论文部分内容阅读
含有序变量的R×C列联表可用对数线性模型来作分析,可给予有序变量每一等级一个分数,建立一个“有序交互作用”模型,据此模型可列出求期望频数的最大似然估计值(MLE)的似然方程,用迭代法可求得这些估计值,并由此求得该模型的拟合优度(一般用统计量X2和G),从而可对两变量间是否存在有序交互作用作出统计推断。以前对医学上常遇到的等级分类资料构成的列联表尚只能用一般的R×C表方法进行分析,现应用本方法可充分反映由序提供的信息。
R × C Contingency Tables with Ordinal Variables can be analyzed using a log-linear model. A score can be assigned to each level of ordinal variables and an “ordered interaction” model can be established. According to this model, the desired frequency can be listed. Likelihood equations of maximum likelihood estimation (MLE), these estimates can be obtained by iterative methods, and the goodness of fit of the model (generally statistics X2 and G) can be obtained, so that two variables can be Whether there is an orderly interaction between the statistical inference. In the past, the contingency table composed of hierarchical classification data often encountered in medicine can only be analyzed using the general R×C table method. The present method can now fully reflect the information provided by the preface.