Introduction
Over the past two decades, scholars have largely embraced the ideational approach to populism and adopted Cas Mudde’s (Reference Mudde2004: 543) definition, which conceptualizes populism as a thin-centered “ideology that considers society to be ultimately separated into two homogeneous and antagonistic groups, ‘the pure people’ versus ‘the corrupt elite’, and which argues that politics should be an expression of the volonté générale (general will) of the people.” While this minimal “thin” definition allows scholars to categorize politicians and parties as “populist” (De Cleen and Glynos Reference De Cleen and Glynos2021: 181), populism conceptualized in such a way lacks programmatic content and political actors invariably attach this “thin” ideology to a range of host ideologies (sometimes termed “thick” ideologies) that signal their policy positions. For instance, a common host ideology adopted by populist political parties on the right is nativismFootnote 1 , which contends that each nation should have its own state and that “non-native people and ideas are fundamentally threatening to the homogenous nation-state” (Mudde Reference Mudde2017) and manifests itself in anti-immigrant rhetoric. In contrast, populist parties on the left often pair it with calls for redistribution (Mudde Reference Mudde2004; Mudde and Kaltwasser Reference Mudde and Kaltwasser2013; Hawkins and Kaltwasser Reference Hawkins and Kaltwasser2017).
Political scientists have therefore been interested in estimating which components of thin populist ideology such as people-centric, anti-elite, or anti-pluralist messages garner support among voters and how the appeal of these factors compares to more traditional host ideology positions that politicians combine with thin populist rhetoric (e.g., Neuner and Wratil Reference Neuner and Wratil2022; Castanho Silva et al. Reference Castanho Silva, Neuner and Wratil2023; Dai and Kustov Reference Dai and Kustov2024). To examine these questions, scholars are increasingly leveraging conjoint experiments (Bansak et al. Reference Bansak, Hainmueller, Hopkins, Yamamoto, Druckman and Green2021a; Druckman Reference Druckman2022). Neuner and Wratil (Reference Neuner and Wratil2022) noted that the concern with previous research was that it was difficult to disentangle any effects of thin and host ideology because these two factors are deeply intertwined within real political actors. The ability to estimate support for fictitious candidates with varying levels of both thin and host ideology is thus the primary benefit of the conjoint method while a drawback is that the repeated, sequential exposure to stimuli does not easily lend itself to the inclusion of manipulation checks.
This leads to a problematic lack of validation that a given treatment truly operationalizes the underlying independent variable of interest (Mutz and Pemantle Reference Mutz and Pemantle2015). For instance, while we might a priori expect that a statement such as “Give voice to the American people” or “Reject the bureaucratic elite” will increase perceptions that a candidate espouses people-centric and anti-elite beliefs, respectively, this has not been systematically validated. Moreover, it is unclear whether survey respondents also make inferences about a candidate’s host ideology based on such statements. Indeed, when a populist appeal invokes “the people” conceived in ethnic terms, this might signal nativist host ideology, or when they rail against “bureaucratic elites” in the European context, it might signal Eurosceptic host ideology.
Thus, in experiments where host ideology is not explicitly operationalized through an attribute and people perceive an association between a thin populist appeal and host ideology, then respondents may be using thin ideology as a proxy for host ideology (Bansak et al. Reference Bansak, Hainmueller, Hopkins and Yamamoto2021b; see also Dafoe et al. Reference Dafoe, Zhang and Caughey2018 on information equivalence). In this case, we would attribute a causal effect to thin populist rhetoric when people are partially responding to inferred host ideology. While designs usually include some type of attribute operationalizing facets of host ideology, they may do so with measurement error or may not capture all relevant dimensions of host ideology, raising the question of what quantity the estimates of thin populist attributes truly represent.
In this paper, I, therefore, test whether people-centric and anti-elite populist treatments successfully operationalize the independent variables of interest and, more importantly, if they do so without also manipulating perceptions of host ideology. In so doing, I also hope to demonstrate the value of pilot studies for validating conjoint treatments.
