The ideational approach has reshaped the understanding of populism as an ideology or attitude. Scholars have leveraged this approach to develop measures for populist attitudes, but have not integrated their inventories with a “thick” host ideology, which must serve as a platform for “thin” populism. The host ideology breathes life into populism, shaping its morphology; the way in which populism materializes leads to different consequences for society and democracy. Because populism typically manifests on the left-right flanks of the political spectrum, we developed measures for left-wing and right-wing populist attitudes. We conducted four studies (two exploratory, N = 341, two confirmatory, N = 1164). Results yielded two factors for left-wing populist attitudes and three factors for right-wing populist attitudes, consisting of anti-pluralism, anti-elitism, and people-centrism themes. For both scales, we provide evidence of their predictive validity (e.g., predicts electoral skepticism, explains additional variance beyond left-right identification), construct validity (e.g., links to authoritarianism, better representation of elites in item content), and internal validity (successful invariance tests, adequate reliability). The differences between left-wing and right-wing populist attitudes, the nature of thin and thick ideologies in scale development, and implications of future research on populist attitudes are discussed.
The rise of populism transformed modern politics. With its ascent, many scholars have investigated populism as an integral feature of contemporary political discourse (e.g., Moffitt & Tormey, 2014; Sakki & Martikainen, 2021), as well as an ideology among politicians, political parties (e.g., Betz, 1994; Norris & Inglehart, 2019) and even individuals (e.g., Akkerman et al., 2014; Castanho Silva et al., 2018; Oliver & Rahn, 2016). Once considered an unidentifiable or incoherent concept (see Canovan, 2004), the ideational approach has clearly defined populism as an ideology (e.g., Mudde & Rovira Kaltwasser, 2013, 2017). Now with a clear understanding of populism, as a belief system, scholars have shown evidence that populist ideas exist as attitudes within the wider population.
The ideational approach conceptualized populist attitudes on a minimal definition, casting a wide net on various instances of populism regardless of political, cultural, or ideological contexts (Mudde & Rovira Kaltwasser, 2017). Whereas much of the literature seems to agree that there are vital features of populism, its manifestations vary considerably. Though there are undoubtable similarities, few would confuse the right-wing populism of U.S. president Donald Trump with the left-wing populism of former Bolivian president Evo Morales. Supporters of either of these two populists differ considerably in their attitudes and beliefs. The stark contrast between supporters of Trump and Morales is determined by a “host” or “thick” ideology – in the case of the above examples: right-wing and left-wing beliefs.
Yet, existing scales in the literature focus on capturing populist attitudes with a minimal definition that is supposed to apply equally to populism on the left and right (e.g., Hawkins & Kaltwasser, 2017). They do not include the thick ideologies that breathe life into populist attitudes. In this sense, there exists a gap between the theoretical conceptualization and the empirical construction of populist attitudes. This lack of inclusion may have led to several challenges in the literature like low predictive validity (e.g., Quinlan & Tinney, 2019), and conceptual issues like over-representing the political elite relative to other elites (Jungkunz et al., 2021).
In this investigation, we developed two new scales for the measurement of populist attitudes: left-wing and right-wing populist attitudes. These scales demonstrated additional variability beyond related constructs like authoritarianism and left-right political identification, and showed evidence of measurement invariance across the nationality, gender, and age. The inclusion of the host ideology also increased the construct validity of populist attitudes by including items concerning the economic and cultural elite.
Ideational Approach
The ideational approach is the leading framework that has defined populism in recent research (e.g., Hawkins & Kaltwasser, 2017; Mudde, 2004, 2007; also see Stanley, 2008). Proponents of the ideational approach argued that populism is defined by its central ideas, not by policies, leadership styles, or as a means to galvanize the masses (Mudde & Rovira Kaltwasser, 2013). Thus, populism can function as an ideology or a belief system, emphasizing a “cosmic struggle” between the people and the elite; this struggle can emerge irrespective of other contexts such as thick ideologies prominent in politics or inter-group dynamics (Hawkins & Kaltwasser, 2017).
The ideational approach has two central ideas: (a) society is divided by a clash between the people and elite, and (b) the will of people must be at the forefront of politics. Through this lens, populism is viewed as a “thin-centered” or “thinly-veiled” ideology which indicates that populism does not exist in its purest form. Rather, populism only materializes when it is paired with “thick” ideologies like nationalism or socialism (e.g., Mudde & Rovira Kaltwasser, 2017; Stanley, 2008). For instance, socialist beliefs shape the identity of the people/ingroup and the elite/outgroup as well as the issues that populists prioritize in left-wing populism (e.g., Otjes & Louwerse, 2015). In this context, socialist beliefs create a Marxist dualism of “haves” (the elite/outgroup) versus “have-nots” (the people/ingroup), while prioritizing working-class issues.
The first core idea suggests that “the people” (the populists’ ingroup) represent a virtuous and pure group who is mobilized by political rhetoric, whereas “the elite” are powerful, but corrupt establishments in economy, culture, or politics (e.g., Mudde, 2004; Mudde & Rovira Kaltwasser, 2017). Left-wing populists generally embrace some form of inclusionary socialism, creating a conception of the people/ingroup based on socioeconomic status (e.g., bottom 99% vs. top 1%). Ethnic minorities and other marginalized groups can be a part of the people/ingroup (March, 2017). Conversely, right-wing populists tend to embrace nativism – an exclusionary conception of the people/ingroup based on immutable features like race or ethnicity, and a conceptualization of the elite as powerful people who support non-natives and other minorities (Mudde, 2004; Staerklé & Green, 2018).
The second core idea indicates that populism is people-oriented, emphasizing the people’s sovereignty, and their centrality to politics (Canovan, 1999; Rooduijn, 2014). The concept of the will of the people speaks to the thinness of populism: the claim to serve the people’s will is often vague, who the people are is open to interpretation. However, this “imprecision” provides populists with great flexibility in mobilizing several social groups against the elite (Laclau, 2005). An example representing left-wing populism, the Occupy movement of 2011/2012, claimed to speak for the economically disadvantaged while maintaining an inclusive understanding of the entire bottom “99%.” In terms of right-wing populism, the Tea Party movement in the U.S. mobilized both conservatives and libertarians against economic bailouts for the elite (bankers, Democrats, and Hollywood) with implicitly racialized notions for “we, the people” (Mudde & Rovira Kaltwasser, 2017).
Populist Attitudes and Summary of Existing Scales
Several scholars have converted aspects of the ideational approach to describe an attitudinal construct, relevant to the general population (e.g., Akkerman et al., 2014; Castanho Silva et al., 2018; Oliver & Rahn, 2016; Schulz et al., 2018). All scales contain the two core ideas of the ideational approach: (1) clash of people and the elite, and (2) the people’s importance in political matters. For example, item content on popular sovereignty, and skepticism toward the elite and liberal democracy in these scales were meant to tap those two core ideas (Castanho Silva et al., 2020). Another commonality among the scales was that they were all designed with the view that populist attitudes are a thinly-centered or thinly-veiled concept consistent with the minimal definition of the ideational approach.
There are several key scales developed in the literature. See Table 1 for a summary of the scales described in the present section which were also used in Pilot Study 2 as an initial item pool. See Table S1 in the supplemental materials for all items used in the Pilot Study 2 exploratory factor analysis (EFA). Stanley (2011) produced one of the first conceptions of populist attitudes (8 items), with a Slovak sample, which was argued to measure four characteristics: (1) the homogeneity of the people/elite, (2) the idea that politics are controlled by the elite at the people’s expense, (3) majoritarianism/sovereignty of the people, and (4) Manichaeanism (good vs. evil or black vs. white thinking). In contrast, Akkerman et al. (2014) developed the most widely cited scale at the time of this writing with a Dutch sample. Their scale comprised of six items emphasizing the following ideas like (1) the people’s will, (2) lack of interest in compromising in politics, (3) homogeneity of the people, (4) the view that politicians fail to represent the people, and (5) Manichaeanism. Other unidimensional measures exist beyond the two mentioned. For instance, Elchardus and Spruyt (2016) created a four-item scale, with a Belgian Sample, that tapped the key themes: (1) anti-elitism, (2) the distinction between the people and elite, and (3) that the people should be listened to more by politicians. Hobolt et al. (2016) proposed a measure of populist attitudes, with a multi-national sample, that contains three themes across 8 items: (1) anti-elitism, (2) majoritarianism/challenges to representative democracy, and (3) attitudes toward outgroups like the rich or immigrants.
