Left Behind and United by Populism? Populism’s Multiple Roots in Feelings of Lacking Societal Recognition

A prominent but underspecified explanation for the rise of populism points to individuals’ feelings of being “left behind” by the development of society. At its core lies the claim that support for populism is driven by the feeling of lacking the societal recognition one deserves. Our contribution builds on the insight that individuals can feel they lack recognition in different ways and for different reasons. We argue that—because of this multifaceted character—the common perception of being neglected by society unites otherwise heterogeneous segments of the population in their support for populism. Relying on data from the German Longitudinal Election Study (GLES) Pre-Election Cross-Section 2021, our preregistered study investigated the multiple roots of populist attitudes in feelings of lacking societal recognition in two steps. First, our results indicate that, from rural residents to sociocultural conservatives or low-income citizens, seemingly unrelated segments of society harbor feelings of lacking recognition, but for distinct reasons. Second, as anticipated, each of the distinct feelings of lacking recognition are associated with populist attitudes. These findings underscore the relevance of seemingly unpolitical factors that are deeply ingrained in the human psyche for understanding current populist sentiment. Overall, by integrating previously disparate perspectives on the rise of populism, the study offers a novel conceptualization of “feeling left behind” and explains how populism can give rise to unusual alliances that cut across traditional cleavages.


Introduction
The debates about the reasons for populism's recent upsurge in many established democracies are ongoing. One lingering puzzle common to populist movements is their success in traversing the traditional cleavages of Western societies in which they managed to assemble a diverse coalition of societal groups that cuts across ideological and socioeconomic divides (Gidron and Hall 2017;Damhuis 2020;Kaltwasser and Hauwaert 2020;Nachtwey et al. 2020;Harteveld et al. 2021). Understanding the commonality that ties together citizens from different backgrounds is thus key to understanding the rise of populism. The existing literature points to attitudinal factors such as xenophobia and sexism, which may have particular value for explaining the success of nativist parties (Ratliff et al. 2019;Young et al. 2019). However, many recent populist movements do not subscribe to such positions. In this study, we propose a more broadly applicable explanation. We suggest that the common perception of being neglected by society unites otherwise heterogeneous segments of the population in their support for populism.
Our argument builds on the popular thesis that supporters of populist parties feel "left behind" (Ford and Goodwin 2014;Goodwin and Heath 2016;Hobolt 2016;Hall 2017, 2020). Despite its popularity, previous scholarship has not fully appreciated that individuals can feel left behind in different ways and for different reasons. Moreover, scientific progress on the left-behind thesis has been hampered by conceptual vagueness. Drawing on the established concept of societal recognition, we therefore propose a more specific, inherently multidimensional conceptualization of feeling left behind as the perception of being denied the societal recognition one deserves. Our synthesis of literature from philosophy, psychology, and the social sciences characterizes societal recognition as a fundamental human desire, the thwarting of which leads to self-defensive reactions with political implications such as populist attitudes. Because there are many ways to the perception of lacking recognition, feeling left behind may unite different groups that have been characterized as supporters of populism: people in routine jobs who perceive a lack of recognition for the work they do (Sandel 2021), conservatives who feel alienated by cultural changes in society (Norris and Inglehart 2019), and those living in "leftbehind" rural places (Broz et al. 2021;Harteveld et al. 2021). In our preregistered empirical analysis, we tested this proposition of lacking societal recognition as the common cause of populist sentiments in diverse segments of society.
To investigate the relationship between the perception of lack of societal recognition and populist attitudes, we drew on a novel four-item battery included in the GLES Pre-Election Cross-Sectional Survey 2021 (GLES 2022) that measured perceptions of societal recognition across different domains. First, we investigated the multiple roots of feeling left behind. We expected that feelings of lacking societal recognition are present in many societal segments, but that different feelings dominate in different segments. To test this, we compared domain-specific feelings of lacking societal recognition across distinct societal groups. We focused on a set of prototypical groups whom we expected to feel denied societal recognition in specific ways but not in others. Second, we tested the hypothesis that the domain-specific feelings of lacking societal recognition are each positively associated with populist attitudes.
The results of our preregistered analyses support our argument of populism's multiple roots to a large extent. First, distinct feelings of lacking recognition are anchored in social characteristics as expected: Individuals with low incomes are more prone to perceive a lack of recognition of their economic needs than those with higher incomes; individuals who work in jobs that are often considered lowskilled are more prone to perceive a lack of recognition for the work they do; those living in rural places are more prone to perceive a lack of recognition of their needs for basic services; and people with socioculturally conservative attitudes are more prone to perceive a lack of recognition of their opinions. While membership in these social groups was thus associated as predicted with feelings of lacking recognition, we also found an unexpectedly outsized role of income and especially of conservatism-which go along with lacking recognition on multiple facets.
Second, as anticipated, each of the distinct facets of perceived lack of recognition predict populist attitudes. Yet, in line with the idea of multiple roots, our results suggest that feeling a lack of recognition regarding only one of the facets may suffice to trigger populist sentiment.
Overall, by pointing to the commonality of lacking recognition, these findings help explain why citizens from different strands of life are all attracted to populist ideas.
Our study adds to the literature on how feelings of being left behind-which previous research has conceptualized and operationalized via subjective social status Hall 2017, 2020;Gest et al. 2018;Kurer 2020;Engler and Weisstanner 2021)-affect support for populism. We propose, and have tested, a refined account that considers that people may feel they are being denied societal recognition in different ways. In doing so, we contribute to the literature on populism more broadly by showing how populism may function as an umbrella that unites citizens who feel left behind for very different reasons.