Conceptualizing the “people” and the “elite”
Building on Mudde’s (Reference Mudde2004) thin ideology definition, research initially focused on classifying parties as populist based on anti-elite and people-centric rhetoric (e.g., Jagers and Walgrave Reference Jagers and Walgrave2007). When the focus turned to the question of what attracts voters to populist politicians, research mirrored this approach by conceptualizing and measuring voters’ populist attitudes primarily based on whether they agreed with people-centric or anti-elite statements (Akkerman et al. Reference Akkerman, Mudde and Zaslove2014; Castanho Silva et al. Reference Castanho Silva, Jungkunz, Helbling and Littvay2020). The core of populism in this tradition thus concerns upholding the will of the “good” or “ordinary” people whose interests are diametrically opposed to those of the “elite” (Mudde Reference Mudde2004; Canovan Reference Canovan1999).
Despite this seemingly straightforward emphasis, the thin ideology definition of populism still requires greater conceptual clarity – most critically the need to disentangle thin ideology from host ideology in general and nativism in particular. For instance, Rooduijn (Reference Rooduijn2019) highlighted that it is easy yet problematic to misclassify nativist rhetoric as populist given its emphasis on how native populations are threatened by “dangerous others” or immigrants. This concern has been echoed by analyses highlighting that many academic and journalistic accounts of populism are concerned with nativism or other host ideologies as opposed to thin populist ideology per se (see Art Reference Art2022; Bonikowski Reference Bonikowski2017; Hunger and Paxton Reference Hunger and Paxton2022).
In experimental research, manipulating populism following the thin ideology tradition most centrally means developing stimuli that (1) operationalize people-centrism and anti-elitism (Castanho Silva et al. Reference Castanho Silva, Jungkunz, Helbling and Littvay2020), while (2) being devoid of information that also manipulates host ideology. This is easier said than done. As Jagers and Walgrave (Reference Jagers and Walgrave2007) note, “thin” populism becomes “thick” populism based on who is conceptualized as the people and the elites. Indeed, while the good and homogeneous “people” and the evil “elites” provide parsimonious conceptual categories, populist actors usually combine elements of thin and host ideology.
This is particularly evident when examining actual populist communication. One discursive analysis of the term “the people” noted that it acts as an empty signifier that can be attached to multiple narratives including the “people as a nation” or the “people as public,” concluding that “while ‘the people’ serve as the legitimizing anchor for all narratives, each mobilizes a different repertoire of additional underspecified constructs” (Pasitselska and Baden Reference Pasitselska and Baden2020: 683). Conceptually, the “people” in populist discourse refers to a “demos that seek to restore popular sovereignty,” whereas in nativist discourse, it refers to ethnos and “national sovereignty” (Venizelos Reference Venizelos2023: 2; see also De Cleen and Stavrakakis Reference De Cleen and Stavrakakis2017). This creates a fine line between populism and nativism that is often blurred in actual rhetoric. For instance, an analysis of press releases from the German Alternative für Deutschland (AfD) and the United Kingdom Independence Party (UKIP) highlighted that both parties’ conceptions of the people adopted a “(more or less exclusionary) nationalist paradigm” (Breeze Reference Breeze2019: 95). Moreover, an analysis of Donald Trump’s inaugural address revealed that he exclusively used the term “the people” to refer to US citizens (Chilton Reference Chilton2017). Similarly, there are various ways of conceptualizing “the elite.” Anti-elite rhetoric, for instance, can target the general political establishment, the intellectual class, or economic powers (Jagers and Walgrave Reference Jagers and Walgrave2007; Wodak Reference Wodak2017). In the aforementioned analysis of AfD and UKIP discourse, for instance, the elites are usually conceived as nationally established parties and the European Union (Breeze Reference Breeze2019; see also Hameleers and Goldberg Reference Hameleers and Goldberg2022).
