摘要
目的 :对 SAS软件中处理 k个相关样本的非参数检验方法进行评价。方法 :利用 SAS软件中的过程经编程模拟计算 ,以 型、 型及 ( 型 + 型 )误差作为评价指标。结果 :给出所有模拟条件下的 型、 型及误差值。结论 :当处理组数 k=3时 ,Friedmanχ2 检验与 RF检验完全等价 ;当 k>3且配伍组数 b不太大时 ,若要控制 型误差应选 Friedm anχ2检验 ,若要控制 型或误差 ,应选 RF检验 ;当 b较大时 。
Objective:Evaluation of the k -dependent samples nonparametric tests in SAS software.Methods:Simulation experiments were made by SAS software,using the type Ⅰ error,type Ⅱ error estimate of tests as the index of evaluation.Results:The type Ⅰ error and type Ⅱ error estimate are given.Conclusion:When the number of treatment groups is three,Friedman test and RF test are equal in value,When the number of treatment groups is over three,and the number of block is small.TO control the type Ⅰ error,the Friedman χ 2 test is selected;TO control the type Ⅱ error or error of test,the RF test is selected.When the number of block is large,two tests are nearly equal in value.
出处
《现代预防医学》
CAS
2002年第1期18-20,共3页
Modern Preventive Medicine
