In article <[EMAIL PROTECTED]>,
Jon Cryer <[EMAIL PROTECTED]> wrote:
>"This may surprise some folks."
>
>It certainly should surprise anyone who has studied statistics!
(I assume you meant "should not surprise")

You can improve the with-replacement variance by "labeling" the
balls and making your estimate a sum over the labels.
Still doesn't beat sampling without replacement though.
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to