The Professional Counselor | Volume 12, Issue 3 255 Based on the findings of Kalkbrenner and Neukrug (2019), we computed a higher-order confirmatory factor analysis (HCFA) to test for a Global Barriers to Counseling scale. As expected, the single-factor RFSV model (see Figure 1, Model 2) revealed poor model fit: CMIN (χ2 [77] = 1,013.71, p < .001, χ2 to df = 13.17); CFI = .61; RMSEA = .19, 90% CI [.18, .20]; and SRMR = .13. Accordingly, the theoretical support for a higher-order model (Kalkbrenner & Neukrug, 2019) coupled with the poor fitting single-factor model (see Figure 1, Model 2) indicated that computing an HCFA was appropriate. Except for the CMIN (χ2 [74] = 257.55, p < .001, χ2 to df = 3.48), the higher-order model (see Figure 1, Model 3) displayed a satisfactory model fit: CFI = .92; RMSEA = .08, 90% CI [.07, .10]; and SRMR = .08. Tests of internal consistency reliability revealed satisfactory reliability evidence of scores on the Fit (α = .84, ω = .83), Stigma (α = .86, ω = .87), and Value (α = .79, ω = .79) subscales and the Global Barriers to Counseling scale (α = .88, ω = .88). STEM students’ RFSV scores were entered into a logistic regression analysis to answer RQ2 regarding the capacity of STEM students’ RFSV scores to predict at least one referral to the counseling center. The logistic regression model was statistically significant, X2(1) = 80.97, p < .001, Nagelkerke R2 = .064. The odds ratios, Exp(B), revealed that a decrease of one unit in STEM students’ scores on the Value subscale (higher scores = less value toward counseling) was associated with a decrease in the odds of having made at least one peer-to-peer referral to the counseling center by a factor of .559. A factorial MANOVA was computed to answer RQ3 regarding demographic differences in RFSV barriers among STEM students. A significant main effect emerged for gender on the combined dependent variables, F(3, 316) = 5.23, p = .002, Pillai’s Trace = 0.05, η2 p = 0.047. The post hoc discriminant analysis (DA) revealed a significant discriminant function, Wilks λ = 0.93, χ2 = 23.60, df = 3, canonical correlation = 0.26, p < .001. The standardized canonical discriminant function coefficients between the latent factors and discriminant functions showed that the Value factor loaded more strongly on the discriminant function (1.10) than the Stigma (0.17) or Fit (−0.62) factors. The mean discriminant score on the function for male participants was 0.40. The mean discriminant score on the function for female participants was −0.19. In other words, the MANOVA and post hoc DA revealed that male STEM students scored significantly higher (higher scores reflect greater reluctance to seek counseling) on the Value barrier when compared to female STEM students. A significant main effect also emerged for help-seeking history on the combined dependent variables, F(3, 467) = 4.65, p = .003, Pillai’s Trace = 0.04, η2 p = 0.042. The post hoc DA displayed a significant discriminant function, Wilks λ = 0.93, χ2 = 24.10, df = 3, canonical correlation = 0.26, p < .001. The standardized canonical discriminant function coefficients between the latent factors and discriminant functions showed that the Value factor loaded more strongly on the discriminant function (1.10) than the Stigma (0.01) or Fit (−0.71) factors. The mean discriminant score on the function for participants without a help-seeking history was 0.25. The mean discriminant score on the function for participants with a helpseeking history was −0.29. In other words, the MANOVA and post hoc DA showed that STEM students without a help-seeking history scored significantly higher on the Value barrier than STEM students with a help-seeking history. Discussion The purpose of the present study was to validate STEM students’ scores on the RFSV Scale and investigate demographic correlates with the Fit, Stigma, and Value barriers. The CFA results demonstrated that the RFSV Scale and its dimensions were estimated adequately with a sample of STEM students. This finding is consistent with the existing body of literature on the generalizability
RkJQdWJsaXNoZXIy NDU5MTM1