TPC-Journal-V4-Issue4

The Professional Counselor \Volume 4, Issue 4 397 conversely, clients feeling less supported by their social network felt less secure financially. The other subjective SES variable, perceived SES, did not significantly correlate with motivation, treatment expectancy or subjective social support. Therefore, the overall pattern of findings did not support the first hypothesis. Hierarchical Multiple Regression Analysis We used hierarchical multiple regression to test the second hypothesis that objective SES variables (e.g., education level, income, health insurance status) and subjective SES variables (e.g., perceived financial security, perceived SES) predicted counseling outcome. In the first step of the hierarchy, we entered OQ pretest scores to control for initial differences in symptoms. In the second step, we entered objective and subjective SES variables. In the third step, we entered psychological variables (subjective social support, treatment expectancy and client motivation) to test whether these variables accounted for additional outcome variance beyond that which SES variables explained. Because we did not have hypotheses about the primacy of specific individual variables’ effects on counseling outcome, we examined semipartial correlations ( sr ) to identify which predictors within each step had the greatest impact on outcome. Results of the hierarchical regression analysis appear in Table 3. Controlling for OQ pretest scores in the first step, results supported the hypothesis that SES variables significantly predicted counseling outcome, Δ R 2 = .05, F (5, 42) = 2.93, p < .05, a small to medium size effect. Taking into account the other predictors, the following two of the six SES variables significantly predicted outcome: education level and health insurance status. The semipartial correlations indicated that education level and health insurance status accounted for 3% and 4% of outcome variance, respectively, small to medium effect sizes. The beta coefficient for education indicated that for every unit increase in education, clients had, on average, a 3.6-point reduction in their final OQ scores relative to their initial level ( t = -2.49, p < .05). Similarly, clients who had health insurance reported an average 8.7 OQ points greater positive change than those who did not have insurance ( t = –2.60, p < .05). Table 3 Hierarchical Multiple Regression Analyses Predicting OQ Posttest Score Predictor r sp B SE B β R 2 F df Baseline – OQ pretest .84** 0.88 0.08 0.84 0.70 111.2 1, 47 Model 1 Δ R 2 .08 Δ F 2.93** 5, 42 Education level –.18* –.63 1.46 –.20 Income .12 1.54 0.93 .05 Health insurance –.19* –8.67 3.34 –.22 Financial security –.01 –0.12 1.05 –.01 Perceived SES –.01 –0.36 3.58 –.01 Model 2 .02 0.90 5, 37 Social support .01 0.71 4.22 .02 Treatment expectancy –.10 –3.30 2.47 –.12 Identified regulation –.06 –3.22 3.84 –.07 External motivation .13 4.06 2.28 .16 Amotivation –.09 –3.42 2.79 –.13 Note. r sp = semipartial correlation coefficient. Initial covariate in the first step was Outcome Questionnaire-45 pretest score. Negative signs indicate lower posttreatment symptoms. OQ = Outcome Questionnaire-45; financial security = perceived financial security; perceived SES = perceived socioeconomic status; social support = Subjective Social Support; treatment expectancy = Treatment Expectancy Scale; health insurance = Health Insurance Status; coding: no = 0, yes = 1. *p < .05. ** p < .01.