These are old questions, and Daniel Kahneman has previously shown that these aren’t a test of your analytical skill. Your performance on these tests can be “primed”. A simple analytical “priming” can vastly improve your performance.
Knowing that performance can be “primed”, I would be very curious if they properly controlled the experiment to verify that no additional interference caused people to perform differently. In fact, I would be willing to bet that they allowed something to interfere BEFORE the questions, not after the questions. I would be willing to be that the experiment is not repeatable.
I submit one of my favorite anecdotes:
