A question regarding statistics - 未名空间MITBBS历史存档

国际科技财经博客移民网络热点娱乐民生时事公众号

Redian新闻

>未名空间

>Environmental - 环境科学与工程

A question regarding statistics

A question regarding statistics# Environmental - 环境科学与工程

i*02008-08-15 07:08

1 楼

I have a question regarding statistics in experiment design. My supervisor
said that the replicates n at least equal 3 in order to publish paper. But
if CV is small enough ( < 5%), dupilicate (n=2) can be acceptable? I will
greatly appreciate it if somebody can give me an answer about this.

a*h2008-08-15 07:08

2 楼

How do you get CV when you have less than 3 samples?

【在 i*****0 的大作中提到】

: I have a question regarding statistics in experiment design. My supervisor
: said that the replicates n at least equal 3 in order to publish paper. But
: if CV is small enough ( < 5%), dupilicate (n=2) can be acceptable? I will
: greatly appreciate it if somebody can give me an answer about this.

i*02008-08-15 07:08

3 楼

CV=Standard Error/mean
standard error = standard deviation/(n)^0.5
n=2 still have standard deviation
what's wrong with this?

【在 a***h 的大作中提到】

: How do you get CV when you have less than 3 samples?

a*h2008-08-15 07:08

4 楼

CV = Cross Validation
我认为 n>=3 是让你的样本数足够多, 不是说能不能得到 standard error。如果n=1
你也能得到 standard error.

【在 i*****0 的大作中提到】

:
: CV=Standard Error/mean
: standard error = standard deviation/(n)^0.5
: n=2 still have standard deviation
: what's wrong with this?

i*02008-08-15 07:08

5 楼

n=1 you can't get standard error, because there is no standard deviation for
n=1

1

【在 a*****h 的大作中提到】

: CV = Cross Validation
: 我认为 n>=3 是让你的样本数足够多, 不是说能不能得到 standard error。如果n=1
: 你也能得到 standard error.