Uncertainty and the Election

Nothing momentous to relate today, but with a few days to go it looks as if the polls have levelled off, so public opinion may have found a new equilibrium.

Just what that new level is remains the subject of debate. Bottom line, all scientific results depend on your assumptions, and in this case they make an important difference. I've split the confidence band (nominally, 95% of polls are expected to lie within the shaded regions on this graphic) into two bits for both Labour and the Conservatives.

I've based the light coloured band on polling by ICM and ComRes, who reckon that youth turnout will not be much different to 2015. The dark band is everyone else, who generally think that the increased number of 18-24 year-olds saying they'll definitely vote will hold up on Thursday.

---

So, what's going to happen? The short answer is 'we don't know'. As I've said before, turning vote shares into seats is really tricky, and there are good reasons to think that this time any kind of 'uniform swing' approach would be risky: there have been big changes in which demographics support which party. What will be crucial is how this affects marginal constituencies; it's not clear to me how much benefit the Conservatives will get from picking up more votes from less wealthy people in these crucial seats, if their base is weakened.

That said, what the data says depends on your assumptions.

If youth turnout is very high...

...there's an outside chance of an irretrievably hung parliament (i.e. one where the Conservatives can't cobble together a working majority with the DUP) and a reasonable prediction would be a very small majority for the Conservatives, no bigger than what they have at the moment.

This depends also on the extra turnout being spread evenly, not concentrated in Labour safe seats (this is something I'd worry about even if all the pollsters agreed on their predictions).

If youth turnout is not high...

...a reasonable prediction from the numbers would be a Conservative majority of about 50. Not the landslide that was anticipated, but a very comfortable majority (more than Labour had in 2005, for reference). However, being scrupulously fair, a swing/polling error similar to the one needed in the previous scenario for a hung parliament (but in the other direction and assuming poor youth turnout) would result in a majority of a little over 100.