Have an idea?

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

Comparability of zero-centered diffs of different CAs

I've run a TCA (CVA), ACA and CBC Analysis with the same attributes and levels in order to show the design differences and the differences in performance.
Now I want to compare the part-worths of the three studies. The part-worths are zero-centered diffs.
Following all the interpretation guidelines, we can just compare differences of part-worths with each other (interval data).
Zero-centered-diffs are scaled and normalized. So in my opinion it is possible to compare part-worths differences from study A with part-worth differences from study B. Am I right?

Thanks for your help.
asked Oct 21, 2014 by Michael

1 Answer

0 votes
Sounds like an interesting study.  Yes, you can compare the zero-centered diffs (or the importance scores) between conjoint studies (that share the same attributes and levels).

You'll probably find that ACA tends to "flatten" the differences in importance across attributes (most important attributes will be a little bit less important; less important attributes will be a little bit more important).

CBC and CVA should be more similar.
answered Oct 21, 2014 by Bryan Orme Platinum Sawtooth Software, Inc. (152,255 points)