Our final forecast was for the Conservatives to win an expected 278 seats (or somewhere in the range of 252-305 seats), Labour to win 267 (240-293), the Scottish National Party 53 (47-57), and the Liberal Democrats 27 (21-33). The actual final results are 330 seats for the Conservatives, 232 for Labour, 56 for the SNP and just eight for the Lib Dems. Even though we took (or at least tried to take) into account the scale of historical poll misses in the U.K., our prediction intervals fell short of including the result for all of these parties except the SNP.
Labour’s performance wasn’t awful in England and Wales. But the party didn’t gain seats where it needed them. And it was completely destroyed in Scotland by the SNP, holding on to just one seat, in Edinburgh South. The Conservatives, conversely, gained a number of seats from the Liberal Democrats and held off a threat from UKIP, which will most likely wind up with just one seat.
Our pre-election forecast, put together in conjunction with electionforecast.co.uk, did project a narrow Conservative plurality. And it was about in line with or slightly more bullish on the Conservatives than other forecasts and betting markets.
However, that’s not much of an excuse. The forecast assigned too little of a chance to an outcome like this one, especially given that there have been significant polling errors in the U.K. before. It’s a good lesson as we begin to plan our coverage for the 2016 U.S. election.
We’re signing off for now — thanks to all of you for reading, and congratulations to David Cameron and the Tories.
To better understand Liberal Democrat voters, I recently traveled to the neighboring constituency of Sheffield Central. In 2010, the party campaigned relentlessly in Britain’s youngest constituency on reducing university tuition fees — only to vote to raise them after going into a coalition government with the Conservatives. Sometimes, voters find it hard to forget. I discovered a wealth of frustration about the Lib Dems, frustration that now explains in part their dramatically bad performance in this election.
The Tories are on track to outperform the polls by about 6 percentage points over Labour. But while this was a big miss for the pollsters, it’s far from unprecedented. In fact, there have been several almost-identical cases in the past.
In the October 1974 election, the Conservatives beat their polls by a net of 5.9 percentage points. In 1997, they did so by 5.7 percentage points, and in 2001, by 5.9 percentage points. The 1992 election was the biggest and most infamous miss of all: The Conservatives beat their polling average by 9.4 percentage points.
Some forecasters accounted for the Conservatives’ tendency to outperform their polls, but only around the margin. Betting markets were somewhat more confident about Tories’ chances than the polls alone implied. And our forecasts, put together by the team at electionforecast.co.uk, projected the Conservatives to win the popular vote by 1.6 percentage points instead of the tie the polls showed.
But neither betting markets nor the forecasts gave the Conservatives much chance at all of achieving an outright majority in Parliament — something that now appears probable.
So were the pre-election forecasts overconfident or just unlucky?
The simple answer is that they were overconfident — after all, errors like this have happened in the polls before.
But the problem was not solely with projecting the vote; it was also with projecting how it would translate into seats. The 90 percent confidence interval on our popular vote forecast had the Conservatives getting as much as 37 percent of the popular vote and Labour as little as 30 percent — within the range of the results it now looks like we’ll see.
Labour’s problems have turned out to be even worse than the national swings imply, however, because it’s gaining votes in areas where it doesn’t help them while losing votes in places where it hurts. Meanwhile, the Tories are gaining some additional seats from the Liberal Democrats, who have been completely obliterated. That means the error in the seat counts has been even greater than the error in the vote counts.
Still, forecasters almost certainly ought to have accounted for a greater possibility of an outcome like the one we saw. Polls, in the U.K. and in other places around the world, appear to be getting worse as it becomes more challenging to contact a representative sample of voters. That means forecasters need to be accounting for a greater margin of error.
Sinn Fein MPs typically don’t attend Parliament sessions and don’t vote. If they won five seats, that would effectively leave 645 votes, meaning 323 would be an effective majority. But UUP’s Elliott is expected to vote, and to attend Parliament, meaning there will be 646 votes, and 324 needed for a majority.
His win also lowers the Conservatives’ chance of holding an effective majority, and our forecast team at electionforecast.co.uk has accordingly adjusted its calculation of the probability of a Tory majority-led Parliament. This wouldn’t have mattered if our model’s forecast of 278 Conservative seats before returns came in had held steady. But Tory gains have kept pushing the party’s projected seat total higher. It’s now most likely to be 327 — and the Conservatives have a 92 percent chance of winning at least 324 seats and an effective majority. That’s up from about 1 in 500 about eight hours ago, just before votes started to be counted.
