prediction in bivariate probitWatch
I have a econometrics question. Say one was using a bivariate probit( 1 and 0). One of the outomes ,say 1, is 5% of the population. The other 95% of the population is zeroes.
The estimation worked safe. But the predictions came out awful. The event capture is very low..or can safely say negligble.
Could somebody suggest a way out of this problem?