Have an idea?

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

Price and feature correlation

Hi! We are having a problem with a conjoint study design. We have 3 factors: Package type, Control and Information Sharing.

Price depends on Package type  "costs" and information sharing "discounts". Also, there is another variable "Estimated Savings", which depends on the control.

I am aware of the "Continuous/Summed Price" approach. However, I am bit confused about how we can conduct this study with "Price" and "Estimated Savings", which are correlated with product attributes. We just have less than hundred respondents, and thinking about conducting a simple CVA type study.  

We thought about using "Price" and "Estimated Savings" in the survey, but excluding them in the analysis. Is it theoretically valid to include some variables in the stimulants and to exclude them in the analysis? Is there a way to handle this problem by including "Price" in the analysis? Any suggestions will be greatly appreciated. Thanks!
asked Jul 15, 2014 by Chathoo (120 points)
retagged Jul 15, 2014 by Walter Williams

1 Answer

0 votes
It's usually a bad thing to show information in the conjoint questions that could affect ratings or choices and then to ignore it during utility estimation.  It leads to unexplained variance that could have been explained.

Regarding dealing with conjoint attributes where the levels are very strong drivers of price differences (package sizes, package types, etc.), ACBC offers "Summed Pricing", CBC offers "Conditional Pricing", and CVA offers...well, nothing.

But, you can "trick" CVA into doing Conditional Pricing just like CBC does...if you format/program the questions yourself to account for the price lookup table...then keep track of that lookup table for the purposes of conducting market simulations.  This can get a bit tricky for new conjoint analysts, so you might want to hire a consultant with experience to assist you.

Additionally, you could use a CVA design and implement summed pricing, then code the design matrix for the continuous price variable, estimate utilities using HB-Reg, and build your own simulator in Excel.  This is a bit tricky and again an experienced analyst could be retained to help you through the details or do the work for you.
answered Jul 15, 2014 by Bryan Orme Platinum Sawtooth Software, Inc. (174,440 points)