Author . | Dimensions & Item content . | Issues reported in the literature . |
---|---|---|
Stanley (2011) | 8 Items (4 reversed) | Lacks predictive & content validity (Akkerman et al., 2014). |
Unidimensional | Poor factor analytic fit & poor reliability (Castanho Silva et al., 2020) | |
Item content: Homogeneity of the people/elite, Anti-elitism, Manicheanism, Majoritarianism | Over represents political elite (Jungkunz et al., 2021) | |
Akkerman et al. (2014) | 6 Items (none reversed) | Lacks homogeneity of the people – content validity (Schulz et al., 2018) |
Unidimensional | Items are double-barreled (Wuttke et al., 2020) | |
Item content: Manicheanism, Anti-elitism, Popular sovereignty, Majoritarianism | Overrepresents political elite (Jungkunz et al., 2021) | |
Lacks predictive validity (see Quinlan & Tinney, 2019) | ||
Elchardus & Spruyt (2016) | 4 items (none reversed) | Low conceptual breadth & predictive validity (Castanho Silva et al., 2020) |
Unidimensional | Over represents political elite (Jungkunz et al., 2021) | |
Item content: Anti-elitism, Popular sovereignty, People-centrism | ||
Hobolt et al. (2016) | 8 items (1 reversed) | Over represents political elite (Jungkunz et al., 2021) |
Unidimensional | Low conceptual breadth & cross-national validity (Castanho Silva et al., 2020) | |
Item content: Anti-elitism, Popular sovereignty, Attitudes toward outgroups | Lacks explanatory power (Castanho Silva et al., 2022; Olivas Osuna & Rama, 2022; Quinlan & Tinney, 2019) | |
Castanho Silva et al. (2018) | 9 items (3 reversed) | Over represents political elite (Jungkunz et al., 2021) |
multidimensional | Not robust to changes in operationalization (Wuttke et al., 2020) | |
Factors: People-centrism, Manichaeanism, Anti-elitism | ||
Oliver & Rahn (2016) | 12 items (none reversed) | Low cross-national validity (Castanho Silva et al., 2020) |
multidimensional | Not robust to changes in operationalization (Wuttke et al., 2020) | |
Factors: National affiliation, Mistrust of experts, Anti-elitism | ||
Schulz et al. (2018) | 12 items (none reversed) | Over represents political elite (Jungkunz et al., 2021) |
multidimensional | Low cross-national validity & conceptual breadth (Castanho Silva et al., 2020) | |
Factors: Homogeneity of the people, Sovereignty of the people, Anti-elitism | Not robust to changes in operationalization (Wuttke et al., 2020) | |
Lacks predictive validity (see Quinlan & Tinney, 2019) |
Author . | Dimensions & Item content . | Issues reported in the literature . |
---|---|---|
Stanley (2011) | 8 Items (4 reversed) | Lacks predictive & content validity (Akkerman et al., 2014). |
Unidimensional | Poor factor analytic fit & poor reliability (Castanho Silva et al., 2020) | |
Item content: Homogeneity of the people/elite, Anti-elitism, Manicheanism, Majoritarianism | Over represents political elite (Jungkunz et al., 2021) | |
Akkerman et al. (2014) | 6 Items (none reversed) | Lacks homogeneity of the people – content validity (Schulz et al., 2018) |
Unidimensional | Items are double-barreled (Wuttke et al., 2020) | |
Item content: Manicheanism, Anti-elitism, Popular sovereignty, Majoritarianism | Overrepresents political elite (Jungkunz et al., 2021) | |
Lacks predictive validity (see Quinlan & Tinney, 2019) | ||
Elchardus & Spruyt (2016) | 4 items (none reversed) | Low conceptual breadth & predictive validity (Castanho Silva et al., 2020) |
Unidimensional | Over represents political elite (Jungkunz et al., 2021) | |
Item content: Anti-elitism, Popular sovereignty, People-centrism | ||
Hobolt et al. (2016) | 8 items (1 reversed) | Over represents political elite (Jungkunz et al., 2021) |
Unidimensional | Low conceptual breadth & cross-national validity (Castanho Silva et al., 2020) | |
Item content: Anti-elitism, Popular sovereignty, Attitudes toward outgroups | Lacks explanatory power (Castanho Silva et al., 2022; Olivas Osuna & Rama, 2022; Quinlan & Tinney, 2019) | |
Castanho Silva et al. (2018) | 9 items (3 reversed) | Over represents political elite (Jungkunz et al., 2021) |
multidimensional | Not robust to changes in operationalization (Wuttke et al., 2020) | |
Factors: People-centrism, Manichaeanism, Anti-elitism | ||
Oliver & Rahn (2016) | 12 items (none reversed) | Low cross-national validity (Castanho Silva et al., 2020) |
multidimensional | Not robust to changes in operationalization (Wuttke et al., 2020) | |
Factors: National affiliation, Mistrust of experts, Anti-elitism | ||
Schulz et al. (2018) | 12 items (none reversed) | Over represents political elite (Jungkunz et al., 2021) |
multidimensional | Low cross-national validity & conceptual breadth (Castanho Silva et al., 2020) | |
Factors: Homogeneity of the people, Sovereignty of the people, Anti-elitism | Not robust to changes in operationalization (Wuttke et al., 2020) | |
Lacks predictive validity (see Quinlan & Tinney, 2019) |
Populist attitudes are not always viewed as a unidimensional construct, contrary to the four scales discussed above. Other scholars conceptualized populism with multiple dimensions (i.e., Castanho Silva et al., 2018; Oliver & Rahn, 2016; Schulz et al., 2018). For instance, Castanho Silva et al. (2018) measured populist attitudes with the following empirical dimensions: anti-elitism, people-centrism (similar to “the will of the people”) and Manichaeanism. Oliver and Rahn’s (2016) conception of populist attitudes also yielded three dimensions: anti-elitism, mistrust of experts, and national affiliation. The latter two dimensions represented skepticism to intellectuals and experts, and identification with nationality, respectively. Finally, Schulz et al. (2018) developed three factors: homogeneity of the people, sovereignty of the people, and anti-elitism.
There are some similarities and discrepancies among the scales. All authors seemed to agree with anti-elitism being an integral aspect of populist attitudes. Yet, item content somewhat differed. For example, the Schulz et al. (2018) anti-elitism sub-scale included low political efficacy (“Politicians are not really interested in what people like me think,” whereas the Castanho Silva et al. (2018) sub-scale emphasized perceived corruption (“Quite a few of the people running the government are crooked”). The other dimensions or item content varied by authors more substantially. Indeed, Oliver and Rahn (2016), Elchardus and Spruyt (2016), and Hobolt et al. (2016) included content concerning national affiliation, anti-intellectualism, or intergroup attitudes, unique to their respective scale. Other differences can be observed: Manichaeanism is a factor only in the Castanho Silva et al. (2018) scale and the homogeneity of the people factor is unique to Schulz et al. (2018).
Although existing scales vary in terms of item content and dimensionality, a broad pattern of themes emerged across seven scales in the literature (Castanho Silva et al., 2020). The three themes shared by existing scales in literature are people-centrism, anti-elitism, and anti-pluralism. Anti-pluralism underlies the idea that outgroups should be excluded from politics or are morally questionable (Castanho Silva et al., 2018; Oliver & Rahn, 2016). Anti-elitism describes the people’s opposition to the government, politicians, and the elite. People-centrism characterizes the people’s will in politics and their admirable qualities like honesty (e.g., Akkerman et al., 2014; Oliver & Rahn, 2016).
Limitations of Existing Scale/Items. Despite great strides in understanding populist attitudes, existing measures face limitations. Jungkunz et al. (2021) argued that existing scales fail to explain populist voting in countries that have a populist in power because these inventories contextualized the elite as solely politicians. Potentially, these scales may not predict antagonism against non-political elites like economists and academics, and lack predictive validity in electoral years in which a populist is already in office. Other scholars like Castanho Silva et al. (2022) and Quinlan and Tinney (2019) suggested that there is lack of explanatory power in populist attitudes; traditional variables or “old reliables” – like left-right identification – were better predictors of support for Donald Trump in the 2016 US election or Jair Bolsonaro in the 2018 Brazil election. Rooduijn (2019) and Castanho Silva et al. (2020) also contested the validity of several scales in that they are not sufficiently distinguished from other attitudinal concepts like political cynicism or even simply anti-elitism (also see Olivas Osuna & Rama, 2022).
A Missing Piece of the Puzzle. We argue that validity concerns of populist attitudes could be reduced if they are integrated with a host ideology. A premise to populism is that it does not normally exist in its purest form and its morphology is dictated by the host ideology (Mudde & Rovira Kaltwasser, 2017). In this sense, existing scales attempt to measure a concept that does not materialize because it is missing a key ingredient – the host ideology. There is a gap between how populist attitudes are theoretically conceptualized relative to how they are empirically constructed. From an attitudinal perspective, without a thick ideology, an individual may hold anti-establishment beliefs, but it would be unclear as to which groups they identify with, who they oppose, and the issues they prioritize.
Our Approach
Despite weaknesses reported in the literature, existing scales/items contain important aspects of populist attitudes, as outlined by the ideational approach. They can be adjusted to include a thick ideology – in this case, the left-right political spectrum. As such, we decided to modify existing inventories to suit the needs of left-wing populist (LWP), and right-wing populist (RWP) attitudes as opposed to developing a completely new set of items.
Predicted Factor Structure
Castanho Silva et al. (2020) observed three themes across seven measures of populist attitudes: people-centrism, anti-elitism, and anti-pluralism, as mentioned above. However, these themes have not yet been tested as factors.
Our goal was to tap those three themes as three separate factors for the left and right scales. Our reasoning was twofold. Our interpretation of the ideational approach would have further muddied the nature of populist attitudes with yet another set of disparate factors (see Kenny & Bizumic, 2023). Second, the themes argued by Castanho Silva et al. (2020) showed that despite disagreements on the operationalization of populist attitudes, there seems to be a consensus on what populist attitudes entail. Thus, Castanho Silva et al.’s (2020) themes provided us with an educated guess on the factor structure of both scales.
We also expected LWP and RWP to be higher-order models as they can best represent populist attitudes as an attitudinal syndrome. Several scholars conceptualized populist attitudes as a latent, multidimensional construct (see Wuttke et al., 2020 for a review, p. 357-358) which suggests that populist attitudes have necessary components that must be present to exist. Therefore, the left-wing and right-wing variants of the three factors must be realized in order for the totality of left-wing and right-wing populist attitudes to manifest.
