Hotelling T^2检验与u检验、t检验之间的关系  

The Relationship Between Hotelling T^2 Test and u or t Test

作　　者：谭瀛[1] 蒋知俭[1] 郭晓伟 

机构地区：[1]哈尔滨医科大学卫生统计教研室,150001

出　　处：《中国卫生统计》2000年第1期11-14,共4页Chinese Journal of Health Statistics

摘　　要：目的　推导反映多元T2统计量与t统计量关系的函数,以揭示T2检验结果与t检验结果差别的本质原因。方法　利用图示方法,在固定t(或r)值条件下,观察r(或t)值变化对T2值的影响,直观地探索T2与t两种检验得到不同结果的可能性。结果　导出T2统计量对t,r统计量的函数形式,并证明T2与t两种检验结果差别存在的可能性。结论　导出的函数说明,不仅仅t会影响到T2,相关系数r对T2的影响也不容忽视,并且主要是r的不同造成了T2与t两种检验的结果有差别。Objective Conducting the function of the statistics which reflects the relationship between T 2 and t to reveal the essentical reason for the differences between the test result of T 2 and t.Methods Under constant r or t values,making use of charts and graphs to observe the effects of change in T 2 value and to explore visually the possibility between different test result of T 2 and t.Results Induced the function of statistic T 2 to t or r,proved that two kinds of test results of T 2 and t have the possibility of difference.Conclusion Induced function which statistic T 2 to t showed that t influences T 2,the coefficient of correlation does also.The action of correlation coefficient can't be neglected as the result is mainly caused by the different r formed.

关 键 词：多元假设检验 Hotilling T^2检验 U检验 T检验 

分 类 号：R195[医药卫生—卫生统计学]