医学随访研究中重复有序分类资料的分析方法

Analytical method of repeated ordered categorical data in medical follow -up study

介绍医学随访研究中重复有序分类资料的分析方法—累积比数模型和相邻比数模型。方法 :采用加权最小二乘法对模型中的参数进行估计 ,并根据剩余 χ2 的大小选择模型。结果 :选择有交互作用的模型对资料的边缘合计进行拟合 ,剩余 χ2 为 5 62。结论 :累积比数模型用于重复有序分类反应资料的分析更有效。

Objective: To introduce the methods used in repeated categorical data of clinical follow up study:cumulative logits model and adjacent category logits model.Methods:Weighted least squares is used in estimating parameters and models are selected by residual χ 2. Results:We allow for interaction with the cumulative logits model to fitting marginal response distribution with the cumulative logits model to fitting marginal response distribution on values of explanatory variables and on the occasion of response.The residual χ 2 was 5.62.Conclusion:It is more effectively applied with cumulative logits than adjacent category logits in repeated categorical response data.