Left-Wing Populist Attitudes
See Figure 1 for our hypothesized three-factor model of left-wing populist attitudes. We modified existing items to have an economic focus, as LWP often entails the support for the working-class against the economic elite such as bankers and Big Business (Akkerman et al., 2017; March, 2017). Anti-pluralism items emphasized the division of social groups (e.g., the working class), and majoritarianism in the context of economic issues. Accordingly, anti-elitism items focused on economic elites like the rich and affluent, and corporations. Finally, the people-centrism factor can be described as the idea of the working class having the most political power.
Right-Wing Populist Attitudes
See Figure 2 for our hypothesized tripartite model of RWP attitudes. Items in this scale focused on the notion of the people as “True [nationality]” as an identity common to right-wing populists who feel socially and culturally threatened and similar to terms used by Donald Trump (e.g., Schertzer & Woods, 2021). Indeed, RWP rhetoric juxtaposes indigenous or native people with the cultural elite (March, 2017). Anti-pluralism items underscored the division of social groups, and majoritarianism in the context of cultural issues. Anti-elitism highlighted opposition to cultural elites like bureaucrats, experts, and academics, and minorities. Finally, the people-centrism factor focused on the idea of “True [nationality]” being at the center of politics.
Parallel Construction
Our approach toward scale development, via modification of existing items in the literature, draws on Conway and colleagues’ (2018) parallel construction approach which was utilized in developing a measure of left-wing authoritarianism (LWA) based on Altemeyer’s (1998) right-wing authoritarianism (RWA) scale. Specifically, these authors developed a new LWA scale by changing the context of Altemeyer’s (1998) RWA scale. For example, items that indicated the desire for a strong traditional leader from the RWA scale were changed to emphasize the need for a mighty liberal leader in the LWA scale. This modification galvanized research on LWA, and their scale performed better than some of the existing measures. The Conway et al. (2018) LWA scale did not have floor effects (B. Altemeyer, 1996) nor was it only applicable to those who hold extreme left-wing views (Van Hiel et al., 2006). Whereas authoritarianism and populism are clearly distinct concepts with different views on the establishment, and political identities and commitment (Jami, 2025), we believed that the parallel construction approach would be fruitful in the development of parallel measures of left-wing and right-wing populism.
Operationalization of Populist Attitudes
Especially in psychology, most multi-dimensional constructs like authoritarianism (e.g., B. Altemeyer, 1996; Conway et al., 2018) or social dominance orientation (e.g., Ho et al., 2015) assume an additive model. That is, items across different factors are added up and averaged for a composite score; a higher score on one factor is averaged with a lower score on a different factor. Wuttke et al. (2020) critiqued this approach by arguing that to characterize a person as a populist, they need to endorse all elements of populism, not just an average across the elements. These authors advocated for a non-compensatory approach based on which they only used the lowest score on all populism factors to characterize a respondent’s populist attitudes. We found Wuttke et al.’s (2020) argument to be compelling, but because results are nearly identical, we report only findings based on the compensatory approach, and refer the reader to our supplementary materials for results based on the alternative operationalization (see Section S10, Tables S8-S10); however, we highlight and discuss any difference in the findings in the main text.
Research Overview
Initially, we conducted two exploratory studies which are only summarized briefly, with full reports being available in the supplementary materials (see Section S1 and S2, respectively). Pilot Study 1 asked respondents as to who they view as the elite to assess if people perceive non-political elites like economic and cultural elites. Pilot Study 2 used EFA to assess whether Castanho Silva et al.’s (2020) themes map onto factors using existing inventories. Pilot Study 2 also helped narrow down the existing item pool which was then modified to create measures for left-wing and right-wing populist attitudes.
Study 1 and Study 2 were more confirmatory in nature. In both studies, we examined key predictive outcomes (e.g., electoral skepticism). Study 2 also established the nomological net of LWP and RWP attitudes and their measurement equivalence across the U.S., Canada, and Australia. For a flowchart of our validation process, see Figure 3.
Pilot Study 1
We conducted Pilot Study 1 to determine the salience of non-political elite among lay individuals. Our motivations to conduct this study were based on the idea that populism often begins with antagonism against the elite (e.g., Laclau, 2005; Mansbridge & Macedo, 2019). We collected 111 American respondents form Amazon’s Mechanical Turk (MTurk), and asked them open-ended questions as to who they believe the elite are. We found that economic elites were mentioned the most, followed by cultural elites and political elites. The latter elite is over-utilized in the populist attitudes literature (see Jungkunz et al., 2021), despite being the least likely to be mentioned in our pilot study. Overall, these findings provided additional justification for our measures of LWP and RWP attitudes. See Section S1 of our supplemental materials for a full report on this study.
Pilot Study 2
Pilot Study 2 (n = 230) entailed an integrative EFA across seven different measures of populist attitudes (see Table 1 for summary of all seven scales), explored the dimensionality of populist attitudes, and selected the best candidate items to be integrated with the left-right political context. We found the themes argued by Castanho Silva et al. (2020) as three factors. Anti-pluralism (Factor 1) describes Manichaeanism (good versus evil), majoritarianism, homogeneity of social groups, and the lack of acknowledgement for other groups or different viewpoints which accounted for 16% of the variance. Anti-elitism (Factor 2) accounted for 11% of the variance that focused on cynical views of government and low political efficacy. People-centrism (Factor 3) explained 9% of the variance, containing notions about the people’s focus on having the most say in political matters (see Section S2, Tables S1-S2 for a full report).
Study 1
We adopted Conway et al.’s (2018) parallel construction by using the same item stems, but varying the ideas, people or groups that the items referred to. The right-wing items contained a collective identity around “true Americans,” and antagonism toward cultural elite which include academics and scientists. In contrast, left-wing items referred to the “the working class” and opposition against the economic elite like bankers and Big Businesses. Thus, terms like the “the people,” were changed to “True American” to fit the RWP context, and “the working class” to fit the LWP context. Moreover, some of items that in the original version used the terms “government” or “politics” were changed to focus on culture for the populist right and on economics for the populist left. We did not change all content for terms, “government” or “politics” because opposition to those groups still matter for any type of populism. An example of transformation can be seen with an item from the Schulz et al. (2018) scale, “People like me don’t have much say in what government does.” To fit the RWP context, we changed this item to, “True Americans don’t have much say in what happens to American culture.” To fit the LWP context, we modified that original item to, “Members of the working class don’t have much say in what happens to the American economy.” Section S3, Table S3 (supplemental materials) contains the original items alongside the right-wing and left-wing variants.
We also note three other design choices. First, we removed redundant items. Both the Akkerman et al. (2014) scale and the Hobolt et al. (2016) scale contained an item about selling out one’s principle, for instance (see Section S4, supplemental materials, for an extended explanation). Second, we added two additional items that did not make the cut-off in Pilot Study 2’s EFA from the Oliver and Rahn (2016) scale (“I’d rather put my trust in the wisdom of ordinary people than the opinions of experts and intellectuals”) and the Schulz et al. (2018) scale (“ordinary people all pull together”). The third design choice was the development of reverse-coded items based on the Pilot Study 2 item pool (Section S5, Table S4 supplemental materials for the reverse-coded items along with their original counterparts). This was done because we noticed a conspicuous absence of such items in current populism scales.
We first conducted an EFA to confirm our three dimensions with the goal of retaining the four best items for each factor based on a combination of the best standardized factor loadings, low cross-loadings, as well as examining if the items correlated with other items, and whether or not they improve or reduce reliability based on Cronbach’s alpha and McDonald’s omega. We then examined several key criterion variables: evaluations of social groups, institutions, and ideas/beliefs, satisfaction with democracy, trust in government, peaceful and violent future activism, whether President Biden should use more executive orders, belief in 2020 electoral fraud, perceived Christian discrimination, and whether President Trump was responsible for the January 6 insurrection.
The first four outcomes were included because both forms of populism entail different evaluations of social groups – negative views of outgroups, encouragement to participate in activism by their populist leader, and negative views on institutions and the status quo (e.g., de la Torre, 2016; Homolar & Scholz, 2019; Norris & Inglehart, 2019). The belief that Joe Biden should utilize more executive orders was intended to assess majoritarianism and illiberalism (a way to bypass the ordinary process of legislation with its checks and balances). The belief that Christians are being discriminated against measured the perceived cultural threat often linked to the populist right (e.g., Betz, 1994; Norris & Inglehart, 2019). Belief in 2020 electoral fraud should illustrate the conspiratorial elements of populism (e.g., Castanho Silva et al., 2017), and finally, whether one believes Donald Trump was responsible for the January 6 insurrection represented a point of separation between the populist left and right. The populist left is expected to believe Trump is responsible whereas the populist right should believe the opposite.
Participants
We acquired an American sample using MTurk with a final sample of N = 255 (52.5% male) after data quality checks. These micro-workers were compensated $1.80 for their participation. With all our samples, we engaged in several data quality checks. Respondents who completed fewer than 50% of the survey items, had duplicates entries, were outside of the U.S. (based on IP address), or failed an attention check were removed from analysis (e.g., Chmielewski & Kucker, 2020; Kennedy et al., 2020). Based on these data quality checks, we removed 119 respondents.
