Saturday, 25 September 2010

Could it just have been YouGov wot lost it for David Miliband?

The Labour leadership was perhaps the least polled recent party leadership contest, but projections that the race looked neck-and-neck, and that Ed Miliband might just prevail were borne out in the result.

However, analysis of the final results (full details) present the intriguing theory - an unprovable hypothesis perhaps - that Parliamentary reactions to the YouGov polling of the race might just have had a decisive influence on the knife-edge outcome.

Ed Balls MPs' split Ed Miliband 26 (60%), David Miliband 15 (35%), with Ed Balls and Yvette Cooper abstaining. That lead of 11 in Parliamentary Ed Balls transfers were in themselves exactly enough to just decide the result: David Miliband would have won the electoral college if they had split 20 all. While Ed Balls chose not to be a "kingmaker" in the election, nor to cast his own second preference vote, his Parliamentary supporters in effect made enough of a difference to change the result.

David Miliband lost all MP transfers 38 (56%) to 29 (44%) after the first round, with Ed Balls transfers were decisive in his being unable to achieve a 50-50 split to maintain the scale of his first preferences lead in that section. He ended 53.5% to 46.5% - a seven point lead - having led by 10 points (41.7% to 31.5%) on the opening round. That failure to quite match his rival in attracting further support cost him a decisive single point in the electoral college.

David Miliband's support did appear to slip slightly in Parliament after the September YouGov poll suggesting a neck-and-neck race with Ed Miliband possibly edging it.

In fact the elder Miliband did six points better (44%) among party members in the election itself than his 38% in that final poll. It projected Ed Miliband at 31%, very close to his 30% share but overstated Diane Abbott with 11%, when she got 7.5% in the ballot. In the end, David Miliband won party members by nine points rather than losing them by 4 points (48-52) as in the final poll.

Had that early September survey found similar shares to the actual members' vote, it would have projected - overall - a very narrow David Miliband victory in the electoral college, again by a 51-49 margin the other way.

Might the Parliamentary votes have then ended up marginally different enough to make all the difference?

The poll shook-up the media coverage and political debate, eventually led to Ed Miliband becoming favourite after voting closed. It is unlikely that it had a dramatic effect on member or trade union voting at that late stage.

Given how few MPs votes would have swung the result, the hypothesis that YouGov helped to lose it for David is at least worth entertaining in the annals of "what if?" political leadership election history.

Where the transfers went

The most impressive achievement of David Miliband's campaign was to win 44% of party members votes on the first round, with Ed Miliband on 30%. That 14% lead closed to 9% by the final round, with the elder Miliband leading 54-45% on the final ballot, because the supporters of each of the other three candidates broke more for Ed than David Miliband.

Ed Balls party member support split 9295 (60%) to 6439 (40%) for Ed Miliband in the final ballot, 15734. (However, that was after 13% of his 18,114 votes and transfers did not express a preference between Milibands).

That 3:2 transfer rate was precisely that which underpinned Next Left's analysis of whether second preferences would decide the election. And so our results day prediction that afour point electoral college lead would probably not be enough was borne out as David Miliband's 3.5% first round lead was overhauled on the final round of Balls' transfers, pretty much as described. This outcome was also correctly called in real-time on twitter by Left Foot Forward's Will Straw, while the BBC's Nick Robinson had the opposite hunch, apparently through going more on instinct than the numbers.

Interestingly, Balls union supporters split by a slightly narrower margin: 16253 (55%) to Ed Miliband and 13377 (45%) to David.

Earlier, Andy Burnham's 24 MPs split 14 to David Miliband (58%), 8 to Ed Miliband (33%) and 2 to Ed Balls. But his member supporters preferred Ed Miliband, with 4521 (40%), and Ed Balls (31.5%) with 3604 to David Miliband 3247 (28.5%).

And Ed Balls won more union transfers from Burnham than either Miliband with just over 9000, compared to 7500 to Ed Miliband and 5500 for David Miliband, reflecting that Burnham won union votes in the north and from the left of the party.

Diane Abbott got only 7 MPs votes in the first round. 2 (Abbott herself and John McDonnell) did not vote for any other candidate, the others splitting Ed Miliband 4 and Ed Balls 1.


Guido Fawkes said...
This comment has been removed by the author.
Guido Fawkes said...

I think there might be something in your hypothesis, I changed my bets on the back of the YouGov poll, if MPs are rational they might decide their self-interest career-wise is in being seen to back the winner.

Am I right in thinking that 7 MPs changing their vote would have swung it?

Sunder Katwala said...

MPs who vote in the final round are worth 33.33/262 so about 0.127% of the electoral college.

The final gap is 1.3% which is just over 10 MPs.

So I make that that if you could switch just 6 from one camp to the other, that would swing 0.76% and so give the college to David M by a millimetre. 7 would give you a swing of 0.9% (so you'd be at 50.25 vs 49.75).

(if abstentions more complicated, as have to reweight everyone)

Guido Fawkes said...

Half a dozen careerist MPs?