YouGov’s Mayoral Polling

(For the record, as well as being a contributor to CentreRight, I am also co-founder of YouGov)

As you can see from Harry’s post, YouGov does have significantly different numbers from the other pollsters on the mayoral race. Obviously we think we’re right, as we were last time, but there is one considerable uncertainty: turnout. Usually the sample that pollsters talk to is reasonably similar to the active voting population, but in elections such as this one, with historically low levels of participation, deciding which (if any) ‘likely-voter’ filters to apply is tricky. With any polling methodology there tends to be strong over-claiming by respondents – not because they are necessarily fibbing about their certainty to vote, they may really believe it, but can’t be bothered on the day.

With high differential levels of support at different likelihoods of voting, this can become highly consequential. The outer boroughs tend to be more pro-Boris than pro-Ken, but tend to feel less identified as Londoners and so have not in past mayoral elections voted in the number that might be expected. Boris may change that. Meanwhile people in inner-London, who tend to be more pro-Ken, may be fired-up by fear of a Tory winning London for the first time in a generation. Turnout will be the deciding factor in this race, but our poll today suggests that Ken has a mountain to climb in the three days that remain.

Please note: contrary to the implication in the graphic accompanying Harry's post, today's is NOT our final poll: one more to come!

Comments

[Kevin Davis]

From what I am hearing on the doorstep the inner London vote is not going to come out and the outer London vote will!

Maybe outer London is feeling empowered to do something in a way it never has before.

[Brian Wood]

It is often forgotten that the weather on the day can be an important factor in determining turnout. If it is raining then only the really determined (often Conservative) voters will turn out but if it is a pleasant day/evening then the turnout is likely to be higher as there is less of an excuse not to go and vote.
Another point is what is on TV, now much harder with all the cable and satellite channels.

[Chad Noble]

"today's is NOT our final poll: one more to come!"

LoL. Let me guess. Would that be the one with seconds to go that predicts a strong 'last minute swing' to Ken....

Seriously though, let's not get the excuses in early. Either this poll should be ignored for being completely unreliable for the reasons you state and thus is little more than pro-Tory propaganda, or it shows some genius that the others have completely missed.

Voting is in just 72 hours, and a double digit win for Boris is a *very* bold prediction.

Without doubt, this mayoral election is a very public test of whether opinion polls hold any merit.

(someone who wanted Boris to win until his woeful performance on QT)

[Ric]

Opinion polls cloud the issue and should be banned for the 2 weeks before an election. In the 2006 N.E.Regionalisation vote ,opinion pollsters kept saying thatthe result was too close to call. The No vote by a 4 to 1 majority. Not very accurate to say the least.

[bluepatriot]

Very difficult to do during this internet age, Ric.

[Stephan Shakespeare]

Chad, that's pretty unfair. The 'last-minute swing' excuse is one that can only be used by companies who don't do an eve-of-election poll; the point about doing a late-as-possible poll is that it DOESN'T allow that excuse!

Nor do I 'state' anything is 'completely unreliable', I just said that it's much more difficult to call a low-turn-out election - that's pretty obvious and standard. Maybe for that reason they should ban polls for two weeks or newspapers shouldn't commission them, it's not my concern one way or another, we just carry out the polls we're paid for as best we can.

[Chad Noble]

Hi Stephan,

The Evening Standard of course had this poll emblazoned across the top of its front page.

Now the question is, do we seriously believe just hours from polling that Boris has a double digit win coming up, or was the poll simply a tool for the ES to beat Ken.

Don't get me wrong, there's nothing wrong with giving Ken a bit of a kicking, but if the polls are simply skewed to meet the client's agenda, no-one can seriously use them for predictive purposes.

[Tim Montgomerie]

Chad: You are close to accusing YouGov of fiddling figures for a client. That's not acceptable.

[CCHQ Spy]

I see that ICM aren't publishing a final poll.

Chickens!!!

At least YouGov has balls.

[Chad Noble]

No I'm not Tim. We all know poll results depend on a million factors, and can obtain any result desired really. The client pays, the client gets.

My point was that, that's fair enough, but to them assign any predictive power to them is as non-sensical as your poll of polls (for all the reasons Graeme has more than ably detailed time after time).

But, let's reserve the post-mortem until after the vote as a double digit win for Boris will clearly prove YouGov's bravery and genius and condemn the other pollsters calling a close vote with slim victory for Ken into the dustbin.

[stephan shakespeare]

"The client pays, the client gets" - completely wrong, Chad, completely crap. This is indeed accusing us of fiddling.

I'll never forget the night when IDS, for whom I was pollster and who remains a close friend, had to read one of my polls in the Telegraph saying "IDS was a mistake, say party members" at the very time he was fighting for his political life.

It is an atrocious accusation to make. Even if you couldn't credit us with professionalism, you might credit us with common sense: we thrive because our polling is accurate, simple as that.

[Chad Noble]

Stephan, I am not accusing you of anything.

I wonder why Tim did not feel the need to criticise in anyway the posters on conhome who made references to the 'union funded' other polls that do not show a double digit win for Boris but a close race slim win for Ken.

Surely the 'union funded' remarks could be equally misread by overly sensitive pollsters to imply the other polls were fiddled? Why is reference to your client's , the ES that has a long-running feud with Ken any different to the 'union funded' remarks?

No-one seems at all concerned at attacking the pollsters who do not call a double digit Tory win. Funny that...

Anway as I said before, it is not worth wasting any energy picking apart your double digit win prediction for Boris (just 72 hours before the vote) until after the vote, as I clearly stated, it could just as easily be genius that vindicates YouGov and condemns the rest to the dustbin.

I'm interested in the accuracy of the pollsters, not the party who wins, and find this clear divide in predictions a great chance to put the pollsters themselves to test.