Statalist


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

st: lrtest and coefficent estimates


From   "Jiang, Frank" <fjiang@ivey.uwo.ca>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: lrtest and coefficent estimates
Date   Wed, 19 Aug 2009 11:15:26 -0400

Dear Statalists, 

I am writing to seek advice on how to interpret the results of a
likelihood-ratio test (lrtest) when they seem to be inconsistent with
coefficient estimates.  

There are several interaction terms in my conditional logit model.  I
test the impact of each interaction term separately by running multiple
specifications. I then perform a lrtest for each specification against
the baseline specification.  The coefficient estimates on two
interaction terms are statistically significant but the results of
corresponding lrtest are not significant.  Do the lrtest results
invalidate the coefficient estimates or vice versa?   Shall I report the
coefficient estimates in spite of the non-significant results of the
lrtests?

Your advice will be highly appreciated. 

Thanks
Frank

*
*   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   |   What's new   |   Site index