The help file for the S+ function loglin is scant. You pass it the model by specifiying a list, e.g. loglin(dat, list(1:2,3)) I'm trying to understand this. Suppose I have a 3D table. Does 1:N mean fit the nth way interaction? Are all lower level interactions and main effects automatically included? In other words, is list(1:3) the saturated model? Then suppose I want to fit everything but the three way-do I want list(1:2,2:3,1:3) ? By the way-is there an S+ newsgroup (not list)? I did a search, and got some newsgroups, but they just contained crap. Thanks- Geoff Collier ================================================================= Instructions for joining and leaving this list and remarks about the problem of INAPPROPRIATE MESSAGES are available at http://jse.stat.ncsu.edu/ =================================================================