ApolloChoiceModelling forum

Dear Stephane and David,

I run a couple of mixed logit models in which I asked respondents to choose between two unlabelled alternatives (i.e. Programme A vs Programme B), described by 6 attributes (each one with 3 levels).

Now that I have the results, I would like to run a post-estimation scenario analysis to investigate what is the value that people place on each potential programme. In other words, I want to come up with a list of programmes (ranked) that would maximize sample population’ utility.

To do so, I think I need to estimate the predicted probabilities for each potential programme (using the estimated coefficients). I don't think I can use the apollo_prediction function in this case. Any guidance on this would be much appreciated.

Thanks!

Best wishes,

Pamela.

Pamela

can you explain a bit more what you're after. In your prediction scenario, would the choice be between all potential programmes, or would you create a number of binary cases, for example?

You can definitely use apollo_prediction for predictions on data different from the estimation data

Stephane