Attitudinal Change in Participants of Partner Assault Response (PAR) Programs: A Pilot Project

3. RESULTS

3.1 Factor analysis of men's attitudes

3. Results

Results are reported first on the development of the attitude and knowledge scales, second on initial attitude differences between men [4] attending Changing Ways voluntarily and those referred to the program though probation or early intervention court, and third on changes in men's attitudes on the constructed measure from pre- to post-intervention.

3.1 Factor analysis of men's attitudes

As mentioned above, men completed a 78-item attitude measure, entitled the "Abuse-Related Attitudes Assessment", tapping a variety of attitude domains including: attitudes towards abuse, attitudes towards women, and attitudes towards intervention. As a first analysis task, men's responses to these items were conceptually grouped and then factor analyzed [5]. This form of analysis essentially asks the question: "Which of these items group together as a measure of the same underlying concept?" Results of this analysis are presented in Table 1 [6]. Four separate factors, or item groupings, were identified. The following labels were given on the basis of factor content: 1) Sexist attitudes towards women; 2) Distrust of, and lack of readiness for, intervention; 3) Lack of personal responsibility for abuse and its effects and; 4) Partner blaming. Items loading on each of these factors are listed, along with a numerical indication of the degree to which they relate to this underlying concept. This numerical value varies from 0 to 1, with higher numbers indicating "better" items. By convention, items loading at .4 or higher are considered "good" indicators of an underlying concept.

In Table 2, items for a fifth subscale, labelled "Denial of expected relationship difficulties," are presented. The nine items on this scale tap the extent to which men deny expected relationship difficulties. It was expected that men would endorse these items unless they were in exceptionally high levels of denial. As such, this subscale was expected to function in a similar manner as social desirability, or lie, scales often used in traditional self-report measures. To inform the development of this scale, the endorsement frequency was examined for each item. A man was judged to be in complete denial if they admitted no relationship problems, for example, by choosing "strongly agree" to a statement such as "I have never wondered if things would be better if I were in a different relationship." Items for this subscale are presented in Table 2 along with the percentage of men denying difficulties in this relationship area.

Table 1: Factor analysis of attitude items

Factor 1: Sexist attitudes towards women
Items Factor Loading
51. Women should worry less about their rights and more about becoming good wives and mothers. .74
54. When a woman is lying she deserves to be put in her place. .71
34. If a husband beats his wife, it is most likely due to her mistaken behaviour. .70
47. Women are sweet until they have you, then their true self shows. .66
27. Women are generally not as smart as men. .65
63. Wife-beating should be given a high priority as a social problem by government agencies. (Reverse scored) .64
25. It would do some wives good to be put in their place by their husbands. .62
23. If I heard a woman being attacked by her husband, I would call the police. (Reverse scored) .56
11. Most women fail to appreciate fully all that men do for them. .53
39. Discrimination against women is no longer a problem in Canada. .47

Factor 2: Distrust of, and lack of readiness for, intervention
Items Factor Loading
24. The only purpose of this program is to blame men for their problems. .83
16. Counsellors here act as if they are better than me. .78
64. Counsellors here can be trusted. (Reverse scored) .71
56. Counsellors here accept me as an individual. (Reverse scored) .62
28. Counsellors here act as though my problems were important to him/her. (Reverse scored) .55
52. If other people know I'm in a treatment program, they'll see me as a failure. .53
60. I believe that my counsellor here does not understand what I am trying to accomplish. .53
44. I don't think I can trust others in this group treatment program. .50
72. If I were to talk about my problems with other people, they would think that I'm crazy or emotionally unstable. .49
48. A mentally healthy person is a person who pretty much keeps his feelings and emotions to himself. .47

Factor 3: Lack of personal responsibility for abuse and its effects
Items Factor Loading
58. I can expect my partner to take a long time to trust me again. (Reverse scored) .77
42. My abusive and hurtful behaviours have had lasting effects on my partner. (Reverse scored) .74
78. My behaviour has made my partner angry, but has had no lasting negative effects on her. .69
1. I need to be the one to end my use of violence. (Reverse scored) .62
10. When men are abusive they do it to gain control over their partners. (Reverse scored) .56
62. When people have been drinking, they cannot be held responsible for their actions. .55
12. I would much rather be somewhere else than in group right now. .50
73. I am 100% responsible for my feelings and behaviour. (Reverse scored) .47
30. There is no justification for my abuse or violence. (Reverse scored) .46
76. My partner has less confidence as a result of my abusive and hurtful behaviours. (Reverse scored) .43

