Predictive discriminant analysis was used to determine the existence of variable subsets that predicted success in practical nursing programs. Chi-square analysis was used to test the significance of differences between program completion rates of remediated and nonremediated groups of practical nursing students. Eight null hypotheses were tested.

The subjects in this ex post facto study were the 362 practical nursing students who entered the Broward County Practical Nursing Program between April 1988 and August 1989. The subjects included 195 Black females, 138 White females, 11 Black males, and 18 White males. Approximately sixty percent (218) of the students completed the program.

Classification accuracy or (hit-rates) for all possible subsets of predictor variables (age, gender, race, ACT, Career Preparation Program sub-test scores, education level, year diploma received, nursing theory score, number of courses completed) were generated for the total sample, completers, and dropouts. The classification accuracy of predictor sets (also called models) were compared to what could be achieved by simply assigning all subjects to the larger group (completers).

Analyses revealed that a number of cross-validated models, or predictor sets, were significantly better at predicting success than both maximum and proportional chance criterion. The model that was the best predictor of dropouts contained the variables age, reading sub-test score, math sub-test score, and average nursing theory score. The best predictor set for the total sample contained the variables age, race, nursing theory score, and number of term one courses completed.

Significant differences (p $<$.05) between the program completion rates of LPN students requiring remediation before program entry and those not requiring remediation were found for all sub-tests except reading. These findings add to the growing body of knowledge that may provide program administrators a better understanding of some of the factors affecting attrition from practical nursing programs.