ApolloChoiceModelling forum

Dear all,

I am a new user of Apollo, working on HB_MMNL model.
I have some questions to interpret the model output.

1. I can see there are some different outputs of estimation. If I understand correctly, "Upper level model results” could be used to parameter estimates. Is it correct?

2. I have a difficulty to interpret the “t-test value for Geweke test” How can I determine the significance of estimated parameters? Is it the same as the general (e.g. 1.96 for 95% etc)?

3. How can I figure out goodness-of-fit of the models? It seems classical rho-squared values are not applicable. Is there any alternative index?

Thanks in advance.

Best regards,
Yeonwoo Jung

Hi

this is really more a theoretical question about Bayesian estimation than about Apollo per se. I suggest you consult the Bayesian literature for the interpretation of results and goodness-of-fit. In terms of the Geweke test, please also have a look at the paper cited in the Apollo manual. This is not for the significance of parameters, but to test for the convergence of the chains.

Best wishes

Stephane