UK parliamentary by-elections 2014

[Filuwaúrdjan]

Whatever can be said about that particular by-election, two things can be noted. First, that Labour's poll share held steady; second, that the issue that allowed UKIP to build up a head of steam was (as you know very well) a local one.

[EPG]

The C2 demographic has been contested since the 80s so one can't simply say that Labour are losing these people; they lost them a long time ago and probably those inclined to the Conservatives were first to jump to Ukip.

[Filuwaúrdjan]

Though I'd point out that there is no such thing as 'the C2 demographic'. One of the principle reasons why polling breakdowns in that category are notably volatile (even when allowing for the replacement of theoretically more reliable survey data with commercial polling aggregates) is because the range of occupations in it is extremely diverse. There are subsections that tend to vote heavily for this party or that party, and subsections that are extremely swing-y. And of course subsections that vote quite differently in different parts of the country. Back when there were still huge industrial employers employing huge numbers of skilled workers, sheer weight of numbers from that source meant that the category was (just about) uniform enough to (just about) make some sort of sense; not so now.

[EPG]

But that describes every demographic group in society, particularly those in the middle of some distribution. Outside theoretical sociological models, no demographic is uniform, not even a social class. Groups don't have to be uniform to be useful or meaningful. You still typically end up with, for instance, C1C2 voters supporting Labour at rates somewhere between AB and DE. They still typically earn less than AB, and do better than DE. So there is something useful going on in the industrial split, even if it means different things today than 40 years ago, and even if there's a better hypothetical alternative. It's just a matter of taking large amounts of evidence on aggregate and not relying on any one, small sub-sample.