论文部分内容阅读
目的探讨综合权重在复杂随机抽样数据线性回归分析中的意义和作用。方法基于蒙特卡洛随机模拟思想,采用SAS中REG和SURVEYREG两个不同的多重线性回归分析过程,分别对同一批复杂随机抽样数据(n=6756)在不同随机抽样率条件下进行回归建模,对所得结果进行比较。结果在未考虑和考虑观测权重与抽样权重的多重线性回归模型拟合的结果中,自变量的偏回归系数、标准误及P值的大小均有所不同。结论在对基于不同抽样率的复杂随机抽样资料,尤其是分层随机抽样调查资料的回归建模中,采用多重线性回归模型拟合资料时,将调查数据的综合权重纳入统计分析,方能更准确、灵敏地进行回归系数的参数估计和对结果变量的统计预测。
Objective To explore the significance and role of synthetic weights in linear regression analysis of complex random sampling data. Methods Based on the Monte Carlo stochastic simulation method, two different multiple linear regression analyzes of REG and SURVEYREG in SAS were used to model the same batch of complex random sampling data (n = 6756) under different random sampling rates. The results were compared. Results In the multiple linear regression model without considering and considering the observation weight and sampling weight, the partial regression coefficients, standard errors and P values of the independent variables were all different. Conclusion When using multiple linear regression models to fit data in complex random sampling data with different sampling rates, especially stratified random sampling data, the comprehensive weight of survey data can be included in statistical analysis Accurately and sensitively estimate the parameters of regression coefficients and predict the outcome variables.