Have an idea?

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

Test attribute importance difference between two attributes within sample

I am preparing a CBC study, where I want to test if the attribute importance (or range) of two attributes differs significantly (e.g., Attribute 1: 8.xx%; Attribute 2: 9.xx%) within one sample.

Is there a usual way to conduct this test (e.g., a t-test to complare the mean range/importance, or HB procedure)? I was wondering if this works similar to testing significant differences between two attribute levels within one attribute?

Furthermore, as I could not find a description of this test in the book or forum, I am wondering if this is a valid approach.

Thanks in advance!
asked Jul 10, 2019 by rrussell (125 points)

1 Answer

+1 vote
Yes, you can use either a between-groups t-test or a Bayesian test, as described here:  https://www.sawtoothsoftware.com/index.php?option=com_content&view=article&id=2033

This is a chapter from our Becoming and Expert in Conjoint Analysis book.
answered Jul 10, 2019 by Keith Chrzan Platinum Sawtooth Software, Inc. (93,025 points)
Thank you Keith. I was wondering if a between-groups t-test works here, because we test within one group.

Is it not common to test if two attribute weights (e.g., brand and price) within one sample differ significantly?

In any case, thank you for the quick response.
I'm sorry, yes I meant to write a single group t-test, just like you would for testing an attribute level.
Okay, great. Thank you!