.-
help for ^hetprob^ (work in progress 7mar1997)
.-
Estimate heteroskedastic probit model
--------------------------------------
^hetprob^ depvar [indepvars] [^if^ exp] [^in^ range], variance(varlist)
[ ^nochi l^evel^(^#^)^ maximize_options ]
^hetprob^ shares the features of all estimation commands; see help @weights@.
^hetprob^, typed without arguments, redisplays previous results.
To reset problem-size limits; see help @matsize@.
Description
------------
^hetprob^ estimates heteroskedastic corrected probits. If you type
. ^hetprob y x1 x2 x3, variance(z1 z2)^
^hetprob^ estimates b0, b1, b2, b3, g1, and g2 of
Pr(y~=0) = F( b0 + b1*x1 + b2*x2 + b3*x3 where
s^^2 = exp(g1*x1 + g2*z2) )
= F( [b0 + b1*x1 + b2*x2 + b3*x3] /
exp[(g1*x1 + g2*x2)] )
Examples
---------
. ^hetprob outcome bp age, v(group1 group2)^
. ^hetprob works educ sex, v(age)^
. ^hetprob works age, v(age)^
Calculating predicted values
-----------------------------
After ^hetprob^ estimation, you can calculate predicted probabilities of
a positive outcome by typing
. ^predict i1^
. ^predict s, eq(ln_var)^
. ^replace s = exp(s/2)^
. ^gen p = normprob(i1/s)^
Authors
--------
William Gould
James Hardin
This is work in progress. Results should be viewed with suspicion. Direct
comments either to the authors or Statalist. When work on this project is
complete, the authors will seek publication in the Stata Technical Bulletin;
see help @stb@.
Code improvements are invited. The two ado-files associated with this
command are
hetprob.ado
hetpr_lf.ado
Optimization is via @ml@ method ^lf^.
Also see
---------
On-line: help for @est@; @probit@, @xtprobit@