Thus, while it is possible to conceptually disentangle populism and host ideology, it may be more difficult to do so empirically as evidenced in recent debates on the measurement of populist attitudes (Castanho Silva et al. Reference Castanho Silva, Jungkunz, Helbling and Littvay2020). For instance, Meijers and van der Velden (Reference Meijers and van der Velden2023) demonstrate that survey items operationalizing people-centrism differed in whether they invoked civic or ethnic constructions of “the people,” further showing that ethnic conceptualizations of “the people” also primed nationalist beliefs (Meijers and van der Velden Reference Meijers and van der Velden2023).
Taken together, when people hear politicians talk about the “people” and the “elites,” there are a range of actors that can come to mind, and people can use parties’ host ideology to infer who is being targeted. As Chilton (Reference Chilton2017: 592) notes, populist rhetoric “works by activating emotion-laden and value-laden schematic concepts and concomitant emotions,” suggesting that such rhetoric can prime these connections even if the people and the elites are not explicitly defined. Moreover, given that politicians bundle their populist and host ideology rhetoric, it is possible that people can use the populist rhetoric to infer host ideology (Bonikowski and Zhang Reference Bonikowski and Zhang2023).
Examples of (thin) populism treatments
I next turn to a brief overview of some of the operationalizations that have been used for people-centric and anti-elite populist appeals in recent experimental studies. I focus on these appeals as opposed to anti-pluralist appeals because published studies focus on the former, and do so in diverse ways, whereas treatments operationalizing anti-pluralism are generally more vague and not subject to the concerns outlined above.
There is a lot of overlap in treatments operationalizing the “people,” but a key distinction is whether the “people” are referred to in abstract terms or explicitly with reference to nationality. Indeed, a common operationalization of the “people” uses phrases such as “American people” or “British people” (e.g., Bakker et al. Reference Bakker, Schumacher and Rooduijn2021; Dai and Kustov Reference Dai and Kustov2024) or explicitly refers to citizens (e.g., Neuner and Wratil Reference Neuner and Wratil2022; Castanho Silva et al. Reference Castanho Silva, Neuner and Wratil2023). In contrast, other operationalizations are less explicit about the reference group using phrases such as “the good people” or “ordinary people” (Ferrari Reference Ferrari2024; Kollberg et al. Reference Kollberg, Lauderdale and Wratil2025). Some studies combine elements of both leading to phrases such as “ordinary Americans,” “honest and innocent British people,” or “hardworking American taxpayers” (Bakker et al. Reference Bakker, Schumacher and Rooduijn2021; Kollberg et al. Reference Kollberg, Lauderdale and Wratil2025; Bonikowski and Zhang Reference Bonikowski and Zhang2023). While all these phrases plausibly operationalize people-centrism, they might do so to varying degrees and may differ in the extent to which they signal host ideology.
Operationalizations of the “elite” are more varied. Some operationalizations are very general and just refer to the “political elite” (Neuner and Wratil Reference Neuner and Wratil2022; Castanho Silva et al. Reference Castanho Silva, Neuner and Wratil2023), whereas others use a variety of prefixes such as “useless politicians” or “out of touch elites” (Kollberg et al. Reference Kollberg, Lauderdale and Wratil2025). Similarly, some operationalizations directly invoke corruption such as “corrupt elites” or “corrupt politicians” (Ferrari Reference Ferrari2024; Neuner and Wratil Reference Neuner and Wratil2022; Dai and Kustov Reference Dai and Kustov2024). Others are context specific, such as, “Washington insiders” or “Washington elites” (Bakker et al. Reference Bakker, Schumacher and Rooduijn2021; Dai and Kustov Reference Dai and Kustov2024). Lastly, some operationalizations more overtly tap into host ideologies such as “globalist elites” (Kollberg et al. Reference Kollberg, Lauderdale and Wratil2025).
