responding to my post - On Fri, 19 Dec 2003 07:23:47 GMT, "jackson marshmallow" <[EMAIL PROTECTED]> wrote: [ ... ] me > > Surely, that should say something like, "proportionate to N". [ ... ] > > > > OK, but you only have to sort once, but if you perform randomization, the > correlation will have to be computed over and over again, so the cost of > sorting will negligible. - Right. I think I wandered off my point, which was, Use the word "proportionate" in there; *do not* assert that the cost is N
[ snip ] > > I think (correct me if I'm wrong) the p-value should be fairly accurate if > we do a limited number of random permutations, say, 1000. > Here is the correction. Well, if you are going to waste the time to Monte Carlo, shouldn't you do enough cycles that you out-perform the normal approximation, which takes no time at all? If you do 1000, then you estimate the 5% level based on the 50th (one tailed). The Poisson will apply, so that anywhere from about 37 to 65 is within reason. (Consider the 2 s.d. range on the square root; the point estimate of 7 has the CI of about (6,8).) I am not sure, but I expect that the approximation is better by an order of magnitude. The Monte Carlo 1% and 0.1% estimates will be worse, if you consider the error as a proportion. [ ... ] > I have series X and series Y. Let Yn = f(Xn). The hypothesis I want to test > is that f is a monotonic function. > As someone else points out, that's not phrased as a statistical question. -- Rich Ulrich, [EMAIL PROTECTED] http://www.pitt.edu/~wpilib/index.html "Taxes are the price we pay for civilization." . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
