choice-metrics.com

Dear Michiel,

I am trying to generate a partial choice set design using your series of spreadsheets that create candidate sets for a partial choice set design.

In your spreadshits you mention that unavailable attribute levels should be defined in case of partial choice sets and at least one attribute must be numerical (e.g., price/cost). However, all the attributes are dummy coded in my project. I would like to know if there is any way I can use the spreadsheets? Or do you have any other suggestion to fix this issue?

Thank you very much in advance.

The external candidate set method cannot be used if all your attributes are categorical. The reason for this is that this method uses a trick where the utility of certain alternatives becomes something like -99 such that its choice probability becomes 0. This only works if it is possible to add a level like 999 or -999 to a numerical attribute. Such a level cannot be added to a dummy coded attribute.

It may still be possible to generate a partial choice set design in Ngene using another method that is not explained in the manual. This method requires rewriting your labelled alternatives to unlabelled alternatives using indicators. This is a bit tricky (hence not explained in the manual) and is yet unpublished. I have a conference presentation that explains this method, I am happy to send the slides of this presentation to you, please email me on michiel.bliemer@sydney.edu.au. If your experiment is not too complicated then it may work.

Michiel