JMP Choice Model- Subject Data & Effects

Question

Dear All,

I have contacted a Discrete choice experiment designed using JMP, and I am now analysing the results. Everything is clear and straightforward if I analyse the attributes only.

However, I don't have a clear idea of what can you do if you use subject data in the model. The interaction effect that I see in the model between the variable at subject level added (e.g. education) and the attribute how can be interpreted? I can't find anywhere in JMP documentation an explanation of what you can do adding subject level variables in your choice model.

Thank you!

gav888 · Answer

Thank you for your reply.

My focus is on the relative importance of attributes. Adding two covariates at the level of the subject, I have got that some attributes become insignificant, while the interaction between the covariates and the attributes are. This is the table  from my JMP report

See attachment

How do I interpret that? Is the attribute used only by people with some scores of the covariates?

Thank you

statman · Answer

Welcome to the community.  I'm not sure I can answer your question with the limited information about the experiment you conducted, but here are my general thoughts:

When analyzing relationships between characteristics of the subjects (e.g., age, sex, education level, etc.) and the resultant choices, you are looking for:

1. Is there a correlation between some characteristic and the choices selected?

2. Is the relationship between the characteristics and the choices consistent?

3. Would the subjects make the same choices at another point in time (or under. another set of circumstances)?

4. For interactions, if the interactions are significant, then the choices depend on the subject characteristics.  For example, if the choice was about what show to watch on TV, that selection may well depend on the age (younger ones might choose Paw Patrol and older adults might choose news) or sex (males might be biased to choose sports or action) of the subject.