Have an idea?

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

How do you test multipart pricing in a conjoint?

We want to do an adaptive choice based conjoint with 10 features with multiple levels. 3 of those features happen to be price. We want to test customers' preference for annual vs. upfront costs and need to display these 3 prices. These prices will have a set range (i.e. $50-$100) and can be incremented based on some of the higher-end feature levels. Are we able to use the software to perform this study?
asked Jun 3, 2016 by John

1 Answer

0 votes
In Adaptive CBC, you have the option to specify price attributes just like any other attribute.  For example, annual cost could have three levels: "$50", "$75", "$100".  So, you could specify three different price attributes.

However, you have an additional complication: you want to change the prices based on other attribute levels in your study.

There are a few things you could try.  First, you could try specifying some prohibitions between the prices and the other non-price attribute levels.  But, you must take care and not prohibit too many combinations.  You can use the "test design" feature within ACBC to test the effect of the prohibitions on the quality of the utility estimates.  But, a warning, is that the test procedure for ACBC requires that you send 100s of "robotic" respondents through the survey before obtaining your results.  The software automates all this and computes standard errors for you, before and after the prohibitions, but this can take 30 to 45 minutes per test.

Another thing to try is to use the alternative-specific design capability within the software.  For example, perhaps when a certain level of a non-price attribute appears, only prices "$75 and $100" can appear.  So, you need to create a new price attribute that only has two levels ($75 and $100) that will be displayed whenever that certain level of the non-price attribute appears.

In all things you try, you must ensure that there is enough independent price variation beyond that which could be predicted by the non-price attribute levels.  Otherwise, the design is deficient and you cannot estimate a separate utility function for price levels independent of the non-price attribute levels.  The "test design" capability of the software will help you assess that.
answered Jun 6, 2016 by Bryan Orme Platinum Sawtooth Software, Inc. (175,715 points)
More information