CHAPTER SIX (Part 3)

The Diathesis-Stress Model of Partner Abuse

(Discussion:  ...continued)

The diathesis-stress model of partner abuse that guided this research hypothesized that constitutional risk factors which form an underlying vulnerability (the diathesis) would interact with environmental risk factors (the stress) to explain the perpetration of current partner abuse by males and females.  Although the patterns of results differ for males and females, the findings emerging from the logistic regressions provided support for the application of the diathesis-stress model of partner abuse in this research. Whereas the life stress events interaction model provided the best explanation of current perpetrated partner abuse by males, the alcohol interaction model did the same for females.

The limited explanatory power of the alcohol interaction model in predicting current perpetrated partner abuse by males may be related to the loss of heavy drinkers through attrition.  It is possible that had these respondents participated in Wave 2 of this study, an increase in the explanatory power of the alcohol interaction model would have been realized.  Developing strategies to overcome problems associated with attrition remains an ongoing challenge to social science researchers.

The limited explanatory power of the stress interaction model in predicting current perpetrated partner abuse by females may in part be related to the type of items included in the stress scale employed in this research.  The inadequacy of stress measures has also been raised earlier in this paper and in other discussions on partner abuse (Marshall & Rose, 1990; Seltzer & Kalmuss, 1988). Sex differences in the measurement of stress has been central issues in these discussions.

Of the twelve stress items assessed in this study, only two were found to differ significantly with respect to the proportion of females reporting partner abuse.  Males on the other hand, were found to significantly differ on five stress items.  It is possible that the measures assessed in this study are more relevant to males than they are to females.  As noted previously, most of the items included in the measure are in some way related to individuals' economic functioning (i.e., getting fired, starting work, retiring, someone moving into the household, having a baby).  A stronger effect for both males and females might have been achieved had events that take into account the different dimensions of mens' and womens' lives been included in the scale (i.e., childrearing, division of household labour, balancing home and work, illness, vacations, legal problems).

The testing of the diathesis-stress model of partner abuse in this study has been very useful.  It has pointed out the possible risk factors that differentiate male and female perpetrators of current partner abuse.  The longitudinal component of the model provided some interesting insights into the issue of partner abuse.  For example, the findings that both high scores on psychoticism and neuroticism predict current perpetrated partner abuse among females supports the stability of personality over time.  The odds ratios provided by the stress and violence in the family of origin predictors in this study are higher than those reported by Seltzer and Kalmuss (1988).

Yet, this model like others, is not without its limitations.  The real value of this model of partner abuse is that it serves as a building block to direct future research, and contributes to a better understanding of the circumstances surrounding the perpetration of partner abuse by males and females.

Implications

The results of this study have demonstrated that the problem of partner abuse is multifaceted and as a result, requires a complex strategy to assess its causes.  The equivalent rates of current perpetrated partner abuse reported for males and females confirms the findings of other researchers (Marshall & Rose, 1990; Stets & Straus, 1989; Straus et al., 1986), and beseech those involved in policy, prevention, and treatment to pay attention to all those involved in abusive relationships.

Equally relevant to program and policy implementation is the finding that indicates that of the 26.3 percent of males and 39.1 percent of females who reported having "ever" perpetrated partner abuse in Wave 1, only 7.1 percent of males and 6.6 percent of females reported perpetrating partner abuse currently.  Using the prevalence rates of partner abuse to suggest current risk for abuse not only exaggerates the problem of current partner abuse but may also lead to inappropriately designed interventions.  By focusing on the smaller percentage of males and females involved in ongoing cases of partner abuse, the limited resources currently available can be directed to those who can really benefit.

As noted previously, the results indicating that three percent of this sample required some form of medical attention as a result of a partner abuse incident, suggest that the physical consequences of partner abuse are not limited to clinical samples alone.  In the absence of research delineating the full extent of partner abuse's impact on the healthcare system or other agencies concerned with victims of partner abuse, we are left to speculate about how the physical and psychological consequences of partner abuse translate into costs associated with the use of these agencies as well as days lost on the job.  In light of the economic challenges facing governments, the issue of partner abuse seems worthy of its consideration.

