Study on Pass-Fail Prediction for the National Radiological Technologist Examination by Discriminant Analysis Using Mock Examinations

Abstract

The objective of this study was to establish a technique for pass-fail prediction for a national examination from the results of intramural mock examinations, for practical use within a school. At the initial stage, we performed pass-fail prediction by discriminant analysis using intramural mock examination and national examination results for the national radiological technologist examination over a period of one year at a certain school. Specifically, we carried out experiments to investigate the optimal attributes and techniques for prediction by discriminant analysis using the Mahalanobis distance. It was found that the correct identification rate was higher when using the scores obtained in 14 subjects than when using the total of the scores of a mock examination as an attribute. Moreover, the method of dividing the national examination results into two or more groups, conducting discriminant analysis, and performing pass-fail prediction based on those results was determined to be effective. The predictive accuracy of discriminant analysis performed directly on two pass-or-fail groups was 84.3%. On the other hand, the accuracy of pass-fail prediction based on the results of discriminant analysis of five groups was 89.3%, providing a higher predictive accuracy than the former method.