ApolloChoiceModelling forum

Hi Stephane!

I would like to use some sociodemographic variable in my utility function in order to analyse whether they have impact for choosing the opt-out alternative.

I saw you use this approach in apollo_example_3:

asc_bus_value = asc_bus + asc_bus_shift_female * female
asc_air_value = asc_air + asc_air_shift_female * female
asc_rail_value = asc_rail + asc_rail_shift_female * female

So in this case you analyse, whether female respondents prefer bus/air/rail relative to car and male respondents.

I tried to use in my case the following interaction effect:

asc_no_choice_value = asc_no_choice + asc_no_choice_shift_female * female

So I would like to analyse in this case, whether female respondents prefer no choice alternative relative to choice and male respondents.

However, what is the difference if I do not creat interactions but I also create a parameter for my sociodemographic variable and use that in my utility function of no choice alternative:

U(no choice) = asc_no_choice + b_female*female

Thank you for your help!

David

David

the two approaches are exactly the same

Stephane

Perfect, thanks for the help!