Probit model not converging - maximize options

[Valeria Zimina]

Hi,

I am running a probit model and it does not converge. As all independent variables are crucial to my analysis, before eliminating the problematic variables I would like to try maximize options. After spending some time researching I am still unable to understand which command to use, when and how. I am specifically interested in "tolerance" and "difficult" - therefore my question is how I should use these two options with my probit model in order to relax convergence on Stata 16?

Thanks in advance!

[John Mullahy]

Unfortunately when a probit model doesn't converge that is typically an indication that one or more of your RHS variables is the cause. A typical case is one where a non-negative RHS variable is non-zero only when your outcome variable is zero (there are other problematic scenarios but this one is common).

Before attempting different maximize options it would be better to check whether such data configurations exist in your sample. If so you won't be able to estimate the probit model with that RHS variable included.