Statalist


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

st: Prediction after censored regression in -gllamm-


From   Eva Poen <eva.poen@gmail.com>
To   Statalist <statalist@hsphsun2.harvard.edu>
Subject   st: Prediction after censored regression in -gllamm-
Date   Thu, 19 Feb 2009 11:38:19 +0000

Hello

Using -gllamm-, I have fitted a model for censored outcomes with three
nested random effects (subjects are nested in group which are nested
in sessions). I only have random intercepts, not random slopes.

I followed the advice by Sophia Rabe-Hesketh in this post
http://www.stata.com/statalist/archive/2004-02/msg00789.html when
setting up my model. My -glamm- command reads

gllamm $yvar $xvars if Matching==1, offset(off) i(SubjectID GroupID
SID) fam(gauss binom) link(ident sprobit)  /*
  */ lv(var) fv(var) from(c) copy adapt

Now I would like to obtain predictions like the ones we can compute
after -tobit- with the ystar(a,b) option, but taking into account the
random effects (BLUPs). My suspicion is that this is not very easy,
since Stata's official -xttobit- only offers ystar(a,b) predictions
where the random effect is assumed to be zero, and I have three random
effects, not one.

I can get the linear prediction, including random effects, via
-gllapred- with the linpred option. Using this, I could obviously
apply the standard formula for the expectation of a truncated normal
distribution, but this sounds a bit too simple in such a complicated
model.

I'd be glad to hear comments about this procedure.

Many thanks,
Eva
*
*   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