By Paul Whiteley, University of Essex and Harold D Clarke, University of Texas at Dallas

The polls showing voting intentions for the 2019 UK election up to 1st December show that the gap between Labour and the Conservatives is about 11%. Labour has benefited from the slow decline of Liberal Democrats and the Conservatives from the collapse of the Brexit Party vote. If we extrapolate this to seat shares, assuming a uniform swing across the country (which is a big assumption) then it shows a Conservative majority of 84 seats in Great Britain. At this stage of the 2017 election campaign, the Conservative lead over Labour in voting intentions was about 7% so the campaign dynamics we observed in 2017 are not the same as now.

The air was sucked out of the campaign to some extent by the YouGov MRP (multi-level regression post-stratification) poll published on 27th November. This technique, which uses a very large sample of more than 100,000 respondents was successful in predicting the 2017 general election. In the present election campaign, it predicts a 68-seat Conservative majority.

But as the YouGov team explained on their website, a problem with this method is that it spectacularly increases the margins of error for the seat predictions. They vary from a 328-seat majority for the Conservatives, which is almost a hung parliament, to 389 – which is a sizeable majority.

At least three things can go wrong in predicting the outcome of the election at this stage. They are differential turnout, tactical voting and late campaign dynamics in party support.

To probe turnout and tactical voting in more detail we can look at the YouGov poll for the Sunday Times published on 1st December. This put the Conservatives 9% ahead of Labour in voting intentions, predicting that the party will win 43% of the vote compared with Labour’s 34%. These figures are of course subject to the usual margin of error of about plus or minus 3%.

To estimate turnout, YouGov includes a question that has been used extensively in surveys in the past, asking respondents to indicate their likelihood of voting on a scale where zero means certain not to vote and ten means certain to vote. Respondents are inclined to exaggerate their likelihood of voting in surveys. This means that the percentage who score ten on this scale is a better guide to what might happen than the average score. In the YouGov survey, this was 64%. In an unpublished survey we conducted just before the 2017 election, which carried the same question, 69% of respondents said they were certain to vote, which was accurate in forecasting the actual turnout. This suggests that turnout is likely to fall this time.

Boris Johnson is targeting Labour Leave constituencies in the Midlands and North with the hope of turning lifelong Labour voters into Conservatives. Undoubtedly, many of these will be cross-pressured by their dislike of Jeremy Corbyn and Labour’s ambivalence on the Brexit issue on the one hand, and a traditional dislike of the Conservatives on the other. An easy way to solve their problem is for them not to vote at all. If so, this would reduce the chances of the Conservatives capturing these seats.

Who could benefit from tactical votes?

Tactical voting means people voting for a party other than their most preferred one with the aim of defeating their least preferred one. The crucial relationship here is between Labour and the Liberal Democrats. If large numbers of Liberal Democrats lend their votes to Labour in constituencies where the latter is the main challenger to the Conservatives, this could change things. The same point applies if Labour voters lend their votes to Jo Swinson’s party in constituencies where her party is the main challenger.

In the YouGov poll respondents were asked the following question: “Thinking about how you will vote at the election….are you voting for the party that is my first choice or a party that is not my first choice for tactical reasons?”. Surprisingly, some 20% of respondents said they would vote for tactical reasons. When asked a follow-up question about which party they were intending to stop by voting tactically, 33% said it was the Conservatives and only 16% said Labour. Some 19% said they were doing this to stop Brexit, which is exactly the same percentage as those who said they were doing it to deliver Brexit. Thus Labour has a clear advantage on tactical voting.

The final furlong can change everything

Yet another possible problem with YouGov’s MRP model is that it uses survey data gathered more than a fortnight before the election. If there are late-campaign dynamics in support for the various parties, the YouGov survey cannot pick this up. These late campaign trends could be consequential. Other surveys published since 1st December suggest that the Conservative lead over Labour is narrowing, so this is clearly an issue for the model.

Overall, the polls are still pointing to a Conservative win with a working majority on 12th December. But if they narrow further over the remaining days of the campaign, the prospects of another hung parliament will loom larger. In 2017, the gap between the two parties closed significantly in the final two weeks of the campaign as Labour support surged upward. It could do so again.

Paul Whiteley is Professor in the Department of Government at the University of Essex. Harold D Clarke is Ashbel Smith Professor at the School of Economic, Political and Policy Sciences at the University of Texas at Dallas