The latest LabourList leadership poll: Breaking down the numbers

By Mark Ferguson and Alex Smith

As we know from the topline results of our latest readers’ poll, published last week, the leadership contest is going to be closely fought. David Miliband is leading first preferences by 34.9% to Ed Miliband’s 30.8%, Diane Abbott’s 20.4% Andy Burnham’s 9% and Ed Balls’ 4.9%. But Ed Miliband (34.1%) is significantly ahead on the crucial second preferences, and Andy Burnham (30.3%) is ahead on third preferences. But below those numbers, there is further interesting analysis that can be done by looking at the answers of all respondents.

Here is the basic breakdown of how our readers placed each candidate.

As you can see, translated to the leadership election, and there are obviously complications and caveats (below) required to understand the validity of our polls, these numbers would mean Ed Miliband would be the next leader of the Labour Party.

The Labour Party uses a preferential voting system to select its leader. Voters place candidates in order of preference from 1 (most preferred) to 5 (least preferred), and as candidates are eliminated votes are transferred between candidates based on how voters have ranked them. So, based on our survey, Ed Balls would be eliminated first, with his second preferences would be divided between the three remaining candidates, and the results re-tallied. Andy Burnham does not gain enough of Ed Balls’ second preference votes to take him above Diane Abbott, so he is eliminated in the next round.

Andy Burnham’s voters who ranked David Miliband, Ed Miliband or Diane Abbott second have their second preferences allocated, and those who placed Ed Balls as their second choice have their third preferences distributed between the three remaining candidates. Those who voted for Ed Balls as first choice and placed Andy Burnham as their second preference have their third preferences divided between the three remaining candidates. The process is repeated in the third round, when Diane is eliminated, and the final run off totals are calculated.

As the field narrows, transfers from Diane Abbott will play a significant role. Ed Miliband takes nearly 85% of Diane’s second preference votes in the final round of voting and, indeed, more second, third and fourth preference votes than David Miliband from every eliminated candidate. The very small sample size means these breakdowns are only suggestive, but they perhaps indicate Ed has a broader appeal than David – although David accrues a significantly higher number of first preference votes.

Nothing remarkable happens to either Ed Balls or Andy Burnham’s second and third preference division, except they both break very slightly favourably to Ed Miliband. Fifth preferences – or most disliked candidate -highlight the division over Abbott’s candidacy, and the apparent unease with Ed Balls, as both are receive a lot of fifth preference votes. Conversely, David and Ed Miliband receive remarkably few.

So far it seems that Ed Miliband is more likely to receive second preferences than David (147 vs. 63). For David to overturn this advantage he either needs to reach out to other candidates or he needs Andy Burnham – the candidate to transfer most of his support to David – to do better. Ed Miliband, on the other hand, needs to make sure he is within sight of David on first prefences, and continue to be an acceptable second choice for Abbott voters.

These polls are a useful snapshot of where we are, but we accept, of course, that they are not perfect. The vagaries of the electoral college system mean they could be wholly irrelevant. If David Miliband, for instance, were to pull 70% of the Trade Union vote, 50% of the PLP support, and the result above, he would win in the first round without any re-allocation of preferences. And given the self-selecting nature of respondents and a sample which is probably not representative of the overall party, this problem is further exacerbated. And the sample size is small – in this case resulting in a +/- 4 margin of error.

Some people, therefore, dismiss our results out of hand as “voodoo polls”. We disagree: while they are not totally scientific or representative of the whole voting labour movement, they are self-evidently a useful reflection of our readership, and a guide to how the campaign is developing. These numbers are ten days old, but the analysis of how votes might transfer offers us a little insight into where we are in an otherwise arcane process.