UK Pollsters, Forecasters Looking for a Win in Snap Election

Will polls of the 2017 United Kingdom "snap" election be better than they were last time around?

George Elliott Morris

5 minute read


This is the second of a five part series on the 2017 United Kingdom general election. Join us every Tuesday until election day (June 8th) for parts 3, 4, and 5!



I began the previous piece by saying “If polls are right, the United Kingdom’s Conservative Party is on a path to a big win in June’s general election.” This begs an obvious question: will the polls be right? (P.S. We now have a handy poll tracker that makes it easy for you to follow along!)


In the past, polls in the United Kingdom have been alright. In fact, if I were giving it a letter grade, I would probably give them a B+ with a comment reading “pretty OK.” The polls have been just that — at least, on average. According to data from professors of political science Will Jennings (at Southampton University) and Christopher Wlezien (at the University of Texas at Austin) the average error in the last week’s polled margin of loss/victory between Labour and Conservatives since 1979 is 4.4%. Not as great as the United States’s 2.5% average error, yet barely better than the 4.8% average error across the Channel in France.

Year Poll Margin (%) Election (%) Error(%)
Error in UK General Election Polls
Source: Christopher Wlezien and Will Jennings
1979 -2.7 -7.0 4.3
1983 -19.1 -14.8 4.3
1987 -10.0 -11.4 1.4
1992 2.3 -7.5 10.1
1997 18.1 12.5 5.6
2001 15.0 9.0 6.0
2005 4.8 2.8 2.0
2010 -7.2 -7.1 0.1
2015 -1.0 -6.6 5.6
2017 -18 ? ?
Average     4.4%
95% Confidence     12%


Of course, polling error could be bigger than it has been in the past. Indeed, that was the story in the runoff round of the France presidential election, where a 10 point error occurred in Macron’s favor.

Accounting for the small sample size of UK elections (9), we can expect the error in Labour’s margin of defeat to be as big as 12 percentage points in either direction 95% of the time. That means we can expect that an error big enough (17%) to deliver a Corbyn victory to happen less than 0.5% of the time, all else being equal. But, because there are other parties in the election (the Liberal Democrats, the UK Independence Party, the Scottish National Party, and the Greens) we should treat that 1% with more uncertainty, perhaps closer to 4 or 5%. Indeed, if I were forecasting the race (and I even wrote a mock model!) that’s where I’d peg the absolute chance of a Labour parliamentary victory. The chance that neither party would get a majority.

At any rate, after adjusting expectations for past error in UK polling, Labour has almost no chance to pull off a national popular vote upset. The chance of winning enough seats for the majority in Parliament that upset would imply is even less. Yet, what if polls are wrong? What exactly could make that happen? Will opinions be better measured than they were in the 2015 “disaster?”

As YouGov’s Anthony Wells wrote after the 2015 surprise, polls simply did not contain representative samples of the population.

The age distribution within bands was off, the geographical distribution of the vote was wrong (polls underestimated Tory support more in the South East and East). Most importantly in my view, polling samples contained far too many people who vote, particularly among younger people – presumably because they contain people too engaged and interested in politics. Anthonty Wells

There is reason to believe that most pollsters have fixed these problems. Most now weight their samples (a process by which the demographic makeup of a sample is adjusted to match the desired population) to match the UK as a whole, adjusting for age, gender, region, and even socioeconomic status.

Aside from “fixing” their errors, pollsters and forecasters have also been bolstered by recent events that point to the same outcomes their measures do. In local elections, we saw conservatives gain while Labour and UKIP lost — a very similar picture to the one being painted by Chris Hanretty’s electionforecast.co.uk.

Credit: The Telegraph


To be clear, though, outcomes of local elections have seldom correlated well with eventual national outcomes. But, they almost always get the leader correct. This is just another data point which points towards accurate polls at this point in time.

There are, of course, some yet unanswered questions regarding polling accuracy. In 2015, it was the third parties that caused polls to be off. This year, UKIP voters left with no candidates in some constituencies may also float away to unpredictable candidates. A recent Telegraph/ORB poll of UKIP voters across the UK suggests that 50% will go to Theresa May’s Conservative party, while 10% will go to Labour and 2% will vote LibDem. This is just a national poll, though, asking a question that is very specific seat-by-seat. I’d say this presents the most uncertainty in the 2017 UK election.

All in all, we can expect polls to be heavily scrutinized this time around, as they certainly should be. Most of the evidence points towards better polls of vote intention than we would otherwise expect, though there still is potential for a big 12 percent polling error. Of course, that would still leave Labour short by 5% based on current polling.

Ultimately, only time will tell.


… . .

Come back next time for part three of five in a series on the 2017 United Kingdom “snap” election!

Thanks for reading today everyone. Tune in to my twitter for updates and make sure you sign up for my newsletter to get notifications of recent posts.

-Elliott


comments powered by Disqus