of life stressors and violence within the family of origin as possible risk factors in partner abuse.

Much of the research attempting to develop a socio-demographic profile of the partner abuser has also focused on husband-to-wife abuse and has been derived from data gathered from clinical and abuse samples.  Furthermore, research of this kind has relied upon the reports of victims of wife assault, the police, or women's shelter workers.

Socio-economic status

Whereas Finkelhor (1983) argued that all forms of family violence are more common in the lower socioeconomic strata (in families who were experiencing unemployment and economic deprivation), this contention has not received universal support from all family violence researchers.  Sommer (1990) suggested that the link between low socio-economic status (SES) and partner abuse may be more related to this group's inability to avoid being caught than their actual involvement in acts of violence against family members. This follows Martin's (1985) and Weidman's (1986) suggestion that when compared to middle and upper class persons, individuals coming from low SES backgrounds are more visible to the criminal justice system, hospital emergency rooms, and social programs where abuse is less likely to be hidden and more likely to be documented.  High SES persons on the other hand, have the opportunity to turn to private clinicians for assistance where identification and documentation are less likely to occur.

The nature of the sampling base also plays an important role in determining the relationship between low SES and partner abuse. When clinical samples have been targeted, one finds support for the association between partner abuse and SES.  For example, Gondolf et al. (1990) found that of persons (n=221) who had visited an emergency ward in a psychiatric research and training hospital, 64 percent of the 139 perpetrators of recent assault (including family and non-family) were unemployed and only 29 percent had achieved greater than a high school education.  On the other hand, research conducted by Stets and Straus (1989) examining abuse among dating, cohabiting and married couples, drawn from a sample of college students and a national probability sample, found no significant main effects for occupational status or education. In this latter case, the association between low SES and abuse was not demonstrated.

When data from a sample of married or cohabiting Calgary residents (derived from a systematic random telephone sampling technique) were analyzed, violence was found to occur in couples from all socio-economic groups (Brinkerhoff & Lupri, 1988). Similarly, Lisonoff and Bitman's (1978) study also failed to find a relationship between low SES and spouse abuse.  These findings challenge those of Kantor and Straus (1987) in which abuse rates were found to be lower among white collar workers when compared to blue collar workers.

Research conducted by Stewart, Senger, Kallen and Scheurger (1987) may provide the basis for yet another explanation for what appears to be an over representation of abuse among the lower class. In a study of middle class college students (n=570), they found that only six percent of students' fathers and three percent of mothers physically beat each other.  However, it was also found that 18 percent of parents neglected each others' emotional needs. This study's findings support the existence of differences in the types of abuse across socio-economic groups.  The authors suggested that while members of the lower class have been socialized to express anger and frustration through physical means, those belonging to the middle class have been socialized to inhibit displays of violence.

The dichotomy in the expression of abuse illustrated above attests to the lack of clarity in the relationship between SES and the perpetration of partner abuse.  In order to resolve the ambiguity surrounding the linkage between these variables, general population based research using an objective partner abuse measure such as the CTS that can assess both psychological and physical dimensions is needed.  As well, the manner in which SES is defined may also have some bearing on the associations drawn.  Like partner abuse, the manner in which SES is assessed must also be clearly delineated so that comparisons with other studies can be appropriately done.


Unlike the research aimed at clarifying the link between partner abuse and SES, studies investigating the relationship between age and the occurrence of partner abuse provide a more consistent pattern of findings.  In general, a negative association between age and abuse has emerged in these studies.  Thus, according to the literature, perpetrators of domestic violence are most likely to be young in age.

Relying upon the information gathered in interviews of 542 women seeking assistance at two women's shelters in Dallas, Texas between 1980 and 1982, Stacey and Shupe (1983) reported that the mean age of the assaulting partner was 33 years. Results of a follow-up study comparing data gathered from male batterers attending three rehabilitative programs in Texas (Shupe & Stacey, 1987) with those of the earlier study, revealed the age of the subjects to be somewhat lower (29 years).

As noted previously, much of the data on spousal violence has been drawn from the reports of battered women.  Information gained from reports of this kind has also been useful in providing insight into the nature of the abuser. If one is to operate under the assumption that couples' ages are concordant, then Schulman's (1981) finding that the highest rates of abuse occurred among women between the ages of 18 and 29 suggests that male perpetrators of violence are likely to fall into the same age range.  Likewise, Macleod (1987) reported that 70 percent of the women staying in transition homes were under the age of 34 years perhaps also suggesting that their partners might be of a similar age group.