Factor 4: Partner blaming
Items Factor Loading
65. I have only ever been abusive towards my partner to defend myself against serious harm. .74
9. My behaviour is not nearly as bad as my partner makes it sound. .65
2. My partner exaggerates negative things I have done in our relationship. .63
6. My partner often brings up conflicts that have already been resolved. .56
37. My partner is as much to blame for what I do during conflicts as I am. .53
17. My partner is trying to manipulate the system to "get" me. .52
41. My partner needs to learn to respect me and listen to me. .51
45. If I had a different partner, I would not behave in hurtful and controlling ways. .48
33. If I don't defend myself, my partner will walk all over me. .42
5. Most of the time when I am angry at my partner, it is because she has "pushed my buttons." .37

Table 2: Endorsement frequencies for items on the denial of expected relationship difficulties subscale
Items Endorsement frequency
13. I am sometimes annoyed when my partner asks favours of me. 43.2%
49. I have taken advantage of my partner or our relationship at least once in the past. 42.1%
50. I have sometimes wondered if things would be better if I was in a different relationship. 30.8%
57. I have never been annoyed when my partner expressed ideas very different from my own. (Reverse scored) 26.3%
14. I have never expected my partner to go out of her way to make sure that my needs were met. (Reverse scored) 23.1%
26. I have never deliberately said something to hurt my partner's feelings. (Reverse scored) 22.5%
22. I have sometimes said or done things in anger with my partner that I wish I could take back. 10.0%
69. I have never said or done anything to my partner that I regret. (Reverse scored) 2.5%
61. There is nothing that I would like to change about the way I behave toward my partner. (Reverse scored) 0

3.1.1 Reliability of attitude domains

Following the identification of these five subscales, the items within each factor were examined to explore the internal consistency within each factor, or subscale. Internal consistency refers to the extent to which all items on a given scale tap the same underlying dimension. Items that are not conceptually linked, or that are measured with considerable error, tend to reduce a scale's internal consistency, or alpha. By convention, scales with alpha values of .8 or above are considered good, and values in the .7 range are considered adequate.

The alphas for each of the five subscales examined are follows: 1) Sexist attitudes towards women (alpha = .86); 2) Distrust of, and lack of readiness for intervention (alpha = .85); 3) Lack of personal responsibility for abuse and its effects (alpha = .83); 4) Partner blaming (alpha = .79); and 5) Denial of expected relationship difficulties (alpha = .66).

3.1.2 Correlations of attitude domains

Correlations between attitude domains were examined to ensure that identified factors were sufficiently distinct. A correlation is a measure of the strength of a relationship between two variables. Coefficients closer to the value of 1 represent stronger relationships. The positive or negative value of the correlation indicates the direction of the relationship under study. Positive correlations denote scales that vary consistently with each other (i.e., higher scores on one scale relate to higher scores on the other). Negative correlations, in contrast, denote scales that vary in opposition with each other (i.e., higher scores on one scale relate to lower scores on the other).

Correlations of the five attitude domains are presented in Table 3. Note that each scale was scored so that higher scores indicated more negative attitudes. It was expected that all attitude domains would be moderately positively correlated. Results were generally consistent with this hypothesis. Examination of this table reveals that sexist attitudes towards women are strongly correlated with distrust of intervention and moderately correlated with partner blame. Sexist attitudes are also negatively correlated with denial at close to significant levels. In other words, men who deny relationship difficulties are more likely to endorse non-sexist attitudes. Denial of expected relationship difficulties was also moderately strongly corrected with lack of personal responsibility, such that men who endorsed high levels of denial also reported low levels of personal responsibility for abusive behaviour.

Table 3. Correlations of attitude domains
Attitude Domain 1 2 3 4 5
1. Sexist attitudes towards women 1.00 .55** .11 .31 -.33*
2. Distrust of, and non-readiness for intervention 1.00 .04 .20 -.23
3. Lack of personal responsibility for abuse 1.00 -.04 .44**
4. Partner blaming 1.00 -.01
5. Denial of expected relationship difficulties 1.00
  • * Attitude domains were correlated at the .05 significance level.

  • ** Attitude domains were correlated at the .01 significance level.


Date modified: