Kristy,
A useful reference is:
Modelling Binary Data, 2nd ed, David Collett, Chapman & Hall.
Other useful ones are
Analysis of Quantal Response Data, BJT Morgan, Chapman & Hall
and, of course, Finney's Probit Analysis, of which I don't have the details
with me.
Here is an example from p 101 of Collett:
. probit response logdose, nolog
Probit estimates Number of obs = 200
LR chi2(1) = 86.86
Prob > chi2 = 0.0000
Log likelihood = -87.25441 Pseudo R2 = 0.3323
------------------------------------------------------------------------------
response | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
logdose | -1.054488 .1328878 -7.94 0.000 -1.314943 -.7940326
_cons | -5.276716 .6441272 -8.19 0.000 -6.539182 -4.01425
------------------------------------------------------------------------------
. disp -5.276716/1.054488
-5.004055
. * ED50 of log(ED50) is -5.004
. vce
| logdose _cons
-------------+------------------
logdose | .017659
_cons | .084379 .4149
. disp sqrt((0.4149 - 2*5.004*0.084379 +
5.004*5.004*0.017659)/(1.054488*1.054488))
.10651964
. * approximate SE of log(ED50) is 0.1065
You could also use Fieller's theorem to obtain confidence intervals for the
ED50 - this is also described in the above refs.
John Plummer
*
* 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/
