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

Re: st: Comparing model fit of logistic regressions with robust standard errors

From   "Michael I. Lichter" <>
Subject   Re: st: Comparing model fit of logistic regressions with robust standard errors
Date   Mon, 12 Oct 2009 14:37:20 -0400

Also see -svylogitgof-

   net describe st0099, from(

and the associated Stata Journal article


svylogitgof performs an F-adjusted mean residual test to assess goodness of fit of a design-based logistic regression model. This command is only valid
   following svy: logit or svy: logistic.

Your clustered model may not require svy: logit, but results should be very similar to what you get with -logit, cluster()- as long as you use -svyset- correctly.


Dirk Deichmann wrote:
Hello everyone,
I am applying a logistic regression model with robust standard errors adjusted for clustering. I know there have been some posts about this but to me it still is not clear whether and if so how I can assess the improvement in model fit using the Wald chi square values.
Can you calculate the change in Wald chi square from a restricted to a full model and then look up whether this value is significant? Would you just subtract one Wald chi square value from the other to get to the change in Wald chi square value? Or is it more meaningful to say that if you assess model 1 with say x1, x2, and x3  (Wald chi2(3) = 196.63) to model 2 with x1, x2, x3, and x4 (Wald chi2(4) = 198.9) and both models show Prob > chi2 = 0 that the latter model shows a better fit since it is still significant even though a new variable has been added? Or how should I think about this?
Many thanks for your kind support, Dirk

*   For searches and help try:

Michael I. Lichter, Ph.D. <>
Research Assistant Professor & NRSA Fellow
UB Department of Family Medicine / Primary Care Research Institute
UB Clinical Center, 462 Grider Street, Buffalo, NY 14215
Office: CC 126 / Phone: 716-898-4751 / FAX: 716-898-3536

*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index