Abstract
In this research, we focused on the normal density type of the continuous response model (CRM; Samejima, 1973, 1974; Wang & Zeng, 1998), and proposed a method to estimate item parameters for this type of CRM by marginal maximum likelihood estimation method with the EM algorithm (Dempster, Laird & Rubin, 1977). The proposed method did not require any numerical approximation in the expectation step. Moreover, we showed that exact solutions could easily be obtained in the maximization step when flat priors of item parameters were assumed. Therefore, our method could be noniteratively completed through each EM cycle. In addition, we examined the performance of the proposed method using simulation data. We then confirmed the accuracy of our method by only taking flat priors into consideration.
