A tidbit from one of the blind taste tests:
Levitt used four decanters. Into the first decanter, he poured one of the expensive bottles of wine. The other expensive wine went into decanter number two. In the third decanter, he poured the cheap wine, which cost around $8. In the fourth decanter, he repeated one of the expensive wines...(If you don't want to listen to the MP3 file, you can click on the button at the bottom of the page marked "Audio Transcript".)
LEVITT: The data could not have cooperated more completely with my hypothesis. So for starters the four wines received almost identical ratings on average. Although there were a wide spread among individuals, on average, tallied up, people did not prefer the expensive wines to the cheap wine.
On top of that, and this was the thing that I was hoping for and dreaming of but didn’t believe would actually come true: It turned out among individuals if you compared how differently they rated any two of the wines that they had, it turned out that by a small margin, people actually rated the same wine from the same bottle but presented in a different decanter as being the most different among the two wines.
So the two wines that were absolutely identical, when you looked at the gap between the ratings that an individual gave to those wines, the gap was bigger than they did between the other wines, which actually were different.