Enhance Conjoint with a Behavioral Framework - Best Paper Sawtooth Software Conference 2021 - Peter Kurz and Stefan Binner (bms Marketing Research & Strategy)

Last updated: 03 May 2021
Enhance Conjoint with a Behavioral Framework

Enhance Conjoint

Our hearty congratulations go to Peter Kurz and Stefan Binner for their winning presentation at the 2021 Sawtooth Software Conference, “Enhance Conjoint with a Behavioral Framework”.  The award was based on a combination of audience judgement (using a “relevant items MaxDiff survey”) and evaluation by the conference steering committee.

Peter and Stefan’s work stands out due to how practical and easy it is to implement their suggestions by everyday Sawtooth Software users. The essential idea involves adding a nine-row paired-comparison question grid to CBC surveys (which may easily be done using Lighthouse Studio's “semantic differential” question type). The nine binary response variables are then used as covariates in HB estimation (via a simple point-and-click within our HB estimation interface). 

Finding useful covariates (variables outside the conjoint task) that can boost the predictive validity of conjoint analysis via HB estimation is a topic that Peter and Stefan have investigated and reported on multiple times at past Sawtooth Software conferences. In their 2021 presentation (delivered remotely from Germany on April 22 in San Antonio, TX), Peter and Stefan found better success than past investigations by using a simple set of nine questions (binary semantic differentials) that can be added to the survey just prior to the CBC questions. The nine questions are based on principles of behavioral economics and include such pairs as: “I think brands differ a lot” vs. “I think brands are more or less the same” (respondents pick the statement they most agree with). 

Peter and Stefan proposed that these simple pairs statements help respondents remember their prior shopping situations and values and prime them to do a more realistic job in answering CBC questions. They showed that hit-rates and out-of-sample predictions could be significantly improved using this framework.

They presented results for nine different CBC studies, demonstrating good improvement in both in-sample and out-of-sample hit rates when leveraging the nine covariates. What was perhaps even more intriguing is that just merely asking the nine questions prior to the CBC section seemed to improve respondents’ performance on the CBC tasks. Even without using the covariates in HB estimation, the act of completing the nine semantic differential pairs improved the predictability of the respondents' utilities for out-of-sample holdouts. In addition to the usefulness of the nine covariates for improving predictive accuracy of the conjoint results, Peter and Steven mentioned their value in developing useful consumer segmentations.

If you missed the conference and want to take advantage of our delayed virtual attendance option which lets you view the videos and slide decks for the 21 presentations delivered at the 2021 Sawtooth Software Conference, you may purchase a one-seat subscription to the content through our Virtual Learning Academy portal at https://academy.sawtoothsoftware.com. The presentations for last year’s Sawtooth Software European conference are also available to subscribe to at the Virtual Learning Academy, as well as other content such as the 3-day CBC/MaxDiff/ACBC training workshop.