Here, it is worth acknowledging one promising approach, namely, estimating effects across a range of operationalizations (see e.g., Blumenau and Lauderdale Reference Blumenau and Lauderdale2024). For instance, Kollberg et al. (Reference Kollberg, Lauderdale and Wratil2025) include multiple different operationalizations of populist rhetoric. Across nine issue areas, they create positively and negatively valenced statements in both a populist and a non-populist format. While this provides an important innovation, it is still critical to validate the underlying treatments, especially given the resource constraints scholars face.
Research questions and hypotheses
The aim of this registered report is to validate both the construct and discriminant validity of treatments operationalizing thin populism (see Chester and Lasko Reference Chester and Lasko2021). First, I examine whether treatments operationalizing people-centric and anti-elite statements lead survey respondents to perceive politicians as more people-centric and anti-elitist. Second, and more critically, I examine the degree to which these statements affect perceptions of host ideology. Thus, I examine the following main research questions:
RQ1: Do the various conceptualizations of “the people” and “the elite” successfully manipulate perceptions of people-centrism and anti-elitism compared to their respective control conditions?
RQ2: Do these various operationalizations differentially affect perceptions of host ideology?
In addition, building on the idea that such terms may clearly be associated with host ideology (Chilton Reference Chilton2017; Bonikowski and Zhang Reference Bonikowski and Zhang2023), I also examine whether thin ideology treatments are less likely to affect perceptions of host ideology when explicit partisan information is provided (RQ3). Lastly, I examine the extent to which all operationalizations operate similarly across two contexts that have been studied in previous research and can be directly compared due to them featuring the same national language and the ability to use the same survey vendor (the US and the UK) as well as across respondent-level characteristics (partisanship and populist attitudes) (RQ4).
Building on the theoretical literature discussed above, I contend that the treatments researchers have been using to operationalize various aspects of thin ideology are clearly capturing elements of people-centrism and anti-elitism and will thus affect perceptions of these dimensions of populism. I thus hypothesize:
H1: Treatments operationalizing people-centrism and anti-elitism will increase perceptions of a politician’s people-centrism and anti-elitism, respectively, vis-à-vis a control condition.
I further suggest that different treatments will differentially affect perceptions of host ideology. In the absence of signifiers, each respondent might infer different host ideologies of the candidate. However, explicit invocations of groups will make respondents more likely to infer the same host ideology. For instance, given contemporary political schemas in the US, explicitly appealing to the “American people” as opposed to a more generic “people” might make respondents perceive the candidate as more conservative. While I examine whether the treatments affect perceptions vis-à-vis the control condition, my key hypothesis is:
H2: Treatments that include explicit references to groups (e.g., “American people,” “bureaucrats”) will be more likely to also affect perceptions of host ideology than more generic invocations of “the people” and “elites.”
The studies outlined above either withhold partisan information (e.g., Neuner and Wratil Reference Neuner and Wratil2022) or consist of primary contests for the respondent’s preferred party (e.g., Dai and Kustov Reference Dai and Kustov2024). Given the possibility that people may use thin ideology statements as heuristics to infer host ideology, reliance on such heuristics should decrease when people are provided with more tangible information that they can use to infer ideology (see e.g., Sen Reference Sen2017; Hiaeshutter-Rice et al. Reference Hiaeshutter-Rice, Neuner and Soroka2023). I thus further hypothesize that thin ideology treatments will have a smaller effect on perceptions of host ideology when people have more information to go by:
H3: Treatments operationalizing people-centrism and anti-elitism will have a smaller effect on perceptions of host ideology when explicit partisan information is provided.
Lastly, while I explore possible heterogeneity in treatments across contexts (the US and the UK) as well as across respondent-level characteristics (partisanship and populist attitudes), I have no directional hypotheses concerning these effects. Rather, I conduct these analyses to assess the extent to which the treatments operate similarly across groups and contexts.
