Dear R--Help, My current research focuses upon sensitivity analyses which require [1] clusterings of patients in a baseline-covariate X space and [2] examining the distribution of within-cluster treatment differences in outcome. I have some primitive R code for this, but I want to be able to interact with graphical displays of within-cluster information (local average treatment effects) to identify individual clusters and/or compare pairs of clusters in terms for the X characteristics of their patients. Is anybody currently working on developing this sort of point-and-click capability in R? I would be very interested in becoming an alpha or beta tester of this functionality. If some computer scientist is looking for a thesis topic involving applications of graphics and statistical methods in health care (randomized or nonrandomized studies), I would be happy to send a recent white paper on the methodology I visualize. On top of my work at Lilly, I am an adjunct professor of biostatistics at Indiana University Medical School and would be willing to participate on a thesis committee. For example, I can perform almost all of the point-and-click operations I am interested in from existing menu items in JMP version 5.1, but I have not been able to develop usable JMP scripts to automate the basic processes. ...BobO. Bob Obenchain, PhD, Research Advisor US Medical - Outcomes Research Lilly Corporate Center, Indianapolis, IN 46285-5024 (317) 276-3150 voice, (317) 276-0427 fax, ochain@lilly.com [[alternative HTML version deleted]]