You know those web-based knowledge surveys, such as those done by Pew?
I never gave this much thought until I read a paper examining how often people "Google" the correct answers on such tests before filling in their response. The paper itself is behind a paywall, but I can see it from my university IP address. Keep in mind this is self-reported cheating, which is important. Still, some interesting results pop up and there is the strong likelihood that cheating leads us to underestimate the positive effect of education on what people know, at least as measured by such popular surveys. One way to figure out if people cheated? How long they took to answer the questions. Such data is available, but it's probably rarely analyzed.