Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: st: RE: Robust instrumental variable regression


From   Nick Cox <n.j.cox@durham.ac.uk>
To   "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu>
Subject   RE: st: RE: Robust instrumental variable regression
Date   Fri, 14 Jan 2011 17:41:10 +0000

Your question was perfectly clear to me. I think that my concerns are orthogonal to yours. I don't know anything that would let me answer a question on 2SLS. 

Considering use of some flavour of robust regression is easy to understand. I do it myself occasionally. More often, when I have outliers, I have skewness too and I have nonlinearity too and working with nonlinear regression or a non-linear link seems more appropriate. But experiences and purposes differ, and others may make different choices. 

I have noticed in the past that some people seem unaware that -rreg- being labelled as robust regression does not mean as much as they want. I doubt very much that you fall into that group, however. This isn't StataCorp's fault. What -rreg- is and what it does are explained perfectly clearly. It is just that some users don't read that documentation carefully. 

Nick 
n.j.cox@durham.ac.uk 

Feiveson, Alan H. (JSC-SK311)

Nick - I guess I didn't state my question clearly. My question had nothing to do with the merits (or lack of) of -rreg-. What I was asking is whether 2SLS has been found to gives reasonable estimates of parameters in an IV setting when some form of robust regression was used in each stage, whether that be -rreg-, median regression, or anything else.

Nick Cox

In addition to other comments, I'd advise against basing anything much on -rreg-. 

The help file has it right: -rreg- is "one version of robust regression". When -rreg- was written the method seemed a good all-round flavour of robust regression, but it is doubtful whether it now looks like _the_ method of choice to anyone in 2011. 

If you ever used -rreg- for real, you'd be obliged to explain it and defend the choice in any serious forum. 

"I used robust regression" means virtually nothing. There are probably hundreds of ways to do robust regression (quite apart from what robustness means). 

"I used -rreg- as implemented in Stata" counts for little outside this community. 

"I used robust regression as codified by Li (1985)" obliges you to explain why you didn't use something more recent (to fad- and fashion-followers) or something else that someone else fancies for some reason of their own. The literature would keep you busy indefinitely. 

Li, G. 1985. Robust regression. In Exploring Data Tables, Trends, and Shapes, ed. D. C. Hoaglin, F. Mosteller, and
J. W. Tukey, 281-340. New York: Wiley.

Outliers could be handled in many different ways. Considering transformations on one or more variables is another way to do that. Wonder whether a linear structure makes sense scientifically is yet another. 

Maarten buis

--- Ramiro H. Gálvez asked:
> I am using stata 11 and I'm having problems with outliers
> in a 2SLS instrumental variable regression. Is there any
> implementation in stata equivalent to rreg for
> instrumental variable regression (like rivregress)?

--- On Fri, 14/1/11, Jan Bryla answered:
> > Could using the vce(robust) option when performing
> > -ivregress- potentially solve you problem?

No, -vce(robust)- is robust in a very different sense than
-rreg-. When thinking of robust in terms of -rreg- you
worry about the effect of outliers on the point estimates.
Robust in the sense of -vce(robust)- has to do with the
influence of deviations from model assumption on the 
standard error. 

*
*   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/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index