The following are the average ages for the candidates from the major parties running for election in 2015:
- Labour Party: 52
- Conservative Party: 50
- UKIP: 50
- Liberal Democrats: 47
- Green Party: 45
- Our pre-election forecast had the Conservatives at 278 seats. The exit poll showed them at 316 seats. We’re now projecting 320. We’re also now giving the Conservatives a 20 percent chance of taking an effective majority of 323 seats. That probability was vanishingly small when the polls closed — around 1 in 500.
- We expected Labour to get 267 seats. The exit polls showed 239. We now expect the party to get just 237 seats.
- We thought SNP would win 53 of 59 seats. The exit poll said they’d win 58. They’ve already lost two, and we now expect them to win 56 seats.
- We projected the Liberal Democrats would win 27 seats. The exit polls showed them taking just 10. We now expect they’ll take 13 seats.
Keep in mind that these remain a snapshot of our most likely projections, as of the latest returns. They’ve kept drifting toward the Conservatives and away from Labour and the Lib Dems, but we’ll continue to update as new returns come in.
It’s an important seat for the party — not only because it’s a win in an evening of relentless disappointment for Labour, but also because it’s a seat that was already more likely to be Labour anyway. Some 47.5 percent of residents in Ilford North are non-white (the median across all 650 constituencies is 5.2 percent), and historical voting data shows that ethnic minorities in Britain are more likely to vote Labour. You don’t have to look far for evidence of that. Just down the road in Ilford South, the non-white population is much higher — 75.8 percent. And the results announced there earlier look quite different to those in Ilford North. In Ilford South, Labour won with 64.0 percent of the vote, and the Conservatives only managed to get 25.9 percent.
Read more about how race affects voting outcomes.
On the whole, live interview polls taken during the final month of the campaign and released in the U.K. media leaned more toward the Conservatives than the Internet polls did. The Conservatives averaged a 2.1 percentage point lead over Labour in the live interview polls, while they trailed by 0.6 percentage points in the online surveys. That’s a difference of 2.7 percentage points, and it’s statistically significant.
Even if both live interview and non-probability Internet polls end up being off, the live telephone polls look closer to the truth. The exception was a poll from the online pollster SurveyMonkey, which wasn’t sponsored by U.K. media.
Polls at the start of the short campaign showed a major swing to SNP from the Great Britain-wide parties, particularly costing Labour. Our model shifted that swing slightly downward. But SNP didn’t regress to the mean. The party retained its polling strength, and even gained. The model shifted accordingly, though perhaps not far enough: It predicted SNP would win 53 of 59 Scottish seats, but now, with returns in, it expects the nationalist party to win all but two seats. The exceptions are Orkney and Shetland, which has been held by the Liberal Democrats, and Edinburgh South, which was held by Labour.
This chart shows how SNP’s probability of winning in just about every constituency grew during the short campaign:
I spoke to Carswell when I visited his constituency in April in an attempt to understand the constituency in the U.K. that was most likely to elect a UKIP MP. I discovered that UKIP voters were unlikely to be found in sufficiently high numbers outside of Clacton — which might explain tonight’s results.
The party so far has failed to take two target seats, Rotherham and Great Grimsby, from Labour. UKIP’s leader, Nigel Farage, won’t have his seat in South Thanet announced until the early hours of Friday (I’m hearing that exceptionally high turnout has further delayed the vote count) — but it’s unlikely that Farage could sustain his position after a sixth consecutive failed attempt to become a member of Parliament.
Already, sources are suggesting that Carswell might be the next leader for UKIP:
Just minutes before the first SNP setback in Scotland was announced, the forecasters we’re working with from electionforecast.co.uk calculated the probability of an SNP clean sweep. They had expected SNP to win 53 of 59 seats in Scotland before any returns came in. Early routs by SNP, by unexpectedly big margins, have reshaped our expectations dramatically. At 10:50 p.m. EDT, we were giving SNP a 75 percent chance of winning 58 to 59 seats — and an 18.7 percent chance of a clean sweep. The discrepancy was almost entirely because of Orkney and Shetland; 58 seats still looks likely for SNP.
I looked to Shaw’s election supplies for answers — not least because they claim to be “one of the oldest established businesses of any description in the world, having been founded in 1750 and trading continuously since then.” These days, they work with the Association of Electoral Administrators to provide signs, ballot boxes and voting screens to polling stations. Not including VAT, they charge 19 pounds (about $29) for 100 HB ballot pencils.
There were approximately 46 million people who could vote in the U.K. election today in around 50,000 polling stations across the country. If every member of the electorate showed up, voted and then stole the pencil on the way out, the country would get through 8.74 million pounds (about $13 million) worth of pencils. It’s a good thing we’re a kind of apathetic, kind of honest electorate.
That would give some people confidence that the final polls were right, but I wondered aloud on Twitter today whether the pollsters were herding together, meaning some pollsters were pushing their polls toward the average. The polls seemed to converge rapidly in the final days of the campaign. Take a look at this local regression of the average Conservative lead by day and the standard deviation around those results:
In no other election since 1979 was the standard deviation of polls taken over the last five days less than plus or minus 1.7 percentage points, and the average over the last five days from 1979 to 2010 was plus or minus 2.6 percentage points.
There may have been one exception: SurveyMonkey. The online pollster released data Thursday evening that showed a 6 percentage point Conservative lead over Labour. No other poll conducted in the final five days had either the Conservatives or Labour winning by more than 3 percentage points.
- Some voters in Hackney, my London borough, were shut out of voting because of IT problems, despite registering before the deadline.
- Overseas voters said they received ballots too late to return them by the deadline — meaning they were effectively disenfranchised for living abroad. An Electoral Commission spokeswoman didn’t exactly promise urgent action, telling the Guardian: “We will look carefully at the evidence shared with us on this when we consider what issues to raise in our statutory election report, which will be laid in the U.K. parliament in the summer.”
- Blue gloves were out in Scotland for election workers to wear while seeking to weed out cases of “personation” — someone filling out a ballot intended for someone else.
- About 90 ballots in Darlington erroneously omitted the UKIP candidate, David Hodgson.
- Bedfont arranged shuttle buses for voters who’d been sent the wrong polling-station location.
Black is not only the new Baby of the House, the unofficial title for the youngest MP, but she’s also the youngest since 1667, a very different time in British politics. That year, Christopher Monck, a few months after his 13th birthday, took his seat in the House of Commons, representing Devon. (It was the House of Commons of England; union with Scotland was still 40 years in the future.) It came as no surprise to Monck’s mother, who told her friends: “My son Kit is for the Long Parliament and the good old cause,” according to the History of Parliament Online.
But if the polls have a poor election, it won’t be the first time. In fact, it’s become harder to find an election in which the polls did all that well.
Consider what are probably the four highest-profile elections of the past year, at least from the standpoint of the U.S. and U.K. media:
- The final polls showed a close result in the Scottish independence referendum, with the “no” side projected to win by just 2 to 3 percentage points. In fact, “no” won by almost 11 percentage points.
- Although polls correctly implied that Republicans were favored to win the Senate in the 2014 U.S. midterms, they nevertheless significantly underestimated the GOP’s performance. Republicans’ margins over Democrats were about 4 points better than the polls in the average Senate race.
- Pre-election polls badly underestimated Likud’s performance in the Israeli legislative elections earlier this year, projecting the party to about 22 seats in the Knesset when it in fact won 30. (Exit polls on election night weren’t very good either.)
At least the polls got the 2012 U.S. presidential election right? Well, sort of. They correctly predicted President Obama to be re-elected. But Obama beat the final polling averages by about 3 points nationwide. Had the error run in the other direction, Mitt Romney would have won the popular vote and perhaps the Electoral College.
Perhaps it’s just been a run of bad luck. But there are lots of reasons to worry about the state of the polling industry. Voters are becoming harder to contact, especially on landline telephones. Online polls have become commonplace, but some eschew probability sampling, historically the bedrock of polling methodology. And in the U.S., some pollsters have been caught withholding results when they differ from other surveys, “herding” toward a false consensus about a race instead of behaving independently. There may be more difficult times ahead for the polling industry.