White / Caucasian

All subjects were followed for 12 months with face-to-face interviews every three months to collect data on primary employment variables and secondary outcomes measured. After this initial 12 months, the follow-up period was extended to 24 months at all sites including telephone interviews every three months. MANCOVAs were used to determine treatment group differences over time on the outcome variables (employment index, perceived barriers to employment, level of disability, quality of life, depression, and sustaining care needs), controlling for specified covariates. This included study site, sex, and age.

Competitive employment was as a primary outcome measure and measured every three months. Competitive employment is "community jobs that pay at least minimum wage (directly by the employer to the employee) that any person can apply for, including full-time and part-time jobs." General rehabilitation outcomes were measured at baseline and 3-month follow-up interviews. Standardized measures used included 1) Alcohol Use Disorders Identification test, 2)The Craig Handicap Assessment and Reporting Technique (CHART), 3) VR-36 that measures health-related quality of life in veterans, and 4) The Quick Inventory of Depressive Symptomatology Self-Report.

Demographic variables were also collected such as age, sec, race/ethnicity, employment history, previous levels of income, educational background, legal history, lifetime employment history, vocational services history, duration of SCI, level of SCE, and type of family structure.

The following data were gathered from VESID central files for fiscal years 2001 (October 1, 2000–September 30, 2001) and 2002 (October 1, 2001–September 30, 2002): case status (26 successfully placed for 90 days, 28 closed after IPE-initiated, 30 closed before IPE initiated, 8 closed before IPE developed), weekly earnings at closure, hours worked per week at closure, cost of case services provided to each consumer, consumer demographic characteristics (gender, level of education, employment status at application, ethnicity, age), and name of the state VR counselor serving the consumer. By using case status, the proportion of successful closures (i.e., case status 26) were compared with unsuccessful closures (i.e., case status 8, 28, and 30) for the PWW counselors and other counselors in the catchment areas served by using Pearson chi-square analyses. One-tailed t tests were used to compare weekly earnings at closure, hours worked per week at closure, and cost of services for PWW consumers versus the matched consumers.

A comparison of individuals who were randomized (n=69) to IPS or the Brokered treatment with those who were not randomized (n=18) reveals no statistically significant demographic differences between the two samples. Similarly the randomized individuals did not differ significantly from those who were not randomized in prior illness indicators or symptom severity at screening.

A focus group was help to identify common themes about health, health care, economic security, and return to work. Analysis were descriptive and qualitative. Transcripts of the group meeting were read along with moderator notes. General themes were pulled from this material.

Data was gathered using administrative databases and other records located at the three centres. Job roles were classified using the UK standard occupational classification system. Data on outcomes was classified into one of the following: paid competitive work, education and training, voluntary work, discharge to other services, client withdrew and discharged for other reasons.

Several assessments were used: diagnosis,
background information, employment outcomes, and nonvocational outcomes. Interviews were completed by one of three research staff and reliability checks were conducted on 15% of the interviews by having a 2nd person review the interviews.

Job satisfaction was rated by the Indiana Job Satisfaction Scale. PANSS was used for nonvocational outcomes and modified to be entirely client self-report.

Overall functioning was rated by the Global Assessment Scale, while social and leisure functioning was rated using one subscale from the Social Adjustment Scale-II: Social-Leisure subscale, as well as a global rating.

Social network information was rated with a variant of the Social Support and Social Network Interview and was requested at baseline, 12 months and 24 months. Quality of Life was assessed through the Brief Version of Quality of Life Interview. Substance abuse was also tracked using the Alcohol Use Scale and Drug Use Scale.

Statistical analyses were conducted as follows:
- "Intent-to-treat analyses of employment outcomes were conducted on the entire randomized sample" (p. 483).
- A second set of analyses were conducted to determine if clients differed in their vocational outcomes.
- Employment outcomes were collected over the entire 2-year period. The 3 groups on time to job were restricted to clients who obtained any work.
-Changes in non-vocational outcomes used mixed-effects regression models, using the vocational program as the independent variable.

Throughout the 2 years of the study information on work, including the type of job, wages, and hours worked, was obtained weekly through brief interviews with clients and vocational staff. In addition, job satisfaction was rated using the Indiana Job Satisfaction Scale 2 weeks after beginning a new job and bi-monthly thereafter as long as clients remain on the job.

Those who consented to participate, completed a Workplace Fundamentals Knowledge Test. Afterwards they were randomly assigned to receive supported employment services alone or to receive supported employment services and the supplementary skills training about workplace fundamentals. Individuals were assigned to the group on an average of 56 days after obtaining a job. Among the 35 participants, 18 received treatment as usual (supported employment services alone) the rest were assigned to the receive the work fundamentals training too. Chi square test and t test indicated no significant difference in the two groups. Workplace knowledge was measured with the Workplace Fundamental Knowledge Test scores at baseline and at nine month intervals. Rates of employment for each month of the study period for individuals who recently went to work was documented by control group and workplace fundamentals group. These rates were compared using a generalized estimating equations analysis. The researchers also compared cumulative time worked, wages earned and job tenure for the first and subsequent jobs held. Mann-Whitney tests were used because that data was skewed. Vocational services used were also analyzed with Mann-Whitney tests.

The data was obtained from the individual client closure reports and RSA-911 national data tape, provided by the Rehabilitation Services Administration (RSA). The authors used two tests of statistical significance: chi-square and logistic regression analysis. Chi-square analysis was conducted for type of mental retardation and work status at closure. In addition the authors conducted a post hoc pairwise comparisons (2 x 2 chi-square) for mild by moderate, mild by severe/profound and moderate by severe/profound. To guard against Type I errors, the authors used the Bonferroni technique, which resulted in an adjusted alpha of .003. Then, the authors used logistic regression analysis to evaluate the linear relationship between vocational rehabilitation services and work status at closure. Then, chi-square tests were used to evaluate the proportions of significant vocational rehabilitation services received by consumers with mild, moderate, or severe/profound cognitive disabilities. Then, the relationship between type of cognitive disability and work status at closure were evaluated for consumers who received vocational rehabilitation services. Those consumers who had not received VR services were then eliminated from the sample. The desktop version of SPSS for Windows, version 11.5 was used in this analysis.

Repeated measures included the Addiction Severity Scale, a structured interview related to problem areas (health, transportation, finances, etc.), and six- and 12-month follow-up interviews regarding employment status. Data analysis used mixed-effects models to describe in-treatment change. These models comprise fixed effects, describing the average change over time, and random effects, describing the components of variation about that average pattern of change. For continuously distributed linear mixed-effects models were used, and for binary responses generalized linear mixed-effects models.