Odd situation there. Why do you have groups with small sample sizes? Is this an observational study?

Paul A. Thompson, Ph.D. Division of Biostatistics, Washington University School of Medicine 660 S. Euclid, St. Louis, MO 63110-1093 314-747-3793 paul@wubios.wustl.edu

-----Original Message----- From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of Steve Denham Sent: Tuesday, May 20, 2008 6:35 AM To: SAS-L@LISTSERV.UGA.EDU Subject: Re: Contrasts, interactions, data step and where

I snipped this down to part of Dale's response because it brings up an issue I am wrestling with.

"Just to summarize, I would not want to be presented with contrast results which are based on restricting the data to some subset of the entire data set. In the best of circumstances (where your model is a cell means model and your contrast is a simple difference of cell means), you will have the correct point estimate for the numerator of your test but you will employ a poor estimate of the variance in the test statistic denominator. In other circumstances (where your model is not a simple cell means model), you will not even get the same point estimate for the numerator as would be obtained from complete data analysis." I take this as support for a position that I have taken (and isn't making me real popular amongst stakeholders). It has been a standard operating procedure in some analyses that we do to drop a treatment group that has less than three observations, reanalyze, and report pair-wise comparisons to the control group (and not do any analysis if the control group is the "dropped" group). Something about this feels very wrong to me, but I haven't been able to find any good references one way or the other. My objection is that the reanalysis does not reflect the randomization to treatments that was actually done, but I see in this paragraph another objection regarding the estimate of the variance. Anyone have any thoughts or references? Steve Denham Associate Director, Biostatistics MPI Research, Inc. Remove spamblock from header, and replace with stevedrd to reply to me.

----- Original Message ---- From: Dale McLerran <stringplayer_2@YAHOO.COM> To: SAS-L@LISTSERV.UGA.EDU Sent: Monday, May 19, 2008 1:38:42 PM Subject: Re: Contrasts, interactions, data step and where

--- Peter Flom <peterflomconsulting@MINDSPRING.COM> wrote:

<rest of message snipped>