Have an idea?

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

Constant which is part of the utility function..

Thanks again, Bryan.  I am sorry to keep on asking the same questions but just out of curiosity, in the webinar named "Intro to Choice-Based Conjoint with Lighthouse Studio: Part 1 Webinar", Brian McEwan and Megan Peitz mention that we don't have to put all price points in the attribute levels but we can extrapolate it later. So, my question was in the same line of thought.
I do understand what you mention about "not extrapolating" but isn't it possible that only after a specific price point, the respondent(s) will be willing to pay for a specific combination of service attributes based on the utilities captured even if it was not shown to them before. So, how would you advise to handle this then?

Secondly, for the constant part of the utility function, for example, I am referring to a specific research paper titled "Measurement of consumer preferences for bucket pricing plans with different service attributes" by Christian Schlereth and Bernd Skiera.
His paper and others in pricing research papers always mention that the utility function as a non-linear function which is described as below:

Utility (service in question, composite good) = a* (number of units of service in question) - (b/2 * (number of units of service in question)^2) + c + (price of composite good * units of composite good)

Here "c" refers to the usage independent utility (for zero quantity).

Also "c" can be considered as the constant  of the regression equation.

I hope I could explain myself a bit. Please let me know if you need more information from my side. Looking forward to your advise.
asked Apr 11, 2018 by sacharya (490 points)

1 Answer

0 votes
I'm pretty sure that in the training class, Megan and Brian referred to Interpolating prices within the range that was measured rather than Extrapolating outside of the range (which in our training slides is mentioned as dangerous and not recommended).

Our simulator uses linear interpolation between the closest two measured points (when you use part-worth discrete part-worth utility estimation).

Regarding the Schlereth and Skiera paper, I'm not familiar with it or that concept of that constant.  Perhaps somebody else listening in with knowledge can add something here on the forum to help you with that question.
answered Apr 11, 2018 by Bryan Orme Platinum Sawtooth Software, Inc. (152,955 points)
edited Apr 11, 2018 by Bryan Orme
Thanks a lot for pointing this out.
I will cross check the webinar and the slides even if i don't doubt your explanations.

For the research paper I mentioned, you may ignore this.
It was just like a constant in a regression equation -

U = constant + u1*x+u2*y+.....; where u1, u2 are part-worth utilities of x, y respectively.
If this makes sense, may be you can advise ? Or you may chose to ignore this question.

Thanks a lot again for all your help.

kind regards,