Populism's Multiple Roots in Feelings of Lacking Societal Recognition
Populism is a complex phenomenon, and scholars have not yet come to a consensus on why some citizens support populist parties or hold populist attitudes. Some scholars interpret populism as a revolt of the economic losers of globalization, trying to explain why blue-collar workers frequently vote for populist parties in their left or right persuasion (Betz 1993;Kitschelt and McGann 1995;Swank and Betz 2003;Kriesi et al. 2006Kriesi et al. , 2008Oesch 2008;Doyle 2011;March and Rommerskirchen 2015;Colantone and Stanig 2018). Others question this narrative by pointing out that both blue-collar workers and small business owners regularly vote for rightwing populist parties despite seemingly contradictory economic interests (Ivarsflaten 2008). With an eye on the overrepresentation of whites and men among the populist electorate, other scholars hint at cultural factors as the source of populist resentment (Norris and Inglehart 2019;Green and Shorrocks 2021). Rather than resolving the debates on the origins of populism, these studies underscore the multifacetedness of sometimes contradictory motives that bring citizens from different strands of life to unite behind populist parties and ideas. However, the question remains: Is there a connection between these segments of the societies that form the populist coalition?
Popular discourse has identified such a link that unites the adherents of populism: the feeling of being left behind. Ever since populism was recognized as a major electoral force, politicians, particularly in Anglo-Saxon countries, have begun catering to the "forgotten men and women" (Trump 2017) who "feel left behind" (Biden 2021). The term features regularly in scholarly writing on populism, too (Ford and Goodwin 2014;Hobolt 2016;Hall 2017, 2020). Yet due to its inherent vagueness, there is neither a broadly shared definition of feeling left behind nor a commonly accepted theoretical framework associated with the term.
The most fruitful attempts to establish a related theoretical framework have been put forth in an emerging body of literature on subjective social status and support for populism (e.g., Hall 2017, 2020;Gest et al. 2018;Kurer 2020). For instance, Hall (2017, 2020) argue that support for the populist right, particularly among white working-class men, is fueled by perceptions of a declining social status, which they understand to reflect "people's own feelings about the levels of respect or recognition they receive relative to others in society" (2017, p. 61). They propose a unifying theoretical framework in which social status perceptions mediate the effect of both economic and cultural developments on support for the populist right.
Building on this emerging literature, we propose a refined account that takes into consideration the different ways in which people may feel left behind. We draw on the established concept of societal recognition, which we argue captures those sentiments well that are associated with the feeling of being left behind. Because societal recognition involves multiple facets, this conceptualization highlights the many ways to feel left behind and may thereby help us understand the heterogeneous populist coalition. 1 Recognition is a central concept in the European history of thought. It took center stage in the works of British thinkers such as John Stuart Mill and Adam Smith, who considered acceptance in a social community a fundamental human motivation. This is similar to German idealism, in which Immanuel Kant and Georg Wilhelm Friedrich Hegel considered recognition a basic condition for developing into a rational, autonomous being (Honneth 2021). To this day, recognition is discussed both in philosophy and, often with other terms, the social sciences (McBride 2013;Fukuyama 2018). The fact that recognition has long occupied Western thinking hints at the deep roots of societal recognition for human thriving.
Human beings evolved as social animals. The unique skill that has given humans an edge over other species is the ability to cooperate with others of our kind (Tomasello 2019). Hence, there is an evolutionary advantage in the evolved desire to seek group interaction, which might be the reason for the universal human tendency to become accepted members of human groups (Baumeister and Leary 1995;Raihani 2021). Hence, the fact that social exclusion elicits profound discomfort and 1 Herein lies a crucial difference from Hall (2017, 2020) and related work. While Gidron and Hall focus on an overall perception of one's standing in the societal hierarchy, our conceptualization and operationalization are concerned with the multiple ways of perceiving a lack of societal recognition. In particular, Gidron's and Hall's use of the "social ladder question" may capture whether citizens feel left behind in general but not the different ways in which they do so. Accounting for these different ways strike us as important for understanding how different social groups may unite behind populist parties. Moreover, it is conceivable that feeling a lack of recognition in only one domain may suffice to trigger populist sentiment, even though such individuals may not necessarily assess their general social status as low. If so, this might explain mixed findings regarding the role of subjective social status (Oesch and Vigna 2021;Richards et al. 2021). adverse reactions in (most) human beings (Baumeister et al. 2007) is indicative of a potentially universal desire for recognition (Taylor 1995;Fraser and Honneth 2003).
Contemporary psychology considers the desire for recognition in concepts such as the basic psychological needs for relatedness and belonging (Baumeister and Leary 1995;Ryan and Deci 2017). Other concepts such as the desire for status (Anderson et al. 2015) or self-esteem (Gebauer et al. 2015) are related but are better understood as derivatives of the desire for recognition (Vansteenkiste et al. 2020;Ryan and Brown, 2003). Recognition is more fundamental, as it does not consider our standing in comparison with others but asks the basic question of whether we are in or out.
Being denied acceptance by a group to which we seek to belong goes to the core of who we are and feels like being denied the dignity we deserve (Fukuyama 2018;Honneth 2021). It has profound implications for one's self-concept, thereby affecting a person's sense of worth and psychosocial functioning (Baumeister et al. 2007). Our need for recognition is not salient when we are treated with dignity, but craving the recognition which one feels is being denied feels pressing and leads to self-defensive reactions. These reactions include narcissistic and aggressive tendencies that overcompensate for the lack of recognition (Honneth 2021); they also include passivity, resignation, and depression (Hilbert et al. 2019). While those clinical consequences underscore the deep roots of the desire for recognition, in democratic mass societies it may also affect how we think and act in our role as citizens, particularly when the reasons for lacking recognition are located in the political domain.
