TPC-Journal-V4-Issue4

The Professional Counselor \Volume 4, Issue 4 395 Client Motivation for Therapy Scale. Motivation, conceptualized using self-determination theory (Ryan & Deci, 2000), postulates six types of motivation along a continuum from intrinsic to external to no motivation (i.e., amotivation). The 24-item Client Motivation for Therapy Scale (CMOTS; Pelletier, Tuson, & Haddad, 1997) has six 4-item subscales that measure each type of motivation while one is receiving therapy. We were interested in two CMOTS subscales that could be used before counseling began in order to assess pretreatment motivation levels potentially associated with SES variables. Those subscales included identified motivation (e.g., attending counseling “because I would like to make changes to my current situation”) and external motivation (e.g., attending counseling “because other people think that it’s a good idea for me to be in therapy”). Participants rated their reasons for participating in counseling on a 7-point scale (1 = does not correspond at all, 7 = corresponds exactly). A summary score for each subscale was created using its arithmetic mean. The CMOTS was validated on 138 inpatient and outpatient clients seeking help for a variety of mental health concerns (e.g., self-esteem, interpersonal problems; Pelletier et al., 1997). Internal reliability coefficients in the present study were acceptable for identified motivation ( a = .76) and external motivation ( a = .80). Treatment Expectancy Scale. Client expectation for positive treatment outcome was measured using the TES (Sotsky et al., 1991). The TES consists of a single item: “Which of the following best describes your expectations about what is likely to happen as a result of your treatment?”, with responses ranging from “I don’t expect to feel any different” (1) to “I expect to feel completely better” (5). Although reliability data was not reported, the TES was one of the strongest client predictors of outcome in the National Institute of Mental Health Treatment of Depression Collaborative Research Program, a large randomized control trial (Meyer et al., 2002; Sotsky et al., 1991). Analyses Data analyses followed the guidelines for outcome research that Ogles et al. (2002) outlined. Primary analyses included correlation and multiple regression techniques, beginning with tests of the assumptions of regression (Cohen, Cohen, West, & Aiken, 2003). A repeated measures t test was used to evaluate pre-post change, and ANCOVAs were used to test the need to include various covariates as control variables in the regression analyses. For each participant, the initial OQ total score was considered the pretest score and the last OQ completed was used as the posttest. Because computing a simple difference score between pretest and posttest is subject to regression to the mean (i.e., highest initial scores change the most), we analyzed outcome by partialing out the OQ pretest scores from OQ posttest scores in the first step of the hierarchical multiple regression analysis (Hill & Lambert, 2004). Before conducting hypothesis tests, we inspected data for potential violations of univariate and multivariate assumptions in multiple regression analyses, including outliers, atypical scores, multicollinearity and assumptions of linearity, normality and homoscedasticity (Cohen et al., 2003). Five cases showed highly atypical scores according to recommended cutoff guidelines (Cohen et al., 2003) in small data sets (i.e., DFFITS > 1) and were removed before hypothesis testing. No further problems were evident. Initial analyses were conducted to determine whether any demographic variables should be included as covariates in the regression model. Aside from age and length of time in counseling, demographic variables were categorical: gender, marital status (unmarried versus married) and employment status (unemployed versus employed). These variables were dummy coded for the analysis. Separate ANCOVAs were run for the three categorical variables with OQ pretest scores entered as the covariate. The three categorical variables were not significantly related to outcome ( ps ranged from .29 to .84). A simple regression evaluating age on outcome with OQ pretest scores partialed out showed no significance ( p = .77). Because the amount of time in counseling may have affected how much change had occurred at posttest, we regressed OQ posttest scores on length of time in counseling, controlling for OQ pretest scores. The regression showed no effect of length of