TPC-Journal-V6-Issue2

The Professional Counselor /Volume 6, Issue 2 194 spent on homework and talks with math teacher outside of class), .00 (number of hours spent on ex- tracurricular activities and talks with math teacher outside of class, and .01 (number of hours spent on homework and number of hours spent on extracurricular activities). Low correlations along with low standard errors (ranging from .06 to .18) among the independents suggest the absence of mul- ticollinearity. Tests for multicollinearity revealed tolerances values and various inflations factors to hover around 1.0, and the highest condition index was 7.9. All observations reveal low risk of multi- collinearity (Cohen, Cohen, West, & Aiken, 2013). For the MLR examining the effects of the three predictor variables, the likelihood ratio test for the overall model revealed that the model was significantly better than the intercept-only model [χ 2 (14, 7271) = 594.63, p < .000]. In other words, the null hypothesis (that the regression coefficients of the in- dependent variables are zero) was rejected. Both the Hosmer-Lemeshow test (Hosmer & Lemeshow, 2000) for model deviance [χ 2 (48)=59.87, p < .117] and the goodness of fit test [χ 2 (48)=58.53, p < .142] failed to reject the null hypothesis, implying that the model’s estimates fit the data at an acceptable level. Furthermore, the likelihood ratio test for individual effects showed that all of the predictor variables were significantly related to the categories of the criterion variable: talks with math teach- er, χ 2 (2) = 14.94, p < .001; hours of homework, χ 2 (6) = 13.50, p < .05; and hours of extracurricular ac- tivities, χ 2 (6) = 533.65, p < .000. Regarding effect size, the Nagelkerke R 2 (Norusis, 2004) in the overall model was .086, considered a medium effect size (Sink & Stroh, 2006). Therefore, the independent variables included in the model explained 8.6% of the variability in college persistence. Table 1 MLR Parameter Estimates and the Effects of the Predictor Variables Upon Postsecondary Education Status. Still Enrolled in Two-Year Institution Still Enrolled in Four-Year Institution VARIABLE β Odds β Odds Talks with Math Teacher Outside of Class No Yes .04 1.04 .21*** 1.24 Hours Spent Weekly on Homework Very Low Low Moderate High .13 .20 .16 .88 1.23 1.17 .08 .24 .18 1.08 1.27 1.20 Hours Spent Weekly on Extracurricular Activity None Low Moderate High -.25* -.12 -.01 .78 .86 .99 -1.6*** -.58*** -.15 .20 .56 .86 Note. Leaver is the reference category for the dependent variable. The comparison categories for the predictor variables were talking to the math teacher outside of class, high (16 or more) number of hours per week on homework, and high (15 or more) number of hours spent in extracurricular activities. AM software (American Institutes for Research, 2003) was used to calculate adjusted standard errors for sampling design effects. Nagelkerke R 2 = .09. * p ≤ .05; ** p ≤ .01; *** p ≤ .001.