Why the UK polls were wrong


As you can see the UK pre-election were wrong. Quite massively wrong. The Conservatives beat Labour by 6% and have won a clear majority. Only one poll of of the several dozen in the last few weeks got close to this. The poll were near unanimous that the Conservatives and Labour would be tied in the vote, and Conservatives would get only a few more seats. Instead they got 98 more.

So why were the polls do wrong? Not one poll, but almost all of them. There three broad plausible explanations – which are not mutually exclusive.

1 – People lie to the pollsters

Someone tweeted that the British have shown the one thing they’re really good at is lying to their pollsters.  A more polite version of this is what the Guardian calls Shy Tories. People don’t like to admit they are voting for a party. One has seen this in the US when one candidate is African-American. Also in NZ to a degree where I suspect one of the reasons NZ First often exceeds the polls, is people don’t like to admit they are voting for them.

But I think it is unlikely this explains most or even much of what happened.

2 – People change their mind

Either the undecideds decide to vote a certain way disproportional to the already decideds, or some decideds change their mind. One reason for this is tactical voting. This is why ACT keep winning Epsom despite poll after poll showing them behind. People only get tactical at the last minute.

Major newspapers published guides as to how to tactically vote to maximise the outcome for your preferred PM. This could have had quite an impact.

However while I think this may have been some of it, I don’t think it was the major factor. Even in seats where there was no ability to vote tactically (no major third party), you saw the Conservatives pick up seats off Labour.

3 – Turnout was different

Turnout was higher than expected in many areas. If one side does better at turning out their supporters, this can have a big impact.

In NZ the impact of Dotcom was to so enrage Government supporters, they advance voted in record numbers – determined to keep him out.

If you look at the motivations to vote in the UK for Conservative and Labour voters, they were quite different. Conservative voters had a pretty strong motivation to vote to keep Ed Miliband out, and to stop a party which wants to dissolve the , from holding the balance of power. A Mliband Government propped up by the SNP was very scary to many.

However if you are a Labour supporter, your best outcome was a Labour minority government that could only govern with the SNP’s votes. This is hardly motivating stuff.

So I suspect (we’ll know more as we get more data) that the major difference was turnout.


Of some interest is that in several elections now, it has been the more right wing parties that have exceeded their polls. In Israel Likud did massively better than the polls, as did the Conservatives in the UK. In the 2014 US mid-terms the Republicans did far better in the Senate than projected. And even in NZ National did better than the polls (but within margin of error). I’m not saying this is significant – just that it could be. Or it could just be chance. In one Victorian state election the Liberal Party did far far worse than the polls, and in NZ in 2011 National did a bit worse than the polls. But

Comments (75)

Login to comment or vote