Most respondents identified as White (78.4%), followed by Asian (7.8%), Black (6.3%), non-White Latino/Hispanic (5.9%), and “other” (1.6%). Respondents were politically moderate (M = 2.87, SD = 1.48) on a 1-5 left-right identification scale and most had attained at least a high school diploma (87.8%).
Procedure and Materials
The survey was hosted on Qualtrics®. Respondents provided consent and then answered demographic items. All outcome variables and populism items were measured on a 7-point Likert-type scale.
Outcomes. Evaluation of various groups was measured with a total of 12 items: six for right-wing groups (e.g., “Your overall view of American patriots”, “Your overall view of law enforcement/police”, α = .88), and another six for left-wing groups (e.g., “Your overall view of immigrants”, “Your overall view of LGBTQ+ members”, α = .90). Trust in institutions was measured with five items (e.g., “To what extent do you trust the Federal Government?”, and “To what extent do you trust the news media?”; α = .93).
Peaceful future activism was assessed with five items (α = .90; e.g., “How likely do you think you will do any of the following forms of political actions?: Attending peaceful demonstrations,” “contacting representatives”). Violent future activism was assessed by a single item that asked respondents to what extent they would be “picking up arms” against the government in the future. A single item was used to measure satisfaction with American democracy. And finally, four other single items assessments asked to what extent respondents agree/disagree on whether President Biden should rely more on executive orders, the legitimacy of 2020 electoral fraud claims, perceived discrimination against Christians in the US, and whether Donald Trump was responsible for the January 6 insurrection.
Analytical Framework
We used exploratory factor analysis to reduce our item pool further along with examining the newly crafted reverse-coded items. Though we expect to find three factors, we utilized parallel analysis, Velicer’s minimum average partial (MAP) and the Kaiser criterion (eigenvalues > 1) to determine the number of factors to extract. Our item selection was based on a combination of minimum factor loading of at least .40, inter-item correlations, and as well how the items might affect reliability/dimensionality. We also considered theoretical and practical considerations (e.g., potentially avoiding some redundant items). Because we expected factors to correlate, we used promax rotation. See Table S5a and S5b in Section S6 of the supplemental materials for the item loadings.
Results and Discussion
CFA and Psychometrics
Left-Wing Scale. Parallel analysis, MAP and Kaiser criterion suggested that 3 factors to extract. Table S5a contains all factor loadings for the LWP items. Unexpectedly, people-centrism and anti-pluralism belonged to one factor. For this combined factor, we decided to keep the three best items that reflect people-centrism and anti-pluralism, respectively, based on high loadings on the primary factor (> .50) and low cross-loadings elsewhere (< .20). This combined factor contains a total of six items (see Table 2a for the finalized set of items, Figure 4 for a revised factor structure of LWP).
Left-wing populism . |
---|
Anti-pluralism/People-centrism factor |
Anti-pluralism facet |
1. You can tell if a person is good or bad if you know their views on inequality. |
2. The will of working people should always prevail, even over the rights of the rich and affluent. |
3. I’d rather put my trust in the wisdom of the working class than the opinions of business professionals and bankers. |
People-centrism facet |
1. The politicians in Congress need to follow the will of the working class. |
2. The working class, not the politicians, should make our most important policy decisions. |
3. The will of the working class should be the highest principle in this country’s politics. |
Anti-elitism factor |
1. It doesn’t really matter who you vote for because the economic elite control both political parties. |
2. The government is pretty much run by a few wealthy business people and bankers looking out for themselves. |
3. Members of the working class don’t have much say in what happens to the American economy. |
4. Corruption such as bribe taking is widespread among business professionals and bankers in the United States. |
Left-wing populism . |
---|
Anti-pluralism/People-centrism factor |
Anti-pluralism facet |
1. You can tell if a person is good or bad if you know their views on inequality. |
2. The will of working people should always prevail, even over the rights of the rich and affluent. |
3. I’d rather put my trust in the wisdom of the working class than the opinions of business professionals and bankers. |
People-centrism facet |
1. The politicians in Congress need to follow the will of the working class. |
2. The working class, not the politicians, should make our most important policy decisions. |
3. The will of the working class should be the highest principle in this country’s politics. |
Anti-elitism factor |
1. It doesn’t really matter who you vote for because the economic elite control both political parties. |
2. The government is pretty much run by a few wealthy business people and bankers looking out for themselves. |
3. Members of the working class don’t have much say in what happens to the American economy. |
4. Corruption such as bribe taking is widespread among business professionals and bankers in the United States. |
For the anti-elitism factor, we also identified the best items based on high factor loadings (>.60) and low cross-loadings with other factors (<.20): Item 2, 8, 9, 21, 28, and 29 all qualified. To reduce the items to the best four, we examined inter-item correlations and whether dropping an item improved reliability or dimensionality. Clark and Watson (1995) recommended that items should be correlated by at least .15 with other items on the scales. This applied to all items except Item 2, which was weakly correlated with Item 20, 26, and 27 in the anti-pluralism/people-centrism factor. Next, we examined reliability and dimensionality of the LWP scale with the remaining items. The current scale produces a Cronbach’s α = .88 and Macdonald’s Ω = .63, but the removal of Item 9 markedly improved dimensionality, Ω = .69 with minimal changes to internal consistency, α = .87. Removal of any other item did not yield any improvement, thus leaving Item 8, 21, 28, and 29 as the best four items.
We did not include any of the items from the third column (Table S5a) as they were reverse-coded items that were meant to load on one of our three planned factors and instead loaded separately. The overall LWP scale contained 10 items across 2 factors: anti-elitism (4 items) and anti-pluralism/people-centrism (6 items).
Right-Wing Scale. Based on parallel analysis, MAP and Kaiser criterion, we extracted 4 factors with promax rotation (Table S5b). For the people-centrism factor, the best four items loaded strongly on their factor (> .65) and weakly with the other factors (<.20); all finalized RWP items can be seen on Table 2b.
Right-wing populism . |
---|
Anti-pluralism |
1. Although true Americans are very different from each other, when it comes down to it they all think the same. |
2. In politics, what people call compromise on religious lifestyle issues is really just selling out on one’s principles. |
3. The will of true Americans should always prevail, even over the rights of racial and sexual minorities. |
4. I’d rather put my trust in the wisdom of true Americans than the opinions of scientists and academics. |
Anti-elitism |
1. Most cultural elites do not care about true Americans. |
2. True Americans are prevented from improving their lives by the actions of unaccountable cultural elites. |
3. The government is pretty much run by a few bureaucrats and so-called “experts” looking out for themselves. |
4. Bureaucrats and so-called “experts” talk too much and take too little action. |
People-centrism |
1. The politicians in Congress need to follow the will of true Americans. |
2. True Americans should be asked whenever important decisions are taken. |
3. True Americans, not the politicians, should make our most important policy decisions. |
4. The will of true Americans should be the highest principle in this country’s politics. |
Right-wing populism . |
---|
Anti-pluralism |
1. Although true Americans are very different from each other, when it comes down to it they all think the same. |
2. In politics, what people call compromise on religious lifestyle issues is really just selling out on one’s principles. |
3. The will of true Americans should always prevail, even over the rights of racial and sexual minorities. |
4. I’d rather put my trust in the wisdom of true Americans than the opinions of scientists and academics. |
Anti-elitism |
1. Most cultural elites do not care about true Americans. |
2. True Americans are prevented from improving their lives by the actions of unaccountable cultural elites. |
3. The government is pretty much run by a few bureaucrats and so-called “experts” looking out for themselves. |
4. Bureaucrats and so-called “experts” talk too much and take too little action. |
People-centrism |
1. The politicians in Congress need to follow the will of true Americans. |
2. True Americans should be asked whenever important decisions are taken. |
3. True Americans, not the politicians, should make our most important policy decisions. |
4. The will of true Americans should be the highest principle in this country’s politics. |
The anti-pluralism and anti-elitism factors had several strong items that loaded at least .50 for the primary loading with low cross-loadings: Item 3,4, 13, 15, 18, 25(R), 26, 27 for anti-pluralism and Item 6, 16, 19, 20, 22, and 30 for anti-elitism. To reduce items further, we examined which items produced especially low inter-item correlations (r < .15). We removed Item 6 and Item 25(R) because they lacked adequate correlations with at least five other items (25% of the scale). We next removed items based on dimensionality: Item 27 and 16 were the two poorest items. The removal of these two items improved omega (.69) while alpha remained high (.92). Because anti-pluralism still had 6 items compared to the other two dimensions which each contained 4 items, we removed Item 15 and 18 which helped produce the highest levels of Ω = .73 while maintaining a strong α = .91. The final column of Table S5b contained our reverse coded item. Given that they were meant to be part of the three hypothesized factors and not to load separately on a separate factor, we did not retain these items.
Correlations
The two scales were highly correlated, r = .51, p < .001. LWP was unrelated to right-wing identification, r = -.08, p = .10, whereas RWP was strongly related to right-wing identification, r = .43, p < .001. Neither scale was correlated with education.
Multiple Regression
Table 3 shows the main results for both scales with standardized betas as well as the adjusted ΔR2 when the two populism scales are included in the model. All the regression models controlled for left-right identification, gender, race/ethnicity, education, and age; see Section S6, Tables S6a-S6c in the supplemental materials for the full tables. The two scales showed evidence of explaining variance beyond typical variables like the left-right identification. After including both scales, the adjusted ΔR2 improved on average by 0.11 (range = 0.06-0.19).