For the most part, data gathered from large general population surveys support the negative association between age and domestic abuse.  Straus et al. (1980) and Kennedy and Dutton (1989) noted a decline in relationship violence with age.  Brinkerhoff and Lupri (1988) however, provided somewhat different results.  Whereas mutual violence was most common among the under 30 years of age group, male perpetrated violence was most common among individuals between the ages of 30 to 45 years of age.  In earlier research (Sommer, 1990), it was suggested that this finding may be explained in terms of this middle age group's vulnerability being associated with the collapse of the Calgary economy at the time of that study's data collection.
    As such, the ensuing stress may have manifested itself in men's maladaptive manner of conflict resolution within their intimate relationships.

In spite of the limited research targeting females as potential perpetrators of violence, there is some indication that being young in age is also relevant to their involvement in abusive episodes.
     For example, O'Leary, Barling, Arias and Rosenbaum (1989) measured violence in couples at three separate time periods (i.e., one month prior to marriage and 18 and 30 months thereafter).  At the time of the first data collection, the mean age of the females was 23.6 years (mean age for males was 25.3 years). The results of their analyses indicated that across all abuse tactics (and when compared to males), the prevalence of abuse was highest for females in the premarriage data collection period.

We need to consider whether the relationship between age and abuse is due to constitutional factors associated with young age (i.e., impulse control, heightened aggression), or is associated with a young person's lack of experience or with cohort differences (Egley, 1991).  As yet, the nature of this relationship has not been clearly discerned.

Marital status

The high rates of violence reported by dating couples and divorced people have prompted researchers to examine variations in abuse reporting across marital status categories.  It has been suggested that marital status and abuse may be associated spuriously because age exerts an influence on both these variables (Stets & Straus, 1989).  In other words, dating partners may be more vulnerable for perpetrating abuse, not so much because of their marital status, but instead due to their young age.  Such a position calls into question an earlier claim that "the marriage license is a hitting license" (Stets & Straus, 1989).

Shield, McCall and Hanneke's (1988) study examining violence within both criminal and familial contexts found that violence in general was most common among men who were married for a short period of time.  Furthermore, Stets (1990) suggested that verbal aggression experienced early in an intimate relationship carries the seed of physical aggression later in a marriage.  While these studies support the contention that abuse is most common among married couples, the effects of age as a covariate were not addressed by any of these authors.

One study that did attempt to discern the effects of marital status on abuse while controlling for a number of demographic variables found a nonsignificant interaction effect for marital status and age (Stets & Straus, 1989). Across age groups cohabiting couples had higher rates of partner abuse compared to either dating or married couples.  Severe abuse was also found to be more common among this group.

Research by Kalmuss and Seltzer (1986) examined differences in the incidence of spouse abuse reporting between first and second marriages. Their findings indicated that spouse abuse is more likely when one or both spouses have been divorced than in never-divorced families.  This finding remained even after controlling for exposure to family violence during one's childhood.  This study supports the notion that divorced adults transfer maladaptive conflict resolution strategies from one relationship to another.  Rather than age being an important covariate in the relationship between marital status and partner abuse, perhaps the influence of a long history of inappropriate communication and conflict resolving strategies should instead be considered.

Similarly, Nisonhoff and Bitman (1978) found that divorced and separated persons were more likely to report being hit and to have hit their spouses when compared to respondents from other marital status categories.  Moreover, it was also found that divorced and separated persons were significantly more likely to know at least one other couple that had engaged in spouse abuse.  The findings of both these studies suggest, that divorced or separated are particularly vulnerable to the experience of abuse not necessarily by virtue of their marital status, but through the persistence of dysfunctional modes of conflict resolution that transcend relationships.

For those couples committed to notion of "marriage till death do we part", there may be a reluctance to report abuse in their relationship.  It may be that these individuals fear that reporting their abuse may threaten their relationship. Because of this, many ongoing cases of abuse remain undisclosed.  Reaching these couples is another challenge facing researchers.

The previous discussion has highlighted some of the difficulties faced by family violence researchers concerned with the relationship between marital status and the occurrence of partner abuse.  Research indicating that partner abuse seems to be an ongoing event first evidencing itself during courtship, and in many cases continuing onto marriage (Flynn, 1987; Roscoe & Benaske, 1985), raises questions about Stets and Straus's (1979) assertion that being involved in a married relationship poses the most danger for individuals at risk for abuse.


