You can generate a bunch of spline variables to produce locally
polinomial terms. I don't see why economists don't use them as much as
they do kernel methods. Splines are much nicer to deal with -- you can
just plug them into -regress- or whatever command you are using,
rather than bother figuring out custom kernel estimators in more
complicated cases. The theory for RD seems to be developed for kernel
estimates, though. So if you just try to see if there is any
difference due to those additional covariates, you can try splines and
see if the results change much. Bias reductions will likely come at a
price though -- I would expect the variances go up, as they would get
the slower non-parametric convergence rates.
On Tue, Feb 17, 2009 at 10:02 AM, Martinez, Erika <[email protected]> wrote:
> Hello All,
>
> I am trying to perform Kernel-weighted local polynomial smoothing with
> multiple independent variables. From what I can find, lpoly performs these
> regressions, however only using a single xvar. How would I implement this
> in STATA?
>
> My reasoning for doing so is that I am trying to use a regression
> discontinuity design and introduce additional covariates. I know that the
> use of these covariates will have little effect on the estimator but
> including these variables may help improve precision and also eliminate some
> bias when observations that are not close to the cutoff are included in the
> RD framework.
>
> I would greatly appreciate any help or suggestions. Thanks.
>
>
> *
> * 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/
>
--
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/
