Here’s a final post on the proposed boundary changes for the time being, I’ve had a chance to look at the marginality of seats. Now, as I said in an earlier post, on the levels of support at the 2010 election the proposed English boundaries would have given the Conservatives 5 fewer seats, Labour 18 fewer seats and the Liberal Democrats 7 fewer seats. That’s a gain of 13 seats for the Conservatives relative to Labour.
However, it doesn’t follow that there would be the same impact if levels of support had been different. What if, for example, Labour notionally lost lots of seats at 2010 levels of support, but the new boundaries produced lots of seats that could be won on a very small swing – it could be that the new boundaries were better for Labour than the current boundaries in a scenario where the Conservatives had slightly less support.
Hence the table below shows what the distribution of seats would be with various uniform swings between Labour and Conservative.
These suggest that while the proposed boundaries are beneficial to the Conservatives under 2010 levels of support, they would be much less beneficial to them compared to the current boundaries if there was a swing to the Conservatives (with a 2 point swing to the Tories – the equivalent of an 11 point GB lead – these boundaries would only see the Conservatives gain 3 seats relative to Labour). On the other hand, in a scenario where there was a swing towards Labour these proposed boundaries would be much better for the Tories. If there was a three point swing to Labour (putting the parties roughly neck and neck in GB support) then the Conservatives would win 2 more seats than on the current boundaries, Labour 26 less – a relative gain of 28 seats.
That particular level of support is most advantageous to the Tories, their gain declines again on bigger Labour swings. The point is, however, that in terms of marginality seats are not evenly distributed, so a particular set of boundaries that is good for a party when they are x% ahead in polls may not be good for them when they are y% behind.
The headline figures of gains and losses at 2010 levels of support can be somewhat misleading, given it ignores whether more seats become winnable marginals or safe seats. The best measure will probably be what percentage leads the two main parties need in order to get a majority on the new boundaries. Currently the Conservatives need about an 11 point lead, Labour about a 3 point lead.
We can’t tell exactly what leads they’ll need on the new boundaries until we see Scotland and Wales, but the Conservatives will need at most a lead of 9 points (since on these boundaries, that size swing would give over 300 seats in England alone) and I expect it will be a bit less unless the Welsh boundaries are truly horrific for them. On the old boundaries Labour could win an overall majority with a lead over the Tories of 3%, but given they win far more seats in Scotland and Wales than the Tories do, we really will need to wait for the other Commissions’ reports before we can make any estimates about how their target will change.
With all that done, below is a spreadsheet of the notional figures for all the proposed seats, carried out the same way – ready for people to crunch and experiment with in their own way.
Notional results for Provisional English Boundaries (excel) (csv)
I’ll add the caveat I provided last time I did this, these notionals are just the product of estimating how general election support is distributed throughout each seat, based up which are the stronger and weaker wards for each party in local elections, and then reallocating the wards to their new seats. There is little or no human judgement here – if just how the numbers stack up once they are all pumped into the spreadsheet, so if it is an area you know well you may very well have a better idea of how the party support in your area actually stacks up. It’s not a perfect system of projecting notional results, and sometimes later election results suggest individual seats were off, but one the whole it performs pretty well. That said, if you spot anything spectactularly odd do mention it – it may be an error.