Re: significance of correlation affected by sample size?
- From: Bruce Weaver <bweaver@xxxxxxxxxxxx>
- Date: Wed, 23 Nov 2005 12:41:44 +0000
2046 wrote:
Hi,
I have read that the sample size can affect the significance of correlation. Can someone explain why is this so? Thanks.
The SE of r = SE(r) = SQRT[(1-r^2)/(n-2)]. The significance of r is tested with a t-test.
t = (r-0)/SE(r)
So, as n gets larger, SE(r) gets smaller, t gets larger, and p gets smaller.
-- Bruce Weaver bweaver@xxxxxxxxxxxx www.angelfire.com/wv/bwhomedir .
- Follow-Ups:
- References:
- Prev by Date: significance of correlation affected by sample size?
- Next by Date: Re: Comparing coefficients
- Previous by thread: significance of correlation affected by sample size?
- Next by thread: Re: significance of correlation affected by sample size?
- Index(es):