st: Standard error calculation in predictnl

[andrew@brandeis.edu]

Dear Statalisters,

I am attempting to calculate the the marginal effects, and their  
standard errors, for interaction terms in a probit model using  
predictnl following Boggess  
(http://www.stata.com/support/faqs/stat/mfx_interact.html). The  
dataset is relatively large (70,000) cases.

While predictnl can calculate the mfx and std. errors for most of the  
interactions, there remain some whose standard errors cannot be  
calculated. I receive the error message "Maximum number of iterations  
exceeded." Changing the scale of the variables such that the mfx  
aren't too small has helped in some cases, but not in all. Also,  
increasing the number of iterations and using the "force" option have  
not worked.

In advice would be appreciated.

Best,

Andrew Ryan
Brandeis University
*
*   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/