We can further generalize this: Suppose we want to constrain parameters such that :
b2 = a * b1 b3 = a * b1 We can do the following: fit.a <- glm( y ~ I(x1 + a* x2 + a * x3), data= .... , ... ) For a fixed value of `a', we compute the log-likelihood of `fit.a'. This is the profile likelihood in terms of a. We then maximize this over `a' . We can use, for example, `optimize' to accomplish this. This is not contrived. I have encountered this situation in a project. Ravi. ____________________________________________________________________ Ravi Varadhan, Ph.D. Assistant Professor, Division of Geriatric Medicine and Gerontology School of Medicine Johns Hopkins University Ph. (410) 502-2619 email: rvarad...@jhmi.edu ----- Original Message ----- From: David Winsemius <dwinsem...@comcast.net> Date: Tuesday, October 26, 2010 11:21 am Subject: Re: [R] Setting constraints in the glm package To: "Viechtbauer Wolfgang (STAT)" <wolfgang.viechtba...@maastrichtuniversity.nl> Cc: "r-help@r-project.org" <r-help@r-project.org>, 'David Smith' <dreamwor...@hotmail.co.uk> > On Oct 26, 2010, at 7:50 AM, Viechtbauer Wolfgang (STAT) wrote: > > >The constraint b1=b2 in a model such as b0 + b1 x1 + b2 x2 + b3 x3 > implies that b0 + b1 (x1 + x2) + b3 x3, so just add x1 and x2 (call > this x12) and fit the model b0 + b1 x12 + b3 x3 and you have imposed > the constraint that b1=b2. To impose the constraint that b3=0, just > fit the model without variable x3 in it. > > > > It may be more convenient to do the addition within a formula without > first calculating x12 : > > glm( y ~ I(x1+x2), data= .... , ... ) > > And the offset argument is available for constraints of the form b3 = > <non-zero-constant>. > > -- > David > > >Best, > > > >-- > >Wolfgang Viechtbauer > >Department of Methodology and Statistics > > >----Original Message---- > > >On Behalf Of David Smith Sent: > >Tuesday, October 26, 2010 12:24 To: r-help@r-project.org > >Subject: [R] Setting constraints in the glm package > > > >>Hi, > >> > >>I would like to set a constraint on the fixed effect estimates in a > >>GLM model, such as b1=b2. Is this possible in the glm package? > >>Similarly I would like to set some to equal zero too. I have tried > >>searching the information with this package, but I can't find > >>anything for this. > > ______________________________________________ > R-help@r-project.org mailing list > > PLEASE do read the posting guide > and provide commented, minimal, self-contained, reproducible code. ______________________________________________ R-help@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide http://www.R-project.org/posting-guide.html and provide commented, minimal, self-contained, reproducible code.