. | Right-wing populism . | Left-wing populism . | Adjusted . |
---|---|---|---|
β | β | ΔR2 | |
Evaluation: left-wing groups | -.41 | .43 | .12 |
Evaluation: right-wing groups | .59 | -.30 | .18 |
Satisfaction with democracy | .35 | -.18a | .05 |
Trust in institutions | .30 | -.14 | .05 |
Peaceful future activism | .03 | .35 | .12 |
Biden executive orders | .07 | .27 | .13 |
2020 electoral fraud | .45 | .07b | .19 |
Perceived Christian discrimination | .30 | .07 | .09 |
Trump responsible for January 6 | -.34 | .21 | .06 |
. | Right-wing populism . | Left-wing populism . | Adjusted . |
---|---|---|---|
β | β | ΔR2 | |
Evaluation: left-wing groups | -.41 | .43 | .12 |
Evaluation: right-wing groups | .59 | -.30 | .18 |
Satisfaction with democracy | .35 | -.18a | .05 |
Trust in institutions | .30 | -.14 | .05 |
Peaceful future activism | .03 | .35 | .12 |
Biden executive orders | .07 | .27 | .13 |
2020 electoral fraud | .45 | .07b | .19 |
Perceived Christian discrimination | .30 | .07 | .09 |
Trump responsible for January 6 | -.34 | .21 | .06 |
Note: Entries reflect standardized regression coefficients. Bold face indicates p < .05; a indicates that result is insignificant if operationalization changes; b indicates that result is significant if operationalization changes. Adjusted ΔR2 indicates variance explained by right-wing and left-wing populism are included in the model after controlling for left-right identification, gender, race, education and age.
As anticipated, RWP predicted a negative evaluation of left-wing groups, a positive evaluation of right-wing groups, violent activism in the future, and the beliefs that there was electoral fraud in the 2020 election, perceived greater Christian discrimination, and that Donald Trump was not responsible for the January 6 insurrection, but was unrelated to peaceful activism. Unexpectedly, RWP was also linked to higher satisfaction with democracy and higher trust in institutions (see Table 3). These surprising findings may have to do with the idea of democracy and related institutions as hallmarks of American culture. That is, right-wing populists may like existing institutions, but may not necessarily be satisfied with those in power, especially those who are perceived to be aiding the elite. These results may also have to do with the finding that conservatives/right-wingers generally seem happier about society compared to liberals (Napier & Jost, 2008). Nevertheless, the majority of the results are consistent with the literature on right-wing populism (e.g., Betz, 1994; Norris & Inglehart, 2019).
As also shown in Table 3, LWP was linked to a positive evaluation of left-wing groups, a negative evaluation of right-wing groups, peaceful activism, and the beliefs that Joe Biden should use executive orders, and that Donald Trump was responsible for the January 6 insurrection. These findings are consistent with some scholars who argued that left-wing populists engage in peaceful grassroots activism (e.g., Ivaldi et al., 2017; March, 2017), but have some illiberal views such as the usage of executive orders (a way to bypass checks and balances). Interestingly, we found a couple of discrepancies when we utilized the non-compensatory approach of Wuttke et al. (2020). LWP was related to beliefs in 2020 electoral fraud unlike with the additive approach presented in the main text and the significant finding for low satisfaction with democracy was not replicated. LWP was unrelated to trust in institutions nor violent activism in the future. Overall, results are supportive of LWP’s predictive abilities.
Replication of Study 1
With another U.S. MTurk sample (N = 200; Section S7, supplemental materials for a full report), the 10-item LWP model showed some evidence for replication, CFI = .960, TLI = .947, RMSEA = .085, 90% [.062, .0108], SRMR = .059 using diagonally least weighted squares with robust standard errors (WLSMV); only RMSEA was a bit high. An 11-item RWP model also produced a good fit, CFI = .950, TLI = .933, RMSEA = .098, 90% [.078, .119], SRMR = .064 with the exception of RMSEA.1 In sum, the replication of the Study 1 models had most indices that were sufficient; still, these results necessitate another replication of our models.
Study 2
In Study 2, we tested the invariance of both models across three countries (USA, Canada and Australia), gender (male/female), age groups (18-24 years, 25-34 years, 45-54 years, and 55 and above), examined predictive validity, and the nomological net (theoretical validity). We chose these three countries for several reasons. First, all three countries are democratic, which most often is the typical context for the rise of populism (Mudde & Rovira Kaltwasser, 2017). Second, each country has some elements of populism or populist discourse. The USA is prominent in populist ideas among the population, highlighted by Donald Trump’s two electoral victories. Though not as widespread, populism does exist in Canada and Australia. For example, in Canada, some scholars describe Jagmeet Singh (New Democratic Party) and Maxime Bernier (People’s Party) as left-wing and right-wing populists, respectively, embracing/utilizing populist discourse to mobilize segments of the population (e.g., Moir, 2023; Turcotte & Raynauld, 2023). Australia, on the other hand, has a longstanding right-wing populist party, One Nation, and a Green Party that engages in some left-wing populist discourse (e.g., Gauja et al., 2023; Kenny & Bizumic, 2023).
The differences in the electoral success of populist actors may relate to the prevalence of populist attitudes in the population. Hawkins et al. (2020) argued that societies with more pressing political issues (e.g., inequality), have populations whose latent populist attitudes are more likely to be activated compared to societies with less pressing issues. In this sense, we can test the equivalence of these scales in a society that has a significant rise in populism (USA), which presumably has major political issues, and two countries with, at least by comparison, more modest instances in Australia and Canada. Thus, we tested if the prevalence of populism across three different democratic contexts does influence the invariance of our scales.
In terms of predictive validity, because populists generally perceive society to be on the decline, we assessed respondents’ confidence that society will cease to exist in 100 years and their confidence that economic inequality will erode society (e.g., Betz, 1994; Elchardus & Spruyt, 2016; Norris & Inglehart, 2019). We also included items on future peaceful activism and beliefs regarding electoral integrity (e.g., votes counted fairly) and media coverage (e.g., journalists are bought).
We also sought to connect LWP and RWP attitudes with key theoretical variables: right-wing authoritarianism (RWA), left-wing authoritarianism (LWA), nativism, ethnonationalism, civic nationalism, relative deprivation, anomie, and low political efficacy. The first five are ideological constructs often associated with populism (e.g., Mudde & Rovira Kaltwasser, 2017; Norris & Inglehart, 2019; Vasilopoulos & Jost, 2020). We predicted RWA and LWA to be linked to RWP and LWP attitudes, respectively. Moreover, we suspected that LWP would be linked to civic nationalism, but inversely to nativism, ethnonationalism, but RWP is anticipated to show the opposite pattern. Finally, relative deprivation, anomie, and low political efficacy should be linked to both types of populist attitudes, as populists have a general sense of cynicism and disenfranchisement (e.g., Jami & Kemmelmeier, 2021).
Participants
Participants were recruited via Prolific, consisting of samples from the U.S., Canada, and Australia and were paid $1.75 USD. About 300 respondents were collected from each country (N = 909). We used the same data checks described in Study 1 and removed 10 respondents total. See section S8 in supplemental materials for descriptive statistics of each country.
Procedure and Materials
The survey was hosted on Qualtrics®. Respondents completed a consent form and then answered questions concerning the same demographic information mentioned in Study 1. All respondents completed both populism scales though the order was randomized. Afterwards, respondents completed a set of criterion variables and theoretical measures which were also presented in a randomized order. To fit the Australian and Canadian contexts, we changed the term “True American,” to the term “True Australian” and “True Canadian,” respectively. We also changed the term “federal government” to parliament when appropriate.
Theoretical Variables
RWA was measured by the very short authoritarianism scale by Bizumic and Duckitt (2018) which contains six items (three reversed; α = .77) such as, “The facts on crime and the recent public disorders show we have to crack down harder on troublemakers, if we are going preserve law and order.” LWA was measured with the LWA-4 scale by Conway et al. (2022) which has four items (α = .87; e.g., “This country would work a lot better if certain groups of Christian troublemakers would just shut up and accept their group’s proper place in society”) with no reverse-coded items. Nativism was assessed by a five-item index (e.g., “immigrants take important social services away from real Americans”) which included one reverse-coded item (α = .87) created by Young et al. (2019). Ethno-nationalism was measured by a six-item measure (α = .88; e.g., “My first loyalty is to the heritage of my ancestors, their language and their religion.”) whereas civic nationalism (r = .68; e.g., “Ours should be a plural nation celebrating diversity and allowing for many different cultures”) was measured using a two-item scale (Saucier, 2014).
We utilized a measure on relative deprivation (α = .82; e.g., “If we need anything from the government, people like me always have to wait longer than others,” Bos et al., 2020 ), and two-item scales for both internal (r = .71; e.g., “I am good at understanding and assessing important political issues,”) and external political efficacy (r = .67; e.g., “Politicians strive to keep in close touch with the people.”), developed by Groskurth et al. (2021). Participants also completed a 12-item measure of anomie with two factors, one regarding society (α = .80; e.g., “People think that there are no clear moral standards to follow”) and another for government/leadership (α = .85; “The government uses its power legitimately” [reversed]) which contained four reverse-coded items (Teymoori et al., 2016).
Criterion Variables
We measured seven items on future political activism (α = .88; e.g., “With current politics in mind, how likely in the future are you to join a strike?”), an item assessing whether they will encourage others to take political action in the future, confidence in society existing in next 100 years and whether economic inequality will ruin society, and five items assessing the perceived legitimacy of the next electoral cycle: votes will be counted fairly, whether TV news and journalists favor the establishment, respectively, perceptions that politicians are bought by special interests, and another on the suppression of non-establishment candidates.
