改进累积logistic回归在生存质量研究中的应用

The Application of Improved Cumulative Logistic Regression in the Study on Quality of Life

目的通常多分类有序变量类别之间的差距并不相等,一般的数据分析方法却将其作等距对待,本文探讨有序变量确实存在间距差异时,对累积logistic模型加以修正的方法。方法若经过检验发现有序变量确实存在间距差异,通过引入虚拟工具变量对累积logistic模型加以修正。结果本文实例中有序反应变量存在间距差异,应用改进累积logistic模型,通过以工具虚拟变量的不同取值为参照,使结果的解释更符合实际。结论若经过检验发现有序变量确实存在间距差异,通过对改进累积logistic模型的实际应用,取得良好效果。

Objective The tratitional analysis method of data treat categorical ordered variables with equal intervals even though the differences among categories are not equal.The paper explores one method to modify cumulative logistic regression if the differences of categorical ordered variables are not equal.Methods If the differences of categorical ordered variables are not equal based on the statistic and test method,instrument dummy variable is used to modify cumulative logistic regression.Results An working example is used to illustrate the method.Letting different values of instrument dummy variable as references make the interpretation of the results more consistent with the practice.Conclusion If there are differences for different intervals,the application of improved cumulative logistic regression is satisfying.