Nate Silver pretends to forget how polling works

The last time we checked in on Nate Silver, the top-flight baseball analyst turned bottom-feeding partisan shill (appropriate for a guy who started out in politics as a Daily Kos diarist) was launching a crusade against Strategic Vision so lacking in integrity or even basic mathematical sense that it left many of us wondering whose payroll he’s on.

The sad part is, though, that his analysis is so bad, it would honestly surprise me if anyone were actually paying for this. Take this attempted broadside from Sunday. It’s full of so much bad math and so little critical thinking that I lack the time tonight to address it all. Here are the highlights, though.

The first objective claim he makes about the Strategic Vision poll in question, after his rambling anecdotal sideshow, is that the results are underdispersed. This claim is entirely unsupported in the most literal sense, in that he neither demonstrates what kind of distribution the data should have followed, nor does he show that the actual variance of the data contradicts that predicted distribution. The technical term for this is ‘hand waving,’ however when our professor waves his hands we at least can check the textbook for confirmation. Silver’s just making this up as he goes along, though.

From there we get some more anecdotal rambling, in which a Democrat politican’s words are recorded with the same kind of blind, unquestioning support that a Hitler Youth would have recorded Der Führer’s own speeches. After all that, we get what is supposed to be a smoking gun: A different poll with different results.

However it’s not surprising that Cannaday’s poll has different results. It is conducted with a different pool of students (high school seniors in his district, not students from all high school grades all across the state). The samples were not random (special education students were picked out, according to Silver). The survey environment was different (students were questioned in a school environment with authority figures present, rather than asked at home by strangers over a telephone).

No amount of special pleading can make the two surveys comparable, especially given the menacing glares of teachers ensuring the students try on the tests, and the teachers themselves under political pressure from a state officeholder.

And again, Nate Silver knows this. Different methodologies testing different pools, with samples drawn using different methods, will produce different results. He chooses to disregard this in order to shill for his Democrat superiors.

I sure hope he’s getting paid, because his integrity was surely worth at least a combo meal at Carl’s Jr, with large fries.