TPC Journal-Vol 9- Issue 2-Full-Issue

The Professional Counselor | Volume 9, Issue 2 161 Data Analysis We employed the Mann-Whitney U Test and chi-square analyses for this study due to the data characteristics. Specifically, each analysis included RAMP status as a nominal and dichotomous independent variable. The dependent variables were nominal with four groups or continuous data. However, the distribution of the continuous dependent variables violated assumptions for normality; thus, we applied non-parametric approaches of data analysis to this data. The Mann-Whitney U Test was used with continuous dependent variables. For the Mann-Whitney U Tests, we interpreted the effect sizes by computing the approximate value of r (Pallant, 2011), which could be interpreted using 0.1, 0.3, and 0.5 for small, medium, and large effect sizes, respectively (Cohen, 1988). We also utilized chi-square tests for independence when the dependent variables were nominal. In the case of a two- by-two chi-square table, we used Yates’ continuity correction statistics for interpretation and the phi coefficient to evaluate the effect size. The phi coefficient can be interpreted in a similar fashion as the r statistic. For analyses with chi-square tables of two-by-four, we studied the Pearson chi-square statistic and the Cramer’s V effect size statistic. We interpreted the Cramer’s V based on criteria for four categories (0.06, 0.17, and 0.29 were small, medium, and large effect sizes, respectively; Pallant, 2011). An initial a priori power analysis for the Mann-Whitney U Test using G*Power with an alpha level of .05, power established at .95, and a moderate effect size of 0.5 (Cohen, 1988) identified a minimum sample size of 184. Similarly, we conducted an a priori power analysis for the chi-square tests for independence using G*Power with an alpha level of .05, power established at .95, and a moderate effect size of 0.3 (Cohen, 1988) and identified a minimum sample of 191. We used a Bonferroni corrected value of .003 as a means to reduce the likelihood of Type I errors. Results General School Characteristics Our first research question examined whether schools whose school counseling programs have achieved RAMP (i.e., RAMP schools) differ in general school characteristics when compared to schools with school counseling programs that have not achieved RAMP status (i.e., non-RAMP schools). We facilitated a Mann-Whitney U Test to compare the total number of students per school for both RAMP and non-RAMP schools. The Mann-Whitney U Test revealed a statistically significant difference in RAMP schools (M rank = 159.90, Mdn = 925, M = 1,201.81, SD = 853.67) versus non-RAMP schools ( M rank = 103.96, Mdn = 575, M = 687.96, SD = 534.56, U = 4,915.50, z = -5.97, p < .001, r = .37). Similarly, we completed the Mann-Whitney U Test to analyze FTEs for both RAMP and non-RAMP schools. The Mann-Whitney U Test revealed a statistically significant difference in FTE for schools that had RAMP ( M rank = 159.20, Mdn = 51.37, M = 69.38, SD = 48.49) and those schools that did not have RAMP ( M rank = 105.80, Mdn = 32.48, M = 41.49, SD = 30.27, U = 5,187.00, z = -5.68, p < .001, r = .35). A chi-square test for independence indicated a statistically significant association between RAMP and geographic location among the schools in this study: χ 2 (3, N = 266) = 22.94, p < .001, Cramer’s V = .29. Table 1 provides a breakdown of the frequency and percentage for each geographical location by RAMP status. Non-RAMP schools were more often located in city, town, and rural settings than RAMP schools, whereas RAMP schools were more often located in suburban locations. A chi-square test for independence indicated no statistically significant association between RAMP and school level among the schools in this study: χ 2 (3, N = 263) = 22.94, p = .06, Cramer’s V = .17 (Bonferroni corrected p value of .003).