[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: Clarification on biprobit
>  . biprobit (p = X s) (s = X p),
> with the constraint that [athrho]_cons = 0.
> 3. Would this be equivalent to running a two separate probits? I ran
> ". probit p X s" and did not get the same results.
Yes, it is the same. The reason why I got different results the first
time is because I ran -probit- instead of -svyprobit-, which I should
have. As I indicated before, I have survey data with pweight.
> 4. -mfx- is able to compute the marginal effects but fails for the
> standard error, returning "warning: predict() expression unsuitable
> for standard error calculation". Can I still be confident in
> interpreting the marginal effects in this case?
First, -mfx- fails to compute the standard errors if I run it after the
constrained -biprobit-. However, if I run two separate -svyprobit-'s
(which, as mentioned, gives me the same results as the constrained
-biprobit-), -mfx- is able to compute the standard errors in both cases.
The marginal probability (p = 1) in the bivariate probit is exactly the
same as the "probability of positive outcome" in the single-equation
probit. I suppose it is no problem if I use the standard errors
reported by -mfx- after -svyprobit-.
The question of confidence about the accuracy of the reported dy/dx's
without standard errors has not been resolved:
The rest of my questions
(http://www.stata.com/statalist/archive/2005-02/msg00797.html) are still
unresolved. I am particularly interested in questions 5-7. Any ideas
would be tremendously appreciated.
* For searches and help try: