x*z
发帖数: 67 | 1 I used proc means and get the results as below.
My question is why the 95% confidence interval is mean+/- 1.96*(STD DEV)?
Thank you! |
m****r
发帖数: 237 | 2 why it's not???
just find some intro statistics book, and look up for constructing confidenc
e interval...
【在 x*z 的大作中提到】
: I used proc means and get the results as below.
: My question is why the 95% confidence interval is mean+/- 1.96*(STD DEV)?
: Thank you!
|
x*z
发帖数: 67 | 3 Sorry, I mean why it is NOT
mean+/- 1.96*(STD DEV)? |
j*****e
发帖数: 182 | 4 Std Dev refers to sigma.
You need std err in the formula, which corresponds to sigma over sample size
. |
j*****e
发帖数: 182 | 5 Correstion,
it should be sigma over sqrt of sample size. |
h******e
发帖数: 1791 | 6 this is the definition.
【在 x*z 的大作中提到】
: I used proc means and get the results as below.
: My question is why the 95% confidence interval is mean+/- 1.96*(STD DEV)?
: Thank you!
|
x*z
发帖数: 67 | 7 非常感谢~ 我也这样想着, 算出来的还是有点差距,所以在这儿问问~
用'sigma over sqrt of sample size-1' 好象差的更远:(
【在 j*****e 的大作中提到】
: Correstion,
: it should be sigma over sqrt of sample size.
|
c******h
发帖数: 355 | 8 because your degree of freedom? It will affect 1.96. |
