复杂抽样数据的logistic回归分析方法及其应用

The Application of Logistic Regression in Complex Sample Survey Data

目的探讨抽样权重在复杂抽样数据logistic回归分析中的重要性。方法采用SAS中PROC LOGIS-TIC和PROC SURVEYLOGISTIC语句对数据进行统计分析,并对结果进行比较。结果在未考虑和考虑抽样权重的lo-gistic回归模型拟合结果中,自变量的偏回归系数和OR值大小及其可信区间都有所不同。结论在logistic模型拟合中,纳入调查数据的抽样权重进行统计分析,从而能更加准确地进行统计推断。

Objective To explore the importance of the sampling weight in the logistic regression on complex sample survey data. Methods The logistic and surveylogistic procedure of SAS were used to analysed the survey data, and the results of which were then compared. Results In the logistic regression model with or without sample design, the regression coefficients and the confidence intervals of odds ratio were different. Conclusion In order to make statistically valid inference for the population, sample design must be incorporated into the data analysis.