Ethnographic research (Cramer 2016;Eribon 2018;Arnade 2019) has repeatedly shown that groups who feel excluded from society exhibit political attitudes that fit these patterns of passive resignation or aggressive overcompensation. One form in which the consequences of lacking recognition could materialize is populist attitudes. Populist attitudes describe a moralistic world view that pits the corrupt elite against the virtuous people (cf. Mudde 2004). To those who feel they belong to the virtuous group of ordinary people, blaming the elites provides a useful narrative to explain one's exclusion from society without having to take any responsibility themselves. A more favorable depiction of the relationship between recognition and populism is to see all social struggle as a struggle for recognition (Honneth and Anderson 1995) and to see populism as an emancipatory process of democratization that strives for the inclusion of the unrepresented into the political process (Manow 2020). From this perspective, populism is the political instrument of the excluded to make their demands heard and to push themselves back into the arena from which they feel excluded. In any case, the feeling of being denied the recognition one deserves plausibly results in populist sentiment.
Up to this point, we were careful to stress the experiential dimension of lacking recognition. Social rejection is a subjective process in which similar objective circumstances increase the probability of feeling excluded but not in a deterministic way. Even more important to the conceptualization of societal recognition is that feelings of lacking societal recognition can have multiple-an almost infinite-number of sources. They can result from acts of formal exclusion such as apartheid regimes K but also from subtler discriminatory acts. Some citizens may react to economic exclusion; others may perceive neglect by elites of their sexual, regional, or religious identities, whereas for some citizens, rejection in terms of their values and attitudes might underlie their feelings of lacking recognition. Although no exhaustive list exists (see also Honneth and Anderson 1995), we can thus understand societal recognition as a multidimensional concept with cultural, economic, or regional aspects as constituent components.
From this perspective, societal recognition may account for why populism appeals to different social segments, thereby synthesizing different prominent strands of research on populism. Among the socioeconomically disadvantaged, a lack of perceived societal recognition may emerge from a widespread perception that society does not care about their financial struggles. Apart from financial strain, there is another, perhaps more consequential, feeling of lacking recognition that may nurture populist sentiment among production and service workers. As stressed by Sandel (2021, p. 198), those working in jobs that require lower levels of formal education have gained the impression that their work is "less worthy of social recognition and esteem", and this has made them susceptible to populist mobilization (cf. Spruyt 2014). In the "cultural backlash" argument of Norris and Inglehart (2019), the rise of (authoritarian) populism reflects a counterreaction of those with socioculturally conservative values to a societal shift in dominant value orientations. For these authors, sociocultural conservatives are "likely to feel that they have become estranged from the silent revolution in social and moral values, left behind by cultural tides that they deeply reject" (Norris and Inglehart 2019, p. 15). In essence, this is also an argument about perceived societal recognition, albeit of a cultural variety. Finally, the tendency of individuals in rural and declining "left-behind" regions to vote for populists (Broz et al. 2021) may reflect a sense that these places' needs are insufficiently recognized and respected, fueling a rural variety of populist resentment (Cramer 2016;Harteveld et al. 2021). More broadly, the seemingly contradictory findings in populism research we have cited above can be resolved once we acknowledge that different segments of society feel, for different reasons, denied the recognition they deserve.
Because the need for recognition is deeply ingrained in the human psyche, we argue that the perception of lacking recognition elicits consequences even when individuals feel excluded from just one central dimension of social life while enjoying high satisfaction in other domains. This proposition helps explain why, for instance, in the United States otherwise privileged white citizens in secure economic positions exhibit strong adverse reactions once they feel rejected by society for one recognition-related facet such as their cultural way of life (Mutz 2018). In other words, societal recognition can be understood as a multidimensional concept in which a lack of recognition in one constituent dimension on its own is sufficient to elicit psychological and political consequences.
Based on this discussion of how distinct feelings of lacking recognition may be present in distinct social groups and how these distinct feelings of lacking recognition may in turn trigger populist sentiment, Fig. 1 visualizes the specific model that we aimed to test in this study. We focused on a set of four prototypical groups who we expected-based on the literature and reasoning discussed above-to (primarily) Fig. 1 The multiple roots of populist attitudes in perceptions of lacking recognition. Theoretical model tested in this study, with hypotheses highlighted feel being denied recognition in one specific way: low-income individuals, lowskilled workers, rural residents, and sociocultural conservatives. Figure 1 associates one distinct feeling of lacking societal recognition with each of these social groups and expects each of these four distinct feelings in turn to contribute to populist attitudes.
Next, we formulated (a) a set of hypotheses on the link between social group membership and distinct feelings of lacking recognition, (b) an exploratory research question on the interrelation between feelings of lacking societal recognition, and (c) a set of hypotheses on the link between feelings of lacking recognition and populist attitudes. Beginning with the first, we expected the following: H1a Citizens with low incomes are more inclined to perceive a lack of recognition of their economic needs than are citizens with higher incomes.

H1b
Citizens working in jobs that require little formal skills are more inclined to perceive a lack of recognition for the work they do than are citizens working in jobs that require higher skill levels.
H1c Citizens living in rural places are more inclined to perceive a lack of recognition of their needs for basic services than are citizens living in urban places.
