Statistical Analysis

Does anyone have a formula for finding the number of samples to take of a population in order to have a specified amount of certainty of a specified error rate?

the only equation I have at the moment doesn't deal with the initial size of the population to sample. So whether I have 5 million in the population or just 5, I should take a sample of 384 in order to have 95% assurance of being correct.
Strange huh?

Something here is bound to help.

It is so funny that you should ask on this last night of ths particular project. Someday I'll explain.

I am misapplying this tool (from our friendly, socialistic, non-Libertarian CDC) at this very moment to run some crude statistical tests. It includes a module (and old DOS program really) called Statcalc that includes a sample size estimator. You may be able to just download the Statcalc bit.

Oh, do I wish we had money for SAS!