CHAPTER SIX (Part 1)

DISCUSSION

The purpose of this study was to examine the problem of male and female perpetrated partner abuse.  The stability of its occurrence as well as its associated factors were investigated through the analysis of data drawn from a random sample of adult residents in Winnipeg, Manitoba that were collected at two points in time.  A diathesis-stress model of partner abuse was used to test the individual and joint effects of environmental and underlying factors related to current perpetrated partner abuse.

The Prevalence and Incidence of Perpetrated Partner Abuse

Wave 2 of this research reported that 17.3 percent of males and 27.4 percent of females perpetrated some form of partner abuse against their intimate partners at some point during their relationships.  It was also reported that 7.1 percent of males and 6.6 percent of females were involved in the same kind of behaviour during the past year.  Both of these annual estimates of perpetrated partner abuse are below that which has been reported in the spousal abuse literature (Bland & Orn, 1986; Brinkerhoff & Lupri, 1988; Schulman, 1981; Smith, 1987; Straus & Gelles, 1986).

There are a number of factors that may explain the low incidence rates of perpetrated partner abuse reported in this study.  This study's longitudinal design may be an important one.  When comparing the prevalence rates of perpetrated partner abuse reported during Wave 1 with those of Wave 2, a 35 percent reduction in the reports of males and a 29 percent reduction in the reports of females is noted. Furthermore, the less than optimal correlations between Wave 1 and Wave 2 CTS scores suggested that some of the respondents may have changed the pattern of their partner abuse reports from Wave 1 to Wave 2.  Post hoc analyses indicated that approximately 18 percent of males and 25 per cent of females recanted previous reports of perpetrated abuse made during Wave 1.

It was also shown that mean Wave 1 CTS scores for these respondents when compared to that of overall sample means for males and females were higher for both males (x=8.04, n.s.) and females (x=8.97, p < .001).  Further analyses involving correlations between Wave 2 CTS scores and EPQL scores also suggested that for females, changes in partner abuse reports may have possibly been made in response to social pressure associated with the interview process. Sensitivity to repeated measurement as reflected by changes in responses is an issue addressed Menard (1991) and Carmines and Zeller (1979).

While some may argue that Wave 1 prevalence rates of partner abuse could be considered over-reporting, it seems unlikely that individuals would falsely respond in the affirmative to an issue as sensitive as partner abuse (Szinovacz, 1983).  The above average Wave 1 CTS scores found among female recanters, as well as the relationship found between partner abuse and social conformity, suggests that their sensitivity toward the issue of partner abuse has been enhanced.  This is consistent with research by Arias and Beach (1987) who reported that social desirability influences self-reports of physical aggression against a partner.  Moreover, females' higher rate of recanting is also congruent with Browning and Dutton's (1986) finding that when comparisons were made of the reports of males and females with respect to perpetration and victimization, females were found to under-report abuse perpetrated against their partner.

Szinovacz (1983) addressed the issue of discrepant reports of abuse by couples.  She noted that couples' inability to recall past behaviours or events in a consistent manner was a factor related to this problem.   Because a person's memory of an event (or lack of it) may be shaped by his or her present reality, events of the past often are often reconceptualized.

With respect to the respondents in this study, it is possible that the recanting of previous partner abuse (or the denial of ongoing partner abuse) may have come about in response to the recent attention given to partner abuse by the media and the Justice Department (Sommer, 1993).  However, in the absence of a measure to assess respondents' attitudes and sensitivities toward partner abuse, the above comment is one of speculation.

The finding of recanted reports of perpetrated partner abuse again raises the issue of under-reporting of partner abuse.  Estimates of under-reporting of perpetrated partner abuse by men and women are reported to range between 10 percent (O'Leary et al., 1989) and 50 percent (Szinovacz, 1983).  However, much of these discussions have centred around the validity of self reports based on the presence of sex differences in abuse reporting (Arias & Beach, 1987; Browning & Dutton, 1986; Jouriles & O'Leary, 1985; O'Leary et al., 1989; Szinovacz, 1983) and differences with the same sample.

Given that previous general population based research has relied upon "one wave" estimates of partner abuse, this research is the first to report recanted reports of abuse previously made by the same individual.  The rates of under- reporting of partner abuse demonstrated in this study fall within the range of estimates noted above.  Although the contention is that reports of partner abuse given during Wave 1 are more accurate, concerns are raised about research that relies solely upon reports of partner abuse derived from single waves of data with no means of cross validation.

A final consideration in the under-reporting of the prevalence and incidence of perpetrated partner abuse is a limitation carried over from the first phase of this study.  Unlike other random surveys using the CTS, this research only reported the more severe items included in the scale.  As a consequence, the less severe abuse items contained in the "reasoning" subscale that carry with them higher rates of disclosure (Straus et al., 1986) were not examined.  The inclusion of these items in this study would likely have resulted in higher rates of partner abuse reporting.

