On the morning of the U.S. presidential election, we pointed out that there were three scenarios for what might transpire that night, each of which were about equally likely. In Scenario No. 1, the polls would be spot-on; Hillary Clinton would win narrowly, with a 3-to-4 percentage point popular victory and somewhere on the order of 300 electoral votes. In Scenario No. 2, Clinton would outperform her polls, leading to a near-landslide victory and possible wins in states such as Arizona and Georgia which had traditionally favored Republicans. And in Scenario No. 3, Donald Trump would beat his polls; because the Electoral College favored Trump, even a small polling error in his favor would probably be enough to make him president. Scenario No. 3 is the one that transpired, but it wasn’t any more or less likely than the other two.
There’s a similar set of scenarios in play for the U.K. general election, which is taking place on Thursday. Conservatives lead by an average of 6 to 7 percentage points in recent polls, but with a wide range — anywhere from a 3-point Labour lead in one survey to a 13-point Conservative lead in another. Some of these polls were conducted partly or wholly before the terror attack in London on Saturday, but it’s not clear what effect the attack has had, with polls differing on whether the Conservative lead is rebounding and expanding slightly or instead continuing to narrow.
We don’t have a model of this election — but in our view, there’s no way around the fact that uncertainty is high and that nobody should be surprised about the outcome unless perhaps Labour wins an outright majority of seats. U.K. polls have not historically been very accurate. And pundit attempts to outguess the polls have often been even worse. (In 2016, pundits and betting markets were notoriously confident that Britain would vote to remain within the European Union even when polls showed a nearly even race.) However, if one takes the polling average but assumes that the error is as high as it has been historically, then it turns out that each of these three outcomes are roughly as likely as one another:
Scenario No. 1: Narrow-ish Conservative majority
In this case, the polling average is fairly accurate (although some individual polls will unavoidably be off). May wins by 5 to 9 percentage points, close to the 6.5-point margin that Conservatives won by in 2015. At the higher end of this range, Conservatives might gain one or two dozen seats in Parliament from the 330 they had (there are 650 seats in total). At the narrower end, they might just barely hang onto their majority.
Either outcome would be disappointing relative to when May called the election in April — when Conservatives were ahead by about 17 points on average and possibly headed for a 400-seat majority — and wouldn’t speak highly of her political skills. But it would also be something of a relief given how much polls have tightened since then and still possibly give them their largest majority since 1987.
Scenario No. 2: Conservative landslide
Since Conservatives lead by 6 to 7 percentage points on average, and since U.K. polls have missed by an average of about 4 points in the past, it’s not at all hard to imagine them winning the popular vote by double digits. Indeed, Conservatives have often outperformed their polls in recent U.K. elections, such as in 2015 when polls implied a hung parliament and they instead won a majority. On Saturday, we made a long argument as to why this doesn’t necessarily imply that they will beat their polls again. The gist of it is that pollsters are seeking to correct for their past errors — in some cases, by applying turnout models that shift the polls by several percentage points in Tories’ favor — so it may be a mistake to apply a mental adjustment on top of the one that pollsters are already making.
But a Conservative overperformance is certainly possible if Labour’s youth turnout doesn’t materialize, if undecided voters worry about the unpopular Labour leader Jeremy Corbyn running the government, or if the terror attack pushes voters to May. Such an outcome might yield somewhere in the neighborhood of 375 seats in Parliament for Conservatives — perhaps more if Labour holds ground in the cities but collapses in working-class areas, as happened to Democrats in the 2016 U.S. election.
Scenario No. 3: Conservatives lose their majority
To be precise, I mean that Conservatives win fewer than 326 seats; we’re making no predictions about whether they’d then find a coalition partner to form their next government or if there would be some sort of Corbyn-led government instead. Betting markets as of Wednesday night imply there’s only about a 15 percent chance of Conservatives failing to win an outright majority.
If you take the polls at face value, however, the chances are surely higher than that. If Labour beats the final polling average by only 1 or 2 percentage points, both sides will start having to sweat out the results from individual constituencies. And if they beat their polls by much more than that, May’s majority is probably toast. If this happens, the adjustments that pollsters made to discount Labour turnout will have proven to be counterproductive, and the lesson for pollsters will be to trust their data instead of making too many presumptions about who is likely to vote. The lesson for May, meanwhile, would be never to call a “snap” election again in a country where public opinion can shift so rapidly.