I doubt that many people are cheating, but the "percentiles" are basically meaningless because the survey obviously self-selects for 1) people who are on the In

I doubt that many people are cheating, but the "percentiles" are basically meaningless because the survey obviously self-selects for 1) people who are on the Internet, 2) people who are on Internet sites where this survey will get reposted 2) people who suspect they have pretty good vocabularies (who is going to take a non-mandatory 'test' that doesn't even earn them any Facebook Credits to inform them they aren't so bright), etc. End result: substantial inflation of scores relative to what you'd see if you gave the same survey to the general population (and assuming everyone is relatively honest).