Have an idea?

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

Volumetric conjoint Unweighted Shares Interpretation

Can the unweighted share calculated from the utilities of a volumetric conjoint be considered equivalent to preference shares from a DCM?   

I understand that we weight respondents by the max volume on any screen to get to the volumetric distribution SKUs tested, and create none based on remainder of the max volume not allocated to the SKUs.

I want to understand if the unweighted share have any kind of interpretation / meaning, and if it can be treated same as preference share.

Do volumetric models tend to have a higher none % relative to DCM?  My gut feeling is they do, because on each screen unless someone is allocating their max volume every time, none always gets a piece of the share, inflating it's utility.
asked Jun 19 by rsaini (190 points)

1 Answer

0 votes
The shares you would calculated from a volumetric conjoint would be based on utilities run from volumetric inputs.  So they would not be the same as you would have calculated had you run a simple choice-based DCE.  

I've never run an experiment where I had a standard choice-based DCE and a comparable volumetric DCE run side-by-side.  Part of me wants to say "yes, the none will be higher for volumetric" but I think the more correct response is that the none simply means a different thing in volumetric than in choice-based - so comparing the two nones probably doesn't make a lot of sense.
answered Jun 19 by Keith Chrzan Platinum Sawtooth Software, Inc. (95,875 points)
If you submit chip-allocation data to our HB or latent class estimation programs, if the chips don't sum to 100%, we divide within each task by the sum so that they do.  So, a person who gives an allocation of 10, 10, and 20 units to concepts A, B, and C in a CBC choice task will have those allocations converted to probabilities of 25%, 25%, and 50% prior to utility estimation.