一种题组反应模型的构建与统计分析

摘要

在一般题组反应模型基础上，通过构造题组判别参数，提出了带有题组判别参数的两参数正态卵形题组反应模型。在贝叶斯理论框架下，基于数据扩充技术，使用Gibbs抽样方法，研究了该模型参数估计问题以及该模型的适应型问题，模拟研究表明，在参数返真方面，新提出的模型在一定程度上优于原有相应的项目反应模型和题组反应模型。%A testlet response theory was proposed by constructing testlet discrimination parameter based on the common item response theory model.Gibbs sampling procedure via a data augmentation scheme was used to investigate the parameter estimation and the adaptability of the new model under the frame of Bayesian theory.Simulation sresults indicate that item parameters recovery is overall batter than that corresponding item response model and testlet response model.