Have an idea?

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

T-Test WTP values

Dear all,

I calculated WTP values based on my Zero Centered Utilities (from CBC HB) like described in the related post. I understand that it is better to report the median of the individual WTP values than the mean, since otherwise the WTP value would be distorted by outliers.

I wonder whether I could do frequentist T-Tests in order to identify WTP differences across certain groups (e.g. Demographics). The problem I imagine is that when I do t-tests means are compared and since the means are affected by outliers it is not a good idea. Do you have a suggestion in this regard?

Thank you and best regards
related to an answer for: Calculating WTP from utilities
asked Sep 14, 2020 by Ida27 (295 points)

1 Answer

0 votes
If I were calculating WTP this way, and as you note in the earlier post, doing so is not ideal, I would use a t-test if I were comparing means, but a medians test if I were comparing medians.  

Wikipedia has a nice introduction to the Median test, which is just a clever use of a chi-squared test, here:  en.wikipedia.org/wiki/Median_test
answered Sep 14, 2020 by Keith Chrzan Platinum Sawtooth Software, Inc. (111,275 points)
Thank you for your quick response!
reason for extreme outliers in WTP