Regression Discontinuity Design (Graphs)

[Mona Sameni]

Hi everyone,
I'm working on plotting a regression discontinuity design. So the running variable is age which has a cut-off point at a specific age. My problem is defining the bins and one single fitted line on the two sides of the cut off point...So far I have been using the following code

cmogram searchnf ageint if ageint>18 & ageint<24, cut(21) scatter line(0) qfitci

"

which produces the following graph. Please let me know how I can have a better sense of the bins and the fitted lines. Any suggestions/comments are appreciated.

[Scott Merryman]

You might want to take a look at rdrobust (Calonico, S., M. D. Cattaneo, and R. Titiunik (2014b): Robust Data-Driven Inference in the Regression-Discontinuity Design, Stata Journal 14(4): 909-946.)

In particular, their rdplot has eight different methods to select the optimal number of bins depending on the type of partitioning scheme and the spacing or polynomial regression estimators used. The Mimicking Variance methods are tailored to represent the overall variability of the data while "the IMSE-optimal number of bins specifically tailored to produce an estimator that approximates the underlying regression function well. In this approach, the optimal number of bins is selected to balance squared bias and variance to approximate the underlying conditional expectation globally."

The site https://sites.google.com/site/rdpackages/rdrobust has a number relevant papers and the most current version rdrobust.

[Mona Sameni]

Dear Scott, thanks to your help, I have been working on this paper for quite a while. rdplot command gives me some nice graphs. I am wondering what causes the following problem in the graph. The cutoff point is age 21, but if you look closely the area between age 20 and 21 is blank. Do you have any idea why this might happen.
Thanks in advance.

[Mona Sameni]

Same thing happens when I use twoway and scatter commands.

[Scott Merryman]

I suspect it is because there are no observations between 20 and the cut off point.

For example:

use "D:\Data\rdrobust_rdsenate.dta", clear
drop  if margin < 0 & margin > -30
rdplot vote margin