GENDER DIFFERENCES IN RISK FACTORS FOR

PHYSICAL VIOLENCE BETWEEN DATING PARTNERS

 BY UNIVERSITY STUDENTS*

 

Murray A. Straus and Rose A. Medeiros

Family Research Laboratory, University of New Hampshire

Durham, NH 03824  603-862-2594  murray.straus@unh.edu

 

 

 

Abstract

 

It is widely believed that partner-assaults perpetrated by women have a different etiology than partner assaults perpetrated by men.  This paper reviews previous research on this issue and reports the results of analysis of the extent to which the 21 risk factors measured by the Partner and Relationships Profile are related to assaulting a dating partner by a sample of 232 men and 334 women university students.  Both the previous research and the new study show that half or more of the risk factors investigated apply to both men and women offenders.  Thus, social scientists who are seeking to explain partner violence, and practitioners engaged in prevention or treatment programs need to be aware of both the ways in which the etiology of partner violence is similar for men and women, and the ways in which it is different.

= = = = = = = = = = = =

 

 

 

            After more than 20 years of controversy the focus of the debate over research showing that women physically assault their male partners at about the same the rate as men physically attack female partners (O'Leary 2000; Straus 1997; Straus 1999) has shifted focus somewhat.  The focus now tends to be one whether, when women physically assault a partner, the causes or motives are different.  However, much of what has been written on differences in causes and motives is based on the beliefs and values of the authors rather than empirical evidence.  For example, (Dobash et al. 1992) assert that men’s and women’s motivation for violence differ, but they base this on differences in the greater injury rate suffered by women.  Neither their own research nor any of the studies they cite provide evidence on this issue.  Similarly, Hamberger and Lohy  (1997) summarize their results by as indicating that “female perpetrations were more likely than male perpetrators to defend themselves…” In contrast male perpetrators reported violence motivations related primarily to dominance and control”   However, we could not find that data in the article and wrote to Hamberger to ask about it.  He replied “For a variety of reasons, we chose not to include that data in the paper (personal communication, 10 October 2001).  “Nazroo (1995) states that “Men’s violence in marriage is very different from women’s”  but the empirical evidence he presents concerns differences in consequences, anxiety, and fear, and the fact that male violence is more intimidating.  None of these are measures of etiological factors, and all could follow from the greater average size and strength of men, rather than from the implied but undemonstrated difference in motives.  The unstated agenda of authors such as Dobash and Dobash and Nazroo is to create a perception violence by men as wrong and malicious, and violence by women as morally courageous defense.

 

            There are many studies that present data on factors and motivations for intimate partner violence, but which provide data only for men or only for women, and therefore cannot answer the question of whether the risk factors are different for women.  For example,  Fiebert and Gonzoles, in one of the relatively few studies which obtained data from women offenders, found that 46% said they hit their partner because he “wasn’t sensitive to my needs,”  44% “to gain my partner’s attention,”  38% because they did not believe their hitting would hurt him, 38% because he was “being verbally abusive to me,” and 43% because the partner “was not listening to me.”   However, the sample was entirely women so that no comparisons with men were presented.

 

            An additional problem is that there does not seem to have been a systematic review of empirical studies of differences between man and women in the etiology of domestic assault.  None of the three textbooks we checked contained a section on this issue, and our search of Sociological Abstracts and PsychLIt did not uncover a review article in any of the journals.

 

            Given the theoretical and practical importance of this issue, such a review is needed, and that is the first purpose of this chapter.  The second purpose is to present the results of a new empirical study of gender differences in risk factors for partner assault.  The new study compares men and women on the degree to which 21 hypothesized risk factors were related to physically assaulting a dating partner.  The risk factors examined fall into two broad categories:  The first category refers to personal psychological characteristics of the partner that are considered to be risk factors for violent behavior such as antisocial personality and attitudes approving violence.  The second category refers to characteristics of the relationship such as dominance of one partner and communication problems.

 

            A clear answer to the etiological issues that are the focus of this chapter requires longitudinal data.  However, the data available to us, like the data in the studies reviewed below, are cross sectional.  Consequently, we use the term “risk factor” (Kleinbaum, Kupper and Morgenstern 1982) because it refers to variables that are associated with the dependent variable, but are not necessarily causes .  For example, inadequate anger management skills could be a cause of violence in relationships for some violent partners.  But for some others it might be an effect of relying on physical force and therefore never having to learn to effectively control anger.  It can be argued that, if inadequate ability to deal with anger is associated with partner violence, regardless of whether it is cause or effect, there should be randomized trials to determine if including anger management training should be one element in programs of primary prevention and treatment of intimate partner violence.

 

PREVIOUS RESEARCH ON GENDER DIFFERENCES IN RISK FACTORS

 

The criteria used to search for relevant research was that the study reported empirical data for both men and women, for example, the percent for each gender, the mean for each gender, and the correlation between an etiological variable and partner-assault for each gender. Qualitative studies were included if they presented empirical data for men and women in a form that permitted comparison.

