On June 23rd Britons will head to the polls to answer a simple question they have not been asked since 1975: “Should the United Kingdom remain a member of the European Union or leave the European Union?” If the answer is “remain”, Britain will continue to integrate with the EU’s 27 other countries. If it chooses to “leave”, the Kingdom may split apart and begin to drift gently into mid-Atlantic obscurity. “Remain” has led in the polls for almost the entirety of the campaign. In early June the “leave” side surged, and briefly appeared to have taken a decisive lead. But the tragic murder on June 16th of Jo Cox, a pro-EU member of Parliament, may have helped swing the polls again in recent days. On the surface, this has restored a narrow edge for “remain”. However, the share of people saying they intend to vote for “remain” has not actually increased. Instead, a sliver of the electorate has simply switched from “leave” to “don’t know”. With just one or two percentage points splitting the two sides, the outcome will depend largely on the 10-15% of voters who say they are still undecided.
In the hopes of anticipating how this group will behave at the ballot box, we obtained a sample of individual poll data covering 5,500 respondents from YouGov, a polling company. The responses were collected from April 29th to June 8th, with the bulk in early June. The undecided have a handful of distinctive characteristics. Their partisan affiliation tends to lean Conservative: they were 16% more likely than the rest of the sample to say they had supported the Tories in the 2015 general election. This makes intuitive sense, as the Conservative party is split over the EU. The “remain” camp, led by David Cameron, the prime minister, is being fought by Boris Johnson, his Oxford chum and the former Tory London mayor. People from lower social classes also appear to be on the fence, as they were 15% more likely to be undecided than the rest of the sample. This group is often less engaged in politics, which makes voting decisions difficult.
Next, we used this information to analyse the correlations between demographic attributes and stated voting intentions in the referendum. The five factors that best predicted the “leave”/“remain” margin were age, education, social class, how people voted in the 2015 general election and the region in which they live. For example, there is about a 65% chance that a 50-year-old Labour voter from the north of England, who finished education at 16 and is employed in routine work will support “Leave”. After measuring these relationships, we then estimated how people who “don’t know” would be likely to vote, on the assumption that they will follow the same patterns as those who do.