On Wed, 23 Mar 2011 15:28:51 -0700, Ray Vickson wrote:
> I think that the OP's problem stems from trying to use the formulas > where they do not apply. He does not have an iid sample, because the > only combinations he looks at are those in which there are no > repetitions; that is, he is sampling *without replacement* (he does not > allow things like (1,1,1); he only allows things like (1,2,3).) That > makes his successive sample values dependent instead of independent, so > the formulas he cites (which assume _independence_) do not apply.
Ahaha! That's what I'm missing, thanks.
Works fine once I swap to sampling with replacement instead.