 

We used four approaches to locate relevant research.  We started with articles and books that were already known to us and met the criteria of being an empirical study (as just defined) of gender differences in risk factors.   The risk factors ranged from basic demographics such as age to motives for violence.   This located 30 studies.  Next we examined those references for references to other relevant research.  This identified 11 possibilities, of which three studies met our criteria, making a total to that point of 33 studies.   Finally, we searched Psychological Abstracts and Sociological Abstracts for the years 1970 to 2001 using combinations of the following search terms: sex differences, gender differences, risk factor, causes correlates, effect, etiology, domestic violence, family violence, partner violence, and battering.  A total of 39 references were identified, but all that met the criteria and had been previously located, with the result that the number of studies remained at 33.

 

            The unit of analysis for the literature review portion of this paper is individual analyses within studies, rather than the studies themselves. For example, a paper may have an analysis of the relationship between being having a moderate income discrepancy and assaulting a partner, and a high income discrepancy and assaulting a partner. In the case of such a study, the two analyses are considered separate rather than one finding. Thus one study may account for more than one of the analyses listed in a given section.

 

Percent Showing Gender Differences in Risk factors

 

(Insert Tables 1, 2, 3, and 4 about here)

 

The first step in the analysis of the previous research was to summarize the studies in the form of the Appendix table giving the sample used, the context, meaning, or motive variables investigated, and the findings on gender differences.  The next step was to tabulate the results for the studies in the Appendix.

 

Appendix Table 1 summarizes the results of twelve studies that compared violent men and women on a total of 56 characteristics.  In 73% of these comparisons no significant difference was found between men and women in the risk factors examined.

 

Appendix Table 2 summarizes 23 studies which provided data on gender differences in risk factors for partner assault.  Sixty percent of the variables showed the same relationship for men and women (e.g. both positive, both non-significant, etc.).  Approximately forty percent showed a significant relationship for one sex and a non-significant relationship for the other sex.  One percent of variables in these studies showed opposite relationships for men and women (one positive, the other negative). 

 

Appendix Table 3 summarizes the results for 28 variables examined in six studies that provided descriptive statistics for men and women, but without testing the significance of the difference. We classified the results as different for men and women if the percentage for the gender with the larger percentage was at least 20% greater.   Using this criterion, of the 26 risk factors evaluated by these six studies, 43% of variables used appear to have similar results for men and women.

 

Table 4 summarizes the number and percentage of analyses which found similar relationships for men and women. The table covers 13 major areas of risk factors, and shows the number of findings that were the same or similar for men and women, the number of analysis overall, and what percentage of total findings found similar relationships.  Some of the studies reviewed actually tested for a difference in relationship for men and women, in these studies a similar relationship means no significant difference. For other studies, which merely tested for relationships for men and women separately, a similar relationship means just that, both negative, both positive, or both non-significant.

 

Gender Differences For Demographic Variables.

 

            Demographic Variables. The first row of Table 4 shows that 8 out of ten of the analysis found similar relationships between the risk factor young age and partner assault, for men and women. The next row shows that 9 of the 12 analysis of the relationship between belonging to a racial or ethnic minority group and assaulting a partner show similar relationships for men and women. Row four shows that 75% of the nine analysis of being of a minority racial or ethnic group found a similar relationship for men and women.

             

            SES and Financial Variables. Over half of the studies that looked at low SES or financial stress found similar relationships for males and females. While half of studies on unequal income show a similar relationship to partner assault for males and females.

 

            Relationship Status. Four of the five studies found a similar relationship between partner assault and cohabiting for males and females. In dating relationships, one of two studies found a similar relationship between seriousness of the relationship for men and women, while two of the five studies of behavioral involvement (e.g. amount of time spent together, number of different types of activities shared) found a similar relationship for men and women.

 

            Violence in Family of Origin. The majority of studies of violence between parents (8 of 10), mother to father violence (1 of 2) and father to mother violence (2 of 3) show similar relationships for males and females. Additionally, half of the 14 studies on the effects of corporal punishment found similar relationships for men and women.

 

            Conflict Related Variables. All of the four studies of the link between relationship conflict and partner assault found similar relationships for males and females. Similarly, verbal aggression towards a partner was similarly associated for males and females. However, only two of the five studies on verbal aggression by a partner found a similar relationship for males and females.

 

            Self-Defense Related. In five of the seven studies similar numbers of men and women cited self defense as a reason for assaulting their partner. Only one of five studies looking at assaults in response to a partner’s assaults or in retribution for a partner’s assault found similar relationships for men and women. All of the seven studies which looked at the relationship between physical assaults by partners and assaulting a partner found similar relationships for men and women.

 

            Control. Controlling or jealous behavior, success in controlling one’s partner, and threats made towards a partner were each similarly associated for men and women in one of two studies.  Dominance was linked to partner assault in four of five studies.

 

            Jealousy. Jealousy was similarly associated for men and women in two of four studies. Two studies examined the relationship between desire for an exclusive relationship and assaulting one’s partner, neither of which found the same relationship for men and women.

 

            Substance Use. Half of the 16 studies of the relationship between substance use and partner assault found similar relationships for men and women.