Analytic Framework
For measurement invariance/multigroup factor analysis (MGFA), we examined model fit via overall model as well improvements in model fit when testing a sequence of nested, yet stringent models. Changes in CFI and RMSEA were used to determine if the assumption of invariance was met (e.g., Chen, 2007; Cheung & Rensvold, 2002). Specifically, a .01 change in CFI and/or a .015 change in RMSEA indicate meaningful differences in model fit. Both scales were subjected to three measurement models testing for configural, metric, and scalar invariance with each model having more stringent constraints than the previous model (e.g., Fischer & Karl, 2019; Putnick & Bornstein, 2016). We conducted correlations between the two scales and theoretical constructs to establish the nomological net, and regression analysis to predict outcome variables for predictive validity.
Results and Discussion
Measurement Invariance and Psychometrics
All measurement invariance tests are summarized in Table 4 with WLSMV.
Test . | SRMR . | RMSEA . | CFI . | TLI . |
---|---|---|---|---|
Country invariance | ||||
LWP | ||||
Configural | .061 | .116 [.106, .126] | .962 | .949 |
Metric | .066 | .096 [.086, .105] | .970 | .966 |
Scalar | .062 | .077 [.070, .084] | .965 | .978 |
RWP | ||||
Configural | .041 | .083 [.074, .093] | .987 | .982 |
Metric | .048 | .068 [.059, .077] | .990 | .988 |
Scalar | .044 | .061 [.054, .069] | .985 | .990 |
Gender invariance | ||||
LWP | ||||
Configural | .058 | .112 [.102, .122] | .963 | .951 |
Metric | .062 | .093 [.084, .103] | .972 | .966 |
Scalar | .060 | .075 [.068, .083] | .970 | .978 |
RWP | ||||
Configural | .040 | .089 [.080, .098] | .983 | .977 |
Metric | .045 | .069 [.060, .079] | .988 | .986 |
Scalar | .043 | .063 [.056, .071] | .985 | .988 |
Age invariance | ||||
LWP | ||||
Configural | .063 | .116 [.106, .126] | .962 | .949 |
Metric | .070 | .091 [.082, .101] | .972 | .969 |
Scalar | .067 | .078 [.071, .086] | .961 | .977 |
RWP | ||||
Configural | .046 | .099 [.097, .109] | .985 | .979 |
Metric | .060 | .093 [.083, .103] | .984 | 981 |
Scalar | .050 | .065 [.057, .073] | .985 | .991 |
Test . | SRMR . | RMSEA . | CFI . | TLI . |
---|---|---|---|---|
Country invariance | ||||
LWP | ||||
Configural | .061 | .116 [.106, .126] | .962 | .949 |
Metric | .066 | .096 [.086, .105] | .970 | .966 |
Scalar | .062 | .077 [.070, .084] | .965 | .978 |
RWP | ||||
Configural | .041 | .083 [.074, .093] | .987 | .982 |
Metric | .048 | .068 [.059, .077] | .990 | .988 |
Scalar | .044 | .061 [.054, .069] | .985 | .990 |
Gender invariance | ||||
LWP | ||||
Configural | .058 | .112 [.102, .122] | .963 | .951 |
Metric | .062 | .093 [.084, .103] | .972 | .966 |
Scalar | .060 | .075 [.068, .083] | .970 | .978 |
RWP | ||||
Configural | .040 | .089 [.080, .098] | .983 | .977 |
Metric | .045 | .069 [.060, .079] | .988 | .986 |
Scalar | .043 | .063 [.056, .071] | .985 | .988 |
Age invariance | ||||
LWP | ||||
Configural | .063 | .116 [.106, .126] | .962 | .949 |
Metric | .070 | .091 [.082, .101] | .972 | .969 |
Scalar | .067 | .078 [.071, .086] | .961 | .977 |
RWP | ||||
Configural | .046 | .099 [.097, .109] | .985 | .979 |
Metric | .060 | .093 [.083, .103] | .984 | 981 |
Scalar | .050 | .065 [.057, .073] | .985 | .991 |
Note: LWP = left-wing populism; RWP = right-wing populism.
Left-Wing Populism. The LWP scale yielded a satisfactory fit, except that RMSEA was high for some of the models. In all LWP models, CFI, TLI and SRMR were adequate. Model fit was not worsened for gender and country as requirements became more stringent which suggests that LWP is invariant across country and gender. The scalar model for age groups had a ΔCFI above .01 which suggests that the LWP model was only suitable to capture metric invariance among the age groups. The failed scalar test might indicate that LWP is sensitive to age or generational differences. We speculate that this might be due to generational changes. Compared to traditional left-wing perspectives, younger individuals are more likely to embrace cosmopolitan cultural attitudes, e.g., pertaining to the LGBT+ community, and messaging from left-wing populists like Bernie Sander (e.g., Gallagher, 2016; March, 2017; Norris & Inglehart, 2019). From a practical point of view, the lack of scalar invariance indicates that future use of this scale with age groups/cohorts needs to be met with caution before mean comparisons are conducted, and researchers should consider conducting an invariance test of the LWP scale before usage. Without a successful scalar test, the LWP scale might not reliably detect true mean differences between age groups or cohorts.
Right-Wing Populism. As far as model fit is concerned, the RWP scale produced mostly satisfactory results except that the RMSEA was high on the age configural/metric models. Beyond that, RMSEA, CFI, TLI and SRMR were acceptable for all other models. RWP also demonstrated invariance across age, gender, and country, as none of the models were significantly worse when metric invariance and scalar invariance were tested.
Both scales were reliable: LWP (α = .85) and RWP (α = .89) with the total sample. The LWP scale reliable for all three countries: USA (α = .87), Canada (α = .85), and Australia (α = .83); this was also the case for the RWP scale: USA (α = .90), Canada (α = .88), and Australia (α = .87). The total sample seemed to hold higher populist left attitudes (M = 4.50, SD = 1.00) compared to their populist right attitudes (M = 3.42, SD = 1.09). RWP was related to right-wing identification, r = .47, p < .001, whereas LWP was negatively related to it, r = -.07, p = .03.
Validity Tests
Table 5 summarizes all correlations with the nomological net. For RWP, we found all significant correlations in the anticipated direction except for its null relationship with internal political efficacy. RWP and left-wing authoritarianism were negatively related, but this anticipated finding only emerged with the non-compensatory approach (Wuttke et al., 2020). The LWP results are not so straightforward; it was linked to LWA, but also had links to nativism, ethnonationalism and RWA. The correlations to these latter three variables were low (r = 0.07-0.14). Unexpectedly, it was unrelated to internal political efficacy and civic nationalism. Moreover, higher scores on both LWP and RWP were related to high relative deprivation, low external political efficacy, and high scores on two conceptions of anomie. This pattern is consistent with the idea that populists feel disenfranchised (e.g., Elchardus & Spruyt, 2016; Jami & Kemmelmeier, 2021). However, there are points of differentiation as well. RWP was much more strongly related to RWA, ethnonationalism, and nativism compared to LWP. And as already mentioned above, whereas LWP was significantly related to LWA, RWP was negatively correlated with LWA with the non-compensatory approach. The final distinction emerges with regards to civic nationalism. RWP was inversely related to civic nationalism, consistent with the RWP’s opposition to inclusion, but no relationship was found for LWP. Overall, findings yielded strong support for our theoretical expectations concerning RWP, but only partial evidence regarding LWP.
. | Right-wing populism . | Left-wing populism . |
---|---|---|
r | r | |
Left-wing authoritarianism | -.06a | .31 |
Right-wing authoritarianism | .51 | .10 |
Nativism | .53 | .07 |
Ethnonationalism | .51 | .14 |
Civic nationalism | -.41 | -.01 |
Anomie: Societal | .39 | .39 |
Anomie: Government | .32 | .56 |
Relative deprivation | .38 | .44 |
Political efficacy: Internal | -.04 | .00 |
Political efficacy: External | -.20 | -.47 |
. | Right-wing populism . | Left-wing populism . |
---|---|---|
r | r | |
Left-wing authoritarianism | -.06a | .31 |
Right-wing authoritarianism | .51 | .10 |
Nativism | .53 | .07 |
Ethnonationalism | .51 | .14 |
Civic nationalism | -.41 | -.01 |
Anomie: Societal | .39 | .39 |
Anomie: Government | .32 | .56 |
Relative deprivation | .38 | .44 |
Political efficacy: Internal | -.04 | .00 |
Political efficacy: External | -.20 | -.47 |
Note: Entries reflect Pearson correlation coefficients. Bold face indicates p < .05; a indicates that result is significant if operationalization changes.
Regression Analysis
Each model controlled for left-right ideological self-placement, country, both forms of authoritarianism: LWA and RWA, gender, ethnicity, education, and age. We report only the main effects on Table 6 (see Section S9, Tables S7a-S7b, supplemental materials for full models). Table 6 also indicates the standardized betas and the adjusted ΔR2 when the two populism scales were included in the model. In general, the two scales added an additional 0.03-0.14 (M = 0.07) of variance explained.
