CHAPTER FIVE (part 6)

Figure 5.

     [Will be shown here when graphics files become available --WHS]
 

Figure 6.

     [Will be shown here when graphics files become available --WHS]

Figure 7.

     [Will be shown here when graphics files become available --WHS]

An examination of the antilogs emerging from the logistic regressions conducted on the female data (based on significant predictors from the main effects model) revealed the following estimates:

1. females who observed their mothers hitting their fathers have odds of perpetrating current partner abuse that are 12.514 times the odds of those who did not, other factors held constant,

2. observing parents' mutual violence decreases the odds of females perpetrating current partner abuse by a factor of .001, other factors held constant,

3. for each increase in an EPQP score, the odds of females perpetrating current partner abuse is increased by a factor of 1.333, other factors held constant, and

4. for each increase in a neuroticism index score, the odds of females perpetrating current partner abuse is increased by a factor of 1.355, other factors held constant.

Table 29.  Coefficients representing the main effects of diathesis and stress measures and their interactions on the log odds of perpetrating current partner abuse among females based on Wave 1 and Wave 2 data

 Note: Only predictors reaching a .05 level of significance or less are presented in this table.

Standard errors are reported under parameter estimates in parentheses.

* p < .05, ** p < .01, *** p < .001

(1) significant main effects for diathesis and stress variables

(2) stress interactions

(3) alcohol interactions

(4) odds ratios for significant main effects

Hypothesis Testing

Descriptive analyses revealed that the incidence of perpetrated partner abuse was 7.1 percent for males and 6.6 percent for females.
     These percentages fall short of the incidence rates reported in the literature.  Hypothesis 1 was not supported.

Descriptive analyses revealed that the prevalence of perpetrated partner abuse reported in Wave 2 was 17.3 percent for males and 27.4 percent for females.  Severe partner abuse was perpetrated by 3.3 percent of males and nine percent of females. Across the three partner abuse indices (i.e., minor abuse, severe abuse and overall abuse) , significantly more females perpetrated partner abuse than males.  Although the reported percentages of perpetrated partner abuse in Wave 2 are lower than those reported in Wave 1, the pattern of sex differences is consistent.  Based on the Wave 2 prevalence data, the most common partner abuse tactic for both males and females was "throwing or smashing something".   Wave 1 data found the most common partner abuse tactic to be "pushing, shoving or grabbing".  To the extent that the pattern of perpetrated partner abuse held with respect to sex differences, Hypothesis 2 is supported.

Pearson's correlations were conducted on the prevalence and incidence of perpetrated partner abuse (Wave 2) and three measures of violence in the family of origin for both males and females. Whereas female perpetrated partner abuse was significantly related to observing "mother hitting father" (r=.11, p < .05), this was not the case for males (r=.09).  On the other hand, the prevalence of perpetrated partner abuse and observing "father hitting mother" (r=.18, p < .001 for males and r=.12, p < .05 for females) and observing "parents' mutual violence" (r=.17, p < .001 for males and r=.14, p < .01 for females) were both found to be significant.  All associations were in the expected direction.

Pearson's correlations conducted on the incidence of perpetrated partner abuse and the violence in the family of origin measures provided mixed results. Whereas the associations between these variables were found to be significant and in the expected direction for males (observe "mother hitting father" r=.16, p < .05, observe "father hitting mother" r=.27, p < .001, and observe "parents' mutual violence" r=.27, p < .001), analyses conducted on the female data revealed no significant relationships for these same variables (observe "mother hit father" r=.07, observe "father hit mother" r=.004, and observe "parents' mutual violence" r=.03).  In light of these findings, Hypothesis 3 was supported by the male data, but was only partially supported by the female data.

Pearson's correlations demonstrated that the relationship between stress and current perpetrated partner abuse was significant and in the expected direction for males and females when testing both the unweighted stress scale (r=.21, p < .001 for males and r=.10, p < .05 for females) and weighted stress scales (r=.25, p < .001 for males and r=.12, p < .05 for females.  Based on these findings, Hypothesis 4 was supported by both the male and female data.

Analysis of variance procedures were conducted on the mean scores for perpetrated partner abuse (Wave 1 and Wave 2 prevalence scores and Wave 2 incidence scores) by drinking level for males and females.
     None of the relationships reached significance, and in general, the findings were not in the predicted direction. Hypothesis 5 was not supported.

Correlations testing the bivariate relationships between the above measures and current perpetrated partner abuse by males indicated that the latter was significantly related to the following measures:

1. witnessing father hitting mother (r=.16, p < .05), mother hitting father (r=.27, p <.001) and parents hitting each other (r=.27, p < .001)

2. perpetrating partner abuse in the past (r=.46, p < .001)

3. being unemployed (r=-.20, p < .001)

4. having a high score on the Neuroticism Index (r=.16, p <.01)

5. experiencing high levels of stress (r=.21, p < .001, unweighted stress scale; r=.25, p < .001, weighted stress scale

Logistic regression analyses tested the main effects of a diathesis-stress model that included demographics, personality, alcohol dependence, past perpetrated abuse and violence in the family of origin, stress and current alcohol consumption.  These analyses revealed that relative to the effects of other variables tested, the following emerged as significant risk factors for current perpetrated partner abuse by males:

1. young age

2. Non-Catholic

3. Past perpetrated partner abuse

4. Observing father hitting mother

5. High stress

To the extent that five of the nine hypothesized risk factors were found to be significantly related to male perpetrated current partner abuse, and that three of these were significant relative to the effects of other variables tested, Hypothesis 6 was partially supported.