 

            Personality of Perpetrator. A similar relationship between prior aggressive behavior and assaulting a partner for men and women was found by only two of the ten studies which examined the relationship.  Three of four studies found a similar relationship between having an angry personality and assaulting a partner for men and women. All of the six studies which looked at anger or provocation as a reason for partner assault found a similar relationship for men and women. Four of the six studies which looked at the association between various other personality traits and partner assault found similar relationships for men and women.

 

            Gender Hostility. Of the two studies which looked at gender hostility, one found a similar relationship between gender hostility and partner assault for men and women. Traditional sex role beliefs were found to be similarly associated with partner assault by men and women in one of two studies.

 

Traditional Sex Role Beliefs. Controlling or jealous behavior, success in controlling one’s partner, and threats made towards a partner were each similarly associated for men and women in one of two studies. Jealousy was similarly associated for man and women in two of four studies. Dominance was similarly linked to partner assault in four of five studies. Two studies examined the relationship between desire for an exclusive relationship, and assaulting one’s partner, neither of which found the same relationship for men and women.

 

 

Conclusion

Although this review of previous research revealed a number of differences between men and women in risk factors for intimate partner violence, considerably more of the risk factors examined were found to apply to both genders.  Thus, statements that the etiology of domestic assaults is different for women need to refer to specific risk factors rather than stating or implying that the etiology of partner violence is generally different for women.  However, with the presently available research it is difficult to identify with confidence which of the risk factors apply to violence by both men and women, and which tend to be important for one gender but not the other. The research to be described in the remainder of this chapter is intended to contribute to the body of additional information that will eventually enable the shared and unique variables to be identified.

 

                                                                       METHOD

 

Sample

The sample consists of 566 undergraduate students (232 men and 334 women) from two  universities who were enrolled in sociology courses or the introductory psychology course in the  1998, 1999, and 2000.  The sample was restricted to those who were or had been in a heterosexual romantic relationship of a month or longer.

 

Data Collection

            The questionnaire booklet consisted of:  (1) A cover sheet explaining the purpose of the study and the participant’s rights, and providing the name of a contact person and telephone number for those who might have questions after the test session was over. (2) The demographic questions. (3) The instruments described in the measures section.

 

            The data were gathered using procedures reviewed by and approved by the university board for protection of human subjects.  The sociology class participants were tested during a classroom period.  The purpose, task demands, and rights were explained orally as well as in printed form at the beginning of each test session.  Participants were told that the test session would involve answering questions concerning attitudes, beliefs, and experiences they may have had.  They were assured of the anonymity and confidentiality of their responses and they were told that the session would take about an hour. In practice, most students completed the survey in 40 to 45 minutes.  The psychology class participants were tested in groups of 20-30.  They were asked to sign written consent forms before completing their questionnaires.  After receiving directions about using the machine scored answer sheets, they worked at their own pace.  A debriefing form was given to each participant as they left the testing room.  It explained the study in more detail and provided names and telephone numbers of area mental health services and community resources such as services for battered women.  Students from the Psychology subject pool received two credits toward the fulfillment of their introductory psychology course research requirement for their participation.

 

(Insert Table 5 about here)

 

Table 5 shows that 40% of the respondents were in their first year at the university. The median age was 20, but some older students were included.  Over half were of European descent (62%).  Their socioeconomic background was relatively high: about two thirds of their parents had had at least some college education and the median income category was $50,000-59,999.

 

Relationship characteristics.  Almost all the students described a dating relationship (90%) as opposed to more committed relationships.  The majority (88%) were not living or had not lived with the partner they were describing.  However, sexual relations were reported to be part of the relationship by 76%. About half the sample had been in the relationship described for between one and 11 months, and slightly more than half the sample described a current relationship, and the remainder described a previous relationship.

 

Gender differences. The male and female students were similar in year in school, median age, in the proportion of parents with at least some college education, median family income, percent dating, percent cohabiting and percent sexually active. Women were more likely to identify themselves as European American, while men were more likely to identify themselves as other.  About the same percentage of men than women reported their mothers had attained a four-year college degree or graduate education.  The women had been in their relationships slightly longer than the men, and more, were in the relationship they described at the time of testing.  The fact that more women than men had relationships of a year or more, and the fact that more of them reported on current relationships may affect both the nature and frequency of experiences and behavior reported and memory for events that took place between respondents and their partners.

 

Measures

            Partner-Assault. The revised Conflict Tactics Scales or CTS2 (Straus et al. 1996) was used to measure physical assault by the respondent.  The CTS has been used in over a hundred studies of both married and dating partners in the past 25 years and there is extensive evidence of reliability and validity  (Archer 1999; Straus 1990).  The three core scales of the CTS2 measure Negotiation, Psychological Aggression, and Physical Assault.  The Physical Assault scale includes subscales for “minor” and “severe” assaults.  The Minor Assault scale includes acts such as slapping or throwing something at the partner.  The Severe Assault scale includes acts such as punching and choking.  The difference between the minor and severe subscales is analogous to the legal categories of simple assault and aggravated assault. See (StrausStraus 1990 and Straus) for a complete list of the CTS questions and for data on validity and reliability.

