British politics and what we’ve learned after the 2017 general election

Posted on July 6, 2017 by The Constitution Unit

Last month’s general election delivered the latest in a series of political surprises, with the Conservatives falling short of a majority when many had anticipated they would win a landslide. On 21 June the Constitution Unit hosted a panel of election experts consisting of YouGov’s Joe Twyman and academics Justin Fisher, Jennifer Hudson, Philip Cowley and Alan Renwick to reflect on what happened. Fionnuala Ní Mhuilleoir reports.

Although we have become used to political upsets in recent years the outcome of the 8 June election nonetheless came as a surprise to many, including the Prime Minister, who saw her majority disappear when she had hoped to increase it substantially. How did this happen? How did the Conservatives manage to lose the massive lead they held at the start of the campaign, and Labour out-perform all expectations? How did the pollsters do after they had failed to call the 2015 election correctly? And what does the result mean for the government’s position in the new parliament, and for Brexit and beyond? These questions were all discussed at a Constitution Unit seminar held on 21 June, chaired by the Unit’s Director Professor Meg Russell. The panel included YouGov’s Joe Twyman, Professor Justin Fisher from Brunel University and Professor Philip Cowley from Queen Mary University of London. Dr Jennifer Hudson and Dr Alan Renwick from the Constitution Unit completed the line-up.

Joe Twyman

Joe Twyman opened the seminar with a brief post mortem on YouGov’s 2015 general election polling, which had predicted that the Conservatives would be the largest party in a hung parliament. The Conservatives went on to win 330 seats, securing a small but workable majority. YouGov subsequently identified three problems in the 2015 polling process: the samples used by YouGov and other polling companies to measure voting intention were not representative; figuring out whether people will turn out to vote is challenging; seat estimation across 650 constituencies is inherently very difficult.

Twyman then described how YouGov has responded to these issues. First, it has invested heavily in targeted recruitment, spending more than £100,000 in the last year to identify and recruit the types of people who were underrepresented in YouGov samples between 2010 and 2015, particularly those who were not interested in politics. Second, YouGov has updated how it analyses turnout. Thirdly, it has also developed a new seat estimation model.

This seat model, as is now well known, correctly predicted a hung parliament. In the run up to June 8 YouGov faced trenchant criticism, both from established commentators and on social media. Twyman reflected on Paul Krugman’s statement after the US election results that economists and commentators ‘truly didn’t understand the country we live in’. Through the efforts of YouGov, according to Twyman, we do now understand the country we live in a little better.

Justin Fisher

Justin Fisher began his contribution by drawing the audience’s attention to how important lead time ahead of an election is for party campaigns. The national and constituency campaigns have merged, with national campaigns now supporting the constituency effort. Lead time gives parties more time to plan targeting, information distribution, spending and fundraising. In a normal election campaign the critical period is the six-to-nine months before the poll.

The snap election left no opportunity for such advance campaigning. One implication is that this is likely to have been a much less expensive campaign than usual. Another is a shift in emphasis in campaigning techniques from direct mail (which requires lead time) to face-to-face campaigning and e-campaigning, which require much less preparation time. Fisher stressed that the evidence needed to confirm these expectations is still being collected.

Though the 2017 election may have accelerated the shift to e-campaigning, Fisher argued that campaigning techniques were partly heading that direction regardless. He also warned of what he called e-campaigning myths. He debunked the myth that micro-targeting of voters had only just been invented: parties have gathered data from phone calls and indirect mail for years. E-campaigning, therefore, represents evolution rather than a revolution. A further myth is the claim that because parties are using e-campaigning it must be effective. In 2015, research found it to be electorally effective, but less so than face-to-face campaigning. This has yet to be examined for the 2017 election.

Continue reading →

EU referendum forecast update: probability of Remain win falls for second consecutive week

Posted on June 9, 2016 by The Constitution Unit

Stephen Fisher and Alan Renwick have developed a method for forecasting the outcome of the EU referendum based on current vote intention polling and analysis of opinion polling from previous referendums in the UK and across Europe. For the second consecutive week the probability of a Remain win has fallen and now stands at 68 per cent.

