Last week, a lot of people got very excited about a YouGov poll showing the Conservative Party breaking into a decisive four-point lead after several weeks in which Labour were publicly divided over anti-Semitism. This showed how much damage was being done by (depending on your perspective), Jeremy Corbyn’s failure to tackle the issue, Boris Johnson being back in the news, or whatever hobby horse the commentator in question happened to be riding.
A week later, a number of new polls have come out, not only from YouGov, but from BMG and ICM, all of which show a return to the normal picture: deadlock with the odd very small Labour lead.
It’s a good reminder of Twyman’s Law: if it’s interesting, it’s wrong. For the most part, a surprising poll is a bit like that weird vibration you feel in your pocket that makes you think your phone is ringing – if you ignore it, and there isn’t another weird vibration along pretty damn quick, it means that your phone wasn’t really ringing at all.
We talk a lot about the margin of error in polling being plus or minus three percentage points – but we forget that doesn’t apply to the gap between the political parties but the vote share of the political parties.
Let’s say the “real” figure for Conservative support at the moment is 39 per cent of the vote. That would mean we would expect polls to throw out a Tory share of anything from 42 per cent to 36 per cent of the vote. Let’s say that the “real” Labour performance is also 39 per cent. Again, we’d expect the odd poll to throw out anything from 42 to 36 per cent. So you’d expect, from time to time, a poll showing a comfortable Labour lead of six percentage points, with them on 42 per cent and the Tories on 36 per cent, and vice versa. I really wouldn’t pay much attention to any one poll showing anything that dramatic, not least because it is the middle of the summer and most people are not paying too much attention to politics as a result.
That said, another good sense check is Populus’ weekly series in which they ask people what, if any, news stories they have noticed. The answer is almost always “nothing at all, or a Kardashian”. It’s a good rule of thumb that if no one has even noticed a story, it hasn’t moved the polls. And the good thing about waiting until Friday (when Populus release their “most noticed” figures) is that the chances you will have more than one poll to draw on as far as voting intention goes.