 

Risk Factor Variables.  The Personal And Relationships Profile (PRP) was used to obtain the data on risk factors.  The PRP is a 21 scale instrument designed for research on partner-assault.  The variables measured by the PRP scales were selected on the basis of a review of the theoretical and empirical literature on the etiology of intimate partner violence.  The steps used to develop the PRP, along with data on reliability and validity, are given in (Straus et al. 1999; Straus and Mouradian 1999).  The instructions for the PRP ask the respondent to indicate whether they agree or disagree that the statement describes themselves, using the following response categories: Strongly = 1, Disagree = 2, Agree = 3, and Strongly Agree =4.  Some of the scales in the PRP measure personal or intra-psychic characteristics of the respondent and some measure characteristics of the relationship as reported by the respondent.

 

The personal characteristics scales are: Antisocial Personality, Borderline Personality, Criminal History, Depression, Gender Hostility,  Neglect History, Post-Traumatic Stress Disorder, Social Desirability, Social Integration, Substance Abuse, Stressful Conditions, Sexual Abuse History, and Violence Approval.

 

The relationship characteristics scales are: Anger Management, Communication Problems, Conflict, Dominance, Jealousy, Negative Attribution, Relationship Commitment, Relationship Distress.

 

Social Desirability Scale.  Criminological research that uses self-report data needs to take into account defensiveness or minimization of socially undesirable behavior.  We did this by using the Social Desirability scale of the PRP.  This is a 13 item version of the widely used Crown-Marlow social desirability scale developed by Reynolds (Reynolds 1982).  The scale measures the degree to which respondents tend to avoid disclosing socially undesirable behavior such partner-assault and other crime.  The need for such a control is indicated by the fact that the Social Desirability scale has a correlation of -..20 with the Partner-assault scale. The correlations of the Social Desirability scale with the PRP scales with

 

Socioeconomic Status Scale.  This scale was created by factor analysis of the education of the respondent’s father and mother (each with a possible score of 1-7) and family income (with possible scores of 1-9). The results indicated a single factor explaining ?? percent of the variance.  The factor score for this factor constitutes the SES scale for this study.  It has a mean of approximately zero and a standard deviation of approximately 1.

 

Data Analysis

We used partial correlation (alpha .05, one-tailed tests) to examine the extent to which the risk factors measured by the PRP are associated with assaulting a partner during the previous 12 months.  The partial correlations controlled for the respondent’s scores on the social desirability response scale and the SES scale.  The correlations were computed separately for male and female respondents.  The correlations were first computed using Minor Violence as the dependent variable and then using Severe Violence as the dependent variable.

 

Transformations.  Five of the PRP Scales were moderately to strongly skewed (defined as a skewness statistic greater than 0.5).  To examine the effect of this skew on the correlation coefficients, three of the variables with skewed distributions were selected for transformations. Two types of transformations were used:  power transformations, and normalized STANINE scores.  Power transformations were done by applying powers of increasing strength until the skewness statistic for the distribution was close to zero. The skewness statistic for the Criminal History Scale was reduced from 1.471 to 0.027 by a power transformation of –2.75 (X -2.75 ). Similarly the Depression Scale’s skewness was reduced from 0.605 to 0.056 using a power transformation of 0.25 (X 0.25), and the Neglect History Scale’s skewness statistic was reduced from 0.828 to 0.028 with a power transformation of –0.5 (X –0.5 ).

 

STANINE transformations create a normally distributed variable with nine categories and a mean of 5 and a standard deviation of 2.  First the variables were rank ordered, then the values assigned by rank ordering were transformed into standard normal variables (z-scores). The z-scores are then grouped into the nine categories, with the lowest negative scores assigned a value of 1 and the highest positive scores assigned a value of 9.

RESULTS

 

Prevalence Rates

            Many previous studies of dating relationships have found that between 25 and 45% assaulted their partner in the previous 12 months (Stets and Straus 1989; Sugarman and Hotaling 1989).  The 33% rate for this sample is consistent with those studies.  Most assaults against dating partners fall within the “minor violence” category and that is also the case for this sample.  22 percent of the sample restricted their violence to minor assaults, and 11% engaged in one or more of the acts in the Severe Assault scale.

 

The rates for men and women are also consistent with many previous studies of violence in dating relationships in that the rates for men and women are about the same:  For Minor violence the rates are 22%  for both men and women.  For Severe Violence the rates are 10% for men and 11% for women.

 

(Insert Table 6 about here)

 

Extent of Concordance In Risk factors

            The first two columns of Table 6 give the partial correlations of each of the PRP scales with minor assaults for men and women, and the second two columns give the correlations of the PRP scales with severe assaults for men and women.

 

            Minor Assaults.  The correlations in the first two columns of Table 6 show that 14 of the 21 PRP scales were significantly correlated with Minor Violence by either male or female respondents, or both.  Half of these 14 were significantly related to minor assaults for both women and men.  Five (36%) were correlated only for men, and 2 (14%) were correlated only for women.  There were no instances of a significant relationship in the opposite direction for men and women.

