Re: Confidence Interval from the covariance
From: Ray Koopman (koopman_at_sfu.ca)
Date: 08/28/04
- Next message: Me: "Re: How do odds relate to payout?"
- Previous message: Osher Doctorow: "Re: Slow vs Fast Variables in the Brain and Hyper-Riccati Systems"
- Maybe in reply to: Paulo Gonçalves: "Confidence Interval from the covariance"
- Next in thread: Woodhouser: "Re: Confidence Interval from the covariance"
- Reply: Woodhouser: "Re: Confidence Interval from the covariance"
- Messages sorted by: [ date ] [ thread ]
Date: 27 Aug 2004 23:11:34 -0700
Jack Tomsky wrote:
> I've done some rough calculations assuming that you're dealing with
> very large samples. The large-sample 95% confidence interval for
> Cov(x,y) is approximately
>
> Sxy +/- 1.96*sqrt((Sx^2*Sy^2-Sxy^2)/N)
That should be +Sxy^2, not -Sxy^2.
See Kendall & Stuart, vol 1, example 13.3.
- Next message: Me: "Re: How do odds relate to payout?"
- Previous message: Osher Doctorow: "Re: Slow vs Fast Variables in the Brain and Hyper-Riccati Systems"
- Maybe in reply to: Paulo Gonçalves: "Confidence Interval from the covariance"
- Next in thread: Woodhouser: "Re: Confidence Interval from the covariance"
- Reply: Woodhouser: "Re: Confidence Interval from the covariance"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
|
