On 12/28/08, [email protected] <[email protected]> wrote:
> I'm new to fractional polynomial models. Is there a longitudinal data option
> for mfracpol or fracpoly?
In what particular way do you think of introducing a longitudinal
aspect to this procedure? To me, -*fracpol*- models are mostly useful
for smooth interpolation leading to prediction, and frankly I won't
see terribly much use for them in causal micro level inference --
that's what you are probably most interested in with longitudinal
data. If you are interested in flexible formulations with longitudinal
data, I'd probably be looking towards say B-splines, and/or their
multivariate generalizations. They allow for the familiar (X'X)X'y
formulae (or whatever you can have for more complex models) in
(generalized) linear models (something you can put into -xtm*- models
as predictors), at the same time producing smooth curves through data.
--
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: I use this email account for mailing lists only.
*
* 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/