 

            Severe Assaults. The two right hand columns of Table 6 show that 14 of the PRP scales were significantly correlated with severe assaults on a dating partner.  Eight of these, or 57%, were significantly related for both men and women; four or 29% were correlated only for men, and two (14%) were significantly correlated only for women.  As was true for minor violence, for severe violence there was no instance of a significant relationship in the opposite direction for men and women.

 

Risk Factors For Both Men and women

 

The following PRP variables were associated with Minor Violence by women as well as by men:  Anger management, Antisocial Personality, Borderline Personality, Conflict with the partner, Communication Problems, Dominance, and Negative Attribution about the partner.

 

The following PRP variables were associated with Severe Violence by women as well as by men: Anger Management, Antisocial Personality, Criminal History, Conflict with partner, Communication Problems, Dominance, and Negative Attribution about the partner and Neglect History.

 

            Altogether, of the 18 risk factors found to be related to either Minor or Severe assaults, 9 or half applied to the women as well as the men in this sample.

 

Risk Factors Unique To Men

 

            The first of the PRP scales in Table 6 that is a risk factors unique to men is Gender Hostility (eighth row of Table 6).  For men, this variable was scored to indicate hostility to women, and for women it was scored to indicate hostility to men. (for scoring method, see\(Straus et al. 1999).  This row shows that for men Hostility To Women was correlated with minor assaults.  However, for women Hostility To Men was not correlated with assaults on male partners.  The correlations in the two right hand columns of this row show that Gender Hostility was not correlated with severe assault by either men or women.

 

            The next PRP scale in Table 6 that is significantly related to violence by men but not women is Post Traumatic Stress Syndrome.   This is correlated with Minor Violence but there is no relationship between PTSD and Severe Violence for either men or women in this sample.

 

            Moving down Table 6, the row for Sexual Abuse History shows that it is related to both Minor and Severe Violence by men, but not related to either level of violence by women. This was surprising to us, and we do not have an explanation for why a history of sexual abuse is related to assaults by men but not by women.

 

            The Stressful Conditions scale, which measures the extent to which the respondent is under ongoing or chronic stress, is related to minor violence by men, but not by women.  On the other hand, the Stressful conditions scale is significantly related to Severe Violence by both men and women.

 

            Finally, the next to the last row of Table 6 shows that the Violent Attitudes scale is related to both Minor and Severe Violence by men, but not to either level of violence by women.

 

Risk Factors Unique To Women

 

            The correlation in the tenth row of Table 6 between the Jealousy scale and severe violence is significant for women, but not for men. While the Jealousy scale is associated with increased minor assaults by men.

           

            The next of the PRP variables to be related to Minor Violence only for women is Neglect History.  However, although this relationship to Minor violence is unique to women, the correlations in this row, in the columns for Severe Violence, shows that Criminal History is related to sever assaults by men as well as women, and it fact, the relationship to Severe Violence is stronger for men than women.

 

            Relationship Distress was found to be related to Minor Violence by women, but not to Minor Violence by men.  For both men and women, the correlation with Severe Violence was not strong enough to be statistically significant.

 

            Social Integration has a statistically significant relation to assault by women, but is not related for men.  This applies to the relation of Social Integration to both Minor and Severe Violence.

 

DISCUSSION

 

            The first part of this chapter reviewed research that provided empirical data that permitted comparing risk factors for intimate partner violence by men and women.  The second part of the chapter reported the results of a new study comparing 21 risk factors for men and women.  Both the previous research and the research reported for the first time in this chapter found that, although there are some risk factors that are unique to men and some that are unique to women, depending on which set of comparisons were tabulated, from 43 to 73% of the risk factors majority apply to both men and women.  Only four were found to be correlated with violence by men but not correlated with violence by women (Hostility to Women, Post-Traumatic Stress Symptoms, Stressful Conditions, and Violence Approval).  Only three were correlated with violence by women, but not correlated with violence by men (Neglect History and Social Integration, and Violent Socialization).  Thus, few of the risk factors investigated for this article are unique to either men or women

 

Limitations

 

            Although there is considerable agreement between the previous research and the new study reported in this chapter in respect to the proportion of risk factors that apply to women as well as men, there is little overlap between the specific variables in the current study and those investigated previously.  Thus, results reported need to be confirmed by other research.

 

            Caution is also needed because, as noted in the introduction, the results reported in this chapter are based on cross sectional data and may not reflect a cause-effect relationship between the risk factors studied and partner assault.

