Re: [Rd] New code in R-devel: Rao score test for glm.
Hi Dr. Dalgaard, I just googled one of your old posts and find it very useful. I am very curious about whether your code which implements the score test have been implemented in the glm function of R (instead of the development version of R). Thanks a lot! Yijuan -- View this message in context: http://r.789695.n4.nabble.com/New-code-in-R-devel-Rao-score-test-for-glm-tp3514262p4664440.html Sent from the R devel mailing list archive at Nabble.com. __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] New code in R-devel: Rao score test for glm.
(R-devel is not Dr. Dalgaard) It's been there since 2.14.0. (So the post was presumably when 2.13.x was current.) -pd On Apr 17, 2013, at 03:08 , yhu30 wrote: Hi Dr. Dalgaard, I just googled one of your old posts and find it very useful. I am very curious about whether your code which implements the score test have been implemented in the glm function of R (instead of the development version of R). Thanks a lot! Yijuan -- View this message in context: http://r.789695.n4.nabble.com/New-code-in-R-devel-Rao-score-test-for-glm-tp3514262p4664440.html Sent from the R devel mailing list archive at Nabble.com. __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel -- Peter Dalgaard, Professor Center for Statistics, Copenhagen Business School Solbjerg Plads 3, 2000 Frederiksberg, Denmark Phone: (+45)38153501 Email: pd@cbs.dk Priv: pda...@gmail.com __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] New code in R-devel: Rao score test for glm.
Thanks for doing this Peter. I'll have to install the development version to try this out. One suggestion though. I'm pretty confident that plain old score test is a more common terminology than anything involving Rao's name (econometricians even call it the Lagrange multiplier test). In light of this, I think that it would be much better to use test = score rather than test = Rao. -- Brett Presnell Department of Statistics University of Florida http://www.stat.ufl.edu/~presnell/ We don't think that the popularity of an error makes it the truth. -- Richard Stallman __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] New code in R-devel: Rao score test for glm.
Economists re-invented the Rao efficient score test, calling it the Lagrange multiplier test. Please check the history of this test. Rao's paper was published in 1947. That being said, score would be more consistent with the survival and rms packages. Frank Brett Presnell wrote: Thanks for doing this Peter. I'll have to install the development version to try this out. One suggestion though. I'm pretty confident that plain old score test is a more common terminology than anything involving Rao's name (econometricians even call it the Lagrange multiplier test). In light of this, I think that it would be much better to use test = score rather than test = Rao. -- Brett Presnell Department of Statistics University of Florida http://www.stat.ufl.edu/~presnell/ We don't think that the popularity of an error makes it the truth. -- Richard Stallman __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel - Frank Harrell Department of Biostatistics, Vanderbilt University -- View this message in context: http://r.789695.n4.nabble.com/New-code-in-R-devel-Rao-score-test-for-glm-tp3514262p3514679.html Sent from the R devel mailing list archive at Nabble.com. __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] New code in R-devel: Rao score test for glm.
On May 11, 2011, at 15:10 , Brett Presnell wrote: Thanks for doing this Peter. I'll have to install the development version to try this out. One suggestion though. I'm pretty confident that plain old score test is a more common terminology than anything involving Rao's name (econometricians even call it the Lagrange multiplier test). In light of this, I think that it would be much better to use test = score rather than test = Rao. It's not like that didn't cross my mind, in fact I started out that way, but... - A column labeled score just looks odd, whereas there is some precedence for labeling tests according to authors (e.g. Pillai). - CR Rao is still around, now 90 years of age, and having been taught from Linear Statistical Inference with Applications, I thought paying a little homage would be appropriate - At least the curator calls it Rao score test: http://www.scholarpedia.org/article/Rao_score_test -- Brett Presnell Department of Statistics University of Florida http://www.stat.ufl.edu/~presnell/ We don't think that the popularity of an error makes it the truth. -- Richard Stallman __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel -- Peter Dalgaard Center for Statistics, Copenhagen Business School Solbjerg Plads 3, 2000 Frederiksberg, Denmark Phone: (+45)38153501 Email: pd@cbs.dk Priv: pda...@gmail.com __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] New code in R-devel: Rao score test for glm.
On 2011-05-11 07:30, peter dalgaard wrote: On May 11, 2011, at 15:10 , Brett Presnell wrote: Thanks for doing this Peter. I'll have to install the development version to try this out. One suggestion though. I'm pretty confident that plain old score test is a more common terminology than anything involving Rao's name (econometricians even call it the Lagrange multiplier test). In light of this, I think that it would be much better to use test = score rather than test = Rao. It's not like that didn't cross my mind, in fact I started out that way, but... - A column labeled score just looks odd, whereas there is some precedence for labeling tests according to authors (e.g. Pillai). - CR Rao is still around, now 90 years of age, and having been taught from Linear Statistical Inference with Applications, I thought paying a little homage would be appropriate - At least the curator calls it Rao score test: http://www.scholarpedia.org/article/Rao_score_test Yes, thanks, Peter, for coding this test. As to the name, my vote is for Rao. Peter Ehlers -- Brett Presnell Department of Statistics University of Florida http://www.stat.ufl.edu/~presnell/ We don't think that the popularity of an error makes it the truth. -- Richard Stallman __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel
Re: [Rd] New code in R-devel: Rao score test for glm.
I agree that homage is nice, but I really think you'll get more confusion from Rao than you would from score. You could always label the column some other way, right? I mean, it doesn't have to be exactly the same as the option. FWIW, I also have another vote for score from Alan Agresti, the only other person with whom I've discussed this. But I don't want to argue the point any more than that. It seems right for the person who wrote the code to have the final say, and I'll be very happy to have this functionality regardless of how it's called. Peter Ehlers ehl...@ucalgary.ca writes: On 2011-05-11 07:30, peter dalgaard wrote: On May 11, 2011, at 15:10 , Brett Presnell wrote: Thanks for doing this Peter. I'll have to install the development version to try this out. One suggestion though. I'm pretty confident that plain old score test is a more common terminology than anything involving Rao's name (econometricians even call it the Lagrange multiplier test). In light of this, I think that it would be much better to use test = score rather than test = Rao. It's not like that didn't cross my mind, in fact I started out that way, but... - A column labeled score just looks odd, whereas there is some precedence for labeling tests according to authors (e.g. Pillai). - CR Rao is still around, now 90 years of age, and having been taught from Linear Statistical Inference with Applications, I thought paying a little homage would be appropriate - At least the curator calls it Rao score test: http://www.scholarpedia.org/article/Rao_score_test Yes, thanks, Peter, for coding this test. As to the name, my vote is for Rao. Peter Ehlers __ R-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-devel