Here is another bug in glm. Occasionally, with log linear models for small tables, there can be more parameters (many aliased) than cells in the table. Then, glm does not work. The problem is in the qr decomposition. Here is a quick hack that works, although not very elegant. In glm.fit, replace l. 254 Rmat <- fit$qr[1:nvars, 1:nvars] by Rmat <- diag(nvars) nr <- min(sum(good),nvars) Rmat[1:nr,1:nvars] <- fit$qr[1:nr,1:nvars] Jim =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-devel-request@stat.math.ethz.ch =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=