History Says Brexit Will Be Beaten

This piece was originally published on Peter Kellner's The Politics Counter blog.

The "Brexit" camp should enjoy its current slight bounce in the polls, for it may not last. If history is any guide, then "Remain" is still heading for victory on June 23. Past referendums in Britain have tended to produce a late move to the status quo. The record from six such contests in the past four decades is striking:

1975 U.K.-wide referendum on the Common Market

Gallup's polls during the final five weeks of the campaign showed that support for leaving the Common Market peaked point with two weeks to go — though the lead for "staying in Europe" was still a massive 28 percent. Gallup's final eve-of-referendum poll put the lead for the status quo at 36 percent (68-32 percent). The result: a 34.4 percent gap (67.2 percent in, 32.8 percent out).

1979 Scottish referendum on devolution

Just over two weeks before the vote, Mori reported a 28 percent lead for the pro-devolution camp (64-36 percent). With one week to go, the margin was 20 percent (60-40 percent). The swing to the status quo accelerated in the final week. Mori's eve-of-referendum predicted a 50-50 percent outcome. The result: a narrow 51.6-48.4 percent lead for devolution—a margin too small to reach the winning line set by Parliament, which insisted that 40 percent of Scotland's entire electorate should support devolution; on the day only 33 percent did so.

1979 Welsh referendum on devolution

With three weeks to go, an Abacus/BBC survey said that 58 percent would vote for the status quo and against devolution. By referendum week, that figure had jumped to 75 percent. On the day, 79.7 percent of voters opted to reject a Welsh assembly.

1997 Welsh referendum on devolution

Three weeks before the vote, Beaufort Research showed a majority of almost two-to-one for devolution. The appetite for change receded significantly towards the end of the campaign; but an eve-of-referendum poll by NOP still indicated a 12 percent lead for devolution (56-44 percent). In the event, Wales did vote to set up its new Assembly, but by the narrowest of margins: 50.3-49.7 percent. Support for the status quo had jumped by 16 points in the final three weeks.

2011 U.K. referendum on the voting system

Three weeks before the vote, telephone polls showed an average 12 percent lead for the status quo (46-44 percent), while online polls showed the two sides neck-and-neck (49 percent for the Alternative Vote, 51 percent for first-past-the-post). With ten days to go, support for the status quo had climbed to 60 percent in telephone polls and 57 percent in online polls. The final online polls showed a 22 percent lead for first-past-the-post (61-39 percent), while the final telephone polls put the margin at 34 percent (67-33 percent). The result: 67.9 percent for the status quo, and 32.1 percent for change—a victory margin of 35.8 percent.

2014 Scottish referendum on independence

The u-turn pattern here was even more dramatic than in the 1975 referendum on the Common Market. For much of the spring and summer, ahead of the September 18 referendum, support for independence hovered around 40 percent. Then, during August and early September, the appetite for independence grew, and with around two weeks to go, three polls by different companies showed the race neck-and-neck.

However, in the final ten days, banks and supermarkets warned that independence would mean dearer mortgages and groceries, history reasserted itself and support for the status quo revived. The eve-of-referendum polls showed an average 5 percent lead for Scotland remaining in the U.K. An on-the-day survey by YouGov recorded a further shift away from independence. The final margin of victory for Better Together was 10.6 percent (55.3-44.7 percent).

The one exception

Alert readers will have noticed one glaring omission from this list: Scotland's devolution referendum in 1997. The campaign polls showed little movement in opinion, and all were close to the final result: 74.3 percent for devolution, 25.7 percent against. How come?

My answer is that this campaign was qualitatively different from the others, in that the proposal for self-government had been hammered out over a number of years. Virtually every major player in Scottish politics, business, the trade union movement, the churches and civil society generally had been involved in the process, and all bought into the deal. The only significant exception was the Conservative Party; but as they had lost all their Scottish seats in the general election four months earlier, their opposition had little impact.

In short, the 1997 Scottish referendum was, in effect, a mechanism for ratifying a national consensus, not a means of resolving a major national dispute. In these circumstances, we should not be surprised by the lack of a late-campaign swing to the status quo.

Plainly the current referendum on the European Union is a "dispute" rather than a "consensus" referendum. So: will history repeat itself? If it does, then if—as I have argued in recent blogs—"remain" entered the final month of the campaign with a modest lead, then we should expect that lead to widen in the days leading up to June 23.

Why the status quo usually gains ground

The prospects for the next fortnight depend partly on why the status quo has generally gained ground in past referendums. Two reasons seem to have contributed to the historical pattern, and both look likely to apply this time.

First, some people who take an interest in the referendum issue are unhappy with the way things are and tell pollsters until the last week or two that they support change. It is a cost-free way to express dissatisfaction. (Much the same often happens in mid-term by-elections, when the government of the day suffers a big adverse swing: voters have the chance to protest without having to worry that their vote will hand power to the opposition. However, when referendum day approaches, they think not just about the status quo and what's wrong with it, but about the alternative. Some voters decide that, on balance, change carries risks that they would prefer to avoid. They draw back from the cliff edge and vote for the status quo.

Second, other people have better things to do than follow referendum news at all until the last few days. One of the most significant poll findings of recent weeks has been a YouGov/Times poll that reported that 45 percent of Labour voters were unaware that most Labour MPs want the U.K. to stay in the EU. I suspect that pollsters could ask other knowledge questions and find that millions of voters are unaware of many basic aspects of the campaign.

Now, many such people will end up not voting at all. But quite a few will vote; however, they are the kind of people who hold no strong views about the EU either way (or the issue at hand in past referendums). Whatever they tell pollsters some weeks ahead of the referendum—if they respond to pollsters at all—most of them end up deciding instinctively that it is safer to leave things as they are than to vote for change.

These are not completely hard-and-fast categories. Some people will be a mixture of both types. The key thing is that their views are less fixed and more risk-averse than people with strong opinions and a passion for political news.

A recent article by Daniel Jackson of the Campaign Group, for The Times's Red Box, makes much the same point in a different way. He divides voters into pioneers (typically young, liberal, pro-EU), settlers (typically older, more tribal, anti-EU) and prospectors. People in this final group tend to be busy, status-conscious, motivated by economic self-interest—and often decide late how to vote. He argues that their worries about prosperity will in the end trump their dislike of immigration, and they will break towards remaining in the EU.

Future imperfect

The words of those financial advertisements should be heeded: "past performance does not guarantee future results". Some recent polls have reported a shift towards Brexit. Maybe this will continue, and the safety-versus-risk battle will play out differently than it has in past referendums. Perhaps immigration will trump economics as the issue that sways those who are still not sure how to vote; that would shift the dial towards Brexit. Perhaps some crisis will erupt, to do with terrorism, refugees or the Eurozone, to make the late deciders judge that we are safer outside the European Union. There is always an element of unpredictability whenever voters are asked to decide their nation's future—thank goodness. It would be a sad democracy that behaved in a wholly deterministic manner.

However, the fact that we cannot be certain what will happen between now and June 23 does not mean we cannot judge the probabilities. And while the record of past referendums does not guarantee a shift to "remain" in the final days of the current campaign, it does suggest that such a shift is more likely than not.

Peter Kellner is a political analyst, having worked for more than 40 years as a journalist, broadcaster and pollster.