Labour needs to be more robust with evidence than this

Sometimes, you read something so defiantly muddle-headed, that you wonder if it is – in fact – satire.  Yesterday, was just such an occasion. I read a post on Labour Uncut which was ostensibly evidence based, yet the holes in it were so big, the piece was more hole than substance.

The piece purported to prove that the theory of the missing 5 million votes that Labour lost between 1997 and 2010 is a core-vote strategy to show with number that it is doomed to failure.

For reasons passing understanding, the author Atul Hatwal chose to do this by comparing the ward-by-ward breakdowns from the London Assembly elections of 2012 with the London ward-by-ward breakdowns from 2010 local elections.

Where to start with why this is wrong?

Well firstly, focusing on London alone is crazy. Labour did rather well in London in 2010, or at least not as badly as we did in the country as a whole. So we were starting from a higher water mark than we would be elsewhere. This – of course – means that Labour has less ground to make up and so the distance it is possible to travel from where we were to where we need to be will be statistically less impressive than the national journey might have the chance to be.

Secondly, there are real issues with comparing the data from assembly elections with the data from either General Elections or even council elections. The elections were not being fought at ward level, and so much more emphasis was given to turning out a high Labour vote in areas we are strongest, and less on fighting in marginal and Tory wards. The Assembly elects through a combination of list and multiple borough constituencies. These give voters more reason to vote for smaller parties and independents than they would in a general election.

If you look at places outside London that had local elections in both 2010 and 2012 the Labour strike rate is very much higher. It was 80% on the same definition in Southampton, 100% in Plymouth (plus an extra ward that was even harder to win), all the wards up to a 10% swing in Reading, everything that meets the Hatwal criterion in Yarmouth, Basildon, Harlow, Ipswich. 80% (plus one extra) in Birmingham, Dudley everything up to 10% swing plus one extra. These are all key targets in mostly Lab/Con, southern, swing seats.

So the geographical basis of the piece – quite apart from laying it completely open to charges of extreme London-centricity – massively skews the findings.

And then there are the glaring and really quite bizarre omissions from the analysis. There is no weighting given at all to the fact that the London Assembly elections – with the best will in the world – took a back seat to the Ken and Boris show. No credence is given – at all – to the idea that Ken Livingstone was an overall drag on the Labour ticket despite the fact that Hatwal himself wrote that this was the case in no uncertain terms after the election. On this occasion I happen to agree with him. It seems that if Hatwal and I are both right about this, it is at least likely that this is more likely to be true in Tory wards than in Labour ones, several of which saw Ken increase his vote.

But the weirdest omission of all is the complete lack of any narrative around the Lib Dems spectacular collapse. They just aren’t mentioned. Which is crazy as there’s a real story to tell about the fact that in some of their seats (for example Brent Central and Haringey and Wood Green, they failed to beat Labour in a single ward). It’s true that the fight is going to be between us and the Tories, but to simply ignore the Lib Dems like their voters aren’t good enough or are already in the bag would be a very stupid thing to do.

Finally, the piece doesn’t even have the courage to extrapolate its own numbers to a general election, even though the implication of the piece is that this proves the 5 Million Votes strategy wrong for a future general election. Why not? Well speaking to excellent psephologist and analyst Lewis Baston, he says that because even under this most pessimistic, flawed and cherry-picked of approaches, the result of Labour winning 51% of seats where we need a swing of 5% or less from the Tories (assuming the very least expected from the Lib Dem collapse) would leave us will approximately 299 seats and put Ed Miliband in number 10. Perhaps not with an overall majority, but certainly as leader of the largest Party. The reality may be better, it may be worse, but that would be the case according to this model. The analysis fails even on its own deeply flawed terms. No surprise Hatwal didn’t want you to know that.

Why does it matter? It’s just another blog on a website not famed for its balanced approach to Labour Party politics. But as someone who believes there might be something to this 5 Million Votes theory, it does matter to me that the theory is tested robustly by those who disagree with me. I’d like Labour to win in such a way that opens up more space to be innovative on our left flank. I’d like Labour to have the strength and ability to mobilse a large group of voters whose loyalty is less testable. But ultimately, I want Labour to win the next election and every election after that. So if those who believe a move away from triangulation will be electorally disastrous then I want them to convince me this is so for the good of Party strategy.

I loathe bad data when I see the Government use it, I loathe it when I see the left use it. If this is the best I’m going to be offered, then I’m afraid you’re a very, very long way from anything convincing a good argument (which is a shame, because – as you might be able to tell from this piece – I love a good argument. One that – as Monty Python taught us – is a connected series of statements intended to establish a proposition (true statements by preference)). Labour must be more robust than this. I know I intend to and I ask those who disagree with me to do the same.