Saturday, 25 September 2010

What to watch for on the leadership scoreboard

Very few people, beyond the candidates, will know for sure who has been elected Labour leader before the announcement. (If Sky News are going to risk a punt before the declaration, let's hope they have learnt from their mistake of calling the deputy leadership for Alan Johnson with a few minutes to go).

Even when the first round results are declared, most of the audience in the hall may still not be able to tell who will win (assuming the candidate's body language does not reveal the outcome). That was certainly the case in 2007 when, even with only the final three three candidates left standing, I had no idea whether Alan Johnson or Harriet Harman had won.

It will take up to five minutes to go through the process of declaring each round, eliminating each candidate, and reprojecting the electoral college. But the first round result which will contain some important clues as to whether supporters of any candidate can expect to be jubiliant or dejected three minutes later.

If you need to know what to look for. Here's Next Left's guide.

* The projected shares will be the electoral college totals

It is worth noting that the projected shares on the first round will refer to the entire electoral college.

So first, it is worth looking at the 2007 deputy leadership first round to get a feel for what it will look like.

All of the numbers will look very low, perhaps leading even some D and E Miliband supporters to panic, until they see what the other candidates have!

Every number will be one-third of the percentage share that candidate has of that section. So a candidate winning 50% of party members (or any section) would appear as 16.67, while a candidate with one-third of a section will put 11% on the scoreboard. If a candidate has 3% of MPs, that will come up as just 1% on the board.


As a benchmark of who is doing enough to win, if the YouGov/Left Foot Forward first round figures were exactly right, these would be the figures on the screen on the first round for the leading two. (You can divide any other numbers by three for yourself!)

David Miliband: 13.67% PLP, 12.67% members, 9.67% affiliates.

total: 36%

Ed Miliband: 10.00% PLP, 10.33% members, 12% affiliates.

total: 32%

Remember, David Miliband must do better than that projection where he lost the electoral college 49-51%. But he can do so either by outperforming these first round shares, or by doing better on transfers than he did in that poll. If Ed Miliband is ahead of those YouGov shares, he would seem unlikely to lose (though again this depends on maintaining a very strong transfer performance from each of the other three candidates).

If there is a magic number in the first round, it is the difference between 40% of the party members vote and the 38% David Miliband got in the final poll. If David Miliband has 13.33% on the CLP section, he is probably going to win the party members section. At 12.67%, YouGov him on track to lose it. The arithmetical gap between triumph and disaster is narrow indeed.

Overall, Four points is probably not enough. But remember, it is much harder to get a 5% of 6% electoral college lead than in a conventional poll or election. That requires an average lead of that scale across all three sections. Being 5% ahead among party members is worth a 1.67% lead in the college. So 5% college lead means being ahead of another candidate by 15 MPs, and 7500 party members, and 37500 affiliate voters, or the equivalent in other sections.

* If Ed Miliband takes the lead on any round, he has very likely won. (A mild caveat: maybe not if Andy Burnham finishes third and so is the final eliminated candidate. His MPs would probably break for David Miliband, though YouGov's last poll found Burnham member and union supporters breaking 3:2 to Ed).

* If Ed Miliband were to win narrowly, he would probably not go ahead until the fourth and final count, particularly if Abbott or Balls were to finish third.

* How large a first round lead is needed?

David Miliband lost the electoral college 49-51 based on a 36-32 lead in the final YouGov poll.
In early September, Next Left estimated that David Miliband might need a 6+ point first round lead to stay ahead if losing transfers 2:3. The second YouGov poll showed a poorer transfer performance, with David losing the members 48-52 on a seven point (38-31) first round membership lead. An electoral college lead of five or six points is in the zone of uncertainty, where either candidate could win, while four points is unlikely to be enough, unless the transfers fall better for David.

The penultimate round lead needed is smaller, but perhaps not by much, especially if it does turn out that Ed Balls is eliminated last. (David Miliband will do relatively better on transfers if and when Andy Burnham is eliminated; he was doing worst with Abbott voters, though can mitigate this if Abbott voters prove somewhat less likely to express further preference than those of other eliminated candidates).

For example, if Balls were third on the penultimate round with 20% of the electoral college, we might anticipate that 90-95% of those votes will transfer (with all existing votes mildly increased in value as some votes - like Balls own PLP vote worth 0.1% of the electoral college - are exhausted). If the split were 3:2 (12:8), then Ed Miliband might go up by around 11% and David Miliband by about 7%: David Miliband would just then hold on if he had a 42-38% lead going into the final round of the electoral college. (But a 2:1 transfer split might close a 6 point lead even at that stage).

If Balls had 15% of the college, a 3 point David lead would be enough to survive a 3:2 (9:6) split, while a 5 point lead on the penultimate round would hold out while losing Balls' transfers by 10:5.


Please remember, the strength of vote transfers can go up as well as down in any candidate's direction, so we take no responsibility for any joy or angst created by these rules of thumb.

In short, nobody can know for certain until you hear the winner's name announced.

No comments: