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, is already up and running.

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

Re: st: rreg and suest

From   Nick Cox <>
Subject   Re: st: rreg and suest
Date   Tue, 25 Sep 2012 19:44:20 +0100

I guess that this requires a fix to -rreg_p- by someone, ideally
StataCorp. That is, -predict- after -rreg- uses -rreg_p-, which it
seems does not allow a -score- option.

On Tue, Sep 25, 2012 at 7:29 PM, Amy Dunbar
<> wrote:
> I would like to use robust regression (rreg)  and suest to compare coefficients in pre vs post models.  I realize I can create an indicator variable and use a single model, but I wanted to show the two approaches get the same results.
> If I use reg instead of rreg, I have no problem.
> sysuse auto.dta
> **********foreign***************
> rreg price mpg weight if foreign ==1
> estimates store m0
> **********domestic***************
> rreg price mpg weight if foreign ==0
> estimates store m1
> ********compare coefficients across models **************
> suest m0 m1
> test [m0_mean]weight = [m1_mean]weight
> error message:
> unable to generate scores for model m0
> suest requires that predict allow the score option
> Is there another way besides using an indicator variable to compare coefficients across equations using rreg?
>  I found the following comment from Nick Cox in a post:
> "I'd use -qreg- or one of the newer user-written robust regression programs rather than -rreg-, if only because I would have a job explaining to myself why there is not a less arbitrary way to do robust regression than -rreg-."
> I haven't tried other estimators because I am trying to replicate a paper.

*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index