David Parsley --
The term "treatment" is often reserved for binary X, hence -treatreg-
and the treatment effects literature apply to binary endogenous X. If
you rephrase as "selection or simultaneity makes X endogenous and its
coef biased" then you will have better luck with the textbooks. There
are a variety of matching (-findit psmatch2- -findit nnmatch- etc.),
IV regression (-findit ivreg2- etc.), and Regression Discontinuity
(http://nber.org/papers/W13039) techniques to estimate a causal impact
in this situation--come to NASUG in Boston in two weeks if you want to
discuss further:
http://stata.com/meeting/6nasug/abstracts.html#nichols
On 7/30/07, David Parsley <[email protected]> wrote:
> I am trying to model the effect of a treatment (x-variable) on an
> outcome (y-variable), where the treatment is a self-selected, CONTINUOUS
> variable (e.g., hours of study, dollars spent, etc.), and the outcome is
> also continuous (performance). I am interested in the effect of the
> treatment on the outcome in the general population - not just the
> treatment sample. I observe y for the entire population (i.e.,
> including those studying/spending, and those not). My treatment is
> either zero, or some positive value (lots of clustering at zero).
>
> I have searched through econometrics texts/articles and (~) 99.99% of
> them focus on self selection of the y-variable. In my case, since the
> selection is on the (continuous) x-variable, including the non-treatment
> population would result in clustering at zero, and therefore biased
> estimates. I have found discussions of treatment effects for dichotomous
> independent variables, but not for 'simple' cases of continuous (though
> strictly positive) treatments.
>
> Treatreg seems to apply only to dichotomous varibles.
*
* 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/
