Have an idea?

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

reason for extreme outliers in WTP

Hi all,

I calculated the WTP according to the simplified method described in one of your papers. In the related post you wrote that we should rather report the median instead of the mean which makes sense due to outliers. I have extreme outliers in terms of WTP and I wanted to ask whether you could explain the reason behind this? When I see extreme cases in terms of WTP is this evidence that the respondent is bad? Does this hurt my CBC results? Or can I ignore it and just report the median WTP and it is fine?

 Thanks a lot and best regards
related to an answer for: T-Test WTP values
asked Sep 23, 2020 by anonymous

1 Answer

0 votes
We prefer not to use the simplified methods, but to use simulation-based methods for estimating WTP.

That said, extreme outliers are known to occur in respondent-level data.  Any respondent who has very small differences in their price utilities will have very large WTPs.  For respondents with utilities of 0 for each price point, WTP will be so large as to be undefined.  It's simply the mathematics of dividing by a number that can sometimes be very, even vanishingly, small.
answered Sep 23, 2020 by Keith Chrzan Platinum Sawtooth Software, Inc. (102,700 points)
Thank you for explaining! I understand it now.