Friday 7 November 2014

'Skewed' American polls: part two


Back in 2012, we made quite a big thing of slaughtering poll charlatans during the US Presidential election of that year. Back then, you'll remember, Republicans of all stripes argued that the polls must be wrong - chiefly because they were 'oversampling' Democrats. If the electorate looked more like that of 2010's midterm Congressional elections, they argued, Mitt Romney would be elected President. They were right about that. But they weren't right - and they were very, very unlikely to have ever been right - about what they thought was the 'right' shape of the electorate. In fact, young people, urban dwellers and African Americans turned up in their droves to vote for President Obama, and he was re-elected rather handily in the end. The fog of battle cleared. The flag of reality still fluttered in the breeze, albeit looking a little bit tattered and shot up.

This blog lambasted them for wishful thinking, rather than use of the actual data that always showed that the Obama coalition's key demographics were indeed likely to turn up at the polls. They might not have been as enthused as they had been in 2008 (four years of governing will do that to any party), but they still thought 'their' man much better than the alternative.

But now the boot's on the other foot. Democrats argued this year that they'd be able to turn out the same people to hold on the Senate. They said that the polls were overestimating Republican strength by insisting on over-tight voter screens that let through only those absolutely certain to vote.

They were wrong, nearly as wrong as the 'skewed polls' advocates of 2012 - something we wanted to register before the moment passed. It won't have escaped your notice that this effort - including the multi-million dollar Bannock Street Project - was a total failure. Even in relatively unheralded and unnoticed Virginia, where a popular and apparently centrist ex-Governor was expected to coast to re-election, Democrats were left clinging to a Senate seat by the tips of their fingernails. Elsewhere, Senators in North Carolina, Alaska and Colorado - all reasonably personable sitting Senators who had done nothing particularly crazy - narrowly lost in their bids for re-election.

Now this year's Democratic doubts about the polls was not couched in the same full-on rhetoric of media-hating fury that the Republicans managed to get themselves into during 2012. It was phrased in the tone of 'things might be rather different on the day', 'the polls might be wrong', and the like - rather than headlines about a 'left-wing media' that was deliberately misleading people.

Still. It's important to say that campaigns that are on the defensive - that are basically praying to hold on, rather than confidently filling the airwaves and the streets with confidence - indulge in a kind of false consciousness or epistemic denial. Republicans didn't want to see what was staring them in the face during 2012. Democrats didn't want to see their likely fate in 2014, a drubbing that almost all statistically-aware experts projected within a seat or two in the Senate. Their voters didn't turn up, while Republicans did. Even where they won, it is possible that their core constituency didn't turn out because they felt that 'their' candidate was so far ahead - while wavering voters thought 'I'll go for that new guy, because it might shake up the incumbent'.

Today's losers are still indulging in some of this groupthink, and it's inevitable. Not many people truly want to look at the reasons why people don't like them. In the Republicans case, it was because they were seen as hostile to ordinary and to new Americans. For the Democrats, falling wages and a lack of leadership made them seem as if they had no strong and determined answers to America's problems - certainly in contrast to the iron-hard (and deeply misleading) certainties of their more conservative opponents. So they have turned instead to questioning the numbers. To saying that reality is not as it is. To ignoring that the world is deeply, truly in flux, and that the bias of this year might be the equal and opposite bias of next year - not seeing that the polls that overestimate Republicans in a presidential year won't do that when the electorate is relatively white, rural and old. More and more entrenched in 'their' areas and among 'their' people, hiding in increasingly-gerrymandered red and blue fortresses, that's a psychological bias that's only going to get worse.

It's hard to step back and see yourself as others do. It's difficult to accept that nothing at all is fixed - in political life, least of all. And that's true whichever side of the aisle you sit on.