According to Oklahoma Rep. Ed Cannaday, he administered the exam to every high school in his rural eastern Oklahoma district, and the results were much, much better than the ones Strategic Vision reported. The number of students who could identify America's first president is 98%, not 23%. Only two questions - the number of judges on the Supreme Court, and the length of senators' terms - got correct responses from under 70% of students.
I can't remember if I posted on this or not, but this is the kind of story that gets around the blogosphere and can really do some damage. People read this and it fits into a narrative of how kids are getting dumber and less politically involved, that our country's going to shit, etc etc. The truth is that kids aren't stupid, that for the most part they are informed and capable citizens.
What confuses me is why Strategic Vision would make up those numbers. The poll, it seems, was commissioned by the Oklahoma Council of Public Affairs, who Silver describes as a "conservative-leaning think tank." But it seems to me that liberals have more of an interest in such numbers than conservatives, since such a poll could easily be used to justify increased government spending on education. Why would conservatives want a poll that shows that high school seniors are leaving schools without any knowledge? How do the results of this poll help conservatives? Silver doesn't explain this - I wish he would, because I'm at a loss.
Update: A commenter on the FiveThirtyEight thread, who works as the communications director for the Oklahoma Democratic Party, explains:
There has been an ongoing Republican-led legislative fight to dismantle public schools and essentially create a charter school system instead. Pushing the notion that public schools are failing would help their misguided argument. I still don't doubt that certain GOP legislators will quote the SV results on the floor next session, sadly.That's certainly plausible - these results would seem to suggest that the schools are failing. But the results could easily be interpreted as an impetus to support increased funding for public schools. So these data, in and of themselves, aren't uniquely useful to conservatives if that's the case.
Note that I doubt OCPA is responsible for cooking the data. SV probably wanted numbers that would play well so OCPA would come back to them for other results. Note also that Cannaday is a Democrat - if the bizarre nature of OK politics means that conservatives want bad numbers while liberals want good ones, Cannaday's survey numbers might be a bit on the high side. Wouldn't explain the entire discrepancy, of course, but Cannaday's numbers might be a little high.