The working alliance: rehabilitation outcomes for persons with severe mental illness

Journal of Rehabilitation,  April-June, 2004  by Chandra M. Donnell,  Daniel C. Lustig,  David R. Strauser

• E-mail
• Print
• Link

Data analysis

For the first research question, "Is there a difference between clients who are employed and unemployed with respect to measured levels of working alliance?", a t-test was conducted on the continuous variable of the Working Alliance Survey (WAS) in order to compare employed clients with unemployed clients. For the second research question, "For employed clients, is there a relationship between working alliance and satisfaction with their current job?", a Pearson r correlation between the continuous variables of the WAS and satisfaction with their current job was calculated for employed clients. For the third research question, "For employed clients, is there a relationship between working alliance and the client's view of their future employment prospects?", a Pearson r correlation between the continuous variables of the WAS and future employment prospects was calculated for employed clients. Finally for research question four, "For unemployed clients, is there a relationship between working alliance and the client's view of their future employment prospects?", a Pearson r correlation between the continuous variables of the WAS and future employment prospects was calculated. An alpha level of .05 was used for hypothesis testing.

Since the study was an ex post facto research design, specific demographic and disability related factors were considered as potential sources of error in the results. Based on previous research (Bolton et al., 2000; Wilson, 2000) two variables were considered: (a) the existence of a secondary disability and (b) the participant's ethnicity. The difference between individuals with a secondary disability and without a secondary disability was not statistically significant, t (281) = -1.76, p > .05; [.sup.~2] < .01. The second variable considered for statistical control was ethnicity.

This variable was collapsed into two groups, Caucasian and non-Caucasian. The difference between individuals who are Caucasian and non-Caucasian was not statistically significant on the continuous variable of working alliance (t (303) = - .33, p > .05; [.sup.~2] < .000). The two variables, secondary disability and ethnicity, were not considered appropriate for statistical control.

Results

Using a t-test, a significant difference was found on the variable of working alliance for the employed group (M= - 1.05; SD = 2.01) versus the unemployed group (M = 1.28; SD = 3.32; t (303) = -6.73,p < .001) with an effect size, as measured by [[eta].sup.2], of .13. For employed clients, the measure of working alliance correlated significantly with satisfaction with current job (r = .23; p < .05), and their view of future employment prospects (r = .70; p < .001). For unemployed clients, the measure of working alliance correlated significantly with the client's view of their future employment prospects (r = .54; p < .001).

Discussion

Four research questions guided this study. First, the results indicated that employed clients measured stronger on the working alliance with their counselor than unemployed clients, as measured by the WAS. According to Cohen (1988), the effect size as measured by eta squared (.13) can be considered a medium effect. Second, the results indicated that for employed clients, the stronger the measured level of working alliance the more satisfied the client was with their current job. The correlation coefficient as measured by r (.23) can he considered a small effect (Cohen). Third, the results indicated that for employed clients, the stronger the measured level of working alliance the more positive the client viewed their employment future. According to Cohen, the effect size as measured by r (.70) can be considered a large effect. Finally, the results indicated that for unemployed clients, the stronger the measured level of working alliance the more positive the client viewed their employment future. This result can also be considered a large effect (r = .54; Cohen).