Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: RE: GLM Binomial family logit link postestimation GOF


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: RE: GLM Binomial family logit link postestimation GOF
Date   Mon, 24 May 2010 11:58:06 +0100

I suggest that degrading your proportions to binary makes very little
statistical sense for such analyses. You are just throwing away most of
the information. 

You can get observations and predictions as proportions. That leaves
scope for various diagnostic plots and even conventional R-square
calculations. 

For example, check out some of the ideas at 

http://www.stata.com/support/faqs/stat/rsquared.html

Nick 
n.j.cox@durham.ac.uk 

Venkatachalam, Alicia

I have just run a GLM under the binomial family with a logit link with a
proportion as my outcome. I am now unsure how best to test the GOF of
this model? I have read in the literature it is often best to rerun the
model as a logistic model and then use the available post estimation
commands, however because my outcome is a proportion and not binary I am
unsure if this is permissible? Should I recode my outcome as binary or
is there a way to run post estimation after my GLM model?


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index