Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | Lukas Borkowski <LukasBork@hotmail.com> |
To | statalist@hsphsun2.harvard.edu |
Subject | st: Quadratics and FD? |
Date | Mon, 9 Jul 2012 23:21:17 +0200 |
Dear all, I use Stata 11.2 and work on a two period balanced panel. I try for ages to plot my key explanatory variable in a regression model in order to show graphically that there might be a turning point where the sign of my key explanatory changes. I want to compare these turning points over three categories of observations (ldc, lldc and sids). Now, the tricky thing is that I want to do that while controlling for unobserved fixed effects using the first differencing method suggested by Wooldridge (2002). I understand that the code for the pooled OLS estimator would be // qui reg lngdp_wb $controls c.pter##c.pter##(ldc lldc sids), vce(cluster id) preserve foreach var in $controls { sum `var' replace `var' = r(mean) } foreach var in ldc lldc sids { qui replace ldc = 0 qui replace lldc = 0 qui replace sids = 0 qui replace `var' = 1 qui predict yhat_`var' } sort pter twoway line yhat_ldc yhat_lldc yhat_sids pter // but when replace the first line with -qui reg lngdp_wb D.($controls pter ldc lldc sids pter_ldc pter_lldc pter_sids pter2 pter2_ldc pter2_lldc pter2_sids), vce(cluster id)- then the plotter scat is total useless due to extrem volatilities. Do you have any suggestions what I can do in order to work that out? Regards # Lukas Borkowski * * 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/