Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


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

Re: RE: st: Beta values in QREG


From   Maarten Buis <maartenlbuis@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: RE: st: Beta values in QREG
Date   Tue, 25 Jun 2013 08:47:42 +0200

On Mon, Jun 24, 2013 at 10:50 PM, Santos Silva, J.M.C. wrote:
> As for Maarten's argument of "clash of logics", I think I do
> not agree with it for two reasons. First, you may be interested in estimating
> quantile regression even if the data have finite second (and higher order)
> moments <snip> Second, even if you are using quantile regression because
> your y does not have finite second moments, that does not mean that your
> regressors also do not have second moments <snip>

I did not say that you would use quantile regression when the moments
do not exist, I only said that you would use quantile regression when
you wanted to avoid using the moments. Particular in skewed
distributions there are good reasons for prefering percentiles over
moments, even if the moments exist. In that case it is weird to use a
technique whose purpose is to avoid those moments, and than use them
again to define the scale of your dependent variable... I made an
argument in the style "Not everything you can technically do makes
substantive sense".

-- Maarten

---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
Germany

http://www.maartenbuis.nl
---------------------------------
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index