Research investigating the association between religion and the occurrence of violence within intimate relationships has been both limited and problematic.  First, very few studies have focused on the relationship between these variables, and when attempts have been made, researchers have been faced with problems associated with the manner in which religion should be measured.
     Separating the effects of religious affiliation from those of religiosity seem to be a central importance in studies investigating the link between partner abuse and religion.

In light of these difficulties, Straus et al. (1980) explored the question of whether or not the lack of religious preference influenced the level of violence in families.  According to their findings, the relationship between religion and family violence varied for men and women when rates of perpetration and victimization were compared.  Women without a religious preference were found to be more at risk for both the perpetration and victimization of abuse while men without a religious preference were only at an increased risk for its perpetration.

Goldsmith (1984) reported the findings of a 1980 study conducted in the Los Angeles Jewish community. Her findings suggested, that while rates of violence among Jews were similar to those found in the general population, sixty-one percent of those surveyed believed family violence was not a problem in the Jewish community.  This perception is better understood by considering Scarf's (1983) suggestion, that within the context of Jewish socialization it is believed that Jewish men do not beat their wives.  In light of this idealized concept of Jewish family life,  women who are abused are prevented from seeking or accepting help. Moreover, family-centred socialization characteristic of Jewish families encourages denial among abused Jewish women while making them feel guilty and responsible for their husbands' actions (Gibson & Wyden, 1969).  In spite of the focus on Jewish women's vulnerability toward abuse, Straus et al. (1980) found that the rate of abuse by perpetrated by Jewish women was higher than that reported by Jewish men, and was highest among all the other religious groups compared.

Straus et al. (1980) suggested that members of fundamentalist religious groups would be most likely to advocate and support certain types of family violence.  Research examining the conflict resolution strategies employed by Quaker families provides an interesting test of this hypothesis (Brutz & Ingoldsby, 1984). In general, findings indicated no significant differences in the rates of male and female perpetrated violence when compared to Straus et al.'s (1980) national survey data. However, differences were found in the pattern of violence (as indicated by the CTS) when comparing the Quaker sample with the national sample.  An examination of individual CTS items and subscales suggested that Quakers are less violent than families nationally.  In other words, Quakers engage in similar rates of violence when compared to the general population, but limit themselves to its milder forms.

Straus et al.'s (1980) earlier claim regarding the greater risk of violence among fundamentalist religious groups, should be considered at the very least tentative and subject to variations associated with other features of religious groups.  In the case of Quakers and other groups with similar philosophies, the commitment to pacifism may have the greater influence on the occurrence of violence than simply a membership in a fundamentalist religion.


The literature examining the relationship between race and the occurrence of relationship violence is also limited.  Data collected in the U.S. have focused primarily on comparing the rates of family violence among white, black, and Hispanic people (Schulman, 1981; Straus et al., 1980).  Research by Schulman (1981) indicates that rates of male perpetrated partner abuse were higher among nonwhites. To date, Canadian based research has failed to provide reliable data testing the relationship between race and abuse.

Research conducted in the U.S. has attempted to gain a better understanding of the relationship between violence and race by exploring the trends in abuse over time and by testing the contributions of other variables such as SES.  In a re-analysis of the National Family Violence Survey data, Hampton, Gelles and Harrop (1989) found, while rates of abuse in general were higher among blacks when compared to whites, further analyses revealed a different pattern of partner abuse for blacks than those reported by Straus and Gelles (1986) for their entire national sample. Whereas Straus and Gelles (1986) reported a stabilization and/or a decrease in male, female and couple violence within their national sample, Hampton et al.(1989) found an increase in female and couple violence, but a decrease in male perpetrated violence within the black subsample of the same national sample with the latter reaching a level of significance.

In an effort to explain this finding, the authors concurred with Straus and Gelles's (1986) conclusion, that in general, the changes in the rates of violence are reflective of an actual change in behaviour such as changes in family structure and economics.  For example, Hampton et al. (1989) found that between 1975 and 1985, black families experienced a $16,500 increase in median family income compared to a $10,000 increase in median family income among white families.  Based upon the advances in black families' lifestyles together with lower inflation, Hampton et al. (1989) speculated that the resulting greater economic prosperity may be responsible for reducing the risk of domestic violence for this group of people.
     This would explain the decline in violence found among black males. They explained the substantive increase in the rate of wife to husband abuse among black women on the other hand as occurring as a result of the increased status of black women.

These findings suggest that research investigating a link between race and partner abuse should consider the effects of social class.  Lockart's (1987) study comparing abuse among whites (n=152) and blacks (n=155) found that while there was no significant difference in the proportion of black women (35.48%) and white women (35.5%) who reported being victims of wife abuse during the previous year, a larger proportion of middle class black women than middle class white women reported victimization by a male partner.