Remain have continued their gentle slide in the polls. Last week our polling average saw Remain drop from 53 per cent to 52 per cent. Now they are on 51 per cent after setting aside don’t knows.

The further one-point drop in our polling average has produced a one-point drop in the forecast share of the vote for Remain, from 54 per cent to 53 per cent.

The 95 per cent prediction interval is still ±12 points. We are now forecasting that Remain will win between 40 per cent and 65 per cent of the vote.

The probability that Remain will win the referendum is now down to 68 per cent.

The method behind this forecast is based on the historical experience of referendum polls and referendum outcomes in the UK and on the EU elsewhere, as discussed here.

Our polling average is constructed by taking the most recent poll from each company within the last two weeks. If a company uses both phone and online modes then both the most recent phone poll and most recent online poll are used. This applies just to ICM this week. The current average is based on the results of eight polls from seven companies, of which three were conducted by phone and five online. All polls are adjusted to account for the tendency for phone polls to be more favourable to Remain. This is done by adding 2.2 to the Remain share for online polls and subtracting the same amount for phone polls.

This post was originally published on Elections Etc. and is re-posted with permission.

About the authors

Dr Stephen Fisher is an Associate Professor in Political Sociology and the Fellow and Tutor in Politics at Trinity College, Oxford.

Dr Alan Renwick is the Deputy Director of the Constitution Unit.

EU referendum forecast update: 79 per cent chance of Remain winning

Posted on May 25, 2016 by The Constitution Unit

Stephen Fisher and Alan Renwick have developed a method for forecasting the outcome of the EU referendum based on current vote intention polling and analysis of opinion polling from previous referendums in the UK and across Europe. Since the last update two weeks ago the probability of a Remain win has increased from 72 per cent to 79 per cent.

There has been a small shift towards Remain in the polls over the last two weeks. Excluding don’t knows, our polling average for Remain has moved from 52 per cent on 10th May to 53 per cent now. This figure is based on the most recent polls from each of seven companies: one from each but two from ICM (one by phone and one conducted online). The Remain share has been adjusted down by 2.15 points for telephone polls and up by the same amount for online polls to account for the relatively stable gap between these different methods in the levels of support they tend to give the two sides.

Using the historical experience of referendum polls and referendum outcomes in the UK and on the EU elsewhere, as discussed here, our latest forecast is for Remain to win 55 per cent of the vote in a month’s time. The 95 per cent prediction interval surrounding this estimate has narrowed very slightly to ±12.5 points. So we are forecasting that Remain will win between 43 per cent and 68 per cent of the vote.

Values closer to the middle of this range are more likely. Overall the probability that the Remain vote will be larger than the Leave vote is now 79 per cent, up from 72 per cent two weeks ago.

This post was originally published on Elections Etc. and is re-posted with permission.

About the authors

Dr Stephen Fisher is an Associate Professor in Political Sociology and the Fellow and Tutor in Politics at Trinity College, Oxford.

Dr Alan Renwick is the Deputy Director of the Constitution Unit.

Introducing our EU referendum forecast

Posted on March 14, 2016 by The Constitution Unit

Stephen Fisher and Alan Renwick have developed a method for forecasting the outcome of the EU referendum based on current vote intention polling and analysis of opinion polling from previous referendums in the UK and around the world. In this post they introduce their model and present their initial forecast.

The UK will have a referendum on whether to remain in or leave the EU on 23 June 2016. We have developed a method for forecasting the outcome based on current vote intention polls and analysis of opinion polls from previous referendums from the UK and across the world.

In the run-up to the 2014 Scottish Independence referendum we wrote about how across various referendums there has been a tendency for the eventual vote for change to be lower than it was in polls over one-month out from referendum day, and for even final polls to overstate support for change. But these observations were based on relatively small samples of referendums.

Now, as part of a broader academic project, we are analysing over 1600 polls from 283 referendums in 41 democracies. For the purposes of forecasting the Brexit referendum outcome, we have used just referendums in the UK or on the EU. This still includes 848 polls from 45 referendums, and for the model specification we draw on the lessons from our analysis of the broader data set.

Continue reading →