A new method to predict the passing rate of the national examination using probabilistic model

Abstract

The pass rate of the national qualification exam is of utmost concern for medical colleges because students enrol with the goal of obtaining the national qualification. We proposed a method that can estimate the passing probability of each student’s national exam and the passing rate of the department, at the time of 1 year before the exam. This is to make it possible to consider whether the same measures as last year are sufficient or whether the measures need to be changed based on specific data. Based on the results of the one-year mock tests of a department with a capacity of 40 students and the actual pass/fail data, the probability of passing how many of the students who scored the mock test at a certain time passed for the national exam was calculated for each score. This probability and change over time were modelled by a line graph. This model was applied to data for other years. As a result, the probability of passing the actual exam when each student made the same effort as last year could be calculated from the score of each mock test. By adding up the probabilities, we were able to estimate how many people could pass in the department. As for the results for 4 years, if the value obtained by dividing the predicted pass rate by the actual pass rate is taken as the reliability, the reliability as of 2 months before each year was 0.86, 1.00, 0.99, 1.13. The ideal scenario is to change the measures if the initial prediction is bad, then the prediction just before the exam will have a pass rate of 100%, and 100% pass on the day of the national exam can be achieved.