They took 12 employees of an ad agency to an NYC bar. The idea was to start with people who are already pretty funny, and see if alcohol made them funnier. McGraw and Warner instructed them to come up with a gag, then have a drink. After every drink, the subjects ranked how funny they thought they were on a scale from "slightly amusing" to "hilarious." They also had to rate how drunk they believed they were, on a scale from "sober" to "sh*t-faced" (there were seven options to choose from).
So, did drinking actually make these people funnier? You may be shocked to read that no, it did not. The jokes were judged by an independent, sober, online panel — a panel that was not in stitches over the ad agency's punchlines. McGraw's theory about humor is that it happens when something "wrong, unsettling, or threatening" takes place within a safe, nonthreatening context. For example, in the venerable Tommy Boy, when Richard hits Tommy with a two-by-four. It's funny because he emerged with just a bruise ("Not here so much, or here, but riiiiiight here"), even though the act itself was fairly violent. By the end of the experiment, however, the ad employees' jokes consisted of a Venn diagram with "cancer" in one circle and "unpoppable pimple" in the other. Yeah, we don't really get it either.
McGraw sums up the experiment perfectly: "Drinking reduces inhibition, but it opens the door to failure, with failure likely to be on the side of going too far." (The Atlantic)