Samples and power
To test these hypotheses, I fielded a study on a broad national sample of US-based respondents from the Prolific online platform in May 2024 (Douglas et al. Reference Douglas, Ewell and Brauer2023). I recruited 1403 respondents and randomly assigned them to one of the two versions of the conjoint design (see below). Second, in March 2025, I replicated the baseline version of the conjoint design on a UK Prolific sample (n=690) to examine potential heterogeneity across contexts (see Appendix A for sample demographics). The studies are powered at 90% to detect the median reported effect size (Average Marginal Component Effect = 0.05) when testing H1 and H2 and powered at 80% to detect slightly larger interaction effects (Average Marginal Component Interaction Effect = 0.061) when testing H3 (see Appendix E; Schuessler and Freitag Reference Schuessler and Freitag2020).
Experimental designFootnote 2
In the conjoint experiment, respondents evaluated six pairs of politicians. The profiles were presented as vignettes as opposed to in a table format to allow a control/baseline level for the people-centrism and anti-elitism treatments in which no information on each dimension is provided.Footnote 3 The various operationalizations of the people-centrism and anti-elitism treatments seek to reflect those of existing published studies and are presented in Table 1. An example of the task is presented in Figure 1. I operationalize people-centrism and anti-elitism with three phrases that differ in the specificity of the groups mentioned. For instance, the generic and vague operationalizations of people-centrism and anti-elitism reference “the people” or “corrupt elites” (see e.g., Neuner and Wratil Reference Neuner and Wratil2022; Ferrari Reference Ferrari2024), respectively, whereas other operationalizations provide additional context such as “the American people” or “out-of-touch bureaucrats” (see e.g., Dai and Kustov Reference Dai and Kustov2024; Kollberg et al. Reference Kollberg, Lauderdale and Wratil2025; Bakker et al. Reference Bakker, Schumacher and Rooduijn2021).
Table 1. Conjoint attributes and levels
Figure 1. Example of conjoint task.
Notes: This example is from Sample 1 in which no partisan information is provided. Only forced-choice outcomes shown.
In addition to these core conjoint attributes, the profiles also included a filler attribute which operationalizes an issue position that is purposely vague to not be solely determinative of ideology. This attribute should not substantially affect perceptions of people-centrism or anti-elitism, but it is important to acknowledge that even though the attribute and levels were chosen not to have clear ideological content, some respondents will likely use it to infer ideology (see e.g., Hiaeshutter-Rice et al. Reference Hiaeshutter-Rice, Neuner and Soroka2023). Lastly, there was a split sample design with half of the respondents (Sample 1; n=699) being randomly assigned to see the conjoint tasks shown in Figure 1 and half (Sample 2; n=704) being assigned to see conjoint tasks that include partisanship as an additional attribute of the politicians (with levels Democrat or Republican) (see Appendix Figure B1). As noted above, this manipulation allows me to test the extent to which any effect of the people-centric and anti-elite attributes on perceptions of host ideology is reduced when more information regarding host ideology is presented, thus providing valuable insight into the scope of this concern.
Following each profile pair, respondents answered a series of manipulation check type questions. To assess perceptions of people-centrism and anti-elitism, respondents were asked which politician is more likely to “say that politicians should always listen to the people” and which politician is more likely to “say that many people in the political class are crooked.” These two statements are drawn from survey scales measuring populist attitudes and thus are suggested to capture key dimensions of thin populist ideology (Appendix C). To assess host ideology, I ask respondents to assess which politician is more conservative and which candidate would support reducing immigration. I did not explicitly ask respondents about their perceptions of populism given the conflation of populism and nativism in popular discourse and laypeople’s understanding (Hunger and Paxton Reference Hunger and Paxton2022; but see Casiraghi et al. Reference Casiraghi, Curini and Nai2024). To gain a more fine-grained understanding of how survey respondents perceive these various attributes, the above questions were asked as both forced-choice comparisons as well as ratings of each profile (for detailed information about the preapproved analytical strategy, see Appendix D)Footnote 4 .