K H1d Citizens with socioculturally conservative attitudes are more inclined to perceive a lack of recognition of their opinions than are citizens with moderate or progressive attitudes. 2 By testing these hypotheses, we aimed to demonstrate that different feelings of lacking societal recognition can be rooted in distinct social characteristics. Our goal was not to deliver an exhaustive explanation of perceived lack of recognition. For this reason, we focused on social groups that we expected to primarily perceive a lack of recognition in just one facet, rather than studying social markers that are likely to be connected to several facets of lacking recognition at once. 3 Furthermore, we did anticipate that there might be some overlap between these social groups. For example, some people might work in low-skilled jobs, receive low incomes, hold conservative views, and live in declining rural regions. In fact, some of these conditions are connected through various mechanisms. Yet our argument rests on the assumption that this overlap is far from perfect. To illustrate, sociocultural conservatives in wellpaid, high-status occupations in vibrant cities may feel denied recognition on the cultural dimension but not in any other way. Further below we describe the empirical tests we conducted to test the validity of these considerations.
Next, we are interested in the interrelation between the different facets of lacking societal recognition. Our argument of multiple roots implies that the different facets of feeling a lack of recognition can arise independently from each other. Thus, in principle, individuals can feel marginalized in one dimension without feeling denied recognition in other domains. While we therefore generally expected low correlations between facets of recognition, some individuals may be subject to multiple simultaneous forces that instill feelings of lacking recognition in more than one dimension. Moreover, it is conceivable that feeling neglected in one way may trigger a general feeling of lacking recognition that spills over into other domains. In line with this reasoning, factor analyses of items on perceived lack of recognition in Bollwerk et al. (2022) support a structure with three domain-specific (economy, culture, politics) factors and one superordinate factor. This lends credence to the assumption-implicit to our argument-that there are distinct (albeit correlated) facets of perceived lack of recognition. To reassess this assumption in the context of our study, we formulated the following exploratory research question: How strong is the association between different facets of lacking societal recognition?
Lastly, we argued above that feelings of lacking societal recognition may trigger populist sentiment. We expected this to hold for each of the four distinct feelings of lacking societal recognition that we studied: H2a Citizens who perceive a lack of recognition of their economic needs are more likely to hold populist attitudes.
H2b Citizens who perceive a lack of recognition for the work they do are more likely to hold populist attitudes.
H2c Citizens who perceive a lack of recognition of their needs for basic services are more likely to hold populist attitudes.
H2d Citizens who perceive a lack of recognition of their opinions are more likely to hold populist attitudes.
Per H2a to H2d, we expected an association of each of the four facets of recognition with populist attitudes. More tentatively, our theoretical arguments also imply a specific functional form of these interrelations, as we anticipated these to operate not just additively in affecting populist attitudes. Rather, when thinking about lack of recognition as a composite concept, the deep roots of lacking recognition in the human psyche imply a logic of sufficient condition in which deficiencies on one of these facets may suffice to elicit psychological consequences. For example, the sense that one's opinion is not recognized may trigger populist sentiment, even for an individual who feels recognized with regard to other facets. And while these elaborations imply the presence of sufficient conditions, it is also plausible that lacking recognition in multiple facets may have even somewhat stronger effects, suggesting a concept structure of sufficient conditions with weak substitutability. We thus expected the following: H3 The contribution of individual facets of lacking societal recognition to populist attitudes is best captured by a logic of sufficient conditions with weak substitutability. 4

Data
We used data from the 2021 preelection cross-sectional survey of the GLES (2022), which were collected via self-administered questionnaires in either online or paperand-pencil mode between August 26 and September 25, 2021. The data set consists of responses from 5116 individuals. The survey included a novel four-item battery on perceptions of lacking societal recognition 5 -part of the GLES for the first time in 2021-that allowed us to test our hypotheses in the context of the 2021 German federal election.
We think of Germany as a potentially representative case. It fulfills possible scope conditions of our argument, but not in any extreme way. For example, with the Alternative for Germany (AfD) there is a relevant right-wing populist party in Germany that engages in populist mobilization of those who feel left behind-but it is not exceptionally strong compared with parties in other Western democracies, and it is also relatively young. Socioeconomic inequalities, at the individual and regional levels, have widened in Germany, but are not extreme in international comparison. While these are grounds to expect our results to travel to other country contexts, this is ultimately a question for future research to address. 6 Next, we discuss the operationalization of the key constructs. In Table A1 of the appendix, we list all our operationalizations, including question wordings and precise coding information for all variables.
Perceived lack of societal recognition was measured with four indicators, based on five-point Likert scales, each of which taps into a constituent component (facet) of recognition: lacking recognition of one's economic needs ("The economic situation of people like me receives too little attention from society"), lacking recognition of one's work ("People like me receive too little recognition from society for the work they do"), lacking recognition of one's needs for services ("Society pays too little attention to ensuring that people like me have access to basic infrastructures and services [e.g., post office, doctors, banks, public transportation, schools, Internet]"), and lacking recognition of one's opinion ("People like me are no longer allowed to freely express their opinions in public"). We coded these items to a 0 to 1 scale, where higher values indicate a perceived lack of recognition (i.e., agreement with the statements). 5 This battery builds on the recently proposed Perceived Societal Marginalization scale (Bollwerk et al. 2022) from which some of the items have been drawn. In line with our focus on perceived lack of recognition, the scale intends to measure "people's subjective perceptions of the insignificance and lack of recognition of their own social groups in the domains of economy, culture, and politics" (Bollwerk et al. 2022, p. 1). 6 Another notable feature of the German federal election in 2021 was the focus of the campaign by the Social Democratic Party (SPD) on "respect," a term that is closely related to our key construct of societal recognition. Interestingly, this master narrative seems to have been inspired by some of the academic work we have also built our argument on, particularly the monograph by Michael Sandel (2020): Olaf Scholz has repeatedly referred to Sandel's work in public (e.g., https://twitter.com/olafscholz/status/ 1319542555860750336), and the SPD Debattencamp in December 2020 featured a debate between Scholz and Sandel (see https://debattencamp.spd.de/). Like Sandel's work, the SPD campaign has put an emphasis on work-related recognition (or respect). Following our reasoning that the indicators tap into distinct, not necessarily strongly correlated components of perceived recognition that have distinct roots and individual effects, we analyzed these subcomponents mostly separately. In addition, we explored the role of recognition as a composite concept. To do so, we aggregated the concept's facets (f) in a way that preserves the concept's assumed internal structure of sufficient conditions with weak substitutability: 7 This operationalization ensures high values on the aggregated target concept when only one of its facets is high (sufficient conditions). However, unlike the maximum, this formula considers the values of more than one subdimension in computing the aggregate score to capture the idea that a lack of recognition in multiple subdimensions can at least partially reinforce or offset each other (weak substitutability).
