Analysis Of Examination Missing Data Based On Bayes Estimation And Rasch Model Fitting

The data were randomly missing Bayesian estimation and Rasch model fitting using Winsteps and Matlab software to validate the 9 single-choice questions of English test of high school high school students in Qiannan Prefecture of Guizhou Province and 2000 subjects as raw data.The results are as follows: First,this study uses Gibbs sampling to process the missing data in English test data closer to the original parameter estimation results of the original data than the missing value control group treated with the general mean substitution method,proving that the missing data is reasonably processed The validity and necessity.Secondly,this study can reflect the characteristics of English test questions by dealing with the non-negligible missing data,and prove the rationality and necessity of dealing with the missing data reasonably.Thirdly,all the statistical indicators of the English test items selected in this study are in line with the standard of psychometric measurement.The subject is set up reasonably and the measure of the English level of the subjects is measured.