It's a problem that is general to social sciences - hard to get large samples, so experiments are very difficult to reproduce.A lot of Kahneman's research including many things in the Thinking Fast and Slow book doesn't replicate but to his credit he was very open about the problem.
Many of their hypotheses have been confirmed independently via different experimental design. Relevant to us, there were studies involving market participants (non-survey based) regarding loss aversion and framing.



