ApolloChoiceModelling forum

Dear Stephane,

I use ol_setting to use the ordered logit model.
However, in the result,.

==================================

Estimation method : bfgs
Model diagnosis : successful convergence
Number of individuals : 988
Number of rows in database : 8892
Number of modelled outcomes : 8892

Number of cores used : 1
Model without mixing

LL(start) : -14154.99
LL(0) : Not applicable
LL(final) : -13279.66
Rho-square (0) : Not applicable
Adj.Rho-square (0) : Not applicable
AIC : 26599.32
BIC : 26741.18
================

It shows like this, Rho-square and LL(0) are Not applicable.
Is this the right function?
How can I get the rho-square, or find the estimation result shows good fitting?

Hi

we have so far not included rho square for ordered logit, as the rho2 in Apollo is against a model with all parameters fixed to 0, which of course would fail with OL. In the next release, we are replacing this by rho2 against a model with equal shares for all alternatives, which will make it appropriate for OL too. It will also include rho2 against a model that replicates the aggregates share in the data.

Stephane