 

            Finally, although the conclusions are based on correlations that meet the criterion of statistical significance, the size of the correlations is from .10 to .23.  It is conventional to think of correlations in the .20 range as not very large.  However, correlations of this size represent an effect size that is typical of studies of risk factors for partner violence, and for that matter typical of the effect size for any one independent variable in social psychological research.  Moreover, these effect sizes take on added importance because, unlike most previous research on risk factors for partner violence, they are partial correlations which controlled for social desirability response set.  The correlations without these controlling for social desirability response set are somewhat larger.  Most important, a correlation of .20 can indicate an effect size that has considerable theoretical and practical importance (McCartney and Rosenthal 2000).  For example, the correlation between Communication Problems and partner assault in this sample is about .20 for both men and women and for both minor and severe assaults.  If low communication skills is a cause of partner violence, a correlation this size indicates that an intervention program that enables participants to change from the high to the low half of the distribution, would increase the proportion who were subsequently successful in avoiding violence from 40% to 60%.

 

Conclusion

 

            Interpreting these results is like trying to decide whether to describe a glass as half full or half empty.  Of course it is both, and both aspects need to be emphasized.  For example, 57% of the risk factors we found to be correlated with severe violence against a partner apply to both men and women.  This indicates considerable overlap in the risk factors for partner violence for men and women.  At the same time, the fact that four of the PRP scales were related to severe violence only for men, and three were related to severe violence only for women, suggests that there are also aspects of the etiology of partner violence that may be unique or particularly salient for one gender.  If these results are confirmed by other studies, intimate partner violence prevention or treatment programs need to be constructed in aware of both the ways in which the risk factors are similar for both men and women, and the ways in which they are different.   This, of course assumes that there are prevention and treatment programs for women as well as men, which is an important next step in ending intimate partner violence.

 


 

 

Table 1: Studies with significance tests for gender differences.

Factor

No significant difference

Male higher

Female higher

Study

#

Study

#

Study

#

Violence used for self-defense

(Dixon, Roth and Cauthen ) (Follingstad et al. 1991) (Cascardi and Vivian 1995) (minor and severe) (Harned 2001)

5

(Barnett, Lee and Thelen 1997)

1

(Felson and Messner 1998) (h)

1

In response to physical attack by partner

 

 

(Follingstad et al. 1991)

1

(Felson and Messner 1998) (h) (Watson et al. 2001) (Stets and Straus 1990)

3

Physical aggression by partner

(O'Keefe and Treister 1998)

(White, Merrill and Koss 1999)

2

 

 

 

 

Unable to express self verbally/letting feelings out

 (Barnett, Lee and Thelen 1997)  (Follingstad et al. 1991)(unable to express verbally) (Harned 2001)

3

 

 

(Follingstad et al. 1991) (to show anger)

1

Anger/coercion

(Cascardi and Vivian 1995) (minor and severe)

2

 

 

 

 

Anger/jealousy/punish partner

 

 

 

 

(Harned 2001)

1

Anger only

(Cascardi and Vivian 1995) (minor and severe)

2

 

 

 

 

Provocation

(Cascardi and Vivian 1995) (minor and severe)

2

 

 

 

 

Don’t know

(Cascardi and Vivian 1995) (minor and severe)

2

 

 

 

 

To feel more powerful

(Harned 2001)

1

 

 

 

 

To get control over partner

(Harned 2001)

1

 

 

 

 

To prove love

(Harned 2001)

 

 

 

 

 

Showing your partner who was boss

(Barnett, Lee and Thelen 1997)

1

 

 

(Follingstad et al. 1991)

1

Teaching your partner a lesson/Punish your partner

(Barnett, Lee and Thelen 1997) (Follingstad et al. 1991)

2

 

 

 

 

In retaliation for emotional pain

 

 

 

 

(Follingstad et al. 1991)

1

To get your own way

 

 

(Barnett, Lee and Thelen 1997)

1

 

 

Because it was sexually arousing

(Follingstad et al. 1991)

(Harned 2001)

2

 

 

 

 

Jealousy

(Cascardi and Vivian 1995) (minor and severe)

2

(Langan and Dawson 1995)(h) (Follingstad et al. 1991)

2

 

 

Personality of perpetrator

(Cascardi and Vivian 1995) (minor and severe)

2

 

 

 

 

Function: Stress

(Cascardi and Vivian 1995) (minor and severe)

2

 

 

 

 

Teasing your partner, just playing around

(Barnett, Lee and Thelen 1997)

1

 

 

 

 

Trying to upset your partner emotionally

(Barnett, Lee and Thelen 1997)

1

 

 

 

 

Trying to get your partner’s attention

(Barnett, Lee and Thelen 1997)

(Harned 2001)

2

 

 

 

 

Unaware of your intentions

 

 

(Barnett, Lee and Thelen 1997)

1

 

 

Race

(Zlotnick et al. 1998) (White, Merrill and Koss 1999)

2

 

 

 

 

Help from friends/family living outside the home

(Zlotnick et al. 1998)

1

 

 

 

 

Justification for partner violence scale (mean score)

 

 

 

 

(Follingstad et al. 1991)

1

Prior aggressive behavior

(White, Merrill and Koss 1999)

1

 

 

 

 

Offender has no violent record

 

 

 

 

(Felson and Messner 1998) (h)

1

Drug use

 

 

(Langan and Dawson 1995) (h) (At time of homicide and in the past)

1

 

 

Alcohol use

(White, Merrill and Koss 1999) (Cascardi and Vivian 1995) (severe)

