Examined are methodologies for modeling students' choices among higher education institutions. A statistical technique called "conditional logit analysis" is applicable to the problem studied. These applications are reviewed and certain weaknesses inherent in the approach are pointed out. Alternative approaches are offered, based on the use of Bayes' Theorum, which is easier to use, more flexible, and less expensive to apply. In empirical tests, it is also observed to have greater predictive power than conditional logit analysis. (Author)