Research On The Relationship Between Correct Pattern Classification Rate And Quality Of Item Bank Usage In Cognitive Diagnostic Computerized Adaptive Testing

Item Selection strategies is the very important part of the cognitive diagnostic computerized adaptive testing,and it is related to test the accuracy,security,efficiency,reliability and validity.Item Selection strategies is commonly used in CD-CAT can result in low utilization rate test and project problems.Good Item Selection strategies should balance measurement accuracy and uniformity of Item Bank Usage requirements.Domestic and foreign scholars on the relationship between the test accuracy and test efficiency theory are discussed.In this paper,from the empirical level to investigate the relationship between the test accuracy and test efficiency,provide a reference for choosing appropriate item selection strategy,and guide the future research and development of cognitive diagnosis computer adaptive test.In order to solve the above problems,this article adopts the Monte Carlo(Monte Carlo)simulation method and a series of studies in DINA model as an example,including a preliminary research and four formal experiments:Preliminary study uses the literature analysis which contained both measuring precision and convenience usage data,in order to explore the relationship between them.A formal study using simulation study,to compare seven Item selection strategies such as KL,SHE,PWKL,PPWKL,MPWKL,HKL,PHKL from two aspects of project exposure control and measurement accuracy to evaluate their performancein the uniformity of item bank usage and Measurement accuracy,at the same time to explore th relationship between them.And to explore different test length,number of attributes,the distribution of the participants and the number of participants how to influence them.In this paper,main conclusions are as follows:(1)Measuring accuracy of MPWKL and SHE is higher;If the comprehens ive utilization of test precision and quality of item bank usage,PHKL,PPWKL is better,the second is MPWKL.(2)On the whole,the use of the test accuracy and quality of item bank usage,PHKL and PPWKL at the same level,for the first tier,the criterion for rate of medium,the question bank utilization rate is highest,the best comprehensive performance;HKL and PWKL at the same level for the second tier,the criterion for rate of medium,general question bank utilization;SHE and MPWKL are the third echelon,criterion: the highest rate,the question bank utilization rate is the worst;KL for the fourth tier,criterion for low rate,convenience usage.(3)The test length is proportional to the test precision and is inversely proportional to the question bank using uniformity.When test length is 16,a variety of topic selection strategy of integrated performance is the best.(4)Sample size has smal er influence on topic selection strategy of test accuracy and test efficiency.The criterion for topic selection strategy first to rise,with a total of 1000 subjects,after changes gently.Uniformity test using comprehensive index down first,after the number of participants is 1000,changes gently.For a number of subjects of 1000 comprehensive performance is the best.(5)Number of attributes is inversely proportional to test accuracy of topic selection strategy.Attribute number is less,integrated performance of selected topic strategy is better.(6)The subjects knowledge state distribution form for negative skewness distribution,test precision is highest.Subjects for normal distribution,SHE,PWKL,MPWKL,PHKL topics such as strategy of question bank the highest utilization rate.