A rumoured coup attempt on Ed Miliband, a reduction in the EU bill…it doesn’t seem like this has been a good wekk for Labour. Do the polls reflect that? Let’s cut to the chase with the last two weeks’ predictions.
That right there is a confusinfg set of polls. We have a dramatic narrowing between Labour and the Conservatives, followed by a sudden recovery, followed by another more gradual narrowing, and a sprinkling of utterly anomalous polls in the middle. Overall, though, one thing does seem pretty clear – things aren’t getting much better for Labour.
The bizarre UKIP results are worth looking into a little further. Here’s the master graph with its latest updates:
Now, let’s filter that a bit by the different pollsters whose results we’re using. The simulations based on YouGov’s results look like this:
Much less variation. How about Opinium?
Still a pretty smooth trend – but notice that UKIP’s results are pretty consistently at around the same level as those of the Lib Dems, whereas YouGov consistently tracks them well below. Finally, let’s see what results ComRes come up with.
Much less consistency here, although the results do start to sort themselves out from about July onwards.
So, two things to note from that: one, the pollster you choose to look at has a massive impact on your predictions, because of their varying methodologies; and two, every pollster occasionally comes up with a bizarrely anomalous result (as you’d expect, given the number of people they’re talking to each week), so it’s much better to look at the general picture than it is to focus on any one result.
Let’s finish off with a look at the seat breakdown based on 1000 runs of today’s UKPR Polling Average (CON 32, LAB 33, LD 8, UKIP 16, GRN 5).
Much the same as last week, Labour are almost always below the winning line (in 97.9% of runs, with a mean shortfall of 20.7 seats), although they would still be in power in coalition with the Lib Dems (or the Lib Dems and the Greens) in 97% of hung Parliaments.
Electobot: The Blog 