Order Restricted Weighted Estimation Method In Mixed Tests

Most of the exam papers are composed of both dichotomous and polytomous items,polytomous items often yield more information than dichotomous items.In this article,an adaptive weighted maximum-likelihood estimation which weighted for each subject respectively is proposed. We consider weighted maximum-likelihood estimation when the weight is proportional to the item information,and meet some kind of coherent constraints. By the istonic regression theory knowledge, we consider the isotonic regression algorithm for preorder and given function, and complete the whole proof. Using this method,we give weight function which is nearest from item information function under Euclidean distance and satisfies given Quasi sequence.A simulation study which use the obtained weight is conducted.The results show that the proposed method has better propertis than the maximum likelihood estimation.