Britain has voted to leave the European Union. The news surprised many people, including the British, who have learned that while brushing off early statistical warnings is tempting, it doesn’t make it any easier when those warnings turn out to be right. Give yourselves a break, I say: Polls are fickle, anecdote is limited, and prevailing wisdom is sometimes impossible to shake. (Though these remorseful Brexit voters don’t have an excuse.)
There’s a silver lining for statistics, however. With the close of Britain’s referendum, political analysts now have a concrete dataset to examine: the actual vote totals in the United Kingdom. This data, when matched with regional demographic information from the U.K. Census, gives insight into who actually voted to leave or remain.
While it’s impossible to know how every person in Britain voted—secret ballot and all that—it is possible to infer how broad demographic groups feel about the Brexit by analyzing vote totals from different areas. For instance, if districts with high concentrations of Jedi knights had consistently higher “Remain” turnout, one can infer that the followers of The Force are in favor of unification.
Setting aside Scotland and Northern Ireland—which voted overwhelmingly to stay in the European Union—leaves more than 300 local authority districts, each with a different popluation mix.
So which demographics matched strongly with Britain’s independence streak?
Turns out that average levels of education of the people in a region correlate quite strongly with their Brexit leanings. People in areas where many residents have college degrees were far more likely to vote “Remain,” particularly in central London, where more than two-thirds of the city population has a bachelor’s degree. Hosting a sizable immigrant population seemed to sway communities against leaving in the European Union, and denser cities tended against “Leave” overall.
Other factors mattered less. The median age of a community, despite the much emphasized youths-versus-retirees clash that many said would define the referendum, ended up correlating only slightly with how the vote actually went. And while median income did predict results as expected—with lower income Britons favoring an exit—it still wasn’t as tight a relationship as some of the other factors.
While age was the dividing line expected by many Brexit watchers before the vote, the bigger split might have been between carefree singles and anxious parents, who might fear for the future of their children. Marital status did show a relatively strong correlation with voting to leave. This suggests a divide that, in retrospect, seems quite sensible: The younger, unattached professionals of London favor European membership, while the poorer families living outside of major cities eye the continent with considerably more suspicion.
Mandatory caveat: Correlation never implies causation. It’s possible some of these matches are a coincidence, particularly in cases in which the statistical relationships are weaker. And extrapolating results from broad geographic data always carries a significant chance of error. But given that the unprecedented nature of the referendum meant there wasn’t much in the way of exit-poll data, this kind of extrapolation might be among the best the bewildered world will get.