Have an idea?

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

Calculation of Utility of a specific combination of service attributes

Hello everyone,

I am kinda confused with the calculation of Utility of a particular set of attributes and its levels combinations for a particular respondent.

I am using CBC-HB estimates with "None" utility being estimated.

As per the Utility definition, it is a sum of part-worth utilities and a constant.

U (total utility for a respondent for a specific attribute combination) = Constant + Sum ( part-worth utility* attribute (nominal = 1 or 0)

From CBC-HB estimates, I have all the part-worth utilities of the attributes and its levels. I am really struggling with the estimation of the constant value for this combination.

Would anyone be able to help me with this?

asked Apr 13, 2018 by sacharya (490 points)

1 Answer

0 votes
Best answer
Sujit,  In most conjoint studies there is no constant.  Occasionally there will be an alternative-specific constant, but unless you have built a particular kind of alternative-specific design to include that (and even then only if you don't have the "alternative" as an attribute) it's not something you need to worry about.
answered Apr 13, 2018 by Keith Chrzan Platinum Sawtooth Software, Inc. (66,225 points)
selected Apr 13, 2018 by sacharya
Thank you so much for this answer, Keith.

My design is Alternative Specific CBC because I had prohibitions integrated to certain primary attribute. But I have balanced overlap and dual response none configured as well.
Is this completely wrong?
No, not wrong.  But sometimes folks make what are called "labeled" designs and the labels need to be coded into the design as alternative specific constants.  But that's an uncommon case for most of our users.
Seems like I just entered into this designing world.. So much to learn :)
Thank you again for this confirmation. So, I will go ahead without constants and analyze accordingly.