摘要
[目的 ] SPSS软件包的固有命令无法拟合条件 logistic回归模型 (CL RM) ,探索以此软件包来拟合 ,将丰富其功能、受到广大用户的欢迎。 [方法 ]以 SPSS软件包固有的 Cox回归模型命令 COXREG,根据 Cox回归模型分层变量控制的理论 ,创造所需新的虚拟变量 (Time和 Strata )来进行拟合。以 6个实例进行运算 ,并与 SAS、STATA软件包运算结果相对照。 [结果 ] SPSS软件包 Cox模型分层变量控制的方法 ,可正确地拟合出条件 logistic回归模型 ,可得到与 SAS与 STATA软件包计算完全相同的结果 ,而且运算结果相互对应的参数也相同。 [结论 ]根据 Cox回归模型分层变量控制的理论 ,创造所需新的虚拟变量 (Time和 Strata) ,以 SPSS固有的 COXREG命令 ,可拟合 CL RM,与SAS、STATA运算结果也相同 ,且编程简便易行 ,可适用于 1∶ 1、 1∶ 2或 n∶
To study the fitting method of Conditional Logistic Regression Model (CLRM) by using SPSS package because the original commands (LOGISTIC REGRESSION) can not do it According to the theorem on strata variable control when fitting Cox Regression Model, new dummy variables (Time and Strata) were created for fitting the model 6 examples were selected for running the model CLRM can be fitted correctly by Cox Regression in SPSS and the same results by SAS and STATA can be obtained [Conclusion] According to the theorem on strata variable control when fitting Cox Regression Model, CLRM can be fitted by means of the original COXREG command in SPSS package And the programming is simple, easily applying and can be used for 1∶1, 1∶2 or n∶m (1∶4) matching comparison in CLRM The correspondence parameters of the results are the same as those by SAS and STATA
出处
《海峡预防医学杂志》
CAS
2002年第6期1-4,共4页
Strait Journal of Preventive Medicine
