Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

Adding two 'none' options to my CBC

I am designing a discrete choice experiment where participants allocate 10 tokens to three concepts and two different opt-out options. (In this case, "don't participate in the activity" and "participate in the activity but somewhere else." I only see an option to have one none option or the dual response, which doesn't fit my needs here.

Thank you for any help!
asked Feb 23 by ashtrudeau (125 points)

1 Answer

+1 vote
Perhaps others have ideas for this, but the solution I'm aware of is to treat your CBC study as an alternative-specific design (described in the documentation).  Typically, alt-spec designs are used in cases like trading off bus, car, train, walk, and "other".   These five alternatives are specified as 5 levels of the first (primary) attribute.  Then, bus attributes are associated with the "bus" level of the primary attribute.  Car and train attributes are associated with their levels of the primary attribute.  The Walk and "Other" categories might not have any attributes associated with them, so they look like two fixed "None" alternatives on the screen that respondents can choose among.
answered Feb 23 by Bryan Orme Platinum Sawtooth Software, Inc. (198,715 points)
Hi Nico,
Looping back to this question, do you have an idea of how I could add the two opt-out options within the 10 token allocation format? (rather than just a select question)
Can you please be a little bit more specific? Do you want two opt-out options with a 10 point Likert scale each?
Sorry, I mean that the respondents are asked to allocate 10 days between 3 alternatives and two opt-out choices. I would like to add the second opt out choice so that respondents can allocate a set number of days between those 5 options that must add up to 10.
Sorry, this goes beyond what I have done so far with regard to choice modeling. However, I am sure that the guys from Sawtooth Software are able to help you on how to implement such an approach in Lighthouse Studio. With regard to modeling, maybe the Multiple Discrete Continuous Extreme Value (MDCEV) model is worth taking a look at (e.g., http://www.apollochoicemodelling.com/files/manual/Apollo.pdf, pp. 64-71).
Thank you!