What we mean by populism
Before getting into the findings, a quick definition. Political scientists generally follow Cas Mudde’s influential framework, which defines populism not as a full political program but as a simple worldview (or what Mudde calls a “thin ideology”). This worldview is based on three pillars: people-centrism (politics should reflect the will of “the people”), anti-pluralism (there is one authentic popular will, not many competing interests), and moralized anti-elitism (elites are not merely wrong but evil). This is what scholars call “thin” populism because it doesn’t tell you much about actual policy. A left-wing populist like Hugo Chávez and a right-wing populist like Trump share the same rhetorical structure, the people versus the elite, but disagree on virtually everything else.
This distinction between populism and its “host ideology” (the actual policy positions a politician holds) turns out to be crucial. Because when you peel them apart, something surprising emerges.
When politicians gamble on populism
Our first paper, “When Do Politicians Use Populist Rhetoric?“ published in Political Communication in 2022, asked a deceptively simple question: if populist rhetoric is so effective, why don’t all politicians use it all the time?
To answer this, we built the most comprehensive corpus of U.S. presidential campaign speeches at the time: 4,314 speeches from 1952 to 2016. We used a novel text analysis method combining active learning and word embeddings to measure how much populist rhetoric each candidate employed across the campaign trail. I (Alex) should say, thanks to the prowess of Yaoyao, we did all that fancy text analysis stuff before it was cool and before LLMs were even around.
The pattern was striking. Candidates who were trailing in the polls consistently used more populist rhetoric, regardless of whether they were Republicans or Democrats, incumbents or challengers. Populism, we argued, is a gamble: a high-risk, high-variance strategy that trailing candidates adopt because conventional campaigning isn’t working. If you’re already behind, why not shake things up?
Think of it like a football team that’s losing in the fourth quarter. You start throwing long passes not because they have a higher expected value, but because safe plays guarantee you lose. Barry Goldwater, George McGovern, and Donald Trump (in 2016, when most polls had him behind) all fit this pattern. They reached for populist rhetoric when they had little to lose.
The (in)effectiveness of populist rhetoric
But does the gamble actually pay off? Our second paper, published in Political Science Research and Methods in 2024, tested this directly with a survey experiment.
We presented U.S. respondents with pairs of realistic campaign messages from hypothetical primary candidates. The messages varied on two dimensions: populist features (people-centric language, anti-elite attacks, anti-pluralist framing) and substantive policy positions (on immigration and other issues). This design let us isolate the effect of populist rhetoric from the underlying policy content, something that is nearly impossible to do when observing real elections, where populism and policy positions come bundled together.
The result was unambiguous: none of the populist features had an independent effect on candidate choice. Not people-centrism, not anti-elitism, not anti-pluralism. Not individually, and not in combination. What did matter, enormously, were policy positions that aligned with voters’ own preferences. Voters chose candidates based on what they promised to do, not on how dramatically they framed the conflict between the people and the elite.
This finding is consistent with other experimental work. When researchers across multiple countries carefully separate populist style from policy substance, the style itself contributes very little to voter decisions.
So: if populist rhetoric doesn’t actually persuade voters, why does it seem to work? Why do populist candidates keep winning?
What populism is actually good for
This puzzle motivated our newest paper, our first ever registered report (where scholars publicly specify their hypotheses before running their experiment), now published at Research & Politics. We hypothesized that populism’s real contribution might not be persuasion but mobilization: getting people who already agree with you to actually show up and vote.
Previous studies, including our own, used what’s called a “forced choice” conjoint experimental design: respondents had to pick one candidate or the other. But in real elections, people can also stay home. To capture this, we ran a large-scale, preregistered survey experiment that added an “abstain” option, a seemingly small change that turns out to matter a lot.
What did we find? First, the basic persuasion result replicated: policy positions still dwarfed populist rhetoric in driving vote choice. Having a policy-congruent candidate increased the probability of voting by a massive 27 percentage points. Populist rhetoric, by contrast, had no meaningful persuasion effect.
But here is the twist: populist rhetoric did have a small but statistically significant mobilization effect. Having at least one populist candidate in a race was associated with a ~1.5 percentage point decrease in abstention. The effect was concentrated among voters who already held populist attitudes and encountered a candidate whose policy positions they liked. In other words, populist rhetoric didn’t convert skeptics; it energized true believers to get off the couch.
by Alexander Kustov and Yaoyao Dai, Popular by Design | Read more:
Image: uncredited