The four prototypical social groups were operationalized as follows: 1. Low-income individuals were defined as those with an equivalized household income in the lowest quintile. 2. Low-skilled workers encompassed those whose current or former occupation was either "unskilled and semiskilled worker" or "employee with simple duties." 3. Rural residents included individuals living in rural villages or single homesteads. 4. Sociocultural conservatives were defined based on a combined factor score from five sociocultural political issues (immigration, assimilation of immigrants, European integration, gender equality, climate change) as the quintile with the most conservative attitudes.
We measured populist attitudes using the established scale by Akkerman et al. (2014). To compute an aggregate score of populist attitudes, we multiplied the concept's subdimensions to account for the necessary conditions of the concept (Wuttke et al. 2020).

Analysis Plan
Our preregistered analysis consists of three parts. First, we studied bivariate associations between our key constructs, i.e., perceived lack of recognition and populist attitudes. Second, we analyzed how perceptions of lacking recognition vary across social groups. Third, we estimated regressions with populist attitudes as the outcome variable.
To assess our exploratory research question on the association between different facets of perceived lack of recognition, we show a correlation matrix of the four facets of societal recognition along with the composite index of societal recognition and populist attitudes (Fig. 2 below). The plot shows the degree to which the perception of lacking recognition in one facet is associated with the feeling of lacking 7 We divided the resulting score by 4 such that it varies between 0 and 1. Conservative Opinion recognition recognition in another facet. Our argument that there are distinct facets of perceiving lack of recognition implies that these correlations are less than close to perfect, yet we did not hold expectations regarding the exact strength of the associations.

K
In addition, the plot shows univariate distributions of all focal variables, giving the reader an overview of the data. 8 As a straightforward test of H1a to H1d, we tested for differences in the particular facet of perceived lack of recognition between the prototypical group and the residual category-i.e., low income vs. higher income, low-skilled occupation vs. other occupation, rural vs. urban, and strongly socioculturally conservative vs. moderate or progressive. We expected to obtain statistically significant differences between the prototypical groups and the residual category in line with H1a to H1d (as summarized in Table 1).
In addition, we also checked which lack of perceived recognition was most affected by which social marker. We expected the difference for each facet of lacking recognition to be largest when comparing its "paired" (see Table 1) prototypical group with the respective residual category. For example, we expected that the difference in perceived lack of recognition of one's economic needs would be larger when comparing the low-income group to other income groups than when comparing sociocultural conservatives to moderates/progressives. Conversely, perceptions of lacking recognition of one's opinion should differ most between (strong) sociocultural conservatives and moderates/progressives as compared to low vs. higher income, low-skilled vs. other occupation, and rural vs. urban. 9 Note that this test goes beyond what is directly stated in H1a to H1d. We thereby aimed to assess the implication that distinct feelings of lacking recognition have distinct social roots from a different angle.
We utilized a dumbbell plot to assess these expectations (Fig. 3). For each panel, i.e., facet of lack of recognition, we listed whether we expected the difference () between the means to be relatively large, medium (see footnote 9), or small. In addition to the visualization, we conducted t-tests to assess whether the means within groups were different from one another (e.g., comparing recognition for work between low-income respondents and all other respondents). We employed a p-value threshold of 0.05. The results are displayed in the appendix (Table D1). 10 To address the potential issue of multiple group membership blurring these associations, we also (a) determined the percentage of people who were members of at least two of the social groups chosen here (see Appendix C) and (b) replicated the t-tests for the differences for respondents who were not members of any of the other three focal groups (see Table D2 in the appendix). 11 Figure 2, the scatterplot matrix, gives insights into the bivariate relationship between each facet of lacking recognition and populist attitudes. To provide further evidence on H2a to H2d, we conducted multivariate linear regressions with populist attitudes as the outcome variable. We ran four main models. The results are reported in coefficient plots along with the models' R 2 values in the main text (Fig. 4) and in regression tables in the appendix (Table E1). All regression models included basic sociodemographic control variables (education, age, gender, east-west dummy differentiating respondents according to their place of residence).
The first model included all four lack-of-recognition measures to test whether each facet of recognition has unique predictive power when the other facets are controlled for. Generally, we expected that all facets would uniquely predict populist attitudes-i.e., we expected statistically significant coefficients for each recognition variable. However, it is conceivable that selected facets of recognition share substantial overlap-particularly the occupation and income facets-which may limit the unique predictive power of individual facets. Formally, we tested whether the coefficients were statistically different from zero with p < 0.05.
