Westminster Blog calls it Wonk Wars. Fair enough: this is wonky. But I think it matters.
Nate Silver has shaken up psephology in the UK by proposing a different approach to uniform swing and instead going for ‘proportional swing’. By this I mean: instead of, say, taking 5% off every Labour vote share in the country and handing it out to other parties, take 15% of their vote. So where they took 60% of the vote before, take 15% of 60 or 9 percentage points from their vote to get it down to 51; if they took 40%, then they lose 6%.
I have quickly bashed around my tactical swingometer to see what a difference this makes. And it certainly can. Here is a proportional version of the sheet. I find that where the Conservatives get 34% of the vote and the other two tie, they rise by 25-30 seats under a rough approximation of the Nate Silver approach. Amber Valley, for example, falls, because I have had to assume 20% of Labour votes going to the LD’s, which under the new system means 9%, rather than 6.5% across the board under the old system.
But which model is accurate? Westminster Blog links to the furious to and fro between Politics Home and 538. I thought I’d add my tuppence by looking at the 1992-1997 collapse of the Tory vote.
The red line is a pure proprtional approach. Yellow is the line of best fit. The black line is the average – the uniform swing. Now when you have to choose between them, what would you do? The initial answer would be ‘a bit of both’, which has my heart sinking.
(I have also removed Tatton). Now as far as I can see, this is an almost flat line: the Conservatives lost 11% or so, plus or minus a large error term. So, if we are in a situation similar to the 1997 landslide in terms of lost Labour votes, I think a flat loss is a reasonably good way of proceeding. Of course, in 1997 the swing was Tory – Labour, whereas now the spoils need to be divided. No precedents.