Font Size: a A A

Discrimination Indices And Dual-objective Item Selection Criteria For Cognitive Diagnostic Computerized Adaptive Testing

Posted on:2021-02-14Degree:MasterType:Thesis
Country:ChinaCandidate:Q TangFull Text:PDF
GTID:2415330623973428Subject:Education
Abstract/Summary:PDF Full Text Request
Computerized Adaptive Testing(CAT)based on Item Response Theory(IRT)can get overall ability level or psychological characteristics of examinees.CAT based on Cognitive Diagnostic Theory(CD-CAT)can report the knowledge structure,processing skills or cognitive process in specific fields of examinees.Actually in the test and teaching situation of real life,the examiners often need to know the ability scores and knowledge states at the same time.But two tests(IRT-CAT and CD-CAT)in one time is time-consuming and labor-consuming,which does not meet the requirements of practice.Therefore,the need for getting overall ability sores and attribute mastery status in one test led to the birth of dual-objective CD-CAT.No matter what kind of CAT,item selection method is the core.There are many basis of item selection criteria,one of which is discrimination index.Discrimination indices can reflect distinguishing ability of item,and directly affect the accuracy of the test.By untangling the researches of discrimination indices and item selection criteria in IRT-CAT,CD-CAT and dual-objective CD-CAT,it is found that no scholar has proposed discrimination indices in dual-objective CD-CAT at present,and the existing item selection criteria are less,and the measurement accuracy needs to be improved.Therefore,this paper put forward two discrimination indices(KLCDI and JSCDI)which can be used in dual-objective CD-CAT,and then test their validity by examining the correlation with other discrimination indices and contribution of paper formation.The result shows that KLCDI has a high correlation with the existing IRT and CD-CAT discrimination indices.The higher KLCDI items have,The higher accuracy rate of knowledge states and ability scores we get.So KLCDI can be used as a dual-objective CD-CAT discrimination index.In addition,based on two ideas: combining item selection strategy and discrimination index,modifying KL information,this study constructs four item selection criteria for dual-objective CD-CAT(IPA-KLCDI,DPWKL1,DPWKL2,DSKL),and It is the first time to test item selection strategies DKL1 and DKL2.Compared with the existing strategies(random,JS,IPA)in different test lengths,the result shows that IPA-KLCDI can improve testing accuracy,but the utilization rate of item bank and test overlap rate of IPA and IPA-KLCDI need to be improved.Compared with JS,DKL1 and DKL2,DPWKL1,DPWKL2 and DSKL all can improve measurement accuracy,among which DPWKL1 has the highest test accuracy,the highest utilization rate of item bank,and the lowest test overlap rate.Moreover,DSKL has the highest measurement accuracy of ability when the test length is 40.In general,this paper has got satisfactory results for the discrimination indices and item selection criteria of dual-objective CD-CAT.We should strengthen the construction of multiple strategies,polytomous attribute scoring,polytomous item scoring and even more complex test conditions,as well as the proposal and practical application of non-parametric item selection criteria of dual-objective CD-CAT in the future research.
Keywords/Search Tags:Cognitive Diagnostic Theory, Computerized Adaptive Testing, item selection criteria, measurement accuracy
PDF Full Text Request
Related items