Research that has traditionally supported program accreditation has tended to be primarily descriptive in nature, focusing on outcomes such as the satisfaction of graduates and passing rates on licensure exams. The study reported here illustrates how research employing multivariate methodologies can go beyond such descriptive reporting to address needs central to effective program management, such as designing selective admission processes to admit students who are most likely to succeed. Using a sample of nursing graduates at a large suburban community college, canonical correlation was used to determine how student intake variables employed in the selective admission process could best predict success in the program. The results showed that the current selective admission process was not making the best use of the student intake variables and also illuminated difficulties that can arise in designing a selective admission process to predict success when multiple measures of success are used.