摘要
在研究列联表的假设检验时,列联表的不同形成过程经常被忽视本文强调了列联表的形成过程(与抽样目的有关)和零假设。
In inference concerning the contingency tables,the differences between table forming processes are often neglected.In this paper,the connection between the sampling schemes (forming processes) and the aim of tests (null hypotheses) are emphasized.Examples are given to show that different forming processes and different aspects of concern about a contingency table can affect the inferences.To reduce computing time,simulation methods are suggested to estimate p values when the sample sizes are large.
关键词
列联表
极大似然估计
假设检验
形成过程
and phrases Contingency table,Homogeneity,Independent,Likelihood ratio (LR) test,Mahalandois’ generalized distance,Maximize likelihood (ML) estimation,Neither marginal fixed,One marginal fixed,Pearson (PS) test,Two marginal fixed.