Based on these findings, the authors agreed with Staple's (1976) conclusion that blacks are not inherently more violent in their intimate relationships than whites.  Rather, higher levels of domestic violence found among blacks are thought to be the result of the particular social predicament in which blacks find themselves in American society.  Lockart (1987) suggested, that the higher incidence of violence found among black middle class couples compared to white middle class couples, may be explained in terms of an adaptation process associated with blacks' newly acquired middle class status.

In a reanalysis of these data, Lockart and White (1989) focused on a subsample of black women to examine the effects of SES.  This time, they found that women from lower class households were most likely to experience more conflicts and more conflicts leading to violence when compared to middle and upper class women.  The authors stated that this finding is consistent with social exchange theory that proposes regardless of race, lower class individuals who seek to control a relationship, but lack legitimate resources, may tend to use violence to enforce dominance.

Women from middle class black families were most at risk for being perpetrators of violence.  Consistent with the explanation provided by Hampton et al. (1989) and Lockhart (1987), Lockhart and White (1989) suggested that the vulnerability of black women to perpetrate abuse may be related to women's movement toward economic equality.  Due to the lack of an appropriate model for black women's economic equality, violence was therefore explained as a form of adjustment to the relatively new male/female power structure.

The studies reviewed suggest the need for more research investigating the relationship between race and domestic violence that includes the influences of other variables.  These studies also support the important role of social class in determining the nature of this relationship.  Another area of inquiry involves investigating the possibility that the race/class and domestic violence relationship may be further affected by regional and cultural differences.

Family of Origin

Of the many socio-demographic risk factors investigated by family violence researchers, that which focuses on examining the influence of an individual's early exposure to violence within one's family of origin has been shown to be a salient factor in its occurrence.  A review of case comparison studies drawn from 400 empirical reports (Hotaling & Sugarman, 1986) on husband to wife abuse attests to the relevance of this variable. Their results indicated that of the 42 risk markers analyzed for female victims of violence, witnessing violence between parents/caregivers while growing up was the only consistent risk. When risk factors for male perpetrators of violence were examined, witnessing violence as a child or adolescent was one of nine consistent risk markers among a total of 38 analyzed.

In a review of 16 studies implicating the relationship between the occurrence of spousal violence and violence in the family of origin, Tolman and Bennett (1990) determined that the incidence figures ranged from 24 percent to 81.1 percent.  They provided evidence that violence within the family of origin discriminates among many types of relationship violence (i.e., severe v. minor forms of violence) and violence prone relationships (alcoholic v. nonalcoholic, wife abusers v. generally assaultive men, and recidivate v. nonrecidivate batterers).

A major concern raised by researchers centres on the issue of, whether it is the child's experience of abuse at the hands of one's parents, or his or her observation of parents' violence that is most important in predicting later violence within an intimate relationship.  In general, when spouse abusers are compared to nonabusers, violence within intimate relationships seems to be related to both direct (i.e., parent to child abuse) and indirect (observation of parental abuse) forms of violence within the family of origin.  Barnett, Fagan and Booker (1991) argued that, "violent individuals have experienced types and levels of childhood abuse by their parents, and have observed forms and levels of parental violence, which are in sharp contrast with the experiences of nonviolent individuals" (p. 235).

Based on data gathered by the 1976 National Survey of Family Violence (Straus & Gelles, 1986),  Kalmuss (1984) and Seltzer and Kalmuss (1988) examined the effects of both direct and indirect forms of violence within the family as reported by adults (n=2143 and n=1436, respectively).  The results indicated that, while both forms of violence in the family of origin were found to be predictive of abuse, the perpetration of spouse abuse was more strongly related to observing parental violence than by parent-child hitting.  Separate analyses were not conducted for male and female respondents, and only the perpetration of male to female abuse was examined.

Gaining an understanding the link between violence in the family of origin and current partner abuse implicates the influence of gender (O'Leary & Curley, 1986).  O'Leary and Curley's (1986) findings demonstrated that men's accounts of spousal violence were related to both observing marital violence and being abused as a child.  Analyses of female data focused on being a victim of spouse abuse and failed to link it to either form of exposure to violence within the family of origin.

