When Group Appeals Backfire: Explaining the Asymmetric Effects of Place-Based Appeals

Abstract

Group-based identities are an important basis of political competition. Politicians consciously appeal to specific social groups, and these group-based appeals often improve the evaluation of parties and candidates. Studying place-based appeals, we advance the understanding of this strategy by distinguishing between dominant and subordinate social groups. Using two survey experiments in Germany and England, we show that group appeals improve candidate evaluation among subordinate (rural) voters. By contrast, appeals to the dominant (urban) group trigger a negative reaction. While urban citizens’ weaker local identities and lower place-based resentment partly explain this asymmetry, they mainly dislike group-based appeals because of their antagonistic nature. If the same policies are framed as benefiting urban and rural dwellers alike, candidate evaluation improves. Thus, people on the dominant side of a group divide reject a framing of politics as antagonistically structured by this divide, even if they identify with the dominant group.

Introduction

Political scientists have recently rediscovered the importance of social identities and group memberships for political behaviour. Based on a revived interest in the importance of social cleavages (Bornschier et al. 2021a; Ford and Jennings 2020; Guth and Nelsen 2021; Hooghe and Marks 2018; Hooghe and Marks 2022), these studies investigate the role of social groups on the demand and supply side of politics. On the demand side, several studies have analyzed the importance of social identities and group attachments for individual political behaviour (Bornschier et al. 2021b; Evans, Stubager, and Langsæther 2022; Mason and Wronski 2018; Zollinger 2024). On the supply side, a growing number of studies investigate how parties use group-based appeals to mobilize different social groups (Dassonneville, Stubager, and Thau 2022; Dolinsky 2023; Horn et al. 2021; Huber 2022; Jacobs and Munis 2019; Robison et al. 2021; Stuckelberger and Tresch 2024; Thau 2019).

The literature on group-based appeals established the relevance of such appeals in party manifestos and campaign communications (Dolinsky 2023; Horn et al. 2021; Huber 2022; Stuckelberger and Tresch 2024; Thau 2019). Moreover, researchers have collected experimental evidence that voters react positively to these types of appeals (Dassonneville, Stubager, and Thau 2022; Jacobs and Munis 2019; Robison et al. 2021). These experimental studies generally paint the picture of a highly effective political instrument: most appeals investigated elicit a positive reaction from the targeted in-group. What is more, even out-groups often react neutrally to such appeals. Thus, group-based appeals seem to have a considerable upside but little downside from the perspective of political parties.

This perception, however, might emerge from existing studies focusing on the disadvantaged groups in major societal divides, such as lower-class or rural voters. Indeed, political appeals are often argued to be especially successful among members of more deserving groups that hold a marginalized position in society (Dassonneville, Stubager, and Thau 2022). So far, limited attention has been paid to how politicians can appeal to the opposing sides of these divides. First indicative evidence from a study on class-based appeals suggests that such appeals may have strongly asymmetric effects as upper-middle-class voters in the US react negatively to an upper-middle-class appeal (Robison et al. 2021, supplementary material Figure OB1). This raises a broader question on how group-based appeals work across both sides of structural societal divides, and how we may explain such asymmetries.

In this paper, we demonstrate that group-based appeals can have systematically different effects, depending on which side of the structural divide they address. We do so by reporting the results of two survey experiments in Germany and the UK. They allow us to distinguish in more detail for whom such group-based