多分类有序反应变量Logistic回归及其应用

Polytomous Logistic Regression for Ordinal Response and It's Appliance

对近年发展起来的多分类有序反应变量Logistic回归进行了补充和完善 ,设计出对样本数据的应用条件做 χ2 检验的方法 ,证明了比数比在分析自变量作用中的实际意义 .随机抽取 2 0 0 2年国家执业医师资格临床实践技能考试 30岁以下本科毕业考生 5 0 0 0人 ,以考生的考试成绩 (3分类有序变量 )为反应变量 ,以考生的性别、工作单位级别、考试评分方法、考官回避制度、考题重复次数、考试合格率控制措施等 9个影响因素为自变量 ,进行多分类有序反应变量逐步Logistic回归分析 .研究结果已为卫生部改革和完善国家医师资格考试方法。

Polytomous Logistic regression for ordinal response was perfected,the mothod of χ 2 test was designed for applicable conditions about sample data,and the effect of independent variable to response variable by odds ratio was proved.5 000 candidates (ages<30) who had the degree of bachelor of medicine taking for the clinical examination for national licensing of doctors in China in 2002 were sampled at random.Polytomous logistic stepwise regression for ordinal reponse was achieved by response variable:the performance of clinical examination and 9 independent variable:sex of candidates,the grades of units that candidates were in,the methods of assessing performance,avoiding measure by assessor,et al.The result of the studies provided scientific evidences for laying down the methods and the measures of the national examination for licensing of doctors in China by the Health Ministry.