Okay, I’ve now done a very speculative model of how the Labour membership vote may go. Please do not take this as a prediction. What I’m trying to do is identify the factors that could influence the vote and see what that might look like. There are so many assumptions involved, that it can not be regarded as any sort of prediction.

The first thing one has to do is work out the relative membership in each electorate. Sadly (but understandably) Labour won’t tell me, so what I have done is make assumptions based on the total party and electorate vote each electorate got at the 2011 election, and proportion them out on the basis of total votes over 9,000 (which is basically your weakest electorate). If someone wants to supply me with actual membership numbers happy to update the assumptions!

I’ve assumed 7,000 members (so average per electorate is 100), but that doesn’t matter, what matters is the relative size of each electorate (assuming they all have same voting turnout).

Next one has to assume how they will vote. Now this is like reading tea leaves, and not much more scientific. But can make some assumptions. Here is what I did:

 Cunliffe Jones Robertson Cunliffe Robertson Other 45.0% 22.0% 33.0% 57.7% 42.3% Auckland 65.0% 14.0% 21.0% 75.6% 24.4% Wellington 24.9% 12.1% 63.0% 28.3% 71.7% Christchurch 31.6% 15.4% 53.0% 37.3% 62.7% Dunedin 31.6% 15.4% 53.0% 37.3% 62.7% Maori 27.7% 52.0% 20.3% 57.7% 42.3%

So for most electorates, I assumed Cunliffe 45%, Robertson 33% and Jones 22%, based on general acknowledgement that Cunliffe has more support from members.

But in Auckland I give Cunliffe a 20% boost, in Wellington Robertson a 30% boost and a 20% boost for Robertson in Christchurch and Dunedin. And Jones gets a 30% boost in the Maori electorates.

Then there is one further adjustment. If the electorate has an MP who is a supporter of one candidate, that gives that candidate an additional 30% boost. That is based on the reality that the local MP will have significant influence on their members. This is not like the UK where the number of voters was hundreds of thousands. In each electorate it may be a few dozen only and the MP will have influence. Some MPs such as Goff and Mallard probably have more influence over their electorates than a new MP, but I’m assuming a 30% endorsement factor for each.

So what that does give us as a model for each electorate.

Now what do you get when you multiply each vote by their assumed strength.

• Cunliffe 45.4%
• Jones 13.1%
• Robertson 41.5%

And if you assumes Jones second preferences flow the same as other’s first preferences:

• Cunliffe 53.0%
• Robertson 47.0%

I’ll reveal tomorrow the updated model for each section (caucus, unions and members) and what the overall model is projecting for the total vote.