The finding that self defence was not a motivation for perpetrating current partner abuse for most men and women in this general population sample suggests that researchers need to rethink earlier explanations of spousal abuse or restrict them to the clinical populations in which they were based (Dobash & Dobash, 1979, Walker, 1979).  While no one can dispute the plight of women who seek aid at battered women's shelters, this study challenges an assumption made by the shelter movement that battered women's experiences of partner abuse are strictly unidirectional.  The association between a past history of partner abuse (either through the exposure to violence in the family of origin and/or past perpetrated abuse) and current perpetrated partner abuse also suggests that the perpetration of partner abuse in the general population may simply be a form of conflict resolution established early in life and transferred to later relationships. Programs aimed at prevention and treatment need to consider the effects of past histories of abuse and make them important parts of intervention strategies.

The different profiles found for male and female perpetrators of current partner abuse have important implications for how partner abuse prevention and treatment programs should be designed and carried out.  The differences found suggest that instituting broad based prevention and treatment programs for partner abusers may be ineffective in dealing with the unique problems of men and women. According to this study's findings, programs for males should focus on stress reduction and overcoming issues related to past histories of abuse, whereas programs for females should focus on past abuse issues as well, but also concentrate on problems associated with excessive drinking and learning more effective interpersonal skills. Because male and female partner abusers have adopted violence as a means to resolve their conflicts, the teaching of more constructive methods of communication and conflict resolution should be important components of all partner abuse programs.  However, given that this has been the first study of its kind to longitudinally examine the profiles of partner abusers, the gender differences that have emerged should be considered tentatively, and should be subjected to replication.

The exposure to violence in the family of origin is an important predictor of current perpetrated partner abuse for both males and females, and because of this, those involved in program development need to also consider the children of abusive parents as targets of intervention efforts.  This research indicates that for current perpetrators of partner abuse, the modelling effects of partner abuse begin early in life.  It is important that intervention efforts focus on identifying those at risk and intervene before partner abuse becomes a well established mode of conduct.

As suggested previously, this study's finding of recanted partner abuse reports challenge current methodologies that rely on single waves of data with no method of corroborating reports given. Recanted reports of partner abuse may be a factor associated with the high rates of case collapse (20%) experienced by prosecuting lawyers (Sinclair, 1993).  The reliance upon uncorroborated data to direct program development and policy making is also of concern.  Until accounts of perpetrated partner abuse can be validated through third party reports or official records, it is possible that efforts designed to alleviate the problem of partner abuse may be misdirected.  The results of this study indicate that researchers' earlier concerns about the reliability of couple reports of partner abuse need to be extended to include the reliability of self reports of partner abuse as well.

Limitations and Suggestions for Future Research

This study has supported some research findings while refuting others.  Much of the strength of this research lies in the methodology employed.  In spite of the problems associated with attrition and recanted reports of partner abuse, this is the first study conducted in the general population to examine longitudinal data on the socio-demographic and individual risk factors involved in the perpetration of current partner abuse by males and females. Nevertheless, it should be remembered that additional measures were added to Wave 2, and as a consequence, the models tested in each phase of this research cannot be directly compared.  Thus, any conclusions about longitudinal trends should be considered cautiously.

In the past, the stability of partner abuse has been estimated by comparing rates of reported partner abuse in unrelated studies conducted at different points in time (eg., Straus and Gelles, 1986). This research, on the other hand, assessed the prevalence of perpetrated partner abuse by analyzing partner abuse data collected at two points in time.  Yet in spite of this contribution to the family violence literature, a number of limitations (beyond the ones already discussed) have also been identified.

Earlier it was stated that self defence was not a motive of partner abuse perpetrated by males and females.  Because of this, it was suggested that cases of perpetrated partner abuse occurring in the general population may differ from those that occur in clinical populations.  While this appears to be a reasonable premise, the likelihood of differences existing between general population and clinical samples with respect to the predictors of perpetrated partner abuse have not been empirically evaluated.  In order to investigate the link between perpetrated partner abuse reported in clinical and general populations, future research conducted on general population based samples might consider investigating whether incidents of partner abuse have ever been reported to police, social services or other agencies.  Administering identical survey instruments to samples drawn from general and clinical populations would also provide an opportunity to investigate the possibility of common factors.  Through these research strategies individuals at risk for partner abuse can be appropriately targeted by policy makers and other helpers.

Although the sampling strategy employed in this study enables generalizations to be made to the general population, there are a number of restrictions inherent in the approach used.  In an earlier paper (Sommer, 1990), it was noted that individuals who were institutionalized, transient, or did not hold an MHSC number were excluded from this study.  Because of this, rates of perpetrated abuse within this segment of the population remain unknown, and any generalizations made must take this limitation into account.

