Comparison And Application Of Item Selection Algorithm Based On Multiple-choice CDM In CD-CAT

Cognitive diagnostic model(CDMs)are models that can diagnose the structural of the potential knowledge,traits or skills of subjects.Especially in the field of future education,the cognitive diagnostic model has a broad application prospect in the diagnosis and assessment of students’knowledge,traits or skills.With the increasing use of computer teaching in the classroom,the combination of computerized adaptive algorithm of cognitive diagnostic testing(CD-CAT)highlights the advantages of more efficient diagnosis.However,multiple-choice items(MCI)is one of the most common types of educational tests nowadays.The original CD-CAT model,which only focuses on the correct answers,obviously ignores the diagnostic information of the incorrect choices and wastes the information of the subjects.The original CD-CAT is not suitable for CD-CAT involving multiple-choice items.The purpose of this study is to explore CD-CAT based on the multiple-choice cognitive diagnosis model,compare the performance of the algorithm of item selection and apply it to the students’mathematical knowledge diagnosis,to provide the fourth grade students with the knowledge structure advantages,disadvantages and guidance information,help students to learn independently and teachers to carry out remedial teaching.This research extended two item selection algorithm based on multiple-choice:MC-SHE and MC-GDI.Then two Monte Carlo simulations were carried out to investigate its performance and an empirical study was conducted.The first study is to explore the performance of the item selection algorithms under different experimental conditions.There were Pattern correct Classification Rate(PCCR),Marginal Match Ratio(MMR),Test Overlap Rate(TOR)and item bank using uniformity index(X~2).The second study is to explore the performance of item selection algorithms under different experimental conditions.The evaluation indexes include the minimum test length,the maximum test length,the average test length,the standard deviation of the test length,the ratio of the maximum test length and the PCCR.In the third study,CD-CAT based on a multiple-choice cognitive diagnostic model,was applied to the diagnosis and assessment of mathematics knowledge structure of grade four students.The results show that:(1)under the condition of fixed length CD-CAT,MC-GDI performs best in the index of PCCR,especially when the cognitive attributes are 3.The comprehensive performance of MC-SHE was better than random selection.(2)under the condition of indefinite length CD-CAT,MC-GDI performed better than MC-SHE in the accuracy of 0.65,0.75 and 0.85.(3)there were 60 items in the multiple-choice CD-CAT applied research,and the average length of the students’tests was 12.94.The subjects were selected from the open item bank of the TIMSS test,and the radar chart of the students’mathematical knowledge attributes mastery pattern was obtained.The proportion of the students’overall attributes mastery was obtained too.In summary,MC-GDI is better than MC-SHE,and can be applied to CD-CAT based on multiple-choice cognitive diagnosis model.The application of multiple-choice CD-CAT provides more abundant diagnostic information,greatly improves the diagnostic efficiency,and provides guidance for the remedial learning of students and teachers.