In the second model, we included the composite score of lacking recognition. We thereby assessed whether the composite score-which incorporates our logic that a lack of recognition in one facet is sufficient to give rise to an overall feeling of lacking recognition-has similar explanatory power as a model in which all four facets of recognition are included separately. To assess this, we compared the R 2 values of models 1 and 2. In addition, we compared the R 2 of model 2 to a model 10 In additional analyses in the appendix, we, first, reproduced Fig. 3 using more lenient cut-off points for the focal groups (see Fig. D2). Second, we reproduced Fig. 3 for three further social background characteristics that may also be associated with a perceived lack of recognition in general: low formal education, immigration background, and residence in the eastern part of Germany (cf. Schneickert et al. 2019). As discussed above for education, we anticipated these background characteristics to be not as clearly linked to just one particular facet of lacking recognition. We thus anticipated that these would not serve as well to support our argument. We therefore refrained from formulating expectations and relegated these exploratory analyses, including a brief discussion, to the appendix (see Fig. D3). 11 As another way to address overlapping group membership, we regressed each of the four facets of lacking recognition on the four social group dummies simultaneously. The results are displayed in the appendix (see Fig. D1 and Table D3), and we made comparisons of regression coefficients analogous to the expectations listed above. We anticipated that in the case of substantial overlap, a multivariate model may better disentangle the unique effect of each group membership on one particular facet of lacking recognition, making the expected difference emerge more clearly. For example, if there are more lowincome individuals among sociocultural conservatives than among moderates/progressives, perceived lack of recognition of economic needs might be more widespread among conservatives (in the bivariate display of Fig. 3) even if conservatism per se is unrelated to perceived recognition of economic needs.
K with a simple mean index of lack of recognition, and also to a model with a score derived from a principal component factor analysis (see Table B1). 12 Following H3, we expected that model 2 with our "sufficient condition" composite score would lead to a better fit than a model with the mean index that assumes the facets of lacking recognition to operate purely additively.
Via models 3 and 4, we aimed to approach the complex relationship between positions in society as reflected in the four prototypical groups, perceived recognition, and populist attitudes. To do so, the third model included the four social group variables (and none of the lack-of-recognition variables) to assess whether social group memberships predict populist attitudes (with multiple memberships being taken into account). The fourth model included both the four recognition variables and the four variables on social group membership. We expected to obtain statistically significant associations between lack of recognition and populist attitudes even when the social background variables were controlled for. We also compared the R 2 values of models 3 and 4 to assess how adding perceived recognition would improve our ability to account for populist attitudes.
We used the poststratification weight provided by the GLES that adjusts key sociodemographics to the German microcensus for our analyses (figures, t-tests, and regressions). 13

Results
Figure 2 conveys much of the information needed to explore feelings of lacking recognition and its implications for populist attitudes. The figure reports the prevalence and correlative patterns of the various facets of societal recognition among German citizens on the eve of the 2021 federal elections.
We first inspected the univariate distribution of the separate indicators, which are depicted in the plot's diagonal. They show that the facets of lacking recognition are distributed rather unequally. Feelings of lacking economic recognition are almost normally distributed. But only a minority of respondents feel that their access to basic services is denied or that their opinion is suppressed in public discourse. This indicates that the facets of lacking recognition are not identical but tap into something distinct. The facets' uniqueness is further underscored when looking at the correlation structure between the subdimensions.
As expected, citizens who feel no respect from society for their work are particularly likely to also perceive a lack of recognition of their economic situation (r = 0.64). On the other hand, feeling excluded for one's political opinions only weakly correlates with other facets of recognition. Several correlation coefficients hover around 0.4. Altogether, the correlation patterns confirm our expectation that the facets of recognition are connected to each other but still reflect distinct experiences of a person's perceived relationship with society. 12 The additional comparison with the factor score was not included in the preanalysis plan. 13 We thus excluded 175 observations with missing values on the poststratification weight.

Fig. 2 Distribution and correlations of feelings of lacking recognition and populist attitudes
We expected that certain positions in society-such as low-skilled workers or residents of rural areas-correlate with certain facets of lacking recognition (H1a-H1d). 14 Figure 3 confirms the expectation and shows that membership in these prototypical social groups predicts feelings of lacking recognition in a systematic way. In all cases, we observe the anticipated group differences (see Table D1 in the appendix): Compared to the rest of the population, citizens with low incomes report higher levels of lacking recognition of their economic situation, low-skilled workers report higher levels of lacking recognition for the work they do, residents in rural areas report higher levels of society lacking recognition of their needs for services, and sociocultural conservatives report higher levels of lacking recognition of their opinions.
These findings are consistent with H1a to H1d, but we also observe unexpected patterns. Two unexpected peculiarities stand out. We expected that one's area of residence would make the biggest difference in whether someone feels denied access to basic infrastructure and services, but other sociodemographic characteristics are much more predictive for this facet of recognition. Moreover, sociocultural conservatism stands out because this characteristic strongly correlates with all facets of lacking recognition and not just recognition of one's opinion.
To rule out that the findings reported in Fig. 3 are blurred by respondents belonging to multiple social groups, such as low-skilled and low-income, we conducted a series of additional (preregistered) analyses. In Appendix C, we show that only Fig. 3 Differences in perceived lack of recognition across facets and social groups. Dots are weighted means. All differences are statistically significant at p < 0.05 (see Table D1 in the appendix). Exp. expected, Obs. observed very few respondents meet the criteria for categorization in three or all four groups. The vast majority of citizens belong to no or only one social group that we considered predictive for the feeling of lacking recognition. When we exclude multiple group members from the comparisons of Fig. 3 (see Table D2 in the appendix), we observe very similar patterns. The differences anticipated by H1a to H1d all remain statistically significant and tend to increase. Multivariate regression analysis (see Fig. D1 and Table D3 in the appendix) on the recognition variables with all social groups as predictors is in line with H1a to H1d, too, but also underscores the explanatory power of cultural conservatism for lacking recognition along with income.
