纵向数据项目反应理论中项目参数的估计方法

发布时间：2018-05-06 16:34

本文选题：多维项目反应理论 + 极大似然　；参考：《东北师范大学》2015年硕士论文

【摘要】：在研究纵向项目反应数据的时候,由于潜变量的维数不是一维的,其中多重积分的问题增加了计算的难度,需要寻找一种处理多维积分的方法,使得参数估计的结果尽可能的精确。在本文中,采用两参数逻辑斯蒂克模型和正态卵形模型,在总体能力参数分布已知的情况下,利用极大似然估计方法对难度参数和区分度参数进行估计,在处理多维积分时,分别介绍了高斯—埃尔米特插值方法和M-H抽样方法,最后利用统计模拟对以上所选取的概率模型,总体能力参数的协方差矩阵以及处理多维积分的不同情况进行检验。
[Abstract]:In the study of longitudinal item response data, because the dimension of latent variables is not one-dimensional, the problem of multiple integrals increases the difficulty of calculation, so it is necessary to find a way to deal with multidimensional integrals. The result of parameter estimation is as accurate as possible. In this paper, the two-parameter logical Steak model and the normal oval model are used to estimate the difficulty parameters and the discriminant parameters by using the maximum likelihood estimation method when the distribution of the total capability parameters is known. The Gauss-Hermite interpolation method and M-H sampling method are introduced respectively. Finally, the probability model selected above, the covariance matrix of total capability parameters and the different cases of dealing with multidimensional integrals are tested by statistical simulation.
【学位授予单位】：东北师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：C81

【参考文献】