We would like to see some "real-world" validations of conjoint analysis at the next Sawtooth Software conference. We are particularly interested in validation of choice simulators versus real world decisions or purchases. Here are two examples of the type of project we are looking for:
- Scanner data for a packaged goods category from a store (or stores) on a given product category could provide validation scenarios for testing predictive validity of CBC survey data collected on the same category in the same geographic region.
- A fast-food restaurant could use CBC to price the meals on its menu. Actual prices could be varied between stores for validation scenarios. CBC interviews could be conducted among customers. Cost information would be incorporated into choice simulations for investigating pricing policies to improve profitability. Ideally, some action would be taken based on conjoint results, and changes in actual profitability would be reported.
With data sets like these, some of the different utility estimation and simulation procedures that were spoken of at this most recent Sawtooth Software conference could be put to the test. Please contact us if you have access to such data, or would like to undertake this sort of project for possible presentation at our next research conference.
In addition to real-world validation studies, we'd like to see practitioner case studies which focus on Disk-By-Mail, telephone, foreign language, Internet and multi-media interviewing. Split-sample studies which compare data collection modes with respect to data quality and cost are always welcomed, as they further our collective knowledge about best practices in our field.