Dear list
This is how to do it.
. gllapred u , u // store means of random effects in um
. replace time = 4 // change the fixed effect to any value
. gllapred mu, mu us(um) // predict probabilities with the original
random effects
The simplicity and beauty in the solution suggest that the good authors
of -gllamm- had anticipated this need. Thanks a lot.
The trick is to predict the random effects before changing the fixed
effects and supply them to -gllapred-. If the random effects are not
supplied, -gllapred- calculates them from data using both independent
and dependent variables which would of course give nonsense predictions
if data had changed.
Thanks to Buzz who came very close to the solution.
On 2004-07-07, at 00.41, Winfield Scott Burhans wrote:
The key to what you are after is to use - gllapred prob, mu us(um) - to
obtain the predicted probabilities conditional on the random effects.
You
will have to work out a scheme to either add observations with the
fixed
effects you want, or change existing observations to have the constant
fixed effects you are after temporarily in order to plot them.
Michael
PS. If you get stuck, what you probably need is a good night sleep.
*
* 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/
