Discriminant Analysis to Predict Graduation – Nongraduation in a Master's Degree Program in Nursing

Abstract

Discriminant analysis was used to predict graduation and two categories of nongraduation from readily available admissions data at the University of Kansas nursing master's degree program. The traditional admissions indices, baccalaureate grade point average (GPA) and Graduate Record Examination (GRE)-verbal and -quantitative scores, were used as predictors. Criterion categories were composed of 102 graduates, 103 individuals who dropped out of the program, and 65 individuals who were not accepted. The first discriminant function was, chi 2 (6) = 87.567, p less than .0001, and extracted 98% of the variance of the discriminant space. Follow-up procedures using one-way ANOVA's and Scheffé multiple comparisons indicated that the baccalaureate GPA and GRE-verbal and -quantitative scores independently differentiated the graduate and dropout groups from the not-accepted group at a statistically significant level (p less than .05). Practical significance of the independent contribution of these variables to group differentiation, as measured by omega 2 was 22% for the baccalaureate GPA, 13% for the GRE-verbal scores, and 10% for the GRE-quantitative scores. Implications for future research are discussed.