[Limdep Nlogit List] multivariate probit

[r.rajan at reading.ac.uk]

hi, Hi, I am estimating a multivariate model with 4 options..however I am 
getting very high (but negative)rho vales (-.94) for two of my options...is 
this okay?

Also, a small change in the model specification (even after accounting for 
the multicollinearlity in the independent variables) seems to change the 
coefficients drastically....so I wonder if the specification is not 
correct?

Anmy suggestion would be greately appreciated.
Thanks
Raji