2

(Langan and Dawson 1995) (h) (at time of homicide) (Cascardi and Vivian 1995) (minor)

2

 

 

Subjective intensity of anger experienced           

(Follingstad et al. 1991)

1

 

 

 

 

Disposition to experience anger                       

(Follingstad et al. 1991)

1

 

 

 

 

Frequency of experiencing anger

(Follingstad et al. 1991)

1

 

 

 

 

History of mental illness

(Langan and Dawson 1995)(h)

1

 

 

 

 

Depression

(Zlotnick et al. 1998)

1

 

 

 

 

Social activity

(Zlotnick et al. 1998)

1

 

 

 

 

Relationship conflict

(O'Keefe and Treister 1998)

1

 

 

 

 

Seriousness of relationship

(O'Keefe and Treister 1998)

1

 

 

 

 

Age

(White, Merrill and Koss 1999)

1

 

 

 

 

Marital status

(White, Merrill and Koss 1999)

1

 

 

 

 

Educational level

(White, Merrill and Koss 1999)

1

 

 

 

 

Geographic location of Primary childhood residence

(White, Merrill and Koss 1999)

1

 

 

 

 

Witnessing violence between parents

(White, Merrill and Koss 1999)

1

 

 

 

 

Physical punishment as a child

(White, Merrill and Koss 1999)

 

 

 

 

 

Childhood sexual abuse

(White, Merrill and Koss 1999)

1

 

 

 

 

Hostility towards other gender

(White, Merrill and Koss 1999)

1

 

 

 

 

Angry/impulsive personality

 

 

 

 

(White, Merrill and Koss 1999)

1

Financial work stressors

 

 

(Cascardi and Vivian 1995) (minor and severe)

2

 

 

Marital stressors

(Cascardi and Vivian 1995) (minor and severe)

2

 

 

 

 

Family/child stressors

 

 

 

 

(Cascardi and Vivian 1995) (minor and severe)

2

Health stressors

(Cascardi and Vivian 1995) (minor and severe)

2

 

 

 

 

Multiple stressors

(Cascardi and Vivian 1995) (minor and severe)

2

 

 

 

 

Other stressors

(Cascardi and Vivian 1995) (minor and severe)

2

 

 

 

 

Verbal aggression by partner

 

 

 

 

(White, Merrill and Koss 1999)

1

Parent’s SES

(White, Merrill and Koss 1999)

1

 

 

 

 

(h) indicates studies of spousal homicide

Highlighted categories were not included in Table 5


Table 2: Studies providing data for males and females, but no tests of significance for gender difference.

Factor

Male ns / Female ns

Male + / Female +

Male - / Female -

Male + / Female -

Male - / Female +

Male ns / Female +

Male ns / Female -

Male + / Female ns

Male - / Female ns

Interparental aggression

 (Stets and Pirog-Good 1990) (minor and  severe violence)

(Riggs, O'Leary and Breslin 1990) (Marshall and Rose 1990) (ever and in past 2 years) (White, Merrill and Koss 1999)

 (Kalmuss 1984)

 

 

 

 

 

(Stets and Pirog-Good 1989) (minor and severe)

 

Mother to Father aggression

(Avakame 1998)

 

 

 

 

(Straus, Gelles and Steinmetz 1980)

 

 

 

Father to mother aggression

 

(Straus, Gelles and Steinmetz 1980) (Straus and Yodanis 1996)

 

 

 

(Avakame 1998)

 

 

 

Corporally punished or physically abused by parents as a child

 (Stets and Pirog-Good 1990) (minor and severe violence)  (White, Merrill and Koss 1999) (Stets and Pirog-Good 1989) (severe)

(Kalmuss 1984) (parent to child hitting)

 

 

 

(Moffitt and Caspi 1999) (Stets and Pirog-Good 1989) (minor)

 

 (Marshall and Rose 1990) (ever and in past 2 years)

 

Corporally punished or physically abused by FATHER as a child

 

 

 

 

 

(Straus, Gelles and Steinmetz 1980) (at age 13+) (Straus and Yodanis 1996) as a teenager

 

(Avakame 1998)

 

Corporally punished or physically abused by MOTHER as a child

(Avakame 1998)

(Straus and Yodanis 1996) as a teenager

 

 

 

 

 

 

 

Interparental aggression and aggression by parents towards child

(Ryan 1998)

(Malone, Tyree and O'Leary 1989) (premarital, 6 and 18 months)

 

 

 

 

 

 

 

Aggression as a child

(Riggs, O'Leary and Breslin 1990)

 

 

 

 

 

 

 

 


Born to an unmarried mother

(Moffitt and Caspi 1999)

 

 

 

 

 

 

 

 

 

Parents’ SES

 

 

 

 

 

 

 

(Giordano et al. 1999) (higher score=lower SES)

(Moffitt and Caspi 1999) (at birth, 7 and 9, and 15)

Negative mother-child interaction at age 3

(Moffitt and Caspi 1999)

 

 

 

 

 

 

 

 

