ApolloChoiceModelling forum

Dear Prof Hess

With reference to your paper titled, "A joint model for vehicle type and fuel type choice: evidence from a cross-nested logit study", I am also trying to estimate a mixed NL and CNL model. But regarding the random specification part, I have certain doubts.

i) For the mixed NL model by fuel type, did you have six columns for each of these random variables - Fuel cost, Maintenance cost, Price corresponding to the alternatives in the lower nest.

ii) https://drive.google.com/file/d/1hKYFdb ... sp=sharing
This is the mixed NL model which I am trying to build with Travel Time as a random parameter. So do I need to have six travel time columns??

iii) or 3 columns corresponding to the three modes will suffice ??

Hi

in any structure like this, the bottom row of your tree are individual alternatives, where you need one utility function for each alternative. Whether or not you need 6 columns in your data for a given attribute depends completely on your case study, and whether the same value for the attribute applies to multiple alternatives or not. This has nothing to do with the model using a mixture.

Stephane

Thanks, Prof Hess for the clarification.