Evaluation of College Students with Application of Latent Class Analysis

Abstract

In social sciences, especially in economy, to reveal relations between variables it’s easy to apply many known statistical tools when we deal with observable (measureable) variables. The problems appear when dealing with latent variables – that are not directly observed and they are of subjective matter. It’s also an important issue to measure relations between latent variables. The example of latent variables are preferences. The preferences play a very important role in economy. Very often real market decisions, choices (or answers in a questionnaire) are described by non-metric variables (nominal and ordinal). These variables are also called qualitative. The latent class analysis allows to reveal hidden relations between observable variables. The observable variables allow, with a specified probability, to find a non-observable phenomenon. The latent class analysis allows to analyze the qualitative data [see: McCutcheon 1987, p. 7; 11; Hagenaars 1993, p. 21–23]. LCA was introduced by Lazarsfeld in 1950 [1968]. The paper presents evaluation of college students with application of latent class analysis. To obtain such a goal data collected (winter recruitment of 2008/2009) by a college in Walbrzych was used.