Jeffreygallup wrote:Is anyone interested in the demographics of Jeopardy! online test passers and contestants? I'm interested in testing various hypotheses: that the selection process for in-person auditions is in fact a random selection, not based on how high the score is once the cutoff is met, or other demographic criteria such as age, location, or gender; that the cutoff is in fact 35; and that the actual contestants fairly approximate (or not) the demographic characteristics of the pool of persons who have passed the online test.
The hard question is whether there is anything which distinguishes, demographically speaking or otherwise, test passers from actual contestants. (personality! stage presence! humor! physical attractiveness!) Perhaps the answer is "nothing in particular," although I've seen a lot of speculation on what it takes to "pass" the in-person audition. If people would like to post their demographic or other characteristics in this year's audition thread, that would allow us to see possible patterns.
I very much respect the desire to settle questions that you and likely many others have wondered, but I don't know how effective a method this will be for doing that. I do know someone who scored below 30 on a test and who received an in-person audition, so either the producers made an error or the cutoff is not always 35; if this has happened to anybody else, I'd be very interested to hear it.
I *do* think one might be able to answer these other questions more empirically, by looking through j-archive and seeing the percentage that various demographics win. For example, I would imagine that men and women who appear on Jeopardy! have an equally likely probability of winning. If it turns out that one gender seems to have a significantly higher win percentage than the other, that would be a sign that there's some demographic selection bias. (This assumes, of course, that scores on the test correlate with performance on the show, and that there's no average difference in buzzer reaction times between men and women; I don't know for sure, of course, but I'd certainly imagine this is the case). You could do the same with other demographics, though I don't think it would work well with age since I do think that's probably inversely correlated with buzzer skill.
Has anybody done this? Would this even answer the question?