A DEA selection system for selective examinations

This paper presents a DEA (Data Envelopment Analysis) selection system applicable to selective examinations on multiple academic subjects, e.g., university entrance examinations in Japan. Using this system, the paper can select ‘masters of an art’ (brilliant in some academic subject but not necessarily in all) as well as ‘jacks of all trades’ (scoring quite well across all the subjects) as successful candidates. This is also a proposal for the potential use of DEA in multi-dimensional evaluation other than the standard DEA efficiency analysis. Defining candidates for examination as DMUs (Decision Making Units) and examination scores as outputs, and considering that every DMU has only one input whose amount is unity, the paper can comprehensively evaluate the candidates in terms of their examination scores using DEA. The results of a selective examination case show good performance of the DEA selection system, i.e., it can select various types of candidates with different featured characteristics so as to fill up the allocation of students exactly. It is concluded that DEA selection, which relatively evaluates candidates in accordance with their individual characteristics, can be considered as an alternative method other than the usual selection on the basis of total score.