Although not a limitation of the sampling technique, the sexual orientation of the respondents was not established by this research. Having conducted this research under the assumption that the sample was heterosexual, the findings reported in this study do not reflect rates of partner abuse perpetrated by gays and lesbians.  In order to fully estimate the rates and patterns of perpetrated abuse occurring in the general population, these groups should be included and identified in the sample surveyed.

This study has restricted the testing of the CTS to six of its more severe conflict tactics items.  As stated previously, had the full scale been employed, the prevalence and incidence rates of male and female perpetrated partner abuse would most likely have been higher. Because this study employed an abridged version of the CTS, the risk factors derived from the analyses conducted can only be generalized to other studies testing the same items.  Because the full CTS was not tested in this research, it is not known whether the predictors of psychological or emotional abuse (included in the full version of the CTS) would be the same as those found here.  In order to make this determination and assess the full extent of partner abuse, future research should consider including the entire Conflict Tactics measure.

The instability of a number of measures employed in this research provides yet another reason to be cautious when interpreting results and when making generalizations.  In particular, the low Cronbach's Alpha coefficients obtained by the EPQP and the MacAndrew scales suggest that respondents may not be responding consistently to the items contained in these measures.  However, the test-retest coefficients computed for these measures (.60 or greater) suggests that respondents have responded consistently across time.  What may be observed as low reliability based on obtained alpha levels may actually reflect the multidimensional nature of these particular measures.

Aside from the issue of generalizability of findings, there still exists a number of concerns left unaddressed by this research.  For example, this study did not distinguish between the following subgroups: former drinkers and current abstainers (with respect to partner abuse), partner abusers who drank during a partner abuse incident and those who did not, partner abusers whose spouses required medical attention following a partner abuse incident and those whose spouses' did not and partner abusers who perpetrated partner abuser in self defence and those who did it for other reasons.  These issues provide the basis for a number of testable hypotheses in future research.

The reliability of reports of perpetrated partner abuse has already been addressed.  It was suggested that in order to overcome this problem, the use of corroborative data would be useful.  In addition to the third party sources named before, collecting partner abuse data from the respondents' partner would be useful.  The use of couple data within a longitudinal design would not only provide the means to assess the reliability of reports, but it would also provide the opportunity to examine "couple" risk factors for partner abuse. This latter issue has not yet been investigated.

Given the evidence supporting the salience of violence observed in the family of origin and past perpetrated partner abuse on the perpetration of current partner abuse, it might be useful to broaden the age categories to be surveyed. By including teenagers at one end and elders at the other, the full spectrum of partner abuse (ranging from courtship violence to elder abuse) can be examined and its developmental sequence can be evaluated.

As noted previously, the testing of the diathesis-stress model in this research should serve as a model to guide future studies. Although not all the hypothesized relationships were supported when testing this model, it did provide the opportunity to evaluate both its strengths and limitations and to make recommendations for its improvement.  Research of the future might consider incorporating the suggestions made in the previous section and replicate its testing in other general population and clinical samples.

Conclusions

The pattern of perpetrated partner abuse and its associated risk factors were assessed in this study through the testing of longitudinal data.  The major contribution of this research to the literature on family violence rests in its demonstration that the experience of perpetrated partner abuse at a given point in time as well as across time is different for males and females.  While some of this study's findings were expected, others were not.  The discovery that many respondents recanted previous reports of perpetrated partner abuse is an example of the latter and affirms the need to employ corroborative measures in future partner abuse investigations.

At the same time, this study's findings confirm what has been reported by other writers, that the problem of partner abuse is pervasive and touches all segments of society.  In light of the equivalent incidence rates reported for perpetrated partner abuse by males and females, it is recommended that the problem of family violence be viewed as stemming from the maladaptive interactions of family members rather than the dysfunctional conduct of an individual.

The intergenerational transmission of perpetrated partner abuse reported in this study also confirms what has been suggested by many and reported by few.  The differential profiles of partner abusers attest to the need for individualized programs not only to meet the distinct needs of men and women, but also to address the diversity among people in general.

The challenge that remains is to find ways of preventing partner abuse before it begins.  Early identification of individuals at risk for partner abuse may be critical to providing effective intervention.  Once accomplished, the means to break the cycle of violence could be in hand.

Although this study has uncovered a number of issues not reported previously by other family violence researchers, others still remain to be disclosed.  As investigations into partner abuse increase, the complexity of this phenomenon continues to unfold.  The findings reported in this study at best provide but one piece to an ever growing puzzle.

Next: Appendix A