Have an idea?

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

Alternative specific design vs. two separate conjoint exercises

I am working on a conjoint for a technology product that can be purchased outright (device purchase) or can be accessed remotely (no device purchase).  Other than the device price, which is not applicable for remote users, all other attributes are the same for both types of users.  

I typically use an alternative specific design for studies like this, but my client wants to know if is possible to use a screener question before the conjoint so that device price is only shown to device purchasers and is not shown to remote users.  That is, remote users would never see device price in a choice task and device purchasers would always see device price (instead of presenting a respondent with a mix of purchase and access options within a given choice exercise).

Is this even possible?  The only other option I can think of is to set up two different conjoint exercises in the same survey, but I assume this means I can't combine the estimates of the common attributes.
asked Sep 30, 2021 by dpatek (380 points)

1 Answer

+1 vote
You could certainly use a screener question and then run them as two separate models IF you believed that there is a 100% probability that respondents are all one or all the other. If there's switching behavior between the two, then you'd want to run them together.

As for the analysis, you are correct that setting them up as two models would not make it easy to combine the results, but it could be done if you exported the data, carefully combined them together, and then ran the full data set through the standalone HB.
answered Oct 12, 2021 by Aaron Hill Gold Sawtooth Software, Inc. (12,245 points)
Thank you, Aaron.  If I combine the data from the two versions on the backend, how would I treat price (which is only relevant for the purchase version)?   Would I create a price column filled with zeros in the design?