Analysis of Profiles of Students Applying for Entrance to Universities

Abstract

Applicants to universities present profiles of performance in a variety of relevant content areas as evidence for selection. Even though profiles of different applicants may involve different content areas, the applicants may be competing for places to the same university and even to the same program of study. In that case, and if there are more eligible applicants than there are places, universities must reconcile these different profiles in order to make comparisons among them. When there is a small number of applicants, then these comparisons may be camed out qualitatively; when the number of applicants runs into the order of thousands and there is a short time in which to make the offers, some quantitative analysis is required. This quantitative analysis usually involves aggregating the components of each profile in order to form a single score from which comparisons among applicants can be made readily. Legitimate concerns can be raised regarding forming simple aggregates from diverse components of profiles, but despite these concerns, the practical problem of making relatively rapid decisions means that these concerns are generally not addressed. The premise of this article is that profiles will be more or less consistent among the components and that although some profiles may not be, a great number of others may be, summarized adequately by a single score. It is shown that by applying the principles of latent trait test theory at the level of tests, it is possible to rank order a set of profiles in terms of the adequacy with which they are summarized by a single score, and that as a result, only a subset of the original profiles may require a qualitative analysis. The application of the procedure is illustrated with a random sample of 577 profiles from a population of 12,314, which were presented for selection into universities in Western Australia in 1986.