Have an idea?

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

What is fewest number of design versions for CBC using HB Estimation?

I am conducting a CBC study with complex prohibitions and am having difficulty obtaining a usable design in Lighthouse Studio. The designs I've been generating have multiple versions (e.g., 300+).

To try and obtain a usable design, I was planning to :
1. Create a design in LightHouse Studio with only 1 version that has some of the required prohibitions (but not all) such that it is of acceptable quality according to the Advanced Design Test (e.g., standard errors for main effects no larger than about 0.05).
2. Export this 1 version design and manually add additional prohibitions.
3. Import the modified design back into LightHouse Studio and use the Advanced Design Test to determine if it is of acceptable quality.
4. I would repeat #2 and #3 adding additional prohibitions as possible to obtain a final 1 version CBC design that accommodates as many of the required prohibitions as possible.

My question is: Is one version of the design acceptable for a CBC study where I will need to run HB analysis to obtain respondent-level utilities OR do I need at least 2 versions or at least 3 versions, etc.? One version would be strongly preferred for simplicity if workable for HB analysis.

Thank you.
asked Mar 22, 2019 by anonymous

1 Answer

0 votes
It should be workable for HB, yes.  As long as your design diagnostics looks good you should be OK.  For one reason for another I've done a number of one-version studies and analyzed them with HB.
answered Mar 22, 2019 by Keith Chrzan Platinum Sawtooth Software, Inc. (92,075 points)
Thank you for the information.