st: micombine and overall model fit statistics

["Claire M. Kamp Dush" <cmk54@cornell.edu>]

Dear statalist community,

I am using micombine to deal with my missing data.  I only have one issue - 
how do I get an overall model fit statistic with micombine?  Is this 
possible?  For instance, I am using:

micombine regress
micombine stcox
micombine logistic
micombine nbreg

Under normal circumstances, I would get an F, r-square, chi-square or some 
other indicator of overall model fit if I was not using the micombine 
pre-fix.  But, I do not get an F, or chi-square, with micombine.  What I 
have done then is run each of my models separately within each of my 5 
imputed datasets using if _j == `num' . . .  And, then I thought I could 
somehow combine the results for an overall chi-square or F.  Of course, I 
am sure I am violating a million rules by averaging these 5 F-values 
together.  Any advice?  I just don't want my reviewers, who don't use 
micombine, to question my results because I don't have the overall fit 
reported, and I don't want to turn my paper into a statistical treatise 
(i.e. report all of the results from all 5 models, etcetera, then compare 
this to the original results), I just want to use micombine as a tool 
because I don't have room in this paper to expand in this way.

Any advice would be great.
Thanks,
Claire


Claire M. Kamp Dush
Institute for the Social Sciences Postdoctoral Fellow
Evolving Family Theme Project
Cornell University
http://www.socialsciences.cornell.edu/0407/evolv_fam_desc.html


*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/