摘要
带有层套关系因子变量的协方差分析模型在实际应用中有广泛的适应性。然而,由于此类模型的特殊性,我们不能利用一些通常的统计软件对它进行模型拟合或参数估计。针对这一问题,本文提出了对该模型进行分层拟合以及由设立哑变量将此类模型转化为通常的协方差分析模型再加以解决这两种可行而方便的办法。本文还提出了与之相关的模型变量优选的方法以及该模型在教育数据分析上的应用。
The covariate analysis model with some nested factors is widely used in practical problem.But it can not be fitted directly by using some general statistical software,due to its speciality.For this problem,this paper propose two methods which are both feasible and convenient.The first is to divide the nested factors into two models and fit the two models separately, the second is to transfer the special model to a normal one by establishing some dumb variables and fit it by the normal method. This paper also introduces a relevant method of variable selection and some application of the model in educational data analysis.
