Research in medical education: some methodological considerations.

The complexities of describing human characteristics with sufficient accuracy to enable successful selection of candidates for medical school demand precise and penetrating analysis. An important tool is statistical analysis; however, current methods have not been sufficient. Methods of multivariate analysis which have been extended only recently to problems in the social sciences have great diversity and power. These methods have been outlined and their application in medical education indicated. The appropriateness of a particular method depends upon four factors: the subjects under study, nature of the data, conditions under which data were obtained, and the purpose of the analysis. A critical consideration of these factors should enable the choice of a method capable of providing meaningful conclusions concerning medical school selection to be made.