It's this study.
Researchers raise concerns of fraud and ambiguity in two studies authored by Dan Ariely, renowned Duke researcher and professor
Fuckin' first Jonah Lehrer, then Dan Ariely. It's almost as if the entire field of behavioral economics is fraudulent.
On a related note, I finished this book today. The author quotes Reinhart and Rogoff extensively, only once in passing mentioning that "there were some repeatability errors in some of the research" rather than noting that the entire thesis of the book turned out to be a fat-finger Excel error.