CHAPTER FIVE (part 3)

Level of alcohol consumption

Perpetrated partner abuse rates for male and female respondents were computed for four levels of alcohol consumption.  These levels included: abstainers, and low, moderate, and heavy drinkers.  The latter three categories were based on equal cutpoints along the distribution of the alcohol consumption variable (measured in ounces consumed per day).  Separate analyses were conducted on Wave 1 and Wave 2 data using the same subjects and compared the prevalence of perpetrated partner abuse (i.e., abuse at some point during the relationship) by drinking level.  A third set of analyses examined current rates of perpetrated partner abuse (i.e., abuse occurring during the past year) by drinking level.

In general, a greater proportion of males and females reporting perpetrating partner abuse at some point during their relationships was found among heavy drinkers.  However, with the exception of Wave 2 females, the proportion of those reporting "ever" perpetrating partner abuse did not significantly differ with the amount of alcohol consumed.  In other words, significantly fewer Wave 2 females who abstained from alcohol perpetrated partner abuse at some point during their relationships compared to the partner abuse perpetrated at any  other drinking level (chi-square=14.09, p < .05).  Analyses examining the relationship between partner abuse perpetrated during the past year and drinking levels provided findings similar to those just described.  For example, the proportion of males and females reporting perpetrating partner abuse during the past year did not significantly differ with the amount of alcohol consumed.

Tables 15 and 16 summarize the results of anova analyses.  Table 15 compares the CTS means for the prevalence of perpetrated partner  abuse across levels of drinking for males and females in Wave 1 and  Wave 2 data.  Table 16 on the other hand, compares the mean number of partner abuse incidents perpetrated by males and females during the past year also across all drinking levels.  In each case, the perpetration of partner abuse did not significantly differ for males or females with respect to the amount of alcohol consumed.
     Altogether, these findings do not support the existence of a curvilinear relationship between the amount of alcohol consumed and the perpetration of partner abuse.

Table 15. Mean CTS scores for the prevalence of perpetrated partner abuse by drinking level for males and females based on Wave 1 and Wave 2 data

Note: Asymptotic Standard Errors are presented in brackets.

Table 16. Mean number of partner abuse incidents perpetrated during the past year by drinking level for males and females based on Wave 2 data

Note: Asymptotic Standard Errors are presented in brackets.

Context and effect of abuse

Frequency analyses conducted on data gathered from Wave 2 married, cohabiting and remarried males and females who reported perpetrating  partner abuse at some point during their relationships revealed the following findings with respect to the context of partner abuse episodes:  1) 16 percent of males (n=8) and 8 percent of females (n=6) drank at the time of an abuse incident,  2) 21.4 percent of males (n=14) and 14.3 percent of females (n=14) reported that their spouse required medical attention as a result of an abuse incident, and  3) 14.8 percent of males (n=9) and 9.9 percent of females (n=10) reported that their actions were in self defence.  No significant sex differences were indicated on any of these items.

Life stress events

Married, cohabiting and remarried respondents' experiences of life stress events during the past two years were assessed by analyzing 15 related survey items drawn from Wave 2 data.  Chi-squares were conducted to assess the relationship between the experience of these events and the proportion of those reporting perpetrated partner abuse.  Three items were eliminated because of insufficient cases (i.e., becoming separated, divorced and widowed during the past two years) .  Results indicated that significantly more males and females who perpetrated partner abuse during the past year had financial problems (33.3% v. 4.73% of males and 20.59% v. 5.14% of females) and stopped school (28.57% v. 6.65% of males and 18.75% v. 6.05% of females) during the past two years.  Significantly more males who perpetrated abuse during the past year lost their jobs (39.13% v. 4.91% , changed jobs or started work (18.03% v. 4.84%) and had a spouse who started work (18.03% v. 5.0%) during the past two years.
     Table 17 provides a summary of chi-square and related item analyses for males and females in Wave 2.

