摘要
目的探究行列数为2-4的各列联表Pearson列联系数的分布及其与相关系数的关系。方法采用SAS9.1.3软件模拟行列数分别为2-4,且行合计固定的各种列联表,计算相关系数和列联系数,绘制两者的散点图,描述其取值变化关系。结果 2×2列联表的相关系数和列联系数成曲线相关,但是其他情况下相关系数和列联系数的关系并非一一对应,此时通过相关系数划分列联系数的关联强度是不合适的。结论对于2×2列联表,可以通过相关系数的强度范围近似定义列联系数的强度范围。当行列数多于2时,不能根据相关系数的强度范围定义列联系数的强度范围。
Objective To explore the distribution of Pearson contingency coefficients and their relationship with correlation coefficients.Methods Using SAS 9.1.3 software to simulate tables with row or column of 2-4 respectively when fixed row totals.Compute Pearson contingency coefficients and correlation coefficients of each table.Draw scatter plots of Pearson contingency coefficients and correlation coefficients to describe their relationship.Results The scatter plot of correlation coefficients and Pearson contingency coefficients is a curve for 2×2 contingency table.However,for others there are no one-to-one correspondence relationship between correlation coefficients and Pearson contingency coefficients.It is impossible using Pearson contingency coefficients to determine association strength.Conclusion One can determine intensity of association for Pearson contingency coefficients through correlation coefficients for 2×2 contingency table.But it makes no sense for the other contingency tables.
出处
《中国卫生统计》
CSCD
北大核心
2011年第3期244-246,共3页
Chinese Journal of Health Statistics
基金
中山大学教学改革研究课题(编号:中大教务〔2009〕108号-26)
关键词
列联表
列联系数
相关系数
关联强度
Contingency table
Pearson contingency coefficient
Correlation coefficient
Association strength
