Re: st: option aweight together with rreg

[Nick Cox <njcoxstata@gmail.com>]

Your first question is easy: -rreg- doesn't support weights. That's a
clear implication of the help http://www.stata.com/help.cgi?rreg and a
glance at the code shows that Stata is being straight with you: there
is no undocumented support either.

As to why, a user has to guess, and here's mine. -rreg- is an ancient
command and it  remains in Stata because some people may want to use
it and no purpose would be served by deleting it. I guess, and I keep
using the word, that no one thought about implementing weights when it
was first written, or just possibly they thought about it and couldn't
see a way to do it that wasn't problematic or contentious. One very
important trait of StataCorp that I appreciate very deeply is that
stuff doesn't get added to Stata without a lot of testing and the
company pulls back on releasing lots of stuff because it doesn't trust
it. (This is the kind of comment frequently made by Stata's developers
at users' meetings or the Stata Conference.)

I'd search the archives for comments on -rreg- and you will find some
positive support for it. Other people can, of course, say exactly what
they want, but I said what is quoted below in 2011

http://www.stata.com/statalist/archive/2011-01/msg00416.html

and the only thing that seems out of date is the date.

% begin
[...] 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.

 % end

Nick
njcoxstata@gmail.com

On 8 August 2013 17:24, Dacil Juif <dacil.juif@uni-tuebingen.de> wrote:

> I have trouble with running a robust regression rreg and including weights
> (aweight=population). Stata tells me that "no weights allowed".
> The regression is:
> rreg cognitive landgini1890 lngdpc1910 tropicshare lnpopdens polity2
> fertil1950 (aweight=pop)

> Does somebody know whether there is another way to use weights or why it is
> not possible to include them?
*
*   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/