Predicting Scotland

The overwhelming problem with a uniform-swing-based model like Electobot is that swing is not uniform across the country. While I’ve tried to mitigate this in various ways, the model is still reasonably dependent on different regions not having wildly different politics.

This is a problem when it comes to Scotland. After a long and intensely-fought independence referendum campaign, there’s not a lot of love for the Westminster-based parties north of the border, and this has had a significant effect on Scottish votes. Fortunately a few pollsters are performing research to find the vote distribution in Scotland, and I’ve recently enhanced Electobot to take advantage of this where possible.

To see the effect, here’s the result of 1000 simulated elections based on YouGov’s poll of the entire UK, fieldwork performed over the 29th and 30th of October (results CON 33, LAB 32, LD 7, UKIP 15, GRN 7).

fixed_ukwide_4_nov

Labour are comfortably the largest party in pretty much all simulated elections, even if they don’t manage to rule by themselves in more than 0.1%; LAB-LD is a feasible coalition in 69.7% of simulations and their average shortfall of 33.7 seats is ot too enormous.

But here’s what happens if we do another 1000 elections using the same input for English/Welsh/Northern Irish constituencies, only this time all Scottish constituencies use the results of YouGov’s poll of Scotland over the 27th-30th of October (SNP 43, LAB 27, CON 15, LD 4, UKIP 6, GRN 4).

fixed_ukandscot_bar_4_nov

Fairly similar graphs, but there’s a key difference: Labour’s lead drops noticeably and both the Lib Dems and the SNP get a boost (probably indicating that Labour’s loss to the SNP swings a lot of Lab-Lib marginals). Even though it doesn’t look like an enormous number of seats, because the results were already so finely balanced it makes a big difference to Labour’s chances of taking power: LAB-LD’s coalition chances drop to 59.6%, with the proportion of simulations where no obvious coalition could take power going up from 25.5% to 35.0%. In short, it seems that the dislike of Labour engendered since the referendum could have more of an impact than they’d bargained for.

Leave a comment