Difficult temperament at ages 3 and 5

(Moffitt and Caspi 1999)

 

 

 

 

 

 

 

 

Mother’s mental health problems at age 7 and 9

(Moffitt and Caspi 1999)

 

 

 

 

 

 

 

 

Conduct problems at ages 7 and 9 (reported by teachers and parents)

 

 

 

 

 

 

 

(Moffitt and Caspi 1999)

 

Family conflict at ages 7 and 9

 

 

 

 

 

(Moffitt and Caspi 1999)

 

 

 

Mother’s mental health problems at age 15

(Moffitt and Caspi 1999)

 

 

 

 

 

 

 

 

One parent absent at age 9

 

 

 

 

 

 

 

(Moffitt and Caspi 1999)

 

One parent absent at age 15

 

(Moffitt and Caspi 1999)

 

 

 

 

 

 

 

Composite of family socioeconomic resources

(Moffitt and Caspi 1999)

 

 

 

 

 

 

 

 

Composite family relations

(Moffitt and Caspi 1999)

 

 

 

 

 

 

 

 

Parent-child attachment at age 15

 

(Moffitt and Caspi 1999)

 

 

 

 

 

 

 

Family conflict at age 15

 

 

 

 

 

(Moffitt and Caspi 1999)

 

 

 

Adolescent identity as “troublemaker”

(Giordano et al. 1999)

 

 

 

 

 

 

 

 

Aggressive delinquency by age 15

 

(Moffitt and Caspi 1999)

 

 

 

 

 

 

 

Juvenile Police contact by age 15

 

(Moffitt and Caspi 1999)

 

 

 

 

 

 

 

Delinquent behavior as an adolescent

 

(Giordano et al. 1999) (Moffitt and Caspi 1999)

 

 

 

 

 

 

 

Composite of problem behaviors, with other risk factors controlled

 

(Moffitt and Caspi 1999) NR

 

 

 

 

 

 

 

Aggression against patents

(Malone, Tyree and O'Leary 1989) (premarital)

 

 

 

 

(Malone, Tyree and O'Leary 1989) (at six and 18 months)

 

 

 

Aggression Against peers and siblings

 

 

 

 

 

(Malone, Tyree and O'Leary 1989) (at six and 18 months)

 

(Malone, Tyree and O'Leary 1989) (premarital)

 

Combined predictors (parental aggression both towards the child and towards each other, aggression against parents, and aggression against peers and siblings)

 

(Malone, Tyree and O'Leary 1989) (premarital)

 

 

 

(Malone, Tyree and O'Leary 1989) (at six and 18 months)

 

 

 

Verbal aggression from partner

(Bookwala et al. 1992)

(Straus, Gelles and Steinmetz 1980)

 

 

 

 

 

(Cano et al. 1998) (studies 1 and 2)

 

Verbal aggression towards partner

 

(Bookwala et al. 1992) (Ryan 1998) (Straus, Gelles and Steinmetz 1980) (Cano et al. 1998) (studies 1 and 2)

 

 

 

 

 

 

 

Threatens partner

 

(Ryan 1998)

 

 

 

 

 

 

 

Physical aggression by partner

(Marshall and Rose 1990)  (ever and past 2 years)

(Bookwala et al. 1992) (Cano et al. 1998) (studies 1 and 2)

 

 

 

 

 

 

 

Acceptance of physical aggression

(Stets and Pirog-Good 1990) (severe and minor) (Cano et al. 1998) (studies 1 and 2)

(Straus and Yodanis 1996)

 

 

 

 

 

(Stets and Pirog-Good 1989) (minor and severe)

 

Controlling/jealous behavior towards partner

(Stets and Pirog-Good 1990) (severe violence)

(Cano et al. 1998) (study 2) (Stets and Pirog-Good 1989) (severe violence)

 

 

 

(Cano et al. 1998) (study 1) (Ryan 1998)

 

(Stets and Pirog-Good 1990) (minor violence) (Stets and Pirog-Good 1989) (minor)

 

Success in controlling partner

(Stets and Pirog-Good 1990) (severe violence)

 

 

 

 

 

 

(Stets and Pirog-Good 1990) (minor violence)

 

Controlling/jealous behavior by partner

(Cano et al. 1998) (study 2)

 

 

 

 

(Cano et al. 1998) (study 1)

 

 

 

Partner makes own decisions

(Ryan 1998)

 

 

 

 

 

 

 

 

Dominant personality

(Riggs, O'Leary and Breslin 1990)

 

 

 

 

 

 

 

 

Wife dominant power

 

(Straus, Gelles and Steinmetz 1980)

 

 

 

 

 

 

 

Husband dominant power

 

(Straus, Gelles and Steinmetz 1980)

 

 

 

 

 

 

 

Wife dominant in family decisions

 

(Straus, Gelles and Steinmetz 1980)

 

 

 

 

 

 

 

Husband dominant in family decisions

 

 

 

 

 

 

 

(Straus, Gelles and Steinmetz 1980)

 

Percent shared decisions

 

 

(Straus, Gelles and Steinmetz 1980)