Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


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

Re: st: Are my -xtmixed- intercepts accurate?


From   Maarten buis <maartenbuis@yahoo.co.uk>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Are my -xtmixed- intercepts accurate?
Date   Thu, 19 Aug 2010 10:37:58 +0000 (GMT)

--- On Wed, 18/8/10, Brandon Olszewski wrote:
> ...I compared these predicted intercepts to the actual
> scores (by teacher using -list-), and also to those
> calculated using a Bayes’  prediction with:
> quietly xtmixed avg time tx || tid: tx, cov(un) || id:,
> cov(uns) mle
> predict eb1 eb2 eb3, reffects
> list id eb1 eb2 eb3 teacher_id if student_id<100
> 
> From what I can tell, both the mle and Bayes’ predictions
> are way off. For instance, at time=0, most of the observed
> lines would cross the y-axis in between 0.2 and 0.5 
> (between 20% and 50% correct for the whole test), but the
> predicted intercepts are in between -0.07 to 0.08.
> 
> 1. How do I interpret these? Are these values transformed
> or standardized in some way? Why aren't they closer to the
> actual values? Moving from time=1 to time=0 should not
> produce such big drops in the predicted scores.

Yes they are standardized to have a mean of 0 (they are after
all a type of residuals).  
 
> 2. I interpret the Bayes’ predictions this way: eb1 =
> predicted intercept for different teachers’ students; eb2 =
> predicted slope for different teachers’ students; eb3 =
> predicted intercept for individual students. Is that
> accurate? If not, how should they be interpreted?

That is not accurate, they are the deviations from the overall
mean.
 
> 3. My goal is to know whether or not my model fits well,
> how to interpret the results correctly, and produce a graph
> showing different regression lines for different teachers
> Rabe-Hesketh & Skrondal produce a couple of these, such as
> Figure 4.3 (pg. 146) and Figure 4.10 (pg. 165). 

Just copy their approach (the lines of code necesary appear 
close by in the text), they predict the outcome directly
using the -fitted- option.

Hope this helps,
Maarten

--------------------------
Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
Germany

http://www.maartenbuis.nl
--------------------------


      

*
*   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   |   Site index