# Re: st: Plotting a Local Polynomial Regression with CIs Accounting for Clustering

 From L S To statalist@hsphsun2.harvard.edu Subject Re: st: Plotting a Local Polynomial Regression with CIs Accounting for Clustering Date Wed, 2 Dec 2009 12:55:59 -0600

Thank you to Maarten for the suggestion on mvrs.  I think this seems
more intuitive / simpler than the fractional polynoimals.  Thanks a
lot also to everyone for the other suggestions.

The goal in doing this is actually not a regression discontinuity
design, but rather to see whether actual data seems to fall along a
line, which is predicted by theory.  That is, I want to plot actual
data against a theoretical prediction, plotting the actual data using
a flexible functional form, and plotting a 95% confidence intervals.

When people do regression discontinuity in Stata, what is the standard
(or some of the standard) way / command for plotting the functional
form on each side of the discontinuity?

On Tue, Dec 1, 2009 at 7:16 AM, Austin Nichols <austinnichols@gmail.com> wrote:
> You can do it with a local polynomial, it's just not terribly easy.
> You have to pick the points at which you want to generate prediction,
> perhaps quantiles of x, then -bsample- clusters, do the estimation,
> and then -post- or otherwise save the estimates, and finally collect
> all results.  A similar strategy would work with -pspline- (-findit-
> will find that option).
>
> As Nick Cox points out, it's also not clear why you want this. Are you
> perhaps contemplating a regression discontinuity estimator?
>
> On Tue, Dec 1, 2009 at 3:09 AM, Maarten buis <maartenbuis@yahoo.co.uk> wrote:
>> You can't do it with local polynomial curves.
> *
> *   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/
>

*
*   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/