Given the risks this phenomenon poses to Western societies’ basic political unity and stability, we can probably never have enough good research into how, exactly, it developed in the first place.

Enter three U.S. social scientists with a subtle new analysis that applies fresh data and a skeptical attitude to conventional wisdom about populism: That is, the notion that it’s a manifestation of voter anger or, as it is sometimes termed, “rage.”

The title of their new study asks a question — “Does Anger Drive Populism?” — and answers it in the affirmative, but with a major caveat: Anger alone cannot account for recent U.S. vote shifts in favor of populist candidates (of both the left and right). Rather, the trends reflect a wider mix of negative emotions such as sadness, stress and worry.

It’s a portrait of populism as an expression of dismay and disenchantment, not just resentment.

The data come from Gallup studies in which pollsters asked up to 1,000 individuals to describe their subjective feelings daily in periods spanning from 2008 until 2017.

These polls generated a huge number of discrete responses (3.5 million), enough with which to estimate average levels of anger, and other emotions, including positive ones, for each county in the country during the relevant periods.

The social scientists, Romain Wacziarg of UCLA, Omer Ali of Duke University and Klaus Desmet of Southern Methodist University, found that a county’s average anger level correlated positively with higher vote shares for Trump, and for socialist Sen. Bernie Sanders (I-Vt.), in the 2016 GOP and Democratic primaries, respectively.

(Interestingly, and consistent with the “Bernie Bro” stereotype, the study found that very liberal independents were the most likely voters to report being angry, ahead of very conservative independents and Republicans.)

The same relationship held for Trump’s vote share in the 2016 general election. Notably, angrier counties also voted more heavily for Trump in 2016 against Democrat Hillary Clinton than they had in 2012 for Republican nominee Mitt Romney against President Barack Obama.

One of the niftier products of the team’s research was a color-coded national map that shows the least angry counties in light pink and the angriest shaded dark red. Appalachia, the Deep South and parts of the Great Lakes region — Trump country — look like someone sprinkled them with ketchup.

However, the authors probed deeper. They adjusted these results for the simultaneous occurrence of other negative feelings in the angriest counties. They found, as their article, issued as a working paper in June by the National Bureau of Economic Research, notes, that “anger no longer operates as a separate channel in driving the populist vote share.” Instead “a more complex sense of malaise and gloom, rather than anger per se, drives the rise in populism.”

This makes sense, theoretically and empirically. There’s no necessary connection between anger and discontent; as the authors note, respondents to Gallup’s daily surveys often reported feeling angry and relatively satisfied with life.

Empirically, some mid-country counties that voted heavily for Trump in 2016 were not among the most angry. And where Trump counties were angrier than usual, they also unfortunately abounded in reasons to feel worried, stressed or sad: such as deindustrialization or heavy opioid and alcohol use, with accompanying addiction and “deaths of despair.”

To be sure, these counties were often places where non-college-educated Whites predominate in the electorate, consistent with analyses of right-wing populism that attribute it to racial resentment.

“Does Anger Drive Populism?” adds nuance to, but does not contradict, those views or the view of political scientists Pippa Norris and Ronald Inglehart, which is that right-wing populism stems from an authoritarian “cultural backlash” by social conservatives who feel besieged by postmodern values.

The authors of “Does Anger Drive Populism?” do not offer any causal speculation themselves, for the sound reason that the Gallup surveys don’t provide any basis to do so: The polls asked people to describe their anger and other feelings in general, not explain why they felt them. Also, conditions — including Trump’s own ever-worsening behavior — have changed significantly since the data series used by the authors ended in 2017.

Still, there is much to be learned from this highly original study, with its provocative implication that voters who propelled Trump to the White House seven years ago did so more in sorrow than in anger.