Politicians who use populist language that emphasizes the will of “the people” and criticizes corrupt elites tend to be perceived as more aligned with certain policy positions—even when no explicit policies are mentioned. That’s one of the central findings of a new study published in the Journal of Experimental Political Science. The research provides evidence that common ways of measuring populist attitudes may also unintentionally affect perceptions of politicians’ broader ideological positions, such as conservatism or support for immigration restrictions.
The findings suggest that what political scientists often call “thin” populist ideology—language that pits a morally pure public against a corrupt elite—can sometimes serve as a stand-in for more substantive positions that researchers refer to as “host” ideologies. This overlap presents a challenge for research that aims to isolate the impact of populist rhetoric from the policy content that typically accompanies it.
Fabian Guy Neuner, an associate professor at Arizona State University, designed the study to address a recurring issue in political science research on populism. While many scholars adopt a definition of populism that focuses on people-centered and anti-elite appeals, these traits are often studied in combination with more detailed ideological positions such as nationalism, anti-immigrant sentiment, or economic redistribution.
“One of the questions in my broader research agenda is why people vote for populist politicians,” Neuner told PsyPost. “My previous work in Germany and the United States has sought to unpack whether people vote for these parties and politicians ‘thin ideology’ (i.e. their people-centric, anti-elite etc.) positions or whether they are attracted to their ‘host ideology’ (i.e. the policy positions such as nativism that populist also espouse).”
“Our work and a number of other papers have used conjoint experiments to disentangle these two potential causal factors. This literature suggests that people are attracted to the host ideology rather than the thin ideology. Thus, while political scientists rely on the ‘thin ideology’ framework to classify parties and voters as populists, this is not the driving factor affecting peoples’ vote choice.”
“A growing number of papers are examining this question using a variety of different experimental treatments to operationalize ‘thin populist’ appeals. For instance, politicians might be described as being ‘for the people’ or ‘for the American people.’ My concern with some of this literature was that some of these experimental manipulations might be inadvertently also manipulating perceptions of “host ideology”, thus making it nearly impossible to causally disentangle the effects of thin and host ideology.”
“For instance, if a voter is drawn to a candidate saying “I am for the American people” is this because of the people-centrism element of the appeal or is it because people make inferences about the candidate’s issue positions (for instance their stance on immigration)? As these papers all use conjoint experiments, they do not include manipulation checks to verify what the treatment is doing.”
“When the Journal of Experimental Political Science issued a call for a special issue on validating experimental manipulations, I saw an opportunity to examine this question,” Neuner said. “The paper thus had two aims. First to validate whether commonly used ‘thin populism’ treatments truly manipulate perceptions of thin ideology. And, second, whether they do so without also manipulating host ideology.”
“Beyond the populism case, this paper is relevant for scholars using conjoint experiments as it highlights the importance of validating experimental treatments and showcases how design choices in conjoint experiments can mask true causal factors (i.e. in this case people using the thin ideology treatment to make inferences about host ideology).”
To investigate, Neuner conducted a series of experiments using a method known as a conjoint analysis. This approach allows researchers to present study participants with pairs of hypothetical politicians who vary across different characteristics, including their language and message framing. Participants are then asked to compare the two figures based on how they perceive them—for example, whether they seem people-focused, anti-elite, conservative, or supportive of immigration.
Neuner’s study involved three large surveys. One was conducted in the United States with over 1,400 respondents in May 2024. Another, with more than 600 participants, was conducted in the United Kingdom in March 2025. These two national samples allowed Neuner to explore whether the findings would hold in different political and cultural contexts. A split-sample design was used in the U.S. study to explore how additional information—such as a candidate’s political party—might influence responses.
Participants read descriptions of hypothetical politicians that varied in their use of people-centric or anti-elite language. Some descriptions referred generally to “the people,” while others included more specific language, such as “the American people” or “hardworking citizens.” Similarly, anti-elite phrases ranged from vague terms like “corrupt elites” to more pointed language such as “out-of-touch bureaucrats” or “Washington insiders.” The study also included control conditions in which no populist language was used.
