Development And Application Of Cognitive Diagnosis Model For Missing And Random Guess

With the development of psychological and educational tests,researchers not only pay attention to the measurement of the ability or trait level of the subjects,but also pay attention to the mastery of specific knowledge structure.Therefore,cognitive diagnostic assessment has become a hot spot in the field of psychological and educational measurement.However,no matter in the general test or cognitive diagnostic test,in the actual test process,the subjects not only make the problem response,namely normal response,but also have some abnormal response due to individual characteristics or item characteristics,especially missing and random guess.It is found that the occurrence of abnormal response will lead to a certain bias in parameter estimation,threatening the reliability and validity of the test.Therefore,how to deal with the common abnormal response is crucial to the analysis of test data.However,most of the current modeling research focuses on the field of item response theory,while the field of cognitive diagnosis is still in the early stage of development.In view of this situation,inspired by the IRTree framework,this paper develops a cognitive diagnosis model that considers missing response and random guess at the same time,so as to enrich the modeling research on abnormal response responses in cognitive diagnosis assessment,and provide ideas and references for future research on abnormal response in the field of cognitive diagnosis.Firstly,this paper summarizes the concepts,theories and related researches involved in the research.Then,the modeling idea and framework of the new model are expounded,and the priori information of parameter setting and the Markov Chain Monte Carlo are explained.MCMC)estimation method;Then,a 2×2×2×2 simulation study was designed to verify the accuracy and robustness of the parameter estimation of the new model,and to compare the results of the new model and the traditional cognitive diagnosis model on the attributes of the subjects.Finally,the new model was applied to the real data of the 2019 TIMSS Grade 8 math test,and the model fit was compared with the traditional cognitive diagnosis model,and the subjects were selected for analysis,to illustrate the advantages and application value of the new model.The results showed that:(1)Monte Carlo simulation showed that the newly developed IRTree-LCDM model in this study had better parameter estimation accuracy and diagnostic accuracy than the traditional cognitive diagnostic model(LCDM).(2)The proportion of abnormal response will affect the accuracy rate of attributes.The higher the proportion of abnormal response,the lower the accuracy rate,but it has little influence on the estimation of other parameters.(3)The new model performs well in the empirical data and has better fitting effect than the traditional cognitive diagnosis model,which has certain practical value.