In spite of the problems just described, Wave 2 data attempted to overcome a number of limitations found in Wave 1.  Of those, the following are relevant to the present discussion: 1) Wave 1 reports of perpetrated partner abuse were limited to married and married, but previously divorced persons, and 2) the recency and the frequency of perpetrated partner abuse were not addressed by Wave 1.

The first or these limitations was dealt with by including the marital status category, "cohabiters" in the analyses of partner abuse data in Wave 2.  In doing so, it was revealed that cohabiters accounted for 9.38 percent of "ever" reported perpetrated partner abuse by males, and 11 percent of "ever" reported perpetrated partner abuse by females.  Moreover, it was also found that male cohabiters had significantly higher number of partner abuse incidents during the past year than other marital status categories.  These findings support the need to survey individuals across marital status groups. While Wave 2 improved upon Wave 1 by expanding the scope of its data collection, it did not gather partner abuse data from individuals who were divorced, widowed or involved in informal (i.e., noncohabiting) intimate relationships.  The retrospective information to be gained from these groups of people may provide some insight into whether abuse perpetrated during past relationships is carried on to current ones.

The latter limitation was addressed by the assessing the last time a partner abuse incident occurred and by establishing the number of times an incident occurred during the past year.  Analyses indicated that on average, the last incident of perpetrated partner abuse occurred approximately six years earlier for males and slightly more than seven years earlier for females.  Thus, for most of respondents who reported perpetrating partner abuse at some point during their relationships, its occurrence appears to be an "event of the past".

For the other 7.1 percent of males and 6.6 percent of females who reported ongoing partner abuse, they reported on average, 3.08 (males) and 3.91 (females) incidents during the past year.  These figures suggest that for this group of individuals, partner abuse is a frequently occurring event.  In light of this, and other factors still to be discussed in following sections, persons experiencing ongoing partner abuse should be the primary targets of partner abuse prevention programs.

Partner Abuse Tactics

Wave 2 prevalence reports of perpetrated partner abuse showed that overall, significantly more females perpetrated partner abuse against their partner.  A higher rate of reporting was also found for females on each partner abuse tactic. For three of these tactics, the differences were significant (i.e., "threatened to throw something", "threw something at partner" and "hit partner").  The most common tactic for males and females was "throwing or smashing something (but not at partner)".

The pattern of these Wave 2 findings is similar to those reported in Wave 1, where significantly more females reported perpetrating partner abuse overall and across all but one tactic.  Although, the most common partner abuse tactic endorsed by females remained the same during Wave 1 and Wave 2, this was not found to be the case for males.  Whereas Wave 1 males endorsed "pushing, shoving and grabbing" most often, Wave 2 "males endorsed throwing or smashing something (but not at partner)" most often.  This too, is an example of the change in the reporting of perpetrated partner abuse.  Whereas the changes in abuse reporting by females is most evident in their denial of partner abuse, for males it appears to be in the type of abuse perpetrated.  The greater endorsement of a "milder" form of partner abuse" in Wave 2 by males also supports the redefining of past behaviour based on one's present reality.

Wave 2 incidence reports of perpetrated partner abuse characterized respondents' use of partner abuse tactics much differently than previously described.  First, there was no overall difference between the reports of males and females.  Moreover, an item analysis revealed only one sex difference, whereby significantly more males perpetrated partner abuse against their partners.  When the most common partner abuse tactics were considered, "throwing or smashing something" was reported most often by males and "pushing, shoving and grabbing" was reported most often by females.

These findings based on ongoing perpetrated partner abuse, support males' and females' equal involvement in the perpetration of partner abuse and reject findings reported earlier indicating  females' greater involvement.  It is possible however, that this shift in findings may be related to respondents' recanting of previous partner abuse.  Recalling that "recanters" had above average CTS scores (indicating the endorsement of more partner abuse tactics more often), it is possible that had Wave 2 reports of perpetrated partner abuse been consistent with those of Wave 1, the sex differences and most common tactics noted in Wave 1 might still have persisted.

While it is wrong to condone any form of partner abuse, it is worthwhile noting that the most common tactics reported by males and females in Wave 1 and Wave 2 are considered among the least coercive. According to Brinkerhoff and Lupri (1988), "pushing, shoving and grabbing" was considered a normal part of family life not perceived as abuse by family members.  These partner abuse tactics also share in common a lessened likelihood for injury compared to other that involve hitting, punching or the use of weapons.  The endorsement of these tactics by individuals in the general populations may prove to be important factors differentiating them from clinical samples. This hypothesis can be tested by comparing these sampling bases while using parallel partner abuse measures.