Neuner then measured how participants rated the politicians on characteristics related to populist thin ideology (such as being people-centric or anti-elite) and on host ideology (such as being conservative or likely to support reducing immigration).
The findings suggest that the treatments worked as intended in shifting perceptions of thin populism. Politicians who used people-centric or anti-elite language were more likely to be seen as embracing those values. This was consistent across different phrasings and across both the U.S. and U.K. samples. For instance, statements like “politicians should always listen to the people” or “many in the political class are crooked” were more often attributed to the candidates who had used populist language.
However, the effects on perceptions of host ideology were more mixed—and more concerning for researchers aiming to isolate populist language. In some cases, participants inferred that politicians using populist language were also more conservative or more likely to support reducing immigration, even when the descriptions made no reference to these issues. This effect was particularly noticeable when the language included explicit group labels, such as “the American people” or “honest, hardworking citizens.” In these cases, participants were more likely to assume that the politician was aligned with conservative positions.
Interestingly, the anti-elite messages showed different patterns. References to “corrupt elites” strongly affected perceptions of anti-elitism but did not consistently influence perceptions of policy preferences. In the U.K. sample, however, the same “corrupt elites” phrasing increased the likelihood that participants believed the candidate would support stricter immigration policies, suggesting that context may shape how these messages are received.
Another part of the study tested whether providing additional information—like a candidate’s political party—would change these inferences. Here, the evidence was more tentative. When participants were told a politician was a Democrat or Republican, the effect of some populist phrases on perceptions of conservatism was reduced. For example, the assumption that a candidate who referenced “the American people” was conservative weakened when the candidate’s party affiliation was also shown. Still, this pattern did not hold consistently across all outcomes.
Neuner also explored whether individual-level traits like partisanship or populist attitudes changed how people responded to the messages. While some variation emerged, the main findings appeared broadly consistent regardless of the participant’s political orientation.
“This is primarily an article about research design targeted at scholars working on populism or those who use conjoint experiments more broadly,” Neuner told PsyPost. “The take-away for those scholars should be to be careful when designing experiments and to validate their treatments.”
“The findings from the larger research agenda about the role of host ideology in driving support for populist politicians and parties are more relevant to the broader public. In terms of this paper, the takeaway for the average person might be to think about that what might sound “populist” (i.e. people-centric or anti-elite rhetoric) to some people may be perceived as much more connected to host ideologies such as nativism for other people.”
As with all experimental work, the study has its limitations. One challenge is that participants were reacting to hypothetical politicians in a survey, which may not fully reflect how people respond to real candidates in real elections. The study also focused on a limited set of phrases, which, while drawn from actual research, do not capture the full variety of populist rhetoric used in the real world.
Another limitation is that perceptions of ideology were measured indirectly, through comparisons and ratings, rather than by asking participants directly whether they thought a message was populist. While this avoids some of the confusion around how people define populism, it may also miss other interpretations of the messages.
Despite these limitations, the study provides important insights for political scientists. It highlights the importance of validating the tools researchers use to study voter attitudes, especially when trying to tease apart the effects of different components of political communication. It also provides evidence that even small changes in wording—such as invoking “American” rather than just “people”—can affect how political messages are received.
Looking ahead, Neuner suggests that more work is needed to understand how thin populist rhetoric interacts with broader ideological assumptions, especially across different national contexts. He also sees a need for better tools to help researchers distinguish between the rhetorical style of populism and the policy positions that often accompany it.
“More broadly, my research is unpacking the drivers of populist voting and comparing the role of thin and host ideology (see e.g., here),” Neuner explained. “In particular, this research seeks to explain why we see large effects for thin ideology (e.g., people’s populist attitudes) in observational research but not in experimental research.”
The study, “Can (Thin) Populism be Manipulated without Manipulating Host Ideology – Evidence from a Conjoint Validation Approach,” was published on September 11, 2025.
