ApolloChoiceModelling forum

Dear all,

After running some mixed logit models, I would like to conduct a root likelihood (RLH) test to check for the consistency of the responses. This is, following:
-Jonker et al (2022) The Sensitivity and Specificity of Repeated and Dominant Choice Tasks in Discrete Choice Experiments. Value in Health.
-Gregor et al (2018) Inflammatory bowel disease patients prioritize mucosal healing, symptom control, and pain when choosing therapies: results of a prospective cross-sectional willingness-to-pay study. Patient Preference and Adherence

To run this test, I need the likelihood of all respondents and then calculate the individual-level root likelihood value (i.e. 𝑅𝐿𝐻𝑖 = (𝐿𝑖 )^ 1/𝑇 where T indicates the number of choice tasks).

Any idea how can I do this after using Apollo?

Many thanks!

Best wishes,

Pamela.

Pamela

so essentially you want the likelihood at the person level, and take that to the power of 1/T where this is the number of tasks per person?

Try this after estimation:

L=apollo_probabilities(model$estimate, apollo_inputs, functionality="estimate")

and then just take the appropriate 1/T exponent

Stephane

Many thanks, Stephane!

Best wishes,

Pamela.