Have an idea?

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

Analyze CVA-like exercise in CBC-HB

I have a full profile conjoint with a single-concept per task with 4 radio buttons (where respondents have to choose one answer indicating their recommendation). I can easily generate the actual design using CBC (as I don't have CVA license) but am not sure how to interpret the data collected (e.g. answers 1/2/3 or 4 for each task) using CBC-HB and report distinct 4 choice preference shares.
Should I just treat these 4 answer options as 4 different conjoints (Answer 1 vs None; Answer 2 vs None, etc.) and calculate 4 sets of utilities?
asked Nov 29, 2017 by kasharaz (140 points)

1 Answer

0 votes
This is a slightly different model than the usual conditional logit model we use in choice-based conjoint analysis.  Technically it's an unconditional logit model and both the analysis and interpretation are different.  We call this kind of experiment a "Situational Choice Experiment" to distinguish it from a more common sort of CBC.

Our MBC software is made to handle this kind of experiment, but you can also trick CBC/HB (lr Latent Class) to run the analysis for you, too.  Send me an email at keith@sawtoothsoftware.com and I can provide you with the details you'll need.
answered Nov 30, 2017 by Keith Chrzan Platinum Sawtooth Software, Inc. (50,900 points)
By the way, we devote a whole chapter in our new book (http://www.sawtoothsoftware.com/81-products/conjoint-analysis-software/1852-new-advanced-conjoint-analysis-book) to Situational Choice Experiments.  And if you're interested, we'll be doing a one hour after-hours session on this topic at our conference in March:  http://www.sawtoothsoftware.com/events-hidden/1839-2018-conference