Re: st: random coefficient model for cross-sectional data?

["Mary E. Mackesy-Amiti" <mmamiti@uic.edu>]

Jian,

First, not to be nit-picky but random coefficient models can be applied 
to cross-sectional data if there are multiple measurements. For example, 
I have cross-sectional data on sexual behavior with questions about each 
sex partner in the past six months.  I can include a random component in 
the analysis because most participants have multiple partners.

To answer your question, no I don't know of any papers that explain 
this.  It seems to me self-explanatory.  I think you should pursue 
Maarten's suggestion to look up "average treatment effect".  I'm not 
familiar with this type of analysis, but I think you might be interested 
in propensity scores.

Mary Ellen


On 3/15/2011 9:18 PM, Jian Zhang wrote:
> Thanks.
>
> Mary, are you aware of some papers that explain why random coefficient
> models can't be applied to cross-sectional data? Intuitively, although
> each unique slope for each individual is not able to be identified, we
> can still assume some kind of distribution for the slope (that is
> treating the slope as a random coefficient).  Given the distribution,
> we could estimate out the parameters (e.g. variance) of the
> distribution with maximum likelihood method.
>
> Jian
>
> On Tue, Mar 15, 2011 at 11:10 PM, Maarten buis<maartenbuis@yahoo.co.uk>  wrote:
> > --- On Tue, 15/3/11, Mary E. Mackesy-Amiti wrote:
> > > In order to have random effects there has to be
> > > variability within the unit of analysis.  If the unit
> > > is the individual then you need to have more than one
> > > observation per individual.  If the unit is city then
> > > you need to have more than one observation per city.
> > --- On 3/12/2011 7:35 AM, Jian Zhang wrote:
> > > > I have a data set on individual wage incomes along
> > > > with education and other individual characteristics.
> > > > It is reasonable that the education gives different
> > > > returns to different people therefore the coefficient
> > > > for education is random coefficient: everyone has an
> > > > unique slope. but I only have one year of data on
> > > > individual wage incomes.  So it is still possible to
> > > > use random coefficient model?
> > As mary explained there is just no way to estimate the
> > model you want.  Such individual level effects are just
> > unidentified.  Even though it is impossible to estimate
> > these individual level effects, it is often possible to
> > get some leverage on the average of these effects.  To
> > see a list of references and programs relevant to Stata,
> > type in Stata: -findit average treatment effect-
> >
> > 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/

--
Mary Ellen Mackesy-Amiti, Ph.D.
Research Assistant Professor
Community Outreach Intervention Projects (COIP)
School of Public Health m/c 923
Division of Epidemiology and Biostatistics
University of Illinois at Chicago
ph. 312-355-4892
fax: 312-996-1450

*
*   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/