Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Jen Zhen <jenzhen99@gmail.com> |
To | statalist@hsphsun2.harvard.edu |
Subject | st: SEs after -rdob- with covariates, & use of covariates in -rd-? |
Date | Fri, 15 Apr 2011 11:37:03 +0200 |
Dear Statalist members, I am trying to implement a Regression Discontinuity Design using Imbens&Kalyanaraman's -rdob- and Austin Nichols' -rd-. I have a number of difficulties and would be most grateful for any advice about these: (1) When, to choose the optimal bandwidth according Imbens&Kalyanaraman, I use -rdob- without covariates, -rdob outcome runningvar, c(0)-, the chosen bandwidth is so small that my estimates are no more statistically significant at the conventional levels. So I would like to see whether I increase statistical precision by controlling for some pre-determined covariates. However, whenever I include covariates, -rdob outcome runningvar covariate, c(0)-, -rdob- does not give me standard errors any more. Nor does -rdob- seem bootstrappable (for reasons unknown to me). So I considered bootstrapping instead the -rd- command, using the bandwidth obtained through -rdob-. However, from what I understand -rd- does not allow for inclusion of covariates? Is there a ensible way around this? (2) Another "convenience" issue I have is that when I use Ben Jann's -esttab- to easily export the results of many -rdob- regressions into Excel, what I get in Excel does no more correspond to the (think correct) output displayed directly after each -rdob- regression. Does that mean that I cannot use these two commands together, or does anyone know a way around this? I used: -rdob outcome1 runningvar covariate, c(0)- eststo est1 -rdob outcome2 runningvar covariate, c(0)- eststo est2 esttab est_* using output.csv, options Thanks a lot, JZ * * 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/