Correlations testing the bivariate relationships between the above measures and current perpetrated partner abuse by females indicated that the latter was significantly related to the following:

1. perpetrating partner abuse in the past (r=.29, p < .001)

2. being young in age (r=-.11, p < .05)

3. having a high score on the EPQP (r=.12, p < .05)

4. having a high score on the Neuroticism Index (r=.14, p <.01)

5. experiencing high levels of stress (r=.10, p < .05, unweighted stress scale; r=.12, p < .01, weighted stress scale)

Logistic regression analyses tested the main effects of the diathesis-stress model that included demographics, personality, alcohol dependence, past perpetrated partner abuse and violence in the family of origin, stress and current alcohol consumption. These analyses revealed that relative to the effects of other variables tested, the following emerged as significant risk factors for current perpetrated partner abuse by females:

1. High EPQP scores

2. High Neuroticism Index scores

3. Observing mother hitting father

4. Not observing parents hitting each other

To the extent that five of the eight hypothesized risk factors were found to be significantly related to female perpetrated current partner abuse and three of these were significant relative to the effects of other variables tested, Hypothesis 7 was partially supported.

Logistic regression analyses testing the interaction effects of significant risk factors from the diathesis-stress main effects model revealed the following two significant interaction relationships emerging from the stress interaction model:

1. Past perpetrated partner abuse by stress

2. Age by stress

The salience of the main effects model was compared with both interaction models (i.e., stress and alcohol consumption).  Results of the log likelihood ratio revealed that the stress interaction model significantly improved upon the explanatory power of the main effects model (X2=21.49, d.f. 4, p < .001).  However, no significant improvement was demonstrated by the log likelihood ratio for the alcohol interaction model and the main effects model (X2=5.94, d.f. 4, n.s.).  To the extent that environmental contributions interacted with life stress events, Hypothesis 8 was only partially supported.

Logistic regression analyses testing the interaction effects of significant risk factors derived from the diathesis-stress main effects model revealed the following three significant relationships emerging from the alcohol interaction model:

1. Past perpetrated partner abuse by alcohol consumption
2. Neuroticism by alcohol consumption
3. Observing mother hitting father by alcohol consumption

Results of the log likelihood ratio revealed that the alcohol interaction model provided the best explanation of current partner abuse perpetrated by females (X2=19.76, d.f. 3, p < .001) and in so doing, improved upon the explanatory power of the main effects model.
     A comparison of the log likelihood estimates for stress interactions and diathesis-stress main effects on the other hand, failed to show any improvement of the interaction model over the main effects model (X2=4.76, d.f. 3, n.s.).  To the extent that alcohol consumption interacted with underlying vulnerabilities in the prediction of current perpetrated partner abuse by females and improved upon the explanatory power of the main effects model, Hypothesis 9 was only partially supported.

Summary of the Results

The prevalence and incidence of partner abuse perpetrated by males and females were examined in several ways.  First, the occurrence of perpetrated partner abuse and its context were explored by way of frequency data.  It was found that the pattern of perpetrated partner abuse remained stable across Wave 1 and Wave 2 of this project with a greater proportion of females perpetrating violence than males in Wave 2 (17.5% compared to 27.6% based on same samples p=.002).  In spite of the stability in the pattern of abuse, the prevalence rates of perpetrated partner abuse reported by both sexes in Wave 2 were proportionately smaller than those reported in Wave 1 (26.3% for males and 39.1% for females).  When the rates of partner abuse perpetrated during the past year by males and females were examined (7.1% for males and 6.6% for females), the difference in the reports of perpetrated partner abuse by males and females disappeared.

Of those who abused their partners during the course of their relationships, 16.0 percent of males and eight percent of females consumed alcohol at the time of the abuse incidents, 21.4 percent of males and 14.3 percent of females reported that their partners required medical attention as the result of a partner abuse incident, and 14.8 percent of males and 9.9 percent of females perpetrated violence in self defence.  Males who abused their partners during the past year had significantly higher levels of stress than females who did the same.  Finally, approximately 1/3 of male and female respondents who abused their partners observed some form of violence within the family of origin.

Correlational analyses were conducted to assess the relationships between the independent measures and past and current perpetrated partner abuse in Wave 2 as well as to assess the stability of these relationships across time.  In general, the relationships between perpetrated partner abuse and demographic and personality variables remained stable for males and females in Wave 1 and Wave 2.  However, the relationships between perpetrated partner abuse and some of the alcohol measures (i.e., alcohol consumption, MAST) lost their salience in Wave 2.  Correlational analyses limited to Wave 2 data demonstrated the following: 1) stress was significantly and positively related to the perpetration of current partner abuse by males and females using both weighted and unweighted versions of the scale, and 2) indices of violence in the family of origin were significantly and positively related to the prevalence of partner abuse perpetrated by males and females, but significantly and positively related to current partner abuse perpetrated by males only.

The final phase of analyses involved a series of logistic regressions that tested the diathesis-stress model of partner abuse. For males, having high levels of stress, observing father hitting mother, perpetrating partner abuse in the past and being young and non-Catholic were the significant risk factors that emerged from the main effects model.  Stress was also found to interact with past perpetrated partner abuse and age.  Significant risk factors for current perpetrated partner abuse by females emerging from the main effects model included observing mother hitting father, not observing parents' mutual violence and having high scores on the EPQP and neuroticism index.  It was also found that alcohol interacted with neuroticism, past perpetrated partner abuse and observing mother hitting father.

For both males and females, observing violence in the family of origin posed the greatest risk for current perpetrated partner abuse. For males who observed their fathers hitting their mothers, the risk of perpetrating current partner abuse was increased by a factor of 4.569.  The risk of current perpetrated partner abuse by females was increased by a factor of 12.514 when they observed their mothers hitting their fathers.  Whereas life stress events significantly improved upon the explanation of current perpetrated partner abuse by males, alcohol consumption was found to do the same for females.

Next: Chapter 6 Part 1