This has happened before. Some accused Labour of expectations management at Oldham, after a “too close to call” narrative developed, only for Labour to win easily. But the truth is, on that occasion, Labour people on the ground genuinely thought they were in trouble.
The problem is, parties no longer seem to have a good idea how well they’re actually doing, especially in by-elections (with the possible exception of the Lib Dems, at least in Richmond). It also seems to be the case in general elections these days – in 1992 the Conservatives were pretty sure they were going to win, in 2015 they weren’t.
Most people are pretty sure that the Tories will win in 2020 – ICM’s Guardian poll released yesterday found only 15 per cent expecting Labour to return to power at the next election. The headline figures, which are pretty typical for ICM (Conservatives leading 42-27) bear that out.
As widely discussed, one of Labour’s biggest headaches is Brexit. Anthony Wells took a look at what the least bad approach might look like, and found that leaving but seeking a close relationship (without going into the specifics) would upset the fewest people that voted Labour in 2015.
On the Brexit vote itself, Eric Kaufman produced an interesting regression based on the new ward-level data that demonstrated that there was no special “London effect”. The strong Remain vote in the capital can be entirely explained by its demographics, in particular the fact that its residents are younger, more ethnically diverse and more likely to have formal qualifications than in other parts of England.
“But…” a few smartarses replied, “…London’s youth/diversity/education IS the London effect”.
No. A “special London effect” would be if, after controlling for (ie taking into account) London’s demographics, it were still more “Remain” than elsewhere. Or put another way, its boroughs would be more “Remain” than other parts of England will similar age, ethnic and qualifications profiles.
That’s simply not the case – in fact Eric found a small but statistically significant London effect in the opposite direction. I’ve got variety of results depending on how the model is built, but the basic point stands – the London effect is extremely small.