85 The Professional Counselor | Volume 13, Issue 2 Data Analysis We conducted all analyses using SPSS Version 27. The first analysis had the four dependent variables of Counselor Trust, Counselor Selection, Advice Following, and Weight Bias, with the independent variable of Weight (overweight, underweight, and average weight). We conducted a series of assumption-checking procedures to draw valid interpretations of the findings. Engaging the Shapiro-Wilk test as a test of normality yielded a significant result (W = 0.92, p < .001) for the overweight counselor, indicating the sample was not normally distributed. After removing outliers, the sample for the overweight survey condition did not meet the assumption of normality. A test for homogeneity of variance yielded a statistically significant Levine’s score of F(2, 168) = .46, p = .013; degrees of freedom were adjusted due to unequal sample sizes for each survey condition. Conducting a MANOVA yielded statistically significant results; however, not meeting the assumptions of multivariate normality and homogeneity of variance required, we used the Welch ANOVA, which is recommended for non-normal distributions. Using a Welch ANOVA is also a best practice when the homogeneity of variances test fails; it controls the type I error and gives more power in many instances (Liu, 2015). Although a parametric test such as a typical ANOVA or MANOVA is known to be more powerful than a non-parametric test (e.g., Welch ANOVA), it can lead to erroneous results if required assumptions are not satisfactorily met (Zar, 1998). Considering unequal variances and sample sizes across groups, we used Games-Howell for post hoc testing. We used G*Power version 3.1.9.7 (Faul et al., 2007) to perform the power analysis. For the Welch ANOVA test, the minimum sample size was 157, with a medium effect size of .25, a desired statistical power level of .8, and an alpha level of .05. Lastly, we measured the effect size using partial eta squared (ꞃ²), showing the strength of association as a proportion of variance in the dependent variable explained by group membership (Coladarci et al., 2011). Table 2 Descriptive Statistics for Outcome Variables n Mean Std. Deviation Std. Error 95% Confidence Interval for Mean Minimum Maximum Lower Bound Upper Bound Trust OW 70 49.70 9.89 1.18 47.34 52.06 16.00 63.00 UW 59 41.15 10.72 1.40 38.36 43.95 17.00 63.00 AW 60 47.48 7.75 1.00 45.48 49.49 29.00 63.00 Total 189 46.33 10.16 0.74 44.87 47.79 16.00 63.00 Selection OW 70 32.41 5.40 0.64 31.13 33.70 19.00 40.00 UW 59 30.00 5.12 0.67 28.67 31.33 21.00 40.00 AW 60 31.42 4.92 0.63 30.15 32.69 16.00 40.00 Total 189 31.34 5.23 0.38 30.59 32.09 16.00 40.00 Advice OW 70 25.76 8.72 1.04 23.68 27.84 8.00 42.00 UW 59 24.73 7.33 0.95 22.82 26.64 8.00 42.00 AW 60 31.93 4.64 0.60 30.73 33.13 23.00 42.00 Total 189 27.40 7.81 0.57 26.28 28.52 8.00 42.00 Note. OW = overweight; UW = underweight; AW = average weight.
RkJQdWJsaXNoZXIy NDU5MTM1