Table 17. Perpetrated partner abuse by stress during past two years

Note: * p < .05, *** p < .001

In order to measure respondents' overall experiences of stress, a scale was constructed combining all the items contained in the previous table.  For each item, the value "1" was assigned to a positive response and the value "0" to a negative response.  The values were then summed to provide an index in which high scores denoted high stress and low scores denoted the opposite.  Two separate scales were calculated; one using weighted factors (using the weights suggested by Holmes & Rahe, 1967) and the other using unweighted factors.

Anova results indicated that independent of weighting, a significant main effect was found for current perpetrated partner abuse such that abusers had higher stress scores than nonabusers.  A significant interaction effect was also found between gender and abuse by stress whereby males who abused their partners during the past year had higher stress levels than females who did the same (based on both versions of the stress scale) (See Tables 18 and 19).  Figures 2 illustrates the interaction effect based on the unweighted stress scale. 

Table 18. Stress by perpetrated partner abuse and gender (weighted scale)

Note: F for model=13.24, p < .0001
           * p < .05, *** p < .001

Figure 2

     [Will be shown here when the graphics file has been received. --WHS]
 

Table 19. Stress by perpetrated partner abuse and gender (unweighted scale)

Note: F for model=10.25, p < .0001
           * p < .05, *** p < .001

Violence in the family of origin

Married, cohabiting and remarried respondents were asked to indicate whether they observed violence in their families of origin. Overall, 6.41 percent of males (n=23) and 6.34 percent of females (n=22) observed their mothers hitting their fathers while 11.73 percent of males (n=42) and 12.68 percent of females (n=44) observed their fathers hitting their mothers.  It was also found that 4.44 percent of males (n=16) and 4.02 percent of females (n=14) observed both parents hitting each other.  While no significant differences with respect to respondent's gender were found, observing fathers hitting mothers was reported significantly more often than reports of mothers hitting fathers (p < .001).

Among those who reported to have perpetrated partner abuse at some point during their relationships, 34.78 percent of males and 40.91 percent of females reported having observed their mothers hitting their fathers.  On the other hand, 30.95 percent of "ever" abusive males and 38.64 percent of "ever" abusive females reported having observed their fathers hitting their mothers.  Mutual violence was reported by 37.50 percent of "ever" abusive males and 35.71 percent of "ever" abusive females.  None of these findings were significant with respect to gender.

Among those who reported perpetrating partner abuse during the past year, 17.39 percent of males and 18.18 percent of females reported having observed their mothers hitting their fathers.  On the other hand, 21.43 percent of currently abusive males and 9.09 percent of currently abusive females reported having observed their fathers hit their mothers (p < .01).  Mutual violence (i.e., observing parents hitting each other) was reported by 25 percent of currently abusive males and 7.14 percent of currently abusive females (p < .001).

Reports of observing violence in the family of origin by nonabusers were as follows: 4.97 percent of males and 4.94 percent of females observed their mothers hitting their fathers, 9.51 percent of males and 10.19 percent of females observed their fathers hitting their mothers and 3.56 percent of males and 3.70 percent of females observed their parents hitting each other.  A greater proportion of males and females who reported having "ever" perpetrated partner abuse, observed violence in the family of origin across all three indices compared to nonabusers. This pattern held true when comparing the reports of current abusers with those of nonabusers except in the case of fathers hitting mothers, where current abusive and nonabusive females provided similar reports.

Correlations Among Variables

Pearson's Correlation analyses were performed to examine the relationship between partner abuse and the following sets of variables: (1) demographic variables, (2) life stress events scale, (3) violence in the family of origin, (3) alcohol consumption and dependence measures and (4) personality measures.  In the case of demographic variables, alcohol consumption and dependence measures and personality measures, comparisons of Wave 1 and Wave 2 correlations based on same samples were also performed.

