A turning point in Quebec’s 1980 referendum came when separatist journalist and commentator Lise Payette said women who supported the federalist side were just “Yvettes,” a traditional characterization of stay-at-home women who always did what their husbands told them. This insult riled up tens of thousands of federalist women who started wearing “Yvette” badges and holding “Yvette” meetings. The federalist side won 59.6 to 40.4 so perhaps Mme Payette’s gaffe wasn’t crucial. But it definitely helped the Non side.

I thought of Lise Payette after reading what European Council President Donald Tusk told a Brussels press conference this week: “By the way, I’ve been wondering what the special place in hell looks like for those who promoted Brexit without even a sketch of a plan how to carry it out safely.” His zinger was aimed at Brexit promoters, not Brexit supporters, but as the high-handedness of European officials is a main complaint of Brexiteers, I wonder if that distinction will be noticed. His declaration certainly doesn’t help the United Kingdom back down gracefully from a bold Brexit if that’s its ultimate intention.

Just after reading about the “special place in hell” I came across a new research paper by four European economists (Federica Liberini, Andrew J. Oswald, Eugenio Proto and Michela Redoano) about who voted for Brexit. “Was Brexit Triggered by the Old and Unhappy? …” its title asks, “…Or by Financial Feelings?” As you might suspect, their answer is “by financial feelings.”

The Brexit ballot was secret, so the researchers couldn’t use actual referendum results to make their inquiries. Instead they looked at a continuing large-scale survey of the U.K., “Understanding Society,” to see which personal characteristics were associated with respondents’ answers to the survey’s question whether Britain should “remain a member of the European Union” or “leave the European Union,” which is a pretty straightforward choice.

They then correlated people’s response to that question with their age, gender, highest educational credential, region of the country and several other variables whose use is standard in such studies, but also with two others that aren’t—how satisfied people were with their lives (on a scale of 1 to 7) and how they felt they were doing financially. This last was measured on a scale of 1 to 4, from, in descending order, “Living comfortably” through “Doing all right,” “Just about getting by,” and “Finding it quite difficult.” The percentage of respondents giving each answer was, respectively, 35.0, 39.7, 19.0 and 4.7, so three-quarters of Brits thought they were living comfortably or doing all right, while just under one-fifth didn’t. (The percentages don’t add to 100 because not everyone answered.)

So, what’s the result?

When you do the multiple regressions and thus control for the influence of all these variables at once, the correlation that persists is the one having to do with how well people feel they’re doing financially. Support for Brexit was statistically significantly higher among those who regarded themselves as having financial trouble.

Age didn’t matter much. Very young people—those under 20 and in their early 20s stood out as strongly supporting “Remain” but from people in their late 20s through to those in their 70s there wasn’t much difference in support for “Leave.” This contradicts raw data that do show support for “Leave” rising fairly steadily with age. But of course the goal of econometrics is to see whether apparent relationships might not be proxies for something else and in this case it appears that’s what’s going on.

Not surprisingly, there are regional differences in support for Brexit even when other variables are controlled for. Scotland, no surprise, is least supportive of “Leave.” Contrary to many observer expectations, however, there’s no obvious urban-rural split. Nor, finally, is people’s satisfaction or dissatisfaction with their lives in general a clue to how they stand on Brexit. People who are happy and people who are sad seem equally likely to support either constitutional option.

In sum, the correlation that persists even as all these other variables are added to the mix is how people feel about their financial situation. An intriguing twist on this result is that how they feel about their financial situation turns out to be more important than what that financial situation actually is, for the researchers are able to include people’s income as they have reported it to the survey-takers and it doesn’t affect the results: The variable about financial feelings remains significant; their actual income is not.

“This seems an important message for economists,” they conclude. “The subject of economics has tended to avoid the study of human feelings in favour of ‘objective’ data.”

That’s the last line of their paper. I doubt it will be the last word on the subject. The study does clarify what has been driving Brexit but it deeply muddies economics. In trying to understand people’s behaviour, it seems we can no longer rely on their actual income as an explainer of their what they do. Instead we have to look into how they feel about their income. And then, presumably, we have to figure out why what they feel is different from what the facts of their income should be telling them. It’s not exactly “fake news” or even “fake views,” but it’s clearly a problem of misperception.

I wonder what special place—and whether it’s in heaven or hell—is reserved for people who turn a discipline upside down!