RE: st: Omnibus effects following xtmelogit with margins

["Ploutz-Snyder, Robert (JSC-SK)[USRA]" <robert.ploutz-snyder-1@nasa.gov>]

Thank you Phil for chiming in.  

Indeed, my one thread has turned into two --one about whether a 2 or 3 level model is best, and the second on my original question about getting omnibus main effects and interactions following xtmelogit.  

I originally tried anovalator prior to my post but I was not using the predict(passthru) option that you used here. That option is mentioned briefly in the anovalator help file, with much more thorough coverage at  http://www.ats.ucla.edu/stat/stata/faq/anovalator.htm  (link referenced in the help file).  

Thank you very much.  This is what I needed for my original question (yes, I caught that missing snip of code telling anovalator to run main effects or 2way) 

...still contemplating Michaels ideas...


Rob

-----Original Message-----
From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Philip Ender
Sent: Friday, September 24, 2010 2:06 PM
To: statalist@hsphsun2.harvard.edu
Subject: Re: st: Omnibus effects following xtmelogit with margins

"Ploutz-Snyder, Robert (JSC-SK)[USRA]" wrote
... I also ran into a snag using -margins- as I usually do after an
-xtmixed- model to obtain omnibus main effects for an ordinal factor
variable.
...
But my use of -margins- to get the omnibus main effect for time (as I
would with xtmixed) failed:

. margins time, asbalanced atmeans
default prediction is a function of possibly stochastic quantities
other than e(b)
r(498);
----------------------
Robert-

I've been following your discussion with Michael Mitchell about the
design of your analysis.  I would like to pick up on your second
question concerning using -margins-, for which I have two comments.

1)  The default predictor for -xtmelogit- is mu which has uses both
fixed and random effects.  If you are interested in having the
predictive margins scaled as a linear predictor, you can use

.   margins time, asbalanced atmeans predict(xb)

If you want to have your predictive margins scaled as a probabilities,
then -margins- can (will) only use the fixed effects of the model,
like this.
.  margins time, asbalanced atmeans predict(mu fixedonly)

2)  In any case, neither of the above is a "true" main effect because
the factor variables use dummy coding.  You can use the -anovalator-
to get main effects (findit anovalator).

.  anovalator time, predict(xb)   /* or */
.  anovalator time, predict(mu fixedonly)

I hope this is not too confusing.

Phil
-- 
Phil Ender
UCLA Statistical Consulting Group
*
*   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/

*
*   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/