The Context and Effects of Perpetrated Partner Abuse

Failure to explore the context of partner abuse has been a major limitation of much of the previous partner abuse research.  In an effort to overcome this serious methodological shortcoming, Wave 2 considered the following questions: 1) Was the respondent drinking at the time if an abuse incident? 2) Did the partner require medical attention as a result of an abuse incident?  3) Was abuse perpetrated in self-defence?

As noted previously, the link between alcohol consumption and partner abuse is well established in the literature.  Less understood however, is the mechanism of alcohol's effect on partner abuse.  In order to gain more insight into this relationship,  respondents were asked to report whether they had been drinking at the time of an abuse incident.  Results demonstrated that more than twice as many males (16.0%) compared to females (8%) drank at the time of an abuse incident.  This sex difference is consistent with those of Gelles (1974), Coleman and Straus (1983), Frieze and Schafer (1984) and Russell, Lipov, Phillips and White (1989) who reported that alcohol consumption was predominantly a problem of violent men.

Kantor and Straus (1987) explored this relationship and reported that 24 percent of males drank just prior to a partner abuse incident.  They also found a significant relationship between the amount of alcohol consumed and drinking prior to a partner abuse incident.  The percentages of male and female respondents reporting drinking at the time of a partner abuse incident in this present research falls short of those reported by Kantor and Straus (1987).

The disparity between other researchers' findings and the ones reported here, may in part, be explained by Wave 2's loss of heavy drinkers through attrition.  Another possible explanation could be related to the substantive difference between "drinking at the time" and "drinking just prior to" an abuse incident.  Perhaps the important issue in the alcohol/partner abuse relationship is the time needed for alcohol to be ingested and metabolized.  In other words, the stronger linkage between alcohol consumption and partner abuse reported by others may reflect reports of perpetrated partner abuse following a bout of drinking.  If this is true, the greater reporting of alcohol just prior to an abuse incident can be better understood. Replication using parallel measures is needed to clarify this issue.

The issue of a spouse or partner requiring medical attention following a partner abuse incident has not been widely researched in general population research.  This study provides the first insights into yet another dimension of this serious social problem for Canadians.  It was reported that approximately 21 percent of males' and 14 percent of females' partners required medical attention as a result of a partner abuse incident.  This represents approximately 3 percent of the total subsample of married, cohabiting and remarried males and females. This percentage is greater than that reported by Straus and Gelles (1990) based on the findings from their national survey (.4% of males and 3% of females needed to see a doctor following a violent incident).  Moreover, while Straus and Gelles (1990) found that significantly more females required medical attention, the results of this study failed to find such a sex difference.

The issue of self defence in perpetrated partner abuse has been discussed by feminist writers who have long argued that when a woman hits a man, it is usually in self defence (Pleck et al., 1978; Walker, 1979).  The findings of this study stand in the face of this argument.  This study has found that only approximately 10 percent of women and 15 percent of men perpetrated partner abuse in self defence.  In other words, for almost 90 percent of women and 85 percent of men, the perpetration of partner abuse was influenced by other factors.  According to these findings, self defence is not a common motivation for the perpetration of partner abuse for males and females in the general population.  Further inquiries into other possible reasons for the abuse are needed.

Predictors of Perpetrated Partner Abuse

The correlational and logistic regression analyses conducted in this study revealed profiles of male and female partner abusers that depart somewhat from those previously described in the literature. The following sections will discuss the underlying and situational risk factors found to be associated with the perpetration of current male and female partner abuse and point out their relationships to other research in the area.

Given that this is the first study to examine the perpetration of partner abuse across time in the general population, the degree to which its findings can be compared is greatly restricted.  Moreover, the literature's primary focus on partner abuse perpetrated by males will further impede comparisons with respect to female perpetrators of partner abuse.

Demographic Risk Factors

The literature examining the demographic characteristics of male partner abusers indicates that while violence and abuse between intimate partners occurs at every level of society (Ontario Association of Professional Social Workers, 1987; Schulman, 1981; Sommer, 1990), there are some groups of individuals who are more at risk than others.  For example, Finkelhor (1983) and Kantor and Straus (1987) argued that members of the lower social strata are more at risk.  These individuals are typically young, nonwhite, having achieved only a high school education, being of a low income bracket and more likely to be blue collar workers.

With the exception of low education and blue colour status, the correlational analyses conducted on current perpetrated partner abuse in Wave 2 support the above demographic profile for males.  It was also found that the religious preference category "other" was also found to be correlated with current perpetrated partner abuse by males.

The results of the logistic regression analyses conducted on the male data limited the salience of some of these variables. Significant predictors emerging from these analyses indicated that being young and non-Catholic were the only significant demographic risk factors of current perpetrated partner abuse by males.

Correlational analyses conducted on the female data failed to reveal any significant relationships between demographic variables and current perpetrated partner abuse by females.  Results of the logistic regression analyses likewise failed to demonstrate any significant relationships between any of the demographic variables and current partner abuse perpetrated by females.  The discussion that follows will therefore focus exclusively on the findings that emerged from the male data.  

