Abstract
In this paper we present statistical analysis of data on subject selection by examinees in NCUEE (National Center for University Entrance Examinations) examination in 2006. In NCUEE examinations, exmaminees can choose subjects depending on the university and the department they are applying. As seen from the well publicized news on skipping world history classes in some high schools, the pattern of subject selection is complicated and depends on many factors, incluing geography and sex. Analysis of influences of these factors is important in discussing the university entrance examination and the education in high shools in Japan. In this paper we deal with geographic factors by incorporating effect of individual cells into hierarchical models of contingency tables. We also estimate the influence of sex on selection of science subjects by conditional likelihood method. For confirming these effects we employ Markov chain Monte Carlo methods, in addition to asymptotic approximation.
