Bayesian Analysis for the English Entrance Examination

• Main Authors: Pei-Yu Luo, 羅培語
• Other Authors: Tzu-wei Cheng
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/h75ebp

碩士 === 中原大學 === 應用數學研究所 === 106 === In recent years, Item Response Theory (IRT) is a contemporary developmentin modern test theory. There are two types of IRT models dealing with dichotomous scoring and polytomous scoring data. In this paper, three dichotomous scoring models are used to analyze the data from the English test of 2013 freshmen class at Chung Yuan Christian University. There are 3138 students participating in the examination. It includes forty-eight multiple choice items. One-parameter model, two-parameter model, and three-parameter are considered appropriate for dichotomous scoring items. Parameter estimation is importantinIRT.One-parameter model provides estimates of item difficulty only. Twoparameter model provides estimates of discrimination. Three-parameter model refer to as a guessing parameter. Using Markov chain Monte Carlo (MCMC) technique and Bayesian method are used to estimate the parameters in IRT. In the paper, OpenBUGS software is used. After simulation many times, the averaged value is used to analyze and explain English test result.