On 6/3/15 6:47 AM, Tsung Fei Khang wrote:
@Aki: Thank you. 1/#iterations is problematic because one could then get arbitrarily small p-values... should converge to some value (however small) as the number of iterations exceeds some threshold, which is dependent on data set.
^^^^^ Note, that is NOT what Aki said. What he said was that the lowest reportable p-value is a function of your number of iterations. If you do 9 random draws and compare with your observed data, the lowest p-value you could possibly get is 1/10=0.1 (e.g., you would never see anything lower). If you did 99, then the lowest would be 0.01, 999 -> 0.001, etc.
Under the null model, on average 10% randomized values would be below 0.1, 5% below 0.05, 1% below 0.01, etc. The actual proportion you get for your data will vary around some value, but that value is determined by your original value compared to the randomized results. It has more precision as the number of iterations increases, but it does not get lower because of those iterations.
-ds -- If a response is important to you, keep trying -> I receive 50-100 msgs/day -- MORPHMET may be accessed via its webpage at http://www.morphometrics.org To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
