On Tue, 25 Jan 2000, Kimes, Dean W. wrote:

>
> John's probably got a better handle on it than I do. I generally go
> overboard and go a magnitude up on the entire range rather than the real
> range. Means when I do a test here at work I'd go up to 1000 occurrences
> minimum on a % situation.

My background is more in statistical mathematics and
information theory, not sampling statistics. I could grind
the equations to give you a formula for number of samples
needed to verify a x% frequency of occurrance with 95%
certainty. I could not have told you off the top of my head
that it's roughly an order of magnitude greater than the
inverse of the frequency of occurrance.

"What the h*ll is pi doing in the equation for a normal
curve?!?" (Yes, pi really does pop up in there.)

--
John H. Kim
kim@...