R****9
发帖数: 815 | 1 有两组，一组只有９人，另一组有90人，怎么能做比较，比如t-test？合理吗? |
a****g
发帖数: 8131 | |
d******e
发帖数: 7844 | 3 9个的应该啥也看不出来，呵呵
【在 a****g 的大作中提到】
: 先看看distribution吧
|
o*******w
发帖数: 2310 | 4 no. too small size.
run power analysis? |
R****9
发帖数: 815 | 5 power analysis不是一般用来计算所需的sample size吗？
人数现在已经在那，想比较两组譬如岁数，患病时间，测试得分等等有没有差别。
【在 o*******w 的大作中提到】
: no. too small size.
: run power analysis?
|
s*******1
发帖数: 146 | 6 You may use Bootstrip sampling.
For example, sample 6 from 9samples 100 times, you get 100 means of this
group;
then sample 6 from 90 samples 100 times, yet also get 100 means of the
second group;
then compare the means of the two groups. |
o*******w
发帖数: 2310 | 7 how big sample size will be needed to run bootstripping?
you cannot use so few ba?
【在 s*******1 的大作中提到】
: You may use Bootstrip sampling.
: For example, sample 6 from 9samples 100 times, you get 100 means of this
: group;
: then sample 6 from 90 samples 100 times, yet also get 100 means of the
: second group;
: then compare the means of the two groups.
|
s*******1
发帖数: 146 | 8 See the article below:
Estimating misclassification error with small samples via bootstrap cross-
validation
Wenjiang J. Fu*, Raymond J. Carroll and Suojin Wang
http://bioinformatics.oxfordjournals.org/content/21/9/1979.full
......................................................................how
big sample size will be needed to run bootstripping?
you cannot use so few ba? |
p***r
发帖数: 920 | 9 2 sample t-test, although the 100 sample is a waste. |
