Have an idea?

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

Analyzing Single Attribute Conjoint

I am testing estimating HB utilities for a single attribute conjoint in CBC/HB.  I've always thought of a single attribute conjoint like a MaxDiff, but in my case it is volumetric, so coded much differently than best/worst.

I have read that analyzing MaxDiff data in CBC/HB requires customizing the prior covariance matrix, but I don't think it's relevant in my case.

Can anyone comment on if there are pitfalls or customizations required for estimating HB utilities for a single attribute conjoint in CBC/HB?

asked Apr 28, 2017 by Joel Anderson Bronze (1,310 points)

1 Answer

+2 votes
Best answer
The prior covariance matrix stuff should be attended to in HB whenever you get attributes that have lots of levels in either effects- or dummy-coding...and when the data are sparse at the individual level (which is usually the case for us).  Luckily, most of the time our CBC/HB software is doing this automatically for you.  But, if you do "user-specified" variables (for example, when you code up a MaxDiff experiment using user-specified variables as our documentation describes), then you indeed have to manage this for yourself.
answered Apr 28, 2017 by Bryan Orme Platinum Sawtooth Software, Inc. (144,240 points)
selected Apr 28, 2017 by Joel Anderson