3.1. Descriptive Statistics
displays the descriptive statistics for the available data prior to imputation. The average treatment program adopted 3.5 wraparound services. The most commonly offered wraparound services included case management (81.3%), mental health counseling (76.9%), and crisis intervention (69.7%). The wraparound services that were least likely to be provided included housing assistance (23.1%), legal assistance (8.7%), and child care (5.3%). About one-fifth of the community-based substance abuse treatment programs offered a women-specific-treatment program. In terms of staffing, the average substance abuse treatment program employed about 15 staff members (FTEs) and a typical counselor caseload was 41 clients. About half of the programs had a female administrator (50.5%) and almost two-thirds (62.4%) employed an administrator with a Master’s degree, JD, PhD, or MD. For treatment administrators, the average tenure in the substance abuse treatment field was nearly 17 years. On average, administrators reported that about half of the program’s counselors had a four-year college degree.
There was some variability in organizational culture. Given that these are treatment organizations, it was not surprising that there was strong endorsement of the rehabilitation scale with the mean approaching the scale’s maximum value (mean = 4.60, SD = 0.46). While above the midpoint for the scale, endorsement of the scale measuring organizational support for effective treatment was lower (mean = 3.83, SD = 0.63). On average, treatment programs offered about 7.6 of the 12 substance abuse treatment approaches which include such approaches as peer-led groups, behavioral management approaches, or motivational interviewing.
Notably, use of the three sources of information was relatively modest. Means fell below the mid-point for the time spent using the internet or publications, using professional development resources, and informally communicating with prison staff to obtain information about substance abuse treatment interventions for offender populations.
There was also variability in the measures of systems integration. The average organization reported a level of community support around the midpoint of the scale. In contrast, there was less evidence of formal relationships with other organizations. The average for this scale, which had a maximum possible value of 35, was only 8.5, suggesting a modest level of formal relationships with other treatment providers, community corrections, and the judiciary. Finally, the averages for the influence of legislative priorities and federal priorities indicated that these influences tended to be somewhat positive.
3.2. Bivariate Analyses of Women-Specific Programming and Organizational Characteristics
Before estimating the models of wraparound services, the measure of having a women-specific program was examined as a correlate for a series of bivariate models of the other organizational characteristics. These analyses were conducted using the ten imputed datasets and the “micombine” command to average the results across the ten datasets. For continuous variables (e.g. organizational size), OLS regression was used. Ordered logit models were used for characteristics that represented ordered categories (e.g. counselors with a four-year degree). When the organizational characteristic was dichotomous, logistic regression was utilized to examine whether women-specific programs were associated with the characteristic. There was only one significant difference at the p<.05 level. Women-specific programs were significantly larger organizations as evidenced by the number of staff (FTEs). Two other variables approached significance (p<.10, two-tailed). There was a trend for organizations with a women-specific program to be more likely to have a female administrator (b = .75, SE = .40, p = .063) than organizations without a women-specific program. In addition, organizations with a women-specific program tended to report somewhat greater community support (b = .25, SE = .15, p = .089). There were no other significant differences.
3.3. Bivariate Analyses of Wraparound Services
Using the ten imputed datasets, a series of bivariate Poisson regression models were estimated to identify organizational characteristics associated with the number of adopted wraparound services (). There was a trend at the bivariate level for organizations with a women-specific program to offer a greater number of services (b = .16, SE = .09, p = .073).
Unadjusted Bivariate Poisson Regression Models of Number of Wraparound Services
Of the organizational structure measures, there was a highly significant positive association between the number of FTE employees and the number of services offered (b = .16, SE = .04, p<.001). None of the personnel characteristics were associated with the number of adopted wraparound services at the bivariate-level. Two of the three measures of organizational culture were statistically significant. Organizations that more strongly endorsed the rehabilitation scale had adopted significantly more wraparound services (b = .20, SE = .09, p<.05). Additionally, programs that reported using a greater number of treatment approaches had adopted a significantly greater number of wraparound services (b = .06, SE = .02, p<.001). The bivariate associations for two of the three measures of information sources trended towards significance. Greater use of internet and publications trended towards greater adoption (b = .13, SE = .07, p = .061). A similar positive association was indicated for use of professional development resources (b = .08, SE = .05, p = .096). Finally, three of the four measures of system integration were statistically significant. Stronger endorsement of the relationships with other organizations scale was positively associated with the number of adopted wraparound services (b = .02, SE = .01, p<.01). There were also significant positive associations for the influence of legislative priorities (b = .07, SE = .03, p<.05) and the influence of federal priorities (b = .09, SD = .04, p<.05).
In addition to considering the number of wraparound services, we conducted a series of bivariate logistic regression models to examine the associations between offering a women-specific program and each of the eight wraparound services. There were three statistically significant differences. First, organizations with a women-specific program were significantly more likely to offer housing assistance (b = 1.102, SE = .373, p<.01). The availability of women-specific programming was also positively associated with the odds that organizations offer mental health counseling (b = 1.430, SE = .620, p<.05). Finally, organizations with a women-specific program were significantly more likely to offer legal assistance (b = 1.296, SE = .518, p<.05).
3.4. Multivariate Analyses
Finally, Poisson regression was used to examine the organizational-level correlates of the number of wraparound services offered by organizations which treat predominantly criminal offenders (see ). Organizational characteristics that were significant at p<.10 were entered into a Poisson regression model. Controlling for other organizational characteristics rendered the measure of women-specific programming non-significant. Two variables, however, were significant in this multivariate Poisson regression model. Larger organizations, meaning those with more FTEs, had adopted a significantly greater number of wraparound services (b = .116, SD - .037, p<.01). Also, there was a positive association between the number of treatment approaches and the adoption of wraparound services. Treatment organizations reporting use of a larger number of treatment approaches had adopted significantly more wraparound services (b = .037, SE = .018, p<.05). Finally, there was a trend that greater endorsement of the rehabilitation scale was positively associated with wraparound services (b = .160, SE -.089, p = .071).
Multivariate Poisson Regression Model of Number of Wraparound Services