Discounted likelihood estimation method for item parameter in nominal response model
QU Shuang1， GU Xiyuan2， ZHANG Yinan3
1.Faculty of Mathematics， Changchun Normal University， Changchun 130000；2.The No. Three Senior High School of Mudanjiang， Mudanjiang Heilongjiang 157000；3.Changchun University of Finance and Economics， Changchun 130000
Abstract：Item response theory (IRT) has been widely used in educational and psychological measurement. Nominal response model (NRM) is one of the IRT models. In this paper， we propose a new parameter estimation method——discounted likelihood method (DLM)， in which a type of quasi-Newton methods， BFGS method， is used to estimate item parameters. Simulation results indicate that the DLM is a less computationally demanding method and can fit the NRM well.
2017, Vol. 51 
