Font Size: a A A

Research On Security Strategy Of CAT And Rule Of Termination In A-Stratified

Posted on:2018-11-12Degree:MasterType:Thesis
Country:ChinaCandidate:W Q YangFull Text:PDF
GTID:2348330512994716Subject:Computer Science and Technology
Abstract/Summary:PDF Full Text Request
Since twentieth Century,based on item response theory(Item Response Theory,IRT)of the computerized adaptive test(Computerized Adaptive,Test,CAT)as compared with the traditional paper pencil test is more flexible and efficient,so more and more people of all ages,more and more extensive application.This paper analyzes several strategys of selected item in GRM model under the indefinite length test in CAT,and finds that the more concentratied the value of a function of information and distinction become,the more uniform the item calls.Therefore,on the basis of He Xiang's strategy,the function of the Fisher information quantity and the discriminant parameter a is transformed into an exponential function with an index less than 1,make its function value distribution is more concentrated,this project when the call is more uniform,so as to improve the efficiency of test and safety test.However,the improvement of the safety performance of the items bank will result in the accuracy of the measurement and the efficiency of the test.In order to take into account the accuracy of the measurement and the safety of the items bank,this paper defines the exponential control index function,which gradually weakened of the influence of the exposure factor in the selection strategy,increase the possibility of selecting the item with large amount of information,and improve the measurement accuracy and the efficiency of the test.Finally,two new selection strategies were applied to the fixed length test and the variable length test of the 3PLM model in the GRM model and the 0-1 score.In addition,this paper also studies the analysis of the indefinite length termination rules of astratified by CAT,and finds that the proportion of the total amount of information in the whole test information is reduced and the layer with higher information is increased The total amount of the total test information contributes to the optimization of the number of items and test efficiencies per capita,so two new termination rules are designed.The new termination rule is optimized in the project call uniformity,the per capita number of items,the test efficiency,and the test overlap rate.In the GRM variable-length model,the uniformity of the project call,the number of items per capita,Efficiency,test overlap rate while taking into account the accuracy of the measurement.
Keywords/Search Tags:Computerized Adaptive testing, Item Response Theory, Item Selection Strategy, Fisher Information, Termination rule
PDF Full Text Request
Related items