摘要
本文以Kappa系数和两总体率差值检验为例,说明一般的假设检验法与用置信区间作检验是不相同的。其根源在于两者的标准误估计是不同的:与假设检验法不同的是,在求置信区间时,计算标准误时不需要假定H0成立。即零假设被否定后,不应该再用假设检验中的标准误去代替置信区间中的标准误。
The article sets forth that standard errors are different sometimes between confidence limits and hypothesis test in the light of examples of kappa coefficient and difference of two population rates.The estimation of standard error in confidence limits needn't the hypothesis of H_0.When their estimations of standard error are different,the standard error in hypothesis test couldn't be substituted for that in confidence limits.
出处
《数理统计与管理》
CSSCI
北大核心
2004年第4期77-80,共4页
Journal of Applied Statistics and Management
关键词
Kappa系数
两总体率差值
置信区间
假设检验
kappa coefficient
difference of two population rates
confidence limits
hypothesis test
