Labour Usually Falls Short Of Its Polling Average

Labour Usually Falls Short Of Its Polling Average

This is the fourth of a five part series on the 2017 United Kingdom general election. Join us next Tuesday for a final update before the UK election!



The United Kingdom General Election is (as my British friends tell me) absolutely barmy. Indeed, the contest is shaping up to be a crazy one. The Conservatives have fallen from a 16 percentage points polling lead in early April to just 5.3% in my latest polling average — an absolutely huge decrease. Combined with the large amount of uncertainty in UK elections this makes any prediction even more fickle. However, if we take a close look at past election results, one thing stands out: the Labour party has a history of doing worse on election day than public opinion polls indicate. In a contest where approval ratings forecast point towards precisely this scenario, it’s worth considering the possibility of Conservatives beating their expectations.


It’s worth noting that UK elections have a long, storied history of being volatile in their late stages (though stable compared to US polling). By my count, polls move 2 percent or so on average in the final week of polls. Often movement is much less, however — yet in some years it is much more. This year is one of those years. Labour’s position in public opinion polls has moved roughly five percent in the past week, setting a trajectory for a very tight outcome on June 8th. Some forecasters are already picking up on that closeness: YouGov is publishing a forecast that has Conservatives up four in the final vote share, but just falling short of an absolute majority in Parliament (317 seats instead of the requisite 326). Of course, there is possible error in the way forecasters move from a polling average to a seat projection.

As I — and the developers of that forecast — have said, a so-called “hung parliament” (where no party gets a majority of seats) in no sure thing. The wide range of polling error means there is a decent chance even Labour could win a large amount of seats.

But, Labour could also lose the election.

In fact, just because “anything could happen” does not mean we cannot provide some insight into what may. Using past election results, we see some patterns emerge that provide inklings of insight into which direction polls may miss the outcome.

Labour Underperforms Its Polling Average
Labour Margin (%)
Year Polling Election Error (%)
Source: Will Jennings and Christopher Wlezien
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 -4.6 ? ?
Average Raw Error -0.3%
Average Absolute Error 4.4%
95% Confidence 12%

According to this data — which takes an average of polls in the final week of each UK election since 1979 and compares them with the actual election results — Labour does slightly worse than polling says, underperforming on average by 0.3%. Usual caveats apply here: absolute error remains 4.4% with a 95% confidence interval of 12%. It is thus both true that polling says there are a wide range of outcomes on June 8 and that Labour has a slight disadvantage in the expectations game.

This comparison of past results is just a one-off, however. I mean to say that a 0.3% underperformance is certainly no reason to think that Labour is going to end up being beat by 10% on election day. However, the data here do say that it’s more likely that Corbyn’s party does worse next Thursday than it is that they overtake May’s Conservatives.

The idea that Labour might underperform expectations clearly cuts against what we could extrapolate from their recent comeback in opinion numbers. But, those hopeful of a Labour upset don’t have to look far for encouragement.


In the article I wrote last week I explored what high dissatisfaction with Jeremy Corbyn could mean on election day. According to the model I built — which takes into account this dissatisfaction plus Labour’s margin of loss last go-around — I estimated that the Conservatives would win by 13.5 percentage points after all ballots are counted. Due to the high confidence I have in that model, I would say it’s very unlikely that Labour could win the election. However, there is some evidence that Theresa May’s approval rating has decreased in the past weeks. If the new numbers are correct, the forecast lead for May’s conservatives could drop near 6-7 percentage points or so — nearly at edges of the prediction’s margin of error.


Whether or not these exact estimates end up being correct, one thing is clear: Labour has increased its probability of victory quite a bit from last week. But, polling says, there is reason to be skeptical of Corbyn scoring an upset victory. Though, anything is possible.


… . .

Come back next time for the final part in my series on the 2017 United Kingdom general 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