Nicole Danna <[email protected]>:
If your outcome is y, your "treatment" is x and other RHS variables
all start with v, then you can calculate a propensity score with
logit x v*
predict p
Then you want to make sure that p does not have positive density near
zero or one, e.g.
ssc inst kdens
sysuse nlsw88, clear
logit collgrad south smsa c_city married never_married
predict p
kdens p, ul(1) ll(0)
*(density should be zero for two intervals that include 0 and 1)
*and compare overlap:
loc w=r(width)
kdens p if collgrad==1, ul(1) ll(0) bw(`w') g(d1 x) nogr
kdens p if collgrad==0, ul(1) ll(0) bw(`w') g(d0) at(x) nogr
line d0 d1 x, name(comp, replace)
Then you can match or reweight to estimate the effect of x on y,
assuming v* are exogenous and selection on x depends only on v*;
see also papers at http://www-personal.umich.edu/~nicholsa/stata
On Thu, Apr 23, 2009 at 1:08 AM, Nicole Danna <[email protected]> wrote:
> Hello.
> I'm writing a undergraduate economic honors thesis and need help calculating
> propensity scores with Stata. I'm looking for step by step instructions on
> how to calculate a P score.
> Thank you!
*
* 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/