. | Right-wing populism . | Left-wing populism . | Adjusted . |
---|---|---|---|
β | β | ΔR2 | |
Society exists in 100 years | .02 | -.20 | .03 |
Inequality harms society | .02 | .40 | .13 |
Peaceful future activism | -.08a | .23 | .03 |
Votes counted fairly | -.24 | -.13 | .07 |
News media favors establishment | .08 | .23 | .05 |
Establishment suppresses outsiders | .17 | .27 | .10 |
Politicians are bought | .08b | .39 | .14 |
Journalists are fair | -.16 | -.13 | .04 |
. | Right-wing populism . | Left-wing populism . | Adjusted . |
---|---|---|---|
β | β | ΔR2 | |
Society exists in 100 years | .02 | -.20 | .03 |
Inequality harms society | .02 | .40 | .13 |
Peaceful future activism | -.08a | .23 | .03 |
Votes counted fairly | -.24 | -.13 | .07 |
News media favors establishment | .08 | .23 | .05 |
Establishment suppresses outsiders | .17 | .27 | .10 |
Politicians are bought | .08b | .39 | .14 |
Journalists are fair | -.16 | -.13 | .04 |
Note: Entries reflect standardized regression coefficients. Bold face indicates p < .05; a indicates that the result does not replicate under the Wuttke et al. (2020) operationalization. b indicates that the result is significant under the Wuttke et al. (2020) operationalization. Adjusted ΔR2 indicates variance explained by right-wing and left-wing populism after controlling for left-wing authoritarianism, right-wing authoritarianism, nationality, left-right identification, gender, race, education and age.
RWP was unrelated to the first two criterion variables summarized in Table 6. It was linked to negative electoral perceptions: votes are not counted fairly, establishment suppresses outsiders, and politicians are bought (a finding that only emerged with the non-compensatory approach). RWP was also associated with the perception that journalistic coverage is unfair. Although RWP was negatively linked to peaceful activism, this finding was not replicated when we applied the non-compensatory approach (see section S10, Tables S8-S10).
LWP was a significant predictor of all outcomes. It was related to the view that society will not exist in 100 years and that inequality will harm society; individuals who scored high in LWP also expressed a greater interest in future activism. Moreover, there was a negative relationship between LWP and perceived electoral and journalistic integrity. Overall, these models showed strong evidence for the criterion validity of LWP, but only partial evidence for RWP.
General Discussion
The goal of the present research was to develop scales for left-wing populist (LWP) and right-wing populist (RWP) attitudes. First, based on a review of the existing literature (e.g., Castanho Silva et al., 2022) and our own pilot studies, we developed factor structures for both LWP and RWP scales. This factor structure was confirmed in Study 1 and replicated twice which includes a cross-national comparison of three English-speaking democracies in Study 2. Our scales are not only internally consistent, but also externally valid. Using a battery of measures, we demonstrated that LWP and RWP had theoretically expected relationships with other variables, establishing a nomological network. Lastly, LWP and RWP contributed variance in the prediction of issues, attitudes and group perceptions beyond “old reliables” such as left-right identification and authoritarianism which addressed the lack of predictive power on the part of existing populism measures (e.g., Castanho Silva et al., 2022; Olivas Osuna & Rama, 2022; Quinlan & Tinney, 2019).
Our research offers a conceptual innovation by bridging the gap between “thin” and “thick” ideologies. Populism authors like Stanley (2008) as well as Mudde and Rovira Kaltwasser (2017) argued that populism as an ideology must attach to other beliefs to materialize, and forms of populism tend to be quite distinct as a function of this marriage of “thick” and “thin” ideology.
Our right-wing populism scale integrated three factors of populist attitudes with a focus on cultural exclusionism and national identity by deploying items that explicitly referenced the notion of “true Americans/Canadians/Australians.” We concede that different targets and different phrases may have been viable here, such as “real Americans/Canadians/Australians,” “Americans/Canadians/Australians patriots,” to name only a few that have been used in the U.S. context (e.g., Schertzer & Woods, 2021). Yet, our RWP encapsulates the patriotism, national focus and exclusion of non-natives that is inherent to most right-wing populist movements across the world (e.g., Norris & Inglehart, 2019). In contrast, our left-wing scale marries populist ideas with an inclusive emphasis on the plight of the proletariat (e.g., March, 2017; Otjes & Louwerse, 2015). One can argue that left-wing populism can include cultural issues, but its prominent themes center around the economics or material wealth of the working class (Mudde & Rovira Kaltwasser, 2013, 2017). Our conceptualization is true to the argument that populism cannot be viewed in isolation (e.g., Mudde & Rovira Kaltwasser, 2017; Stanley, 2008). We add to this: that populist attitudes should not be measured in isolation.
Our work also helps flesh out the psychology of populist left and right. Only a handful of studies have examined both forms of populism in attitudinal/psychological research (Akkerman et al., 2017; Jami, 2023; Vasilopoulos & Jost, 2020). Unlike those previous studies which used proxies like vote choice or interacted ideology with populist attitudes to assess LWP and RWP, the present research has directly measured populist attitudes from both flanks. With the LWP and RWP scales, some previous research can be also revisited. For instance, Norris and Inglehart (2019) argued that the rise of right-wing authoritarian-populism occurred because of the resistance to cultural change by members of society who hold traditional values. In their research, they used mistrust of institutions as a proxy of populist attitudes instead of an actual measure of populist attitudes. With our new measure of RWP attitudes, the link between right-wing populism to cultural change can be tested much more precisely.
Our scales have advantages over existing measures, beyond issues regarding predictive validity. First, the inclusion of the host ideology may reduce the number of items needed in a survey. A researcher would need separate measures of economic or cultural attitudes in addition to an inventory of populist attitudes to assess populist left and populist right sentiments, respectively. Relatedly, separate measures that assess economic or cultural attitudes like ethno-centrism or support for the working class would assume populist attitudes are orthogonal to such ideas even though populist attitudes need those beliefs to be integrated with them (e.g., Mudde & Rovira Kaltwasser, 2017; Stanley, 2008). After all, thick ideologies like socialism or conservatism and populist attitudes must be intertwined. Third, our measures have items that include multiple elites which tackle the issue of political elites being overrepresented (Jungkunz et al., 2021). Finally, there have been concerns about the best approach to operationalize populist attitudes with previous scales producing different results based on the specific operationalization (Wuttke et al., 2020). We demonstrated that Wuttke et al.’s (2020) innovative non-compensatory approach only created minor variations: Only five of 56 coefficients were (non-)significant when operationalization changed. This suggests that in most instances it is immaterial for a study’s conclusion whether a researcher adopts a compensatory or non-compensatory approach.
What is the Dimensionality of Populist Attitudes?
The dimensionality of populist attitudes is not quite clear in the literature, as some found a singular factor (e.g., Akkerman et al., 2014), but others posit that there are three dimensions (e.g., Schulz et al., 2018). Our empirical work with regard to RWP confirmed the conceptualization of populism as a three-factor model. However, for LWP, a two-factor model yielded a much better fit, with anti-pluralism/people-centrism as one factor, and anti-elitism as a standalone factor. We speculate that the anti-pluralism/people-centrism factor for LWP may suggest that at least on the political left, the idea of people-centrism goes in lockstep with the Manichaeanism or majoritarianism that is often inherent in anti-pluralism. However, these two tendencies seem to be comparatively distinct for the populist right. That is, the populist right does not view mobilization and out-group prejudice as inseparable for instance.
The Link Between the Populist Right and the Populist Left
Given that political movements on the left and right tend to be at odds with each other, we were surprised to find a substantial correlation between RWP and LWP. This association warrants further investigation, as does the surprising link between LWP to ethnonationalism, nativism, and right-wing authoritarianism. First, it is possible for people to entertain both forms of populism. In the 2016 U.S. electoral cycle, about 6-12% of the voters who initially supported Bernie Sanders switched to Donald Trump once Sanders lost the Democratic primary (Schaffner & Ansolabehere, 2017; Sides, 2017). And in the 2020 USA presidential cycle about 15% of Sander’s supporters planned to vote for Trump over the Democratic nominee, Joe Biden (Sinozich, 2020). Perhaps some individuals may seek to join any populist movement, as both the populist left and the populist right have a common enemy: the political elite. This speculation may draw a parallel to the idea of “wild-card” authoritarianism which describes those high on both LWA and RWA (B. Altemeyer, 1996); thus, there may be “wild-card” populists, high on populism in general.
Because of the high correlation, we included all items/factors in both scales into a CFA model using data from Study 2. This model tested whether all items from our LWP and RWP scales might be understood as contributing to a single conception of populist attitudes. Results showed that the 22-item CFA model produced relatively poor fit, CFI = .851, TLI = .830, RMSEA = .139 [.135-.144], SRMR = .109. This implies that, despite the high correlation, RWP and LWP function better individually as opposed to being combined to represent a singular, broad view of populist attitudes.
A second possibility is that one can hold nativistic cultural beliefs and worker-focused economic views simultaneously. In other words, it is possible for people to endorse both ideas such as those who may be in favor of welfare chauvinism – the idea that robust welfare programs should exist, but in exclusionary terms based on nationality, citizenship or race/ethnicity (e.g., De Koster et al., 2013; Persson & Widmalm, 2024). We speculate that this may explain the relationship between LWP to right-wing authoritarianism, ethnonationalism and nativism found in Study 2.
