Study 1
Tan and colleagues (1995) conducted a randomized controlled trial in a medium-sized midwestern city, to examine the impacts of the Community Advocacy Project (CAP). The study consisted of 141 women recruited from a domestic violence shelter who were eligible (i.e., they had stayed at the shelter at least 1 night and had not moved out of the general area, regardless of whether they returned to their assailants). Participants were randomly assigned to either receive free services from trained advocates for the first 10 weeks through CAP after leaving the shelter (intervention group, n = 71) or to no services at all (control group, n = 70). Participants were interviewed upon leaving the shelter (preintervention), after the intervention period (10 weeks), and 6 months postintervention.
Of the 141 participants, 46 percent were white, 43 percent were Black, 8 percent were Hispanic, 1 percent were Asian, and the remainder were Native American, Arab American, or of mixed heritage. The women had an average age of 28.5 years, and 78 percent had at least one dependent child living with them. Prior to living at the shelter, 33 percent of battered women were married and living with their assailants, 45 percent lived with their assailants but were not married, 6 percent were romantically involved with their assailants but were not living with them, and 15 percent were no longer involved with their assailants at the time of the last assault (i.e., separated, divorced, or no longer dating). On average, participants’ length of stay at the shelter was 17 days. There were no statistically significant differences between the intervention and control groups on baseline characteristics.
Outcomes of interest included the women’s relationships with their assailants, satisfaction with social support, experience of further violence, and effectiveness in obtaining resources. This CrimeSolutions review of this study focused on all outcomes at the 6-month follow up, except for the outcome measuring effectiveness in obtaining resources (data for this outcomes was only available at postintervention). Relationship with assailant was measured through a woman’s current level of involvement as follows: 1) married and living together; 2) living together, unmarried; 3) romantically involved, not living together; or 4) dating. Social support was measured using a 9-item scale instrument with a 7-point Likert-scale (1 = terrible, 7 = extremely pleased) that focused on two dimensions of social support (functional and structural). Experience of physical abuse was measured using the modified Conflict Tactics Scale. Study authors found that the scale results were highly skewed at postintervention and 6-month follow up; thus, they dichotomized values to reflect the existence or absence of further abuse. Effectiveness in obtaining resources was measured using a Likert-type scale (1 = very ineffective, 4 = very effective) that concentrated on types of resources such as housing, employment, material goods and resources, child care, social support, and finances. Multivariate analyses of variance (ANOVAs) and a bivariate analysis (chi-square) were used to compare treatment and control groups on outcomes of interest. The study authors did not conduct subgroup analyses.
Study 2
Bybee and Sullivan (2002) conducted a randomized controlled trial that recruited and randomly assigned 278 women from a midwestern domestic violence shelter to either the free advocacy service intervention group (n = 143) or the control group (n = 135), which received no services at all. To be eligible, women had to spend at least 1 night at the shelter, agree to stay in the area for the first 3 months after leaving the shelter, and agree to participate in the study for at least 3 weeks. Women in the advocacy intervention group received free advocacy services through CAP for 10 weeks immediately after leaving the shelter, for about 4 to 6 hours a week. All participants were interviewed six times in 2 years: immediately (i.e., within a week) upon leaving the shelter; at 10 weeks after leaving the shelter; and then at 6-, 12-, 18-, and 24-month follow ups. The CrimeSolutions review focused on intervention effects at the 24-month follow up.
In terms of race/ethnicity, 45 percent of the women were African American, 42 percent were European American, 7 percent were Latina, 2 percent were Asian American, and the remainder were Native American, Arab American, or of mixed heritage. The average age was 29 years, and 74 percent had at least one child living with them. Twenty-seven percent were married to the men who had abused them, 42 percent had been living with but not married to their assailants at the time of the abuse, 7 percent had been intimately involved with the men who had abused them but were no longer living together, and 20 percent were no longer involved with their partners at the time of the last assault (i.e., separated, divorced, or no longer dating). On average, the women’s length of stay at the shelter was 19 days. In the 6 months prior to entering the shelter, the women had experienced violence that led to such injuries as cuts and bruises (85 percent), broken bones (19 percent), dislocations (10 percent), and miscarriages or pregnancy complications due to abuse (11 percent). There were no statistically significant differences between the intervention and control groups on baseline characteristics.
Outcomes of interest included perceived social support, access to community resources, re-abuse by partners or ex-partners, and quality of life. Quality of life measured the respondents’ feelings about themselves, their interpersonal relationships, neighborhoods, and overall well-being through a 25-item assessment that was selected to predict overall quality of life. Longitudinal latent structural equation modeling was used to compare treatment and control groups for the outcomes of interest. The study authors did not conduct subgroup analyses.
Study 3
Allen and colleagues (2004) conducted a randomized controlled trial with the same sample population from Study 2 (Bybee and Sullivan 2002) but looked at the variety of needs that the women reported 6 months after they left the domestic shelter and what they did to address those needs. Participants were demographically the same as those described in Study 2.
At the first interview, researchers looked at types of resources needed by selecting from a list that concentrated on items such as housing, education, employment, transportation, legal assistance, health care, social support, financial support, material goods and services (e.g., furniture), child care, and issues related to their children. Women were also asked if they had other needs not listed. A posttest interview (10 weeks after the initial interview) was conducted, during which the women were asked which of the 11 needs they had worked on without being reminded of the needs they had indicated previously. Outcomes of interest at the posttest included the number of activities that the women engaged in and their effectiveness in accessing resources. The number of engaged activities was measured through activity-level scores, which ranged from 0 to 7, for the 11 types of resources (asked at the initial interview) that women engaged in to address their needs. The effectiveness in accessing resources was determined using the Effectiveness in Obtaining Resources (EOR) Scale, which measured how efficient the women had been in obtaining the resources they needed (1 = very ineffective, 4 = very effective) across all the areas they had worked on. General linear models were used to compare program effects between women in the intervention and control groups. Subgroup analyses were conducted to determine the differences in amount of activity between the intervention and control groups.