Age

The results of correlational and logistic regression analyses examining main effects indicated that being young in age was significantly related to current perpetrated abuse by males.  This finding is consistent with other general population research in this area (Kennedy & Dutton, 1989; Schulman, 1981; Straus et al., 1980). Thus, when age is considered independent of the influences of other variables, the perpetration of current partner abuse is most likely to occur among males who are under 34 years of age.  

Religion

Being non-Catholic was found to be a significant main effect in predicting current perpetrated partner by males based on the results of the logistic regression analyses.  This finding is particularly difficult to interpret because of the limited research investigating the relationship between perpetrated partner abuse and religion. Furthermore, much of the available research on this topic is overridden by problems associated with measurement and definition. Given the likelihood that individuals who are non-Catholics belong to any number of other religious denominations,  making comments about these unknown groups is difficult.

Nevertheless, these findings do seem to suggest that the tenets of Catholicism may insulate individuals from perpetrating partner abuse compared to those adhering to non-Catholic principles. Further inquiries into rates of perpetrated partner abuse among members of non-mainstream religious groups are needed. However, given that only nine to ten percent of males and females reported membership in the "other" religious preference category, such investigations may not prove to be entirely fruitful.  An alternative to this approach may be to investigate the religious practices of respondents (i.e., church attendance, religious observance of holidays).

History of Abuse

Included under this heading are variables related to abuse within the family of origin (perpetrated by the mother, father or mutually) as well as respondents' reports of past history of perpetrated partner abuse as measured in Wave 1.

Exposure to violence within the family of origin

Compared to persons without a history of abuse within the family of origin, a significantly greater proportion of males and females exposed to some form of violence in their family of origin were involved in both past and ongoing partner abuse.  Of interest, is the finding that the greatest proportion of males and females exposed to violence in the family of origin are those who were involved in the perpetration of partner abuse at some point in the relationship.  Given that the exposure to violence in the family of origin was found to be a less salient factor in the incidence of perpetrated partner abuse, it appears that observing parents' violence may not be a sufficient factor in the perpetration of partner abuse, in general.  It is also possible that for many, exposure to more appropriate models of conflict resolution has a positive influence on how individuals resolve conflicts in their current intimate relationships.

When the exposure to violence in the family of origin was separated into observing mother hitting father, father hitting mother and mutual violence and checked for sex differences, some interesting distinctions emerged.  For example, the strongest correlations between exposure to violence in the family of origin and past and current perpetrated partner abuse were found for father hitting mother and mutual violence among male respondents.  For female respondents, correlations were weaker overall, and little distinctions were made between the type of exposure to violence in the family of origin and either past or current perpetrated partner abuse.

Based on the results of these correlations, it appears that the linkages between past and current perpetrated partner abuse and the modelling of fathers' aggression toward their mothers, as well as the modelling of parents' aggression toward each other, are more important for males than they are for females.  In the absence of measures assessing respondents' attitudes toward the perpetration of partner abuse, one may speculate that the lack of association for females may be related to differences in males' and females' perception of their parents' conflict.

Results of the logistic regressions characterize the relationship between exposure to violence in the family of origin and the perpetration of current abuse somewhat differently than what has been described above.  The magnitude of the coefficient estimate for the main effect, observing "mother hitting father" among females was found to be greater than that for the main effect, observing "father hitting mother" among males.  Moreover, while the latter increased the likelihood of current perpetrated partner abuse by males by a factor of 4.569, the former increased the likelihood of the same by a factor of 12.514.

The gender related modelling effect demonstrated for the exposure to violence in the family of origin and its association with current perpetrated partner abuse is worthy of some consideration.  While past research has supported the link between violence in the family of origin and partner abuse,  the mechanism of this linkage has been disputed (Sommer, 1993).  While some argue that the association between exposure to violence in the family of origin and the perpetration of partner abuse is gender specific  (O'Leary & Curley, 1986; Simon et al., 1993), others contend that this association is role specific (Kalmuss, 1984). This research refutes the latter claim with respect to the perpetration of current partner abuse only by demonstrating that when males and females observe the same sex parent hitting the other parent, they are at greater risk for perpetration of partner abuse in their present relationships.

Another strong predictor of current perpetrated partner abuse by females is observing parents' mutual violence.  Contrary to the previous findings, the likelihood for females' perpetrating current partner abuse is lessened by the influence of observing their parents' mutual fighting relative to other variables included in the model.  In the absence of other research to support this finding, one might speculate that parents' mutual abuse may be perceived as balanced, where neither mother or father is viewed as the sole victim or perpetrator.  On the other hand, women observing reprisals received by their mothers may provide less of an incentive to be aggressive themselves.

Next: Chapter 6 Part 2