ApolloChoiceModelling forum

Dear Concern,

With reference to the HCM model with the OL example, I have a question regarding the interpretation of the dummy coded variable.

For example:

in the example file, it is mentioned that,

regular_user dummy variable for regular users > 1 for regular users, 0 otherwise
university_educated dummy variable for university-educated > 1 for university educated, 0 otherwise
over_50 dummy variable for age over 50 years > 1 for age over 50 years, 0 otherwise

While obtaining the estimates, we got :

lambda 0.640796
gamma_reg_user -0.959511
gamma_university -0.598421
gamma_age_50 0.480009

1. Does it interpret that, regular users are more negatively perceiving the sensitivity to the (non-branded) vaccine, while others positively perceive it? and also the interpretation for the other 2 estimates?

2. Does the model consider 0 as the base level while choosing the dummy variable column?

And why does the lambda value = 1 in the apollo beta function considered, and then we are estimating again? What does the interpretation of the lambda (parameter for the structural model of the latent variable) tell us?

Hi

most of this would be answered by looking at the manual and the paper, as well as standard literature.

1. Does it interpret that, regular users are more negatively perceiving the sensitivity to the (non-branded) vaccine, while others positively perceive it? and also the interpretation for the other 2 estimates?

No, these are impacts on the latent variable, not on the individual parameters

2. Does the model consider 0 as the base level while choosing the dummy variable column?

You decide what is the base, in this case male for gender foe example.

And why does the lambda value = 1 in the apollo beta function considered, and then we are estimating again? What does the interpretation of the lambda (parameter for the structural model of the latent variable) tell us?

apollo_beta is a vector of starting values for the estimation. the actual values then change during estimation. lamba measures the impact of the latent variable in the utility - again, this is explained in the manual/paper

Stephane