Stan Brown <[EMAIL PROTECTED]> wrote in
news:[EMAIL PROTECTED]:

> Do any others have comments?

Well, using a fixed rule for significant digits isn't *always* a
good idea. For example, if you have a bunch of bernoulli variables
(e.g. with success probability p = pi/10), all observations will
be 0 or 1 (one significant digit), but the mean will usually lie
close to .31415926..., and the extra digits *do* contain useful
information on p.

-- 
Karl Ove Hufthammer
.
.
=================================================================
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