Conditional Augmented Data EM Algorithm For Two Parameter Normal Ogive Model

Item response theory(IRT)is a potential trait theory.At present,it is mainly used in the field of educational statistics and psychological measurement,and it is also widely studied and used in other fields,such as pedagogy,measurement and evaluation,medical health and so on.Whether the parameters in the model can be estimated accurately and efficiently is the premise of the practical application of item response theory.In this thesis,conditional augmented data EM(CADEM)algorithm is used to estimate the parameters of the two parameter normal ogive(2PNO)model.The detailed steps of CADEM algorithm are given.This algorithm overcomes the problems of EM algorithm in calculating complex integrals and improves the computational efficiency.Finally,through simulation experiments,CADEM algorithm is compared with EM algorithm.For the evaluation indexes of the results,standard error(SE),bias and root mean square error(RMSE)are considered respectively to verify the effectiveness and high efficiency of this algorithm.