Re: st: Modeling % data

[Austin Nichols <austinnichols@gmail.com>]

Maarten--

I don't see how data approaching the boundaries is a problem in
-qreg-, as long as the fraction at the boundary itself is not too
large (though that in itself is more an indictment of the outcome
measure than a necessary problem for quantile regression).  If 10% of
the outcomes are at the lower boundary (zero) for low X and 10% of the
outcomes are at the upper boundary (100) for high X, how is that a
problem for estimating how the conditional median changes with X?

On Wed, Sep 22, 2010 at 11:51 AM, Maarten buis <maartenbuis@yahoo.co.uk> wrote:
> --- On Sep 22, 2010, at 8:03 AM, Marlis Gonzalez Fernandez wrote:
>> My outcome variable is a % (% error in a language test).  We do have
>> many 0 and 100.  I need to be able to do a multiple regression to
>> adjust for known predictors of the variable vs. the predictors of
>> interest.
>>
>> It was suggested that I use qreg.  I've done so and it seems to
>> work.
>
> This all depends on how close your dependent variable gets to the boundaries of 0% and 100%. If the data stays well within the range of
> 20%-80% than I would have no problem using either -qreg- or just regular
> -reg-. However, when you have observations that get close to these
> boundaries, you'll probably want to take them into account. For that
> there is a whole suite of commands available, which I discussed at the
> last German Stata Users' Group meeting:
> <http://ideas.repec.org/p/boc/dsug10/04.html>
>
> (You'd have to look at your variable as an proportion rather than a
> percentage. But that is trivial, just divide by 100.)

*
*   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/