(cross-posted from my OkCupid journal)
I like the interpretation better here than their usual fare–fewer assumptions and more raw data. If I have one quibble, it would be that I think their interpretation of what is a private question is flawed because it checks for privacy on a per-question rather than a per-answer basis. By way of example:
Do you have rape fantasies?
I would guess (and guessing is all that I can do, as I don’t have access to the data, oh, if only I had access to the data, what fun we would have, the data and I, but in any event, the preceding phrase to this overly lengthy parenthetical in which you’ve forgotten what I was saying was “I would guess”) that this question is much more private to the people who say yes than the people who say know. And even if I’m wrong about this question, I would guess with a very high confidence level that there are other questions which do cut that way, where people are very open about one answer and very private about another.
I love this idea, though, and wish they would apply it more broadly. It would be fascinating to see which questions correlate to what ostensibly unrelated data. Of course, having too much information about those correlations warps the data; now that this article is out there, people who read it are going to be thinking of the semiotics of their answers in ways that they weren’t before.
So, one thing that I didn’t say in the OkCupid journal, which wouldn’t have been irrelevant there, is that I’d like to see which questions had a high correlation with the “available” status, because one could then possibly use that to cut between very high matches who are interested in non-monogamous relationships despite not having “available” as their status nor having answered the explicitly poly/open match questions.