Thus far, our findings reveal that all social positions go along with recognition in the predicted direction. But some social positions have surprisingly wide-ranging ramifications. They play a role for multiple facets of lacking recognition. For example, holding conservative views not only correlates with higher levels of lack of recognition of one's opinion but also predicts the other three facets. Because of this, the observed results in Fig. 3 often do not match our expectations regarding which social marker predicts gradients in a particular facet of lacking recognition (see the deltas). However, we can adopt a different perspective and ask instead where a given social group makes the greatest difference. For example, in Fig. 3, the difference between conservatives and others is relatively large for all four facets, but it is largest for opinion recognition. In these comparisons, the anticipated pairing of social groups with particular facets of lacking recognition tends to emerge, particularly when taking account of multiple group membership, either by excluding multiple group members (Table D2 in the appendix) or by running regressions (Fig. D1 and Table D3 in the appendix). Here, we find that conservatism exhibits its largest association with opinion recognition, that low income displays its largest associaton with economic recognition (alongside recognition of the need for services), and that rural residence tends to reveal its largest association with recognition of the need for services. Thus, while low income and conservatism drive lack of recognition in more general ways than we anticipated, we also obtain some support for the idea of multiple roots, in that being a member of one of the prototypical groups is especially connected to one particular facet of lacking recognition.
Altogether, the picture that emerges from the empirical data up to this point largely conforms to our theoretical arguments on societal recognition as a multidimensional concept with interrelated but distinct facets. Whether and how strongly citizens feel left behind in one of these facets has traceable roots in a person's position in society, and membership in the four prototypical groups affects lack of recognition in the respective domains as hypothesized. However, this relationship is not mechanistic and is not as clearly structured as anticipated. In particular, sociocultural conservatism and low income play an unexpectedly significant role in shaping feelings of lacking recognition in its various forms.
Does a feeling of lacking societal recognition give rise to populist attitudes? Figure 1 gave first hints at the answer, showing modest positive correlations between populist attitudes and all four facets of recognition. The correlation is slightly stronger for recognition of one's opinions than for the other three facets. For the composite score that combines all facets into one aggregate index, we observe the strongest correlation (r = 0.49) with populist attitudes, suggesting that feeling deprived of recognition in any one facet is most predictive for whether a citizen endorses populist ideas.
To assess whether each facet of recognition has unique predictive power when the other facets are controlled for, Fig. 4 shows the results of multivariate analyses. In line with H2a to H2d, we find that each facet of recognition has a role to play in explaining variance in populist attitudes. Even when mutually controlled for, each facet of perceived lack of recognition is associated with populist attitudes in a statistically and substantively significant way. These results further support our argument of populist sentiment being rooted in multiple feelings of lacking recognition.
Yet it is worth noting that the explanatory power is much stronger for recognition of one's opinion and economic situation, and it is weaker for recognition of one's work and need for basic services. In combination, the four facets of recognition account for about a fourth of the variation in populist attitudes (R 2 is 0.24 in a model without any control variables; see Table E1 in the appendix), indicating that this one Fig. 4 Coefficient plots for linear regression models with populist attitudes as dependent variable. Shown are point estimates of coefficients with 95% confidence intervals. All models additionally include controls for gender, age groups, east-west residence, and low education and employ a poststratification weight adjusting the sample to the microcensus. LoR lack of recognition concept alone would often reliably predict whether someone supports populist ideas or not.
Having established the relevance of recognition for populist attitudes, we engage with the questions of how to best capture the multidimensional concept of lacking recognition as a composite score. When we use the composite score based on the aggregation rule of sufficient conditions with weak substitutability (model 2, Fig. 4), the explained variance is identical to the model that includes all facets separately (R 2 is 0.26 in both models, with control variables included). Remarkably, but as anticipated, using this parsimonious representation of lacking recognition does not come with a loss in explanatory power. This is different when using a simple mean index instead (R 2 = 0.24). 15 As anticipated, model 2 with the "sufficient condition" composite score fits better than a model with an index that assumes the facets of lacking recognition to operate just additively. These results support H3, according to which the contribution of individual facets of lacking societal recognition to populist attitudes is best captured by a logic of sufficient conditions with weak substitutability. 15 A factor score, built from a principal component factor analysis (see Table B1 in the appendix), fares even worse (R 2 = 0.23; see Table E1 in the appendix).
In a final set of analyses (models 3 and 4), we approach the complex relationship between positions in society, perceived recognition, and populist attitudes. In model 3, when recognition is not included in the analysis, we find that membership in three of the four prototypical social groups is statistically associated with more populist attitudes, with those in low-skilled occupations being the exception. Cultural conservatives stand out yet again for their comparably large association with the outcome variable. Adding the four lack-of-recognition items to this specification (in model 4) increases the model fit, indicating how feelings of lacking recognition improve our ability to account for populist attitudes. Moreover, the coefficients on the four group variables all become smaller, with only conservatives being significantly more inclined to hold populist attitudes once we condition on lack of recognition. While these analyses on cross-sectional data cannot reliably trace complex causal mechanisms, the results may be interpreted as indicative evidence that low-income individuals, sociocultural conservatives, and rural residents tend to score higher in populist attitudes because of specific feelings of lacking recognition, in line with the theoretical model depicted in Fig. 1.

Discussion and Conclusion
Our preregistered study confirmed the overarching idea of populism having multiple roots in feelings of lacking societal recognition. Distinct feelings of lacking societal recognition each contribute to populist attitudes. This relationship seems to follow a logic of sufficient conditions, whereby a perceived lack of recognition in one domain suffices to trigger populist sentiment. Distinct feelings of lacking societal recognition are, in turn, systematically rooted in distinct social characteristics.
