Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: How to compute SE for linear or nonlinear combination of params?


From   Stas Kolenikov <skolenik@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: How to compute SE for linear or nonlinear combination of params?
Date   Fri, 22 Oct 2004 11:59:30 -0400

Note that things are not so simple with nonlinear tests. Depending on
the curvature of transformation, you can get different results. See
Drukker's paper in STB-54 on that.

Also, if the slope is not statistically significant, then you are
essentially diving by zero, and thus any reasonable method should give
you something like infinite variance of your ratio.

If you are interested in something like the point of zero crossing
(judging from the form of your nonlinear combination), then you can
use inverse regression / regression calibration approach to construct
it. See my -calibr- procedure that I hope is downloadable from
somewhere on the Net.

Stas

On Fri, 22 Oct 2004 09:31:10 -0500, Richard Williams
<richard.a.williams.5@nd.edu> wrote:
> At 10:19 AM 10/22/2004 -0400, you wrote:
> >Frank Zhang wrote:
> >
> >>Dear Statalisters,
> >>If I estimated a regression equation like this,
> >>y=b0hat+b1hat*x1
> >>If I want to get the standard error of (-b0hat/b1hat),
> >>how can I do it?
> >>It seems easy in TSP and LIMDEP, but I don't know how
> >>to get it with STATA? Your help is highly appreciated.
> >
> >It is easy in Stata, too. See testnl.
> >Dave Harless.
> 
> Not sure, but since he wants the s.e. I think he wants
> 
> nlcom (_b[_cons]/_b[x1])
> 

-- 
Stas Kolenikov
http://stas.kolenikov.name
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



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