Have an idea?

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

Price shock for an ACBC with 1 priced attribute

If I have an ACBC with a summed price, and only one attribute has a price set, do the prices/shock need to be set so that all levels could potentially overlap?

Say this attribute has 3 levels with prices of $100, $150, and $200, no base price, and a 30% shock. We expect most people would prefer the higher prices levels all else being equal, but this is by no means an given and some people might prefer the first level even without considering price. Would this design give us enough information to estimate the utilities for this attribute independent of price?
asked Feb 12, 2014 by Nolan Kaplan Bronze (4,090 points)

1 Answer

0 votes
Our experience (based on simulated data) with Summed Pricing is that the +/-30% shock will indeed give you enough information in the design matrix to estimate the price slope (sensitivity) independent of the attribute you are attaching level-based prices to.  In the BYO question, respondents will see the $100, $150, etc. prices you specified on each level.   But, in the rest of the survey, random shocks will occur.  

Some researchers have been willing to do less random shock (say +/- 20%) because they are willing to give up some precision in the utilities because they maybe have large enough sample size to still obtain good enough precision in terms of aggregate measures (shares of preference from simulations, for example).

For more on our simulation results, please see https://www.sawtoothsoftware.com/downloadPDF.php?file=price3ways.pdf
answered Feb 13, 2014 by Bryan Orme Platinum Sawtooth Software, Inc. (163,515 points)