Chapter 5 RDRobust
There is a bias-variance tradeoff when selecting bandwidth size. The smaller the bandwidth window, the lower the bias, but fewer observations which increases the variance. The larger the window, the higher the bias, but more observations, which decreases the variance. RDRobust optimizes the tradeoff by choosing the optimal bandwidth sizes *Which may vary to the left or right of the cutoff.
Use local polynomial point estimators with bias correction
We can estimate the \(LATE\) with the RDrobust command.
Mass points detected in the running variable.
Sharp RD estimates using local polynomial regression.
Cutoff c = .5 | Left of c Right of c Number of obs = 13577
-------------------+---------------------- BW type = mserd
Number of obs | 5480 8097 Kernel = Triangular
Eff. Number of obs | 2112 1895 VCE method = NN
Order est. (p) | 1 1
Order bias (q) | 2 2
BW est. (h) | 0.086 0.086
BW bias (b) | 0.141 0.141
rho (h/b) | 0.609 0.609
Unique obs | 2770 3351
Outcome: score. Running variable: demvoteshare.
--------------------------------------------------------------------------------
Method | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------------+------------------------------------------------------------
Conventional | 46.491 1.2405 37.4775 0.000 44.06 48.9227
Robust | - - 31.4254 0.000 43.2929 49.0524
--------------------------------------------------------------------------------
Estimates adjusted for mass points in the running variable.