Research on test equating usingitem response theory and development of its software

点击翻译按钮获取中文摘要

(1)Development of the software for evaluating the stability of test equatings using item response theoryThe objectives of this research are to provide methods of test equating using item response theory (IRT), which is a current method in test theory, and to develop the software to evaluate the stability of various methods of IRT equating. The following are the methods of equating dealt with in this project.(a)Equating using moments of the parameters in IRT with the equating design of common examinees : M/S (mean/sigma) method, M/M (mean/mean) method and M/GM (mean/geometric mean) method(b)Response function methods of equating with the equating design of common examinees Item response function method and test response function method(c)Equating methods by least squares with the equating design of common examinees : Unweighted least squares method, weighted least squares method and generalized least squares method(d)True score IRT equating(e)Observed score IRT equating(f)Estimation methods of IRT parametersIn the above methods, (c) to (e) are the methods developed in this research project.(2)Release of the software and the final reportThe software is coded in Fortran 90 and takes a form of a set of subroutines, which is called EL Version 1.0 whose manual is released in the homepage of the project manager. The source codes are given upon request to the manager. The software was presented at the 54^<th> Session of the International Statistical Institute at Berlin, Germany.