I am running some choice models in JMP. I am using the one-table method rather than the method with the separate profile, response, and subject tables. When I run my model with the Firth bias adjusted estimates box checked, the model fails with the note "cannot calculate objective at starting value." When I uncheck the Firth box, the model runs and I get output that appears reasonable. Couple questions:
1) Is it essential to use the Firth bias adjusted estimates?
2) Is there something that I could do to make the Firth bias adjusted estimates run?
I am wondering if it might be worth trying to run the model with the profile data separated from the response and subject data.
Any advice would be appreciated.
Thanks.