The Relationship Between Populist Attitudes and Liberal Democracy
We found anti-pluralism content in both the LWP and RWP scales which may speak to the complicated relationship between populism and liberal democracy. Our measures of anti-pluralism (on both scales) might be related to the idea that populists emphasize majoritarianism – the view that the majority/dominant group in society should have the most say in society, and that alternative viewpoints can be disregarded (e.g., Castanho Silva et al., 2020; Mudde & Rovira Kaltwasser, 2017). And this idea is not at odds with all aspects of democracy, but with the liberal components of democracy (e.g., maintaining unalienable rights even if the majority wished to undermine those rights) as well as dividing people into two camps of good vs. evil instead of recognizing diversity of political thought (Castanho Silva et al., 2018, 2020; Oliver & Rahn, 2016). Most anti-pluralism items are certainly anti-liberal, as they emphasize the disregard of other viewpoints, (e.g., “You can tell if a person is good or bad if you know their views on inequality,” and “In politics, what people call compromise on religious lifestyle issues is really just selling out on one’s principles,”) or the majoritarian aspect of populism (emphasis on the will of True Americans/working class over the rights of minorities/the rich).
Yet, our data may also be taken to support the claim that populist attitudes are linked to anti-democratic tendencies. For example, RWP was linked to expressing interest in violent future activism and downplaying Trump’s role on January 6 in Study 1, and the belief that votes are not counted fairly in Study 2. In contrast, we also found RWP to be linked to higher satisfaction with democracy and trust in government. Though left-wing populism (LWP) was linked to Biden’s use of executive orders (which is a way the president can pass laws without allowing congress a vote on them) and the view that votes are not counted fairly, it was also related to peaceful activism. Therefore, there are some pro-democracy elements in populist attitudes as well.
Our results can contribute to the ongoing discussion about the complex relationship between liberal democracy and populism. Although we concur with Mudde and Rovira Kaltwasser (2017) that populism includes a strong anti-liberal component, we also acknowledge Moffitt’s (2018) argument in that populism is not as anti-democratic as made out by anti-populists. Others like Akkerman et al. (2014) and Mudde (2004) also describe elites (those who often prefer to maintain the European Status quo and who Moffitt [2018] may consider as a significant segment of the anti-populists) as a “mirror image of populism” with similar anti-liberal, anti-democratic tendencies. In this sense, the anti-liberalism and anti-democratic tendencies are commensurate among populists and the elite.
Discrepancies Between Operationalization Strategies
The present research found five (out of 56) instances in which results (coefficients) were noticeably discrepant between the additive approach relative to the non-compensatory approach (Wuttke et al., 2020). First, the finding of LWP predicting low trust in democracy only emerged when we deployed the additive approach. Yet, LWP was statistically linked to views of electoral fraud during the 2020 U.S. presidential election with the non-compensatory approach. Still, the lack of replication based on operationalization indicates that either it needs to be investigated further in future studies, and that potentially electoral fraud perceptions are more likely to be features of the populist right, but an inconsistent feature of the populist left.
The other three findings that differed across operationalization strategies were the link between (1) peaceful activism, (2) left-wing authoritarianism, and (3) the view that politicians were bought by special interests and RWP. Peaceful activism was not significant with the non-compensatory approach, but was significant with the additive approach in Study 2, but not in the direction that was anticipated. Peaceful activism might not be commonplace among the populist right, but perhaps a staple for the populist left as the latter is more likely to engage in grassroots activism (see Ivaldi et al., 2017). The view that politicians are bought by special interests was linked to RWP with the non-compensatory approach, but not with the additive approach. This may suggest that the populist right does not always believe that special interest influence elections and it could be because their side has had several recent electoral successes like Donald Trump’s second successful presidential campaign. Finally, the inconsistent link between RWP and left-wing authoritarianism may speak to the idea that the right-wing and left-wing do share some ideas that are not diametrically opposed, as mentioned earlier in the discussion.
The very fact that two findings were no longer significant with Wuttke et al.’s (2020) non-compensatory operationalization whereas the other three finding materialized because of it indicates that neither the additive nor the non-compensatory view is necessarily a more conservative operationalization strategy. Moreover, the convergence of results for most analyses (91.3%) also addresses an issue that Wuttke et al. (2020) posited. They argued that operationalization strategy can have a profound effect on the utility of a populist attitude inventory. Whereas we concur that this can be the case, we showed that results are generally consistent in spite of differences in operationalization.
Benefits of Multi-dimensional Measures of Populist Attitudes
There are several benefits to having multi-dimensional measures of populist attitudes. First, multiple items or dimensions capture more information which allows for fine-grained analyses (e.g., Bergkvist & Rossiter, 2007). For example, scholars can examine if certain dimensions, like anti-elitism, might explain greater proportions of the variance for some outcomes like electoral skepticism whereas anti-pluralism may capture negative inter-groups attitudes better than the other components. Moreover, scholars may also focus on specific factors and not use the entirety of the scale, potentially reducing the chances of participant fatigue. A scholar might want to only use the items that capture people-centrism which is well defined in our study whereas it is less clear in other scales like Akkerman et al. (2014) – which has been argued to be double-barreled (see Wuttke et al., 2020). Another advantage is that greater psychometric qualities can be detected; both the entire scale and its factors can be assessed for reliability, for instance.
Limitations
The present research has at least three limitations. First, a possible consequence of convenience sampling is the over-representation of certain groups. Our samples disproportionately included highly educated respondents. Given that education is associated with reduced radicalization (e.g., Sas et al., 2020), effects reported in this research may underestimate actual effects sizes—if populism exhibits dynamics similar to radicalization.
Second, the present research has solely focused on Anglo-Celtic cultures. Though the political cultures of the U.S., Canada and Australia are clearly distinct, future validation studies are needed in other societies. For example, further research may determine if RWP can be used in non-Christian majority countries like India and Turkey that do have right-wing populists in power (Norris & Inglehart, 2019). Indeed, Erdoğan (Turkey) and Modi (India) engage in religious & nationalist based populism that share some similarities to Trump (e.g., Norris & Inglehart, 2019; Tepe & Chekirova, 2022; Yilmaz, 2018).
Our scales also need to be tested if they can be used in other non-Western societies like those in Latin America. For example, similar nationalist framing can be seen with both Donald Trump (USA) and Jair Bolsonaro (Brazil), as argued by Wehner (2023). Moreover, a recent trend of right-wing populist rhetoric on immigration, resembling Trumpism, has emerged in in Chile and Uruguay (e.g., Kestler, 2022). By the same taken, economic inclusion is common in both Latin American and North American/European left-wing populism (March, 2017; Mudde & Rovira Kaltwasser, 2013). Another limitation is whether these scales can be used in non-adult or non-general populations. For instance, we are unsure if they are appropriate for adolescents, as our samples consisted of adults.
Future Research
This study paves the way for new research. For instance, can these two new scales predict actual populist vote shares during election times? This would further address the poor explanatory power found in electoral data (e.g., Castanho Silva et al., 2022; Quinlan & Tinney, 2019). Another research question is whether populist attitudes that are specific to different sides of the political spectrum can predict radical group membership such as ANTIFA (anti-fascist) on the left, or the Proud Boys on the right. Though scholars have alleged links between populism and radicalism/authoritarianism, many studies have not utilized precise measures of populist attitudes (e.g., Norris & Inglehart, 2019). Populism may have both positive and negative effects on democracy, such as giving a voice to underrepresented groups while also hindering minority rights (Mudde & Rovira Kaltwasser, 2017), which suggests that the connection between radicalism/authoritarianism and populism is unclear.
Future studies are also needed to successfully implement these scales in non-Western contexts. There will be a need for some localized changes for these scales to work in different contexts. In terms of RWP, the biggest change would be to the social identity of the ingroup from “True American/Canadian/Australian” to something that highlights the exclusionary conception of RWP in those contexts. Igreja (2021) argues that terms like “true” or “false” highlight the binary “us vs. them” found in the populist right in Brazil. Similarly, the populist right in Hungary would often consider themselves as representing the “true” Hungarian (Murer, 2015). A more generic term the RWP measure could use is “true patriots” to refer to their ingroup.
The LWP scale may require fewer changes. For example, economic inclusion is a hallmark of both Latin American and North American/European left-wing populism (March, 2017; Mudde & Rovira Kaltwasser, 2013). Yet, notions of multi-culturalism should be infused in the working-class identity in some contexts, as left-wing Latin American populists do emphasize multi-culturalism more explicitly than their Western Counterparts; the latter is still fairly inclusive even though they are not as focused on race and ethnicity as part of their identity (see Algan et al., 2018; March, 2017; Mudde & Rovira Kaltwasser, 2017). Yet, invariance tests and deployment/translation of these scales, among other validity tests, will be needed before there is confidence in the applicability of these scales in Latina American countries and other non-Western cultures.
Conclusion
Populism is a growing force around the world, affecting nascent and longstanding democracies. We deployed two exploratory and two confirmatory studies to elucidate the nature of right-wing and left-wing populist attitudes. These two new scales offer scholars a tool to determine which populist attitudes/facets might contribute to the deconsolidation of democratic foundations as well as the ones that can empower the marginalized.
Data Accessibility Statement
Data are available here: https://osf.io/m69yv/
Ethical Statement
All studies were reviewed by the Authors’ institutional review board. All participants were given and reviewed the informed consent form prior to participating in the study.
Competing Interests
The Authors have no conflicts of interest to declare.
Acknowledgements
The Authors would like to acknowledge James T. Richardson and the Ozmen Institute for global studies for supporting this study.
Footnotes
Item 4 for the anti-elitism factor was unavailable for this analysis.