Statalist


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

st: comparing loglikelihoods of ols dummy regression model vs random intercept model


From   Marcelo Jenny <mjenny@uni-mannheim.de>
To   statalist@hsphsun2.harvard.edu
Subject   st: comparing loglikelihoods of ols dummy regression model vs random intercept model
Date   Mon, 27 Apr 2009 15:57:25 +0200

Could somebody please point me to the reason why a bare model with dummy
variables does better in terms of the loglikelihood of the model than
the respective random intercept model?

The data structure is n=121 at the individual level, n=9 at the group
level. The number of cases at the group level varies between 4 and 23.

I'm using xtmixed for the hierarchical models specifying either mle or reml.

The ols dummies model gives a loglikelihood of -196, the hierarchical
model with reml gives -201 and the hierarchical model with mle gives a
loglikelihood of -215.

Thanks in advance,
Marcelo Jenny


-- 
Mag. Marcelo Jenny
Lehrstuhl f. Politische Wissenschaft III
Prof. Wolfgang C. Müller
Fakultät für Sozialwissenschaften
Universität Mannheim
Seminargebäude A5, Zi. 338
D-68131 Mannheim
Tel. +49/621/181-2077
Fax: +49/621/181-2080
Email: mjenny@uni-mannheim.de
*
*   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