I think there is an insidious tendency to want to post-hoc rationalize things that are just random noise in the kinds of surveys that Ariely et al were doing. I

I think there is an insidious tendency to want to post-hoc rationalize things that are just random noise in the kinds of surveys that Ariely et al were doing. I can imagine over time that it must be tempting for these kinds of people to just start faking stuff. One of my undergrad psych research projects was focus on testing some of the Kahneman & Tversky anchoring/adjustment stuff. We were able to replicate some things related to the theory itself based around economic estimates, but also had some "control" questions in our surveys (one example: "estimate the average weight of a human brainL). Our experience was that some populations just sucked worse at the kinds of estimation tasks we were using. We had no good theory for gender-based differences in estimating the weight of a brain, and in retrospect it was probably just people confusing ounces with pounds as a half-remembered number for weight... but no one gets published by admitting that their experiments were probably just poorly designed.