The sex difference reported by O'Leary and Curley (1986) contradict earlier findings by Kalmuss (1984) suggesting that the transmission of family violence across generations, while not sex specific (i.e., observing father hit mother increases the likelihood that both sons and daughters will be victims as well as perpetrators of marital aggression), tends to be role specific (i.e., the observation of violence between parents teaches children the appropriateness of such behaviour).  The idea that children model specific violent behaviours of parents in their own intimate relationships was supported in research by Bernard and Bernard (1983) indicating that individuals "indulge in the same forms of abuse as they experienced or observed in their families of origin" (p. 286).

Cappell and Heiner (1990) added further insight by examining how levels of violence in the respondent's current family were associated with the presence or absence of violence in the respondent's family of origin.  According to the authors, a vulnerability toward interpersonal family violence (through the observation of parents' marital violence) for both males and females may instead be transmitted rather than a transference of specific roles in networks of aggressive relations.

Finally, the influence of personality on the partner abuse/violence in the family of origin linkage has also been noted. In examining differences in personality and family of origin among alcoholic and nonalcoholic agency identified batterers, community identified batterers and a nonviolent comparison group, Hamberger and Hastings (1991) found that alcoholic batterers were most likely to report having witnessed parental violence and having been abused as a child.  This finding is consistent with that of Jaffe, Babour and Fishbein (1988) who found that early experiences of aggression were the best predictor of drinking related aggression (including spouse abuse) later in life.

The studies reviewed in this section highlight some of the general trends in research concerned with violence in the family of origin and its effects on later interpersonal violence.  There is little dispute as to whether a relationship exists between these variables.   However as noted, researchers remain unclear about the nature of the linkage of these variables.   For example, Seltzer and Kalmuss (1988) pointed out that many adults who perpetrate abuse against their partners come from homes that have not experienced violence while others exposed to violent behaviour in their childhood families fail to carry this mode of conflict resolution into later relationships.  According to the authors, these discrepancies in the transmission of marital violence do not negate the effects of social learning theory, but simply implicate the influence of other important modelling agents outside the family such as peers and the media (Check & Malamuth, 1985; Gwartney-Gibbs, Stockard & Bohmer, 1987; Williams, 1990).


Stress has been variously defined by family violence writers (Farrington, 1986; Porter, 1985; Schinke, Schilling, Barth, Gilchrist & Maxwell, 1986).  Yet in spite of minor differences in interpretation, these definitions have shared in common the idea that, when environmental demands exceed the capabilities of people, the resulting response can be described as stressful.

Stressors or life stress events have been identified as both positive and negative events such as unemployment, mobility, marriage, birth of children, divorce, aging, and death (Makepeace, 1983).  There is general agreement among writers and practitioners in the area that a relationship exists between the occurrence of violence between intimate partners and the experience of life stress events.  However, as with the other risk factors addressed, the relationship of stress and violence between intimates is complex and as a result, not completely understood.

The experience of stress has been measured in a number of ways. Researchers have often relied upon validated scales appropriate for their sample or have developed their own measures.  Because of this, variability in the measurement of stress has been common throughout the partner abuse literature.

Neidig, Friedman and Collins (1985) examined the relationship between male perpetrated spouse abuse and stress using the Social Readjustment Rating Scale developed by Holmes and Rahe (1967). Their findings indicated that this measure differentiated batterers from nonbatterers with the former scoring higher. Seltzer and Kalmuss's (1988) analysis of Straus and Gelles' national survey data extended these findings.  Using a subset of eleven items taken from the same scale, they reported that the joint effects of early exposure to family violence and recent stressful experiences on spouse abuse were additive rather than interactive.  In other words, adults who had been exposed to violence in their families of origin as well as to recent stressful experiences and chronic economic strain were more likely to exhibit violence against their partners than individuals exposed to fewer factors.

The research described above supports peoples' experiences of stress as being idiosyncratic in nature.  Consistent with the comments of Seltzer and Kalmuss (1988), it appears that an individual's subjective perception of events may in part determine his/her experience of stress.  Just as the degree of violence perpetrated and experienced varies in terms of frequency and severity across subjects, so does the experience of stress.  The development of measures for use with particular populations (Makepeace, 1983; Marshall and Rose, 1990) indicates an initial attempt by researchers in acknowledging the uniqueness of peoples' experiences.  However, in order to establish a clear relationship between stress and the perpetration of violence, researchers need to:

  1. consider both positive and negative life stressors as possible predictors of partner abuse, and

  2. distinguish between the effects of using different instruments and the actual correlates of stress and partner abuse.


Next: Chapter 2 Part 2

2001 02 10 (format changes)
2003 10 01 (format changes)