Wave comes in, wave goes out

The quantum entanglement of the quants over the 2014 midterm elections in America has collapsed, in the form of a wave for the Republicans. Sam Wang and Nate Silver were locked in a polarised battle of spin over which side had momentum. Wang's position was that the polls all showed that Democrats might hold the Senate. Silver pointed to the correlation of the "fundamentals" of midterms which all favoured low turnout which would lead to a particular result favouring the GOP.

Silver tries desperately not to run up the score, Bill Belichick style, in his declaration of victory, but the numbers he quotes (graphed above) do not lie: the poll of polls was off by four points. Only one US federal election in the last twenty years has been skewed to the Democrats at anything like that level.

The post-mortems are coming thick and fast with all sorts of post-justifications. This one from Lawyers, Guns and Money is representative, if only for this sarcasm-laden retort from the comments section:

The objectively smartest political strategy for the Democrats to pursue is to push for a platform that just happens to line up with the preferences of the author. I don't believe I've ever read a piece like this before on the internet.

Yes, there is a lot of talking one's own book in these post-mortems, which makes finding worthwhile analysis more difficult. There have been the Captain Obvious pronouncements about the adverse map in mostly red states, the Boomer-heavy demographics of the turnout, and the lack of motivation to vote for hope and/or change. In fact, it may be that not much at all can be concluded for the long term, unless a Supreme Court justice dies in the next two years. Waves come and go, as Markos Moulitsas explains with palpable disgust. There have been some weaksauce attempts to pin the losses on Hillary Clinton, but the map is much more favourable in 2016, and the tide will inevitably turn.

I think digby digs a little closer to the reason for that four point discrepancy.

Nobody seems to be talking about it today but ISIS and ebola and the Border and Ferguson were all huge stories in this campaign and I have to think these stories may have inspired the GOP turnout more than they're being given credit for. They are classic fear motivators for the conservatives and the timing was very good. I would not be surprised if they played a part in making the GOP victory as substantial as it was. And this could be a factor in 2016, so stay tuned.

My suspicion is that the polls aren't skewed as such, they just don't capture very late shifts in mood. Just as the wingnuts complained about Hurricane Sandy being a late boost to the Democrats in 2012, the left could have a justifiable whinge that the Ebola outbreak, building on the aforementioned scare stories, led to a more scared and conservative electorate, and/or scared off people who self-identified to polling outlets as "likely voters" into discouraged non-voters. Why didn't it show up in the polls before election day? How can I prove this conjecture? I can't, I suppose, unless I had control over some expensive polling infrastructure. That Schrodinger's cat will have to stay in its box.