Demographic Variables and Perpetrated Partner Abuse

Correlation coefficients illustrating the relationships between perpetrated partner abuse and demographic variables based on Wave 1 and Wave 2 data are provided for males in Table 20, and females in Table 21.  Prior to conducting Pearson's Correlations, transformations had to be performed on a number of these categorical variables.  In so doing, the variables "religious preference", "race" and "employment status" were dummy coded to form the following new variables: Catholic, Protestant, other religious preference, White (nonwhite), employed (unemployed).  The variable "education status" was transformed as follows to reflect  "years of education": 1=3, 2=6, 3=9, 4=12, 5=14, 6=16, 7=18 and 8=20. Separate correlational analyses were performed on the prevalence  and incidence of perpetrated partner abuse.  For the most part, the associations between the prevalence and incidence of perpetrated partner abuse and demographic variables were weak.  However, the following significant relationships were found in Wave 2 data:

For males,

1. Age was negatively correlated with the incidence of perpetrated partner abuse
   (r=-.17, p < .01).

2. Income (i.e., total family income) was negatively correlated with the prevalence (r=-.15, p < .01) and incidence of perpetrated partner abuse
   (r=-.15, p < .01).

3. Employment status (employed v. unemployed) was negatively correlated with the prevalence (r=-.15, p < .01) and incidence of perpetrated partner abuse
   (r=-.26, p < .001).

4. Race (white v. nonwhite) was negatively correlated with the prevalence of perpetrated partner abuse
   (r=-.11, p < .05).

For females,

1. 1) Age was negatively correlated with the prevalence of perpetrated partner abuse
   (r=-.15, p < .01).

2. 2) Employment status (employed v. unemployed) was negatively correlated with the prevalence of perpetrated partner abuse
   (r=-.12, p < .05).

Z scores were computed to assess the significance in proportions between correlations at Wave 1 and Wave 2. The following are the demographic variables found to differ with respect to the perpetration of partner abuse based on prevalence of abuse data:

     For males,

1. The association between employment status (employed v. unemployed) and perpetrated partner abuse was significantly stronger in Wave 1.

2. The association between income and perpetrated partner abuse was significantly stronger in Wave 1.

3. The association between race (white v. nonwhite) and perpetrated partner abuse was stronger in Wave 1.

4. The association between Protestant and perpetrated partner abuse was significantly stronger in Wave 1.

With the exception of the relationship between race and the incidence of partner abuse, Kendal's Tau-b values approximated those obtained by Pearson Correlation coefficients.  Bonferroni T tests (p=.05) conducted on Wave 2 demographic variables with respect to the incidence of perpetrated partner abuse by males indicated that only employment status and income showed significant differences between groups.  This finding suggests that there is an increased likelihood for a Type 1 error in the associations between partner abuse and age, other religions and race.

For females,

1. The association between age and perpetrated partner abuse was significantly stronger in Wave 1.

2. The association between income and perpetrated partner abuse was significantly stronger in Wave 1.

3. The association between Catholic and perpetrated partner abuse was significantly stronger in Wave 2.

4. The association between Protestant and perpetrated partner abuse was significantly stronger in Wave 1.

5. The association between other religions and perpetrated partner abuse was significantly stronger in Wave 2.

Kendal's Tau-b values assessing the relationship between demographic variables and the incidence of perpetrated partner abuse approximated those obtained by Pearson Correlation coefficients. Bonferroni T tests (p=.05) conducted on Wave 2 demographic variables with respect to the incidence of perpetrated partner abuse by females indicated no significant differences between groups, suggesting an increased likelihood for a Type 1 error in the associations between partner abuse and demographic variables.
 

Table 20. Pearson Correlation Coefficients: Demographic variables and perpetrated partner abuse comparing Wave 1 and Wave 2 data based on the same sample of male respondents.

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

+ Z Scores were derived from the prevalence data and demographic measures measured in Wave 1 and Wave 2, respectively.

Asymptotic Standard Errors are provided in brackets.

Partner abuse was based on the full measure prior to any transformations being conducted.

Next: Chapter 5 Part 4