--- On Tue, 2/2/10, V. Martini wrote:
> I'm estimating a survival analysis parametric model.
> The dataset includes different countries, so I have
> introduced a dummy for each country.
>
> Now, I would like to introduce frailty shared at country
> level in my model, after having removed the dummies for
> country: does it make any sense, or are the frailty
> parameters simply going to "replace" the dummies?
>
> I other words: introducing in my model
>
> -country dummy or
>
> -no country dummy + frailty (shared are country level)
>
> are equivalent procedure?
Similar but not the same, the frailty model replaces
the dummies with the idea that these come from a
distribution. The disadvantage is that we must asume
this distribution is initially independent from the
other observed variables. The advantage is that we
estimate less parameters, instead of all the dummies
we estimate one variance. This is particularly relevant
when you want to study the influence of variables that
remain constant within countries, e.g. their size.
These variables can be entered in frailty models, while
with the country dummies these would drop out of the
model due to multicolinearity. The distinction is
similar to the distinction between fixed effects and
random effects models.
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/
