用SAS软件实现因变量为多值有序变量的多重logistic回归分析  

Multiple logistic regression analysis of multi-level ordinal dependent variables with SAS software

作　　者：胡良平[1] 孙日扬[1] 

机构地区：[1]军事医学科学院研究生部生物医学统计学咨询中心,北京100850

出　　处：《药学服务与研究》2014年第4期258-263,共6页Pharmaceutical Care and Research

基　　金：国家科技重大专项重大新药创制资助项目(2011ZX09302-006-01)

摘　　要：因变量为多值有序变量的多重logistic回归分析是一种研究多值有序的因变量与一组自变量之间依赖关系的有效方法。该方法对自变量的性质几乎没有限制,但要求有较大的样本量并且样本的总体代表性好。回归系数具有明确的实际意义,可依据回归系数得到优势比的估计值。本文结合实例,详细介绍如何用SAS软件拟合此回归方程,并进行结果解读。Multiple logistic regression analysis of multi-level ordinal dependent variables is an effective method for the research of the dependent relationship between multi-level ordinal dependent variables and a group of independent variables. It has few restrictions on data property of independent variables, but requires a large sample size and the sample generally has good representation. With clear practical signficance, the estimated value of odds ratio could be obtained by practical regression coefficients. With a practical example, this article introduces the solution of multiple logistic regression by SAS software and explains the results.

关 键 词：统计学 LOGISTIC回归分析 多值有序变量 SAS实现 

分 类 号：O212.6[理学—概率论与数理统计] R195.1[理学—数学]