I’m off to a wedding this afternoon so I’m not around to see the results of the London mayoral election (I’m going to be an even sadder version of those people who have to keep popping out of wedding receptions to check the football scores, popping out to check election results). It matters not just for the parties and for who runs London, but in this little parochial corner of the internet, for the reputation of the polling companies.
I’m not about, so please use this thread to discuss the results as they come in.
YouGov’s eve-of-poll figures for the London election - the only pollster to produce a true eve-of-poll effort with fieldwork within a few days of actual voting - has topline figures of JOHNSON 43%, LIVINGSTONE 36%, PADDICK 13%. Minor party support is split Green 2%, BNP 2%, UKIP 1%, Alan Craig of the CPA 1% and Lindsey German of the Left List 1%.
After second preference votes have been re-allocated Boris Johnson is projected to win by 53% to 47% for Ken Livingstone. This suggests that the second preference votes have broken in favour of Ken Livingstone.
Several people including Nick Sparrow and Mike Smithson have rightly pointed out that polling figures for how second preferences split aren’t particularly useful to us because of the small sample - if you’ve only got 170 people saying their are voting for Brian Paddick, the margin of error on how they divide between Ken and Boris is huge. Still - if Boris Johnson does have a 7 point lead on first preferences, the second preferences would have to break in favour of Livingstone to an absurdly unrealistic degree for him to overcome it, given the number that are unused or given to minor candidates.
Voting intention in the assembly stands at CON 40%, LAB 33%, LDEM 14%. Others include the Greens on 4%, BNP on 3%, UKIP 2%, Christian Choice 1%, Left List 1%, Respect 1%, Abolish the Congestion Charge 1%. In an earlier YouGov poll the assembly question was asked about only the constituency vote, not the more interesting list vote. I’ve sadly no idea which these were asked about.
The Electoral Commission have released a study by NOP into how many people in London aren’t actually registered to vote. NOP found 20% of people in London between 18 and 25 weren’t registered and 27% of people between 25 and 34 weren’t. In comparison, only 5% of over 65s and 8% of 55-64s are unregistered.
It looks as though these are mostly likely to be disproportionately votes that would otherwise go to Ken Livingstone. Older people are the most pro-Boris group and registration also tends to be lower amongst people in rented accomodation and people from ethnic minorities - a mainstay of Ken’s support.
Some polls - most notably Ipsos MORI’s last poll for Unison - did ask whether people were registered to vote and excluded those who weren’t, but this doesn’t actually go that far to addressing the question because it assumes people who aren’t on the register know they aren’t. The NOP study cross-references the answers they got with the electoral register, and found that 16% of respondents who thought they were registered weren’t actually on it.
Two days to go. Candidates, as ever, are here and polls are here
Despite the report in today’s Telegraph, which managed to publish ICM’s polls from a couple of week’s back as a brand new one, there will not after all be another ICM poll on the mayoral elections. That means YouGov’s “morning-of-poll” effort is the only one left to go before the results on Friday.
YouGov’s weekly mayoral election poll for the Evening Standard is out - this will the last Monday poll, though there may yet be a final eve-of-poll effort - and with four days to go shows Boris Johnson with a solid lead. The first round voting intention figures are JOHNSON 46%(+2), LIVINGSTONE 35%(-2) (Brian Paddick’s figure doesn’t seem to be in the Standard’s report). After second preferences are reallocated Boris leads Ken by 55% to 45%, suggesting that second preferences are very marginally in favour of Ken, but not by enough to have a significant effect.
With so little time to go Boris Johnson must be almost assured of victory… if the poll is correct. In contrast to YouGov’s figures the final MORI and mruk polls of the campaign showed very small leads for Ken. It boils down to either Johnson having a secure lead if you believe YouGov, or Livingstone barely ahead if you believe MORI and mruk. I’ll be putting up a post looking at some of the reasons why the polls may be showing such contrasting figures later today.
What I assume is mruk’s final poll on the mayoral election is published in the Sunday Times tomorrow. The topline voting intention figures, with changes from last week, are JOHNSON 43%(-1), LIVINGSTONE 44%(-1), PADDICK 9%(nc). Once second preferences are re-allocated Ken Livingstone is projected to win by 2 points over Boris Johnson, 51% to 49%.
In practice this has Ken and Boris almost neck and neck, an even tighter race than MORI indicated in their final poll of the campaign a few days ago, which had Ken 3 points ahead on the first round and four points ahead on the second round (though MORI had a higher level of support for others and for Brian Paddick - the latter difference almost certainly because mruk did not include Paddick in the question prompt).
In the week ahead there should still be London mayoral polls by ICM (who showed Johnson 1 point ahead in a poll almost a month back now) and YouGov (who have shown large, but narrowing, Boris Johnson leads up until now). It remains to be seen whether all the pollsters converge together as they did before the 2005 general election, or whether they continue to project contrasting results.
A London voting intention poll by Ipsos MORI for Unison has voting intentions has topline voting intentions of JOHNSON 38%, LIVINGSTONE 41%, PADDICK 12%. Once second preferences are reallocated Ken Livingstone leads by 6% over Johnson. It was conducted between the 23rd and 24th April.
