Polled Over

My final thoughts on the UK general election as the folks back home go to the polls today...

---

Hot on the heels of his barnstorming 71.4 percent Clinton victory prediction, the American pollster Nate Silver has come to titillate us once more with his range of scenarios for the UK general election today. Unbelievably, the election has become quite unpredictable of late, but one thing that is for sure is that the Conservatives have retained a lead of varying degrees throughout the campaign.Silver has provided a neat table of the latest polls:

Credit: Nate Silver, fivethirtyeight.com

Silver makes no prediction but sets out only a range of possible scenarios (once bitten, etc.), each of which he considers equally likely: a narrow Conservative victory, a Conservative landslide, and a hung parliament with the Tories as the largest party.

Generally, the poll that has shown the biggest Tory lead in the campaign has been ICM, and the smallest has been Survation. This is virtually corroborated by Silver's tabulation. ICM gives the Tories a 12-point lead (which would give Mrs May a majority of 100+ seats); Survation gives the Tories a one-point lead (which is hung parliament territory). But why the stark difference in possible outcomes? It's because of polling methodology.

Ever since the Great Opinion Poll Debacle of 2015, in which no one saw the Tory victory coming, the pollsters have been desperate to correct their models to take better account of crucial factors such as demographics and turnout. This has led to a considerable divergence in the methods of different organisations. ICM uses a turnout (TO) model to make its prediction, by which anticipated turnout among different age groups, social grades and regions is estimated using historical precedents. For example, in the TO model, over-65s vote in virtually twice the numbers of under-25s. Survation, meanwhile, uses the self-reporting (SR) model, by which the expected turnout is weighted according to voters' self-reported likelihood to vote on a scale of 1 to 10. The SR model thus suffers from a considerable potential drawback – I say "potential", because we won't know until the results are in.

Voters in different demographics have a propensity to over-egg their own likelihood to vote. The crux of the matter is that this is especially a problem for the youth vote. And the youth vote will make or break Labour. If the 18-24 bracket turns out in sufficient numbers, it could put Jeremy Corbyn in Downing Street. And this explains the closeness of the Survation polls. In fact, the SR model suggests young people will turn out in the same numbers as older people. If this were to be repeated at the ballot box, it would be astounding because that never happens. It is unlikely. The 18-24 bracket is particularly prone to apathy, and we have seen this in elections ever since Corbyn became Labour leader. Even if the youth turnout is high, it simply won't match the over-65 vote. When you consider the TO model, voters' own assessment of their likelihood to vote is discarded in favour of what history tells us, and that implies young people won't turn out. Ultimately, the ICM poll is likely a better reflection of what's actually going to happen.

One of the phenomena that contributed to the unexpected Tory majority in 2015 was the "shy Tory" effect, where people didn't tell pollsters they were voting Tory but did so anyway in the privacy of the ballot box. This led to an undersampling of Tory voters of roughly 2.5 percent. Pollsters have tried to correct for this by increasing sampling in traditionally-Conservative demographics. Time will tell if they've hit the sweet spot or overshot the mark. If we start from the assumption that the TO model is a better predictor of turnout than the SR model, then we can allow for a certain amount of "oversampling" of Tory voters and it still gives Mrs May a healthy majority. Even if the overshoot is 2.5 percent off the ICM poll, it still puts the Tory lead slap-bang in the middle of what the other polls are saying, leading to a majority of roughly 50.

This could all transpire to be a perfect storm of misfortune for Labour. Watch out for the youth turnout. It will be a clear indication of Labour's performance. My gut feeling is that 18-24 voters will do what they always do, and that potential oversampling of Conservative voters will not be enough to overturn the Conservative lead. Jeremy Corbyn's good campaign will be just that: a good, losing campaign.

---

By:

The Imperial Orange,

8th June 2017