Dear list, I am looking for a way to fix the scale parameter when fitting a generalized linear model with gamma errors and log link. Is there something like "SCALE" such as in GLIM? As always thanks a lot. Peter
On Mon, 17 Mar 2003, Dr. Peter Schlattmann wrote:> > Dear list, > > I am looking for a way to fix the scale parameter when fitting a > generalized linear model with gamma errors and log link. > > Is there something like "SCALE" such as in GLIM? >No. Because it doesn't affect the answers we don't provide a way to specify it. It does affect standard errors and the other output of summary.glm, so this function does have a dispersion= argument. -thomas