Results
Populist thin ideology outcomes
First, I examine whether the various thin populism treatments produce their intended effects (RQ1/H1: confirmatory). Figure 2 shows marginal means for perceived people-centrism (left panel) and perceived anti-elitism (right panel) for all respondents in Sample 1 using the forced-choice binary outcome measures.Footnote 5 The dashed line indicates the grand mean (0.5 probability of a profile being selected) such that estimates to the left indicate a decrease and estimates to the right indicate an increase in selection probability.Footnote 6 In line with H1, compared to their respective control conditions, the treatments operationalizing people-centrism increase perceptions that the politician would state that “politicians should always listen to the people” while the treatments operationalizing anti-elitism increase perceptions that the politician would state that “many in the political class are crooked” (see Figure F2 for average marginal component effects).
Figure 2. Marginal Means for Perceptions of People-Centrism and Anti-Elitism (Binary Outcomes, Sample 1).
These results are replicated using rating outcomes (Figure F1), in Sample 2 (Figures F3–F4), as well as in the UK replication study (Figures I3–I5). The findings are consistent across all three samples: (1) the three people-centrism treatments affect perceptions of people-centrism to a similar degree, and (2) perceptions of anti-elitism are most affected by the “corrupt elite” treatments (average marginal mean of 0.64), while the “out of touch bureaucrats” treatment slightly increases perceptions of anti-elitism relative to the grand mean and the “Washington/Westminster insiders” treatment hovers around the grand mean.
Host ideology outcomes
Second, I examine whether the “thin” populism treatments differentially affect perceptions of host ideology (RQ2/H2: confirmatory). Figure 3 presents marginal means for perceived immigration position (left panel) and perceived conservatism (right panel) in Sample 1 in which no partisan information is included. Here, evidence for H2 is mixed. On the one hand, the “Washington insiders” and “out-of-touch bureaucrats” treatments do not affect perceptions of immigration stance or conservatism compared to the “corrupt elite” treatment. In contrast, in line with H2, both the “American People” and “Honest hard-working citizens” treatments increase perceptions that the politician is conservative and in favor of reducing immigration compared to both the “people treatment” and the “no people-centrism” conditions (average marginal component effects are presented in Figure F10)Footnote 7 . While the hardworking citizens treatment does not follow this pattern in the UK replication, the effect of the “British people” treatment is even more pronounced, and the corrupt elite treatment consistently increases perceptions that the politician favors reducing immigration in the UK context (see Figures I6–I11).
Figure 3. Marginal Means for Perceptions of Immigration Stance and Conservatism (Binary Outcomes, Sample 1).
Differences in host ideology outcomes when partisanship is included
Third, I examine whether the impact of thin ideology treatments on perceptions of host ideology differs when explicit partisan information is included (RQ3/H3: confirmatory). Here, I leverage Sample 2, which was fielded alongside Sample 1, and which included an attribute indicating which party each politician was affiliated with. First, nested comparison tests of whether any of the interactions between the conjoint attributes and an indicator variable for whether respondents were provided with partisan informationFootnote 8 (Sample 1 vs. Sample 2) are significant indicate that out of the four models (two outcomes for binary and ratings outcomes), the test only yields significant results for the Perceived Immigration Position ranking outcome (p=0.012). This suggests that including partisan information does not generally moderate the effect of thin ideology treatments on the host ideology outcomes.
However, Figure 4 paints a slightly more nuanced picture. It reproduces Figure 3 but adds estimates for Sample 2 as well as differences in marginal means between the two samples. The main finding that stands out is that the marginal mean for the “American people” treatment is significantly reduced for the conservatism outcome in Sample 2 compared to Sample 1 and is indistinguishable from the grand mean in Sample 2 (though note it is still significantly different from “the people” treatment; see Figure F14). The immigration rating outcomes (Figure F15) provide additional suggestive evidence that some people-centric appeals are less connected to host ideology when partisan information is included (see Figure F15 for ratings outcomes and Figures F16–17 for differences in average marginal component effects). Taken together, while the results are inconsistent across items, there is some suggestive evidence of reduced effects of thin ideology treatments on perceptions of host ideology when explicit partisan information is provided (H3).
