Research On Scoring And Classification Methods And Application Of Multiple-choice Items Cognitive Diagnosis

Among the test items for cognitive diagnosis,multiple-choice items(hereinafter referred to as the MCIs)are one of the most commonly used test item forms.Currently,the existing MCIs of cognitive diagnosis refer to multiple-choice questions with only one correct answer.The MCIs in this study refer to multiple choice items with distinctive characteristics of Chinese exams.Different tests have different scoring methods of MCIs and carry different information,so it is necessary to study the scoring methods of MCIs for cognitive diagnosis.The MCI cognitive diagnostic test can better estimate the true mastery of the subjects.So far,however,no research has developed an MCI cognitive diagnostic model for this question,nor is there an MCI for cognitive diagnosis to study the scoring method.In this study,five multi-choice cognitive diagnosis models and corresponding scoring and classification methods are proposed,and the five models are compared and applied.This study was divided into three parts.In part 1,we discussed the performance of the analytical model(I-DINA and O-DINA model)under different conditions(test length,test quality,number of subjects).In part 2,we primarily discussed the performance of the analytical model(I-DINO,O-DINO,and O-DINAO model)under different conditions in option units.In part 3,we prepared a multiple-choice test of physics(volume 1,grade 8),applies research on five multiple-choice cognitive diagnosis models,and compared the performance of them and their scoring and classification methods in empirical data.The results were as follows:(1)As the test quality declined,the result of Attribute Correct Classification Rate(ACCR),Average Attribute Correct Classification Rate(AACCR),Pattern Correct Classification Rate(PCCR)of the accuracy of the I-DINA,I-DINO,O-DINA,O-DINO and O-DINAO models were all in a downward trend,indicating that model and attribute criteria were greatly affected by project parameters.(2)As the number of participants increased,the guess parameters “g” of the I-DINA,I-DINO,O-DINA,O-DINO and O-DINAO model and the error parameter“s” were closer to the true value.That illustrated that the estimation of projectparameters was affected by the number of respondents.When the number of respondents increases,the project recurrent value was better.(3)Under the 4 attributes as well as different subjects,test lengths,and test quality,the score estimation results of the O-DINAO model are consistent with O-DINO model,and better than the other three models.In this study the O-DINAO model and O-DINO model was more recommended.(4)Empirical research showed that the O-DINAO model proposed in this study has higher parameter estimation accuracy and better model data fitting than the other four models.In practical applications,it is recommended to use the O-DINAO model for the MCI recognition diagnosis.