Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | graupel75@gmail.com |
To | Statalist <statalist@hsphsun2.harvard.edu> |
Subject | st: -poisson postestimation- chi2 discrepancy |
Date | Thu, 27 Sep 2012 09:38:51 +0800 |
Dear 'listers, I am working on Poisson models, and using -estat gof- I obtain: Deviance goodness-of-fit = 301.3502 Prob > chi2(269) = 0.0852 Pearson goodness-of-fit = 1135.374 Prob > chi2(269) = 0.0000 My data do not appear to be horribly over-dispersed (alpha from -nbreg- is 3.39e-07), (variance = 3.6, mean 1.8). I suspect the discrepancy could be due to the very large df (it's a collapsed Cox model), but the Stata manual for -poisson postestimation- doesn't provide any clarity on when one gof might be more apppropiate. Alternatively, the model could be very poorly specified, but only options I know of are SSC's -fitstat-. Any suggestions? cheers- Andrew Lover ______________________________________________ Epidemiologist Centre for Infectious Disease Epidemiology Research (CIDER) Saw Swee Hock School of Public Health National University of Singapore * * 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/