Figure 4. Marginal Means and Differences in Marginal Means between Sample 1 (no partisan information) and Sample 2 (partisan information included).
Heterogeneity by partisanship and populist attitudes
In the Appendix, I plot sub-group marginal means by partisanship and populist attitudes for all analyses relevant to RQ1/H1 and RQ2/H2 (RQ4: exploratory; Figures G1–G16 and I12–I19). While nested model comparison tests show significant interactions in 11 out of 24 models for partisanship and 8 out of 24 models for populist attitudes (at p<0.05), the figures suggest that this heterogeneity does not substantively change conclusions and that evidence supporting H1 and H2 emerges for both populist and non-populist respondents and across partisan groups.
Conclusion
This paper sought to validate whether the types of populism treatments recently used in conjoint experiments affect perceptions of (thin) populism without also affecting perceptions of host ideology. A few findings stand out. First, while the generic “the people” treatment affected perceptions of people-centrism to the same degree as treatments invoking the “American” or “British” people, it did so without affecting host ideology (i.e., it was equivalent to the “no statement” condition). Second, the “corrupt elite” treatment was superior to the “Washington/Westminster insiders” and “out-of-touch bureaucrats” treatments in moving perceptions of anti-elitism without – at least in the US case – differentially affecting perceptions of host ideology.
Taken together, these results highlight the need for scholars to consider and ideally validate whether their chosen treatments successfully manipulate populist thin ideology and to be cognizant that people can use thin populist rhetoric as heuristics for host ideology. In particular, the results suggest that people-centrism treatments invoking specific (i.e., “American” or “British”) people should be avoided as they cannot isolate thin ideology from host ideology. The differences in results between the US and the UK highlight the benefits of generic treatments while also cautioning that the effect of thin ideology treatments on perceived host ideology can differ across contexts. Lastly, while beyond the scope of this paper, the unexpected finding that the “filler” attribute affected perceptions of people-centrism and host ideology suggests both that perceived people-centrism is affected by issue positions and that even relatively bland statements can affect perceptions of thin and host ideology.
In conclusion, this paper illustrates the difficulties involved in designing populism treatments to disentangle the effects of thin and host ideology on populist voting. More broadly, the findings suggest that conjoint experiments can suffer from information equivalence concerns (Dafoe et al. Reference Dafoe, Zhang and Caughey2018) while highlighting the promise of using pilot conjoint experiments to validate conjoint experimental treatments to ensure they successfully manipulate the intended independent variables.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/XPS.2025.10018.
Data availability
The data and code required to replicate all analyses in this article are available at the Journal of Experimental Political Science Dataverse within the Harvard Dataverse Network, at: https://doi.org/10.7910/DVN/JIUPOU (Neuner Reference Neuner2025). Additional study materials are provided at https://doi.org/10.17605/OSF.IO/NUAVW. See Appendix Section N for detailed information on reporting standards.
Acknowledgments
The author is grateful for feedback from the editors and the four anonymous reviewers as well as participants at the ASU School of Politics and Global Studies Democracy lab and Bruno Castanho Silva, Diogo Ferrari, Jennet Kirkpatrick, Steven Smith, and Christopher Wratil.
Competing interests
The author declares there are no conflicts of interest. See also Appendix Section L.
Ethics statement
The studies adhere to APSA’s Principles and Guidance for Human Subjects Research. The studies were reviewed and deemed exempt by Arizona State University’s Institutional Review Board (STUDY00020157). Stage 1 materials are deposited at https://doi.org/10.17605/OSF.IO/NUAVW. The studies were preregistered at https://doi.org/10.17605/OSF.IO/ZMUHQ (US Studies) and https://doi.org/10.17605/OSF.IO/RCDTX (UK Study). Additional information is provided in Appendix Sections L, M, and N.
Open access
