Item Selection Strategy Of Computer Adaptive Testing Based On Content Balancing

The computer adaptive testing is a testing method that can estimate the current ability level of the examinee in time according to the response of previous test items,and select the item that best meets the current ability level from the item bank.The development of computer adaptive testing provides a new method for educational evaluation.Compared with traditional testing methods,adaptive testing has the advantages of higher test efficiency and can get more accurate results.Because the core of computer adaptive testing is to select the most suitable item according to the current ability estimation of the examinees,therefore,how to select the test item,that is,the item selection strategy is one of the core issues in the research area of computer adaptive testing.Although many strategies have been proposed,there is little concern to the control of contents balancing.However,contents balancing has a great impact on the test accuracy.Therefore,in this thesis,a contents balancing item selection strategy named c-STR-ST is proposed,and Monte Carlo simulation is used to verify the test results of c-STR-ST.Compared with the existing CCAT,MMM and STR-C strategies,the results of Monte Carlo show that the c-STR-ST strategy is superior to the other three strategies in terms of test accuracy,exposure rate,and test overlap.In addition,this thesis designed and developed a computer adaptive testing system based on the c-STR-ST strategy,and built an item bank of mathematics in the first year of senior high school.This system and item bank are applied to the evaluation of mathematics for senior middle school students.The test results include students’ mastery of knowledges,cognitive diagnosis of each student,which provide teachers and students a more detailed and individualized evaluation results.