An Equivalence Study Of Multi-cluster Unanchored Item Tests With Post Hoc Introduction Of Multiple Linking Groups

With the announcement of the "Decision of the Central Committee of the Communist Party of China on Several Major Issues Concerning Comprehensively Deepening Reform",various regions across the country had proposed corresponding guidelines for "deepening comprehensive reform in the field of education".Guided by these guidelines,the reform of the examination system had begun.Taking the Zhejiang College Entrance Examination for English as an example,the examination plan of two exams per year has been adopted.When comparing the true scores of two tests,it is necessary to equate the two tests.In the actual process of educational evaluation,there are often situations where there are no common items or participants.In this situation,it is necessary to find alternative solutions to equalize the tests without anchors.One of the effective solutions is to join the linking group to participate in the estimation of test parameters after the event to achieve the purpose of equivalence.At the same time,with the gradual development of education quality monitoring,it is often necessary to monitor the quality of a large number of schools.The traditional design of linking group can not meet the requirements of the proportion of anchor questions and can not adapt to the equivalent situation of multiple groups without anchor questions.Moreover,the use of different equivalence theories or design methods may cause deviations in the estimated scores.Analyzing these factors that may affect the accuracy of equivalence,and controlling the variations caused by these factors in a targeted manner,thereby improving the accuracy of equivalence,is of great significance.This study investigated the equivalence performance of multi linking group design under various conditions,and explored the impact of sample size,group characteristics of the test group,and anchor question characteristics on multi linking group design,in order to provide some theoretical reference for selecting effective equivalence design and equivalence theory under different actual measurement scenarios.This paper is divided into four studies.Study 1 is to explore the effectiveness of multiple linking group design in the context of multiple equivalence groups;Study 2explored the influence of the number of subjects in the linking group on the equivalent design;Study 3 explores the impact of the group’s abilities on equivalence design;Study 4 explores the influence of anchor question parameters on equivalent design and obtains several main conclusions:(1)Multi linking group design shows stronger applicability.(2)With the increase of the number of linking group,the equivalent effect of linking group design is better.(3)When the linking group subjects’ ability is relatively moderate,the linking group design has the best equivalent effect.(4)Compared with midi anchor test and random anchor test,the effect of mini anchor test extraction on the equivalent design of linking group is better.