【摘要】 该文对于教育与心理测量理论中常用的IRT模型的未知参数提出了一种新的参数估计方法———Berkson参数估计方法,并通过大量的实验考察了该方法的性能,发现研究结果比BILOG算法的估计结果略好,这将有利于更好地进行题库建设,并且有较好的恢复能力,较好的适应性和较好的收敛性.更多还原

【Abstract】 This paper proposes a new method of estimating parameters discovered by Berkson which is often applied in estimating the unknown parameters of IRT model that can be usually found in educational and psychological estimation theory.Furthermore,we carry out many experiments to review the performance of this method,finding that results are better than that of BILOG algorithm,which is good for constructing item bank.The new method has good ability of recovering,adjusting and converging.更多还原