However, a few findings were surprising. Some social characteristics (low income and, especially, conservatism) and some facets of recognition (opinion, economic) had a disproportionately large influence on populism relative to our expectations. Hence, populism has multiple roots in distinct feelings of lacking recognition that vary in prevalence across social segments, yet some of these roots seem more important than others. However, the deviations from some of our expectations do not change our main conclusions.
Our findings point to the explanatory power of societal recognition which, despite established bodies of literature across scientific disciplines, has not received systematic attention in political science. The concept has proven useful to reconceptualize the vague term of "feeling left behind" and to unite previously disparate perspectives on the drivers of populism. More generally, the feeling of lacking societal recognition as a concept that is seemingly remote from the political domain but deeply ingrained in the human psyche can serve as a theoretical bridge to connect macrolevel socioeconomic transformations with individual-level reactions. Based on our work, we see a vast avenue for future research to engage our theoretical framework and address some of the limitations of our article.
First, further work should define the conditions under which feelings of lacking recognition give rise to populist attitudes. Populist ideas are fundamentally about politics. It is an attitude about how society is and should be organized. For percep-K tions of recognition, which may feel very personal and remote from contemporary political debates, these feelings of lacking recognition must become politicized. In other words, citizens must establish a mental connection between the feelings of recognition and populist attitudes. Elite communication is likely key to this process (Zaller 1992). In order for the envisaged association to materialize on the level of ordinary citizens, politicians need to offer citizens a frame of interpretation of how elites are responsible for the lack of recognition that individuals perceive. For example, we suspect politicization is one reason why recognition of one's opinion stood out in our study with higher explanatory power whereas economic and regional aspects were less important. Here, the context of our study may play a role, as the alleged marginalization of culturally conservative opinions features prominently in current political debate in Germany, particularly among right-wing populist politicians. By contrast, contemporary political discourse in Germany rarely links populist tropes and socioeconomic conditions. Second, and related to the first point, future research should study populism's roots in feelings of lacking societal recognition in other countries to better understand the role of the context and elite communication. Such research may well reveal that the roots vary in importance across space and time. That being said, the prominence of the linkage between cultural conservatism and perceived lack of recognition of one's opinion, which we found in our study, features prominently in many other developed democracies, where the right-wing variety of populism that is tied to cultural issues dominates over left-wing varieties of populism that tie it to economic issues (e.g., Norris and Inglehart 2019).
Third, some of our results might also be sensitive to how we operationalized key constructs. This could be an issue with regard to the two prototypical groups for which we found weaker connections to perceived lack of recognition, i.e., lowskilled workers and those living in rural places. In the first case, we lacked detailed occupational information that would have allowed us to work with modern class schemes (e.g., Oesch 2006). 16 Alternative measures could help identify individuals who feel their work is not valued by society. In the second case, we deliberately decided to focus on rural locations as such, but it is conceivable that feeling a lack of recognition for one's needs for basic services results from rural deprivation, i.e., the combination of rurality and poor socioeconomic conditions at the local level. Moreover, and as indicated above, there may be other important facets of perceived lack of recognition that our nonexhaustive operationalization of this complex multidimensional construct failed to tap into.
Fourth, in our article we focused on populist attitudes rather than voting. Thus, the question remains how lacking societal recognition ultimately maps to vote choice through the support of populist attitudes. In the German case we studied, populist attitudes are strongly associated with voting, especially for the AfD (Jungkunz et al. 2021;Loew and Faas 2019;Steiner and Landwehr 2018). Thus, a path running from perceived lack of recognition and populist attitudes to populist voting is plausible in the German case. In line with this, Sachweh (2020) finds that feeling excluded by society in general strongly increases the likelihood of an AfD vote. However, other studies show weak or contingent connections between populist attitudes and voting for parties standing on populist platforms (Jungkunz et al. 2021;Quinlan and Tinney 2019). At the same time, the societal groups we linked to lack of perceived recognition often disproportionately support populist parties, albeit this strongly varies across contexts as well (Rooduijn 2018). Our research thus indicates new pathways that may link both sociodemographic profiles and populist attitudes to vote choice through a more complex path, which may help explain the varying outcomes in studies addressing the electoral support for populist parties.
Finally, our study remains limited in terms of causality. We were careful not to interpret the empirical evidence in a causal way because the available cross-sectional data does not permit strong inferences on the causal relationship between the variables of interest. In ex post analyses, shown in Table F1 of the appendix, we computed the robustness value to assess how sensitive the reported associations between populist attitudes and lack of recognition are to potentially omitted confounders (Cinelli and Hazlett 2020). These analyses show that if we had overlooked an unobserved confounder as strong as education (which is observed), this confounder would attenuate the estimated association between recognition and populism only from 0.49 to 0.47. While it is thus unlikely that a single unobserved confounder would completely wash out the observed correlation, we remain cautious about causal claims because our model may still have overlooked a bundle of multiple, interacting confounders. Moreover, with the available data we cannot rule out issues of reverse causality (see Traunmüller and Menzer 2022 in this special issue).
In sum, it has often been suggested that populism appeals to those who feel left behind by the development of society. Yet a theoretical framework centered on the insight that individuals can feel left behind in different ways and for different reasons has been lacking. Drawing on the established and inherently multifaceted construct of societal recognition, we aimed to contribute to the development of such a framework. The results of our preregistered study indicate that this framework holds considerable potential to better understand how populism can unite heterogeneous segments of the population. In the context of the 2021 German federal election, we found distinct feelings of lacking recognition to dominate in different social segments and observed that each of these distinct feelings of lacking recognition in turn contributed to populist sentiment. Given its promise to better understand how populism may unite seemingly unrelated groups to form powerful electoral and societal coalitions, we strongly encourage other researchers to apply this framework to explore contextual dependencies and other features we did not focus on in our study.
mons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4. 0/.