Re: comparing theta's

From: "Nick Holford" <n.holford@auckland.ac.nz> Subject: Re: comparing theta's Date: Mon, 29 Mar 1999 22:04:05 +1200 Rik, >In reply to Nick's reply, I think you should be extremely cautious in using empirical Bayes estimates for subsequent >statistical comparisons whether it be using parametric or non-parametric tests. I accept your concerns but it does depend on what one might be interested in e.g. if you have 2 samples of patients and you want to know if the patient samples are similar the comparison of the post hoc estimates seems to make sense (conditional on the model and the design etc). But Peter asked about comparing the thetas and in that case I suppose one is not much interested in comparing the patient samples. In a separate email to Peter I suggested that one could bootstrap each of the data sets and thus obtain an empirical distribution of the population estimates of the each theta. The bootstrap empirical distributions could then be compared using the KS procedure. This would retain the generality of comparing the distribution of the thetas (rather than just the mean as with a typical ANOVA) and would be less dependent on the peculiarities of the individuals which influence the post hoc estimates. -- Nick Holford, Dept Pharmacology & Clinical Pharmacology University of Auckland, Private Bag 92019, Auckland, New Zealand email:n.holford@auckland.ac.nz tel:+64(9)373-7599x6730 fax:373-7556 http://www.phm.auckland.ac.nz/Staff/NHolford/nholford.html