Discrete Choice Experiment: None of these options

[#1]

Dear community, 

Just a quick question considering the formatting of the "No choice" option in JMP 17.

If we have Option A, Option B, and Option C, where Option C = none of these options. Do we give three rows of data to JMP? Or would we give two rows of data in binary format, and when both option A = 0 and Option B =0, it will use this information that no option was selected? 

In the images below see in particular choice set 5 and choice set 7, where the respondent has selected the "No Choice" option. So, which of these coding formats is correct? 

Kind regards, 

Declan 

 

P.S. I have tried to find an existing solution in this community, but the one I've found was a work around for JMP 13. Alternatively the Design of Experiments Guide (V12) does not address the "No Choice" option as far as I could find. 

[#2]

Hi @Declan_JMP,

I might have not a direct answer to your question, but I have found some litterature describing the "Do not know" or "Status Quo" option in Discrete Choice Designs.
Here are two links describing this situation :
https://academic.oup.com/heapol/article/24/2/151/591265
https://www.emerald.com/insight/content/doi/10.1108/FER-04-2020-0005/full/html

In these two references, the possibility to add a "Status Quo" option is mentioned (the second link provides more info on how to deal with this third option in the analysis). This enables to reject both alternatives and looks like a real world scenario, even if it may reduce information on how participants trade-off some attributes levels in the alternatives.

There might be no definitive answers to your questions, it's more a question of what you expect and want to discover/analyze from the data.
Should the individuals always choose for the better option (even if there seem to have none for them, which means you treat the results as 0-0 for both options if they can't choose), or do you offer them a third option (that might be chosen often by default, and risks to "hide" partially some positive/negative attribute levels effects) ?

Hope these ressources will help you,

[#3]

Hi @Victor_G

Thanks for the fast reply, much appreciated. I think as you mention: "There might be no definitive answers to your questions, it's more a question of what you expect and want to discover/analyze from the data. Should the individuals always choose for the better option (even if there seem to have none for them, which means you treat the results as 0-0 for both options if they can't choose), or do you offer them a third option (that might be chosen often by default, and risks to "hide" partially some positive/negative attribute levels effects)?", is really a key point when choosing one over the other method for analysis. Also, especially Koemle & Yu (2020), is useful for understanding this. Thanks for pointing me in the right direction!