Have an idea?

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

Multidimensional attributes in TCA, ACA and CBC

I’m conducting a comparison of conjoint methods for my master thesis at the moment (with Sawtooth Software). I want to analyze the outcomes of different CA Methods for peoples' willingness-to-pay for smartphones.

Smartphones are complex products and you can easily find 50 attributes which may influence the buying decision of people or their willingness to pay. That's why I want to find attributes which reflect those different needs in very general words.

One example is the attribute "lifestyle": Some people buy their smartphones because they like the brand or the image which goes along with the product. Further I would summarize the appreciation of friends/family for the holding of the product under "lifestyle".

So all in all the word lifestyle summarizes "brand", "image of the product" and "appreciation for the product".

In literature I don't really find any suggestions whether it's possible or not to have multidimensional attributes, explaining the different dimensions in each level. Of course I will get results just for the summarized attribute and not for each sub-attribute/dimension.

Do you see any issues or problems with "multidimensional" attributes especially in the context of different CA methods?

Thanks a lot for your help!


asked Aug 18, 2014 by Michael

1 Answer

0 votes
You'll need to worry about the issue of independence of attributes used in conjoint analysis.  Most conjoint analysis approaches assume an additive model where each attribute is entirely independent of the others.  Also, most conjoint analysis uses "hard" attributes such as specific feature levels or prices.  As soon as you start to use "soft" attributes such as brands with images, perceptions, etc. then you may particularly run into independence of attributes issues.  When attributes that have overlap in meaning are both included in a conjoint experiment, then double-counting of the importance of that attribute can begin to occur.
answered Aug 18, 2014 by Bryan Orme Platinum Sawtooth Software, Inc. (171,240 points)
Hi Bryan,
thank you for your answer.
Is it possible, that you mean "then you may particulary run into DEPENDENCE of attribute issues"?