Let's be fair: that study was measuring the fraction of people that say they'd take an imaginary $500 over an imaginary 15% chance at an imaginary $1 million.

I doubt that most respondents were deliberately messing with the survey results, but I do think that people may use different decision-making resources for amusing hypotheticals vs. for the real world. E.g. the percentage of people getting the Wason Selection Task correct can jump from under 10% to over 70% when you change the task context from more abstract to more concrete. I suspect that for lots of people imaginary money counts as too abstract.