Ipsos MORI also asked a question about whether people were actually registered to vote, though bizarrely the implication in this Guardian article is that they were still included in the topline figures, since once they were excluded Livingstone’s lead fell to 4% on second preferences.
No other details yet. There should be more mayoral polls from YouGov, mruk, ICM and - possibly - another Ipsos MORI to come before polling day on Thursday. Unless other pollsters also show a Livingstone lead the race is looking close enough to produce a real embarrassment for at least one pollster.
UPDATE: it appears the poll was leaked to the Guardian, and shouldn’t have been published yet. It’s the Guardian who emphasised the 6 point lead that included people who aren’t eligible to vote, MORI I assume will be correctly excluding them from their topline figures! Over at Mike Smithson’s site he also says that, in response to his comment that this is the poll MORI will be judged upon for the mayoral election, MORI are pointing out “that the fieldwork was carried out more than a week before the election and a third of respondents said they might change their mind.” MORI are quite correct in this - in all fairness a poll does need to be done within a couple of the days of the election to really be compared to the result…but it does sound like they are hedging their bets already. Whether it is entirely fair or not, if polling companies’ final figures are showing different winners for the London election, it is going to reflect very badly on the companies who get it wrong.
The mruk poll in the Sunday Times also included a voting intention question for the Westminster elections. Strictly speaking it’s not a proper mruk poll of London voting intentions, since it there is no proper likelihood of voting filter (we can’t be certain that people who say they are likely to vote in the London mayoral election would also say they are likely to vote in a general election, though I suspect they probably would), but it’s all we’ve got. The figures, with changes from the general election, are CON 41%(+9), LAB 35%(-4), LDEM 18%(-4).
If repeated at a general election the Conservatives would gain 14 seats in London, 11 from Labour and 3 from the Lib Dems: Finchley & Golders Green, Croydon Central (notionally), Battersea, Carshalton and Wallington, Harrow East, Sutton and Cheam, Richmond Park, Hendon, Brentford and Isleworth, Eltham, Westminster North, Poplar and Limehouse, Tooting, Hampstead and Kilburn.
YouGov’s weekly Mayoral election tracker for the Evening Standard has topline voting intentions of JOHNSON 44%(-1), LIVINGSTONE 37%(-2), PADDICK 12%(nc). Once second preferences are reallocated the result is JOHNSON 53%, LIVINGSTONE 47%.
Johnson’s lead has fallen from the heights it reached earlier in the campaign (my guess was that this was becase of Boris’s less than impressive performances in the televised debates, though it could be because the earlier large leads were when the Lee Jasper affair was prominent in the London media). However, this poll is pretty consistent with YouGov’s figures from last week (and indeed with MORI’s figures from last week) suggesting the position has stabilised. There are now only 10 days to go until the mayoral election, so while this race has moved about a bit over the last few weeks, there isn’t much time for Livingstone to turn round a 6 point deficit and Boris looks set to win.
All this depends, of course, on YouGov’s polls being right. They have been the most regular pollster during the London mayoral campaign, but other companies have tended to show a tighter race. In their most recent poll Ipsos MORI, the only other company to have produced more than one mayoral poll in this election, also showed Johnson with a 6% lead in the first round, but showed a narrower contest as second preferences split overwhelmingly in Livingstone’s favour.
Low turnout elections do tend to be tricker for pollsters to predict correctly (though YouGov did manage it in 2004), but judging by the polls Boris does seem to have a significant lead as the candidates turn the corner into the final straight.
There are two polls in the Sunday papers. Firstly there is a nat-rep Populus poll for the Sunday Mirror. The topline voting intention figures, with changes from Populus’s last poll, are CON 40%(+1), LAB 30%(-3), LDEM 19%(+2). The poll was conducted between 16th and 17th of April.
Populus’s last poll showed a significantly smaller Tory lead than polls from other companies and in contrast to the figures from ICM who use almost identical methodology. Now Populus, ICM and YouGov are all showing double point Tory leads. ComRes and Ipsos MORI continue to show tighter figures.
Secondly there is a mruk Cello poll on the London mayoral race, published in the Sunday Times. It shows first round voting intentons of JOHNSON 44%, LIVINGSTONE 45%, PADDICK 9%. With second preferences re-allocated, the second round figures had the two lead candidates equal on 50%.
It is very hard to put this poll into any context: mruk Cello have no track record of voting intention polling outside of Scotland, and up there they overestimated Labour support compared to other pollsters prior to the 2007 election. While, like all polls, the figures were weighted to be representative, we don’t know if that includes any political weighting (without which phone polls given much higher levels of support to Labour) nor how likelihood to vote was taken into account.
UPDATE: the mruk Cello poll was weighted by past vote (though we don’t yet know to what figures so can’t judge the effect). It wasn’t prompted by party name, probably explaining that low figure for Brian Paddick. The turnout filter included all those 8+/10 likely to vote, so rather more generous than MORI’s filter. This covered 75% of respondents. Thanks to Mike Smithson for getting the info.