General election 2019: Can we trust the polls?
We realise British politics can all feel a bit much sometimes. But this upcoming election is pretty important, so we're going to be publishing this regular briefing.
This campaign will be a barrage of information. We want to cut through the headlines and tell you what really need to know.
This is an extended version of the BBC's new election newsletter, Outside The Box. If you'd like to receive the newsletter by email, sign up here (UK users only).
We also want to hear from you about the issues you want us to cover: click here to send us your questions.
My first long-term job at the BBC was as Brexit researcher in the Analysis & Research (A&R) team. I loved it - spending days reading briefings and poring over statistics, helping brief reporters in Westminster and beyond. My boss there was Peter Barnes. His full title is Senior Elections and Political Analyst. He's a real expert on polling - you can keep up to date with his poll tracker throughout the campaign. The industry has a bit of a bad reputation these days, so I asked him what we know, and to what extent we can trust the polls - because you will be seeing a lot of them over the next few weeks.
Q: Hi Peter. Who is leading in the polls?
A: The Conservatives are definitely ahead in the polls since Boris Johnson became prime minister. It's been pretty clear they've been ahead of all the other parties and particularly ahead of Labour.
Having said that, there is a massive difference between the different polling companies, so in some polls the Conservative lead is fairly small - certainly in single digits - whereas with other polling companies they're more than 15 points ahead.
The decision by the Brexit Party not to contest Conservative-held seats should give Johnson a further boost in national polls, though the impact on seats may be smaller.
Q: In 2017, Labour were really far behind in the polls but ended up doing better... what happened?
A: For practically the first time ever, there was a massive shift in the polls during the campaign period. It used to be widely thought that the campaign made no difference at all to how people voted, and that the parties' campaigns would effectively cancel each other out. So on polling day itself, the votes would basically go as the polls were at the beginning of the campaign.
But 2017 proved that isn't true because there was a massive shift during the seven-week period of the election campaign. Labour went from being perhaps 20 points behind to being only a couple of points behind on election day.
Q: What's happening with the Lib Dems and the Brexit Party?
A: It's always hard to judge from opinion polls what the result of an election is going to be, but it's particularly hard of the moment, because we've got four parties across Britain polling in double digits. That's been the case in Scotland for some time, so the party system there has become very complicated. But in England it has tended to only be three parties who are strong at any one time, but now we've got four.
We've got the Liberal Democrats who have had a massive resurgence this year, really since the time when Brexit was first delayed, then they did really well in the local elections and the European Parliament elections.
Since Jo Swinson became leader they seem to have had a slightly higher profile, so have been up in some polls even ahead of Labour - although since the election was confirmed they seem to have fallen back a little.
Then there's the Brexit Party. They also did incredibly well at the time of the European Parliament elections. They have since declined quite a lot, but are still polling somewhere in the low teens. So we've got this very volatile situation where there are four parties and it makes it impossible to know exactly what will happen at the election.
Q: A lot of people will just say 'the polls have got it wrong so many times, we can't trust them'. What would you say to that?
A: It is very difficult for the polls at this stage of an election campaign to accurately predict what the result is. The 2017 case shows voters' opinions can change so much in the last few weeks, so what the polls say now won't necessarily be a good guide to where the election is going to be in a few weeks' time.
Having said that, the polls did get quite close to the right result in 2017. By the end of the campaign they were showing Labour and the Conservative Party fairly close together. They were also showing that UKIP had fallen way back compared to where they were in 2015, and the Lib Dems doing very badly. Both of those things came true, and the Conservatives and Labour were indeed close.
The polls slightly underestimated Labour, so the result was still a surprise to most people. But it wasn't that far away from the outcome of the election. The polls were much, much worse in 2010, where the polls really were seriously bad. They were pretty bad in 2015.
Although it's true that the polls have often given us a misleading picture of the overall result, in some recent elections they've actually been relatively accurate. In the European Parliament election again, they didn't quite get it exactly right. But they did show the Brexit party was doing incredibly well and that the Lib Dems are doing incredibly well. They basically got the main headlines of that election right.
So sure, don't expect the polls to tell you exactly what's going to happen. But they're probably still the best guide we've got, they're relatively speaking not too bad at telling you what approximately is going on.
Q: It seems like people change their mind more often these days. Do we have evidence for that?
A: Yes. There's been a recent academic study by the British Election Study, a group of academics based in Manchester and Oxford. They looked through data that they collect from voters in each election and found that between 2010 and 2017, nearly half of voters switched parties. So voters are very, very volatile now.
Perhaps half the voting population are what we call 'floating voters', people who are prepared to vote for more than one party. That is massively different from the past. Only 13% of voters switched between two elections in the 1960s. So there are just a lot more voters who are prepared to change party.
Q: Do we know the reasons why?
A: We have some ideas. One reason is this long-term trend of people just not feeling closely identified with a party. Party membership on the whole has dropped, although the Labour Party does currently have quite a lot of members. But more people don't feel strongly associated. They don't feel they have a natural home.
The other side of it is more short term, it's what the academics call 'electoral shocks', where very dramatic events have changed people's motivation for voting. The financial crisis, the Scottish referendum in 2014, and most of all, the 2016 EU referendum. Those political shocks changed the reasons that are guiding how people vote, and have made them more inclined to switch from one party to another.
Q: OK, and just finally, you've worked on lots of elections. How does this compare to previous ones?
A: Well David Butler, who is the doyen of election studies, he's been in the business since just after the Second World War. He said he's never been so perplexed by what's going on in politics. He's never found it so difficult to predict the outcome of the election. I think to some extent that has got to be true.
This issue of voters becoming more volatile over time does mean it's much harder to be confident about what the result of an election is going to be. Even if polls get it entirely correct in terms of the share of the vote, that still doesn't really tell us who's going to win, because knowing what the share of the vote is doesn't necessarily tell you how the seats are going to be distributed, especially where we've got this situation with lots of parties doing well in the polls.
Translating vote shares into number of seats in the House of Commons is a pretty difficult business, and it's never going to be totally accurate.
Q: To end on a positive note... the polls aren't completely useless and we should pay some attention to them?
A: Yes, I think that would be my summary. Don't think the polls are going to tell you exactly the right figures. But it will be surprising if the polls are completely upside down.
If you'd like to receive Outside The Box by email, sign up here (UK users only).