t*******e 发帖数: 1 | 1 To compare two unpaired groups:
what is the difference between "assuming equal variance" and "assuming
unequal variance"? How should I know which one to use? Thanks. |
i********f 发帖数: 206 | 2 你可以用Levene's test检测是不是equal variance.
如果n1和n2一样或者差别不大,两种方法给的结果应该很接近吧
【在 t*******e 的大作中提到】 : To compare two unpaired groups: : what is the difference between "assuming equal variance" and "assuming : unequal variance"? How should I know which one to use? Thanks.
|
o******6 发帖数: 538 | 3 F test,看SAS OUTPUT
【在 t*******e 的大作中提到】 : To compare two unpaired groups: : what is the difference between "assuming equal variance" and "assuming : unequal variance"? How should I know which one to use? Thanks.
|
s*****n 发帖数: 2174 | 4 主要就是如何计算整体variance的问题.
如果可以认为equal variance, 那么整体variance就可以用pooled variance. 本质就
是样本variance的加权平均.
如果不能认为equal variance, 那么就要用别的公式. 本质上就是把mean(X2)- mean(
X1) 看成样本的一个线性组合, 硬去求它的variance(近似)分布.
不过实际上, 如果不能认为equal variance, 做这个t-test本身是否有意义就有争论.
在不同方差的分布之间, 比较均值, 实际意义并不明确.
【在 t*******e 的大作中提到】 : To compare two unpaired groups: : what is the difference between "assuming equal variance" and "assuming : unequal variance"? How should I know which one to use? Thanks.
|
R*********r 发帖数: 225 | 5 use welch's method, never ever use student's two sample t-test, according to
JZ Neyman.
【在 t*******e 的大作中提到】 : To compare two unpaired groups: : what is the difference between "assuming equal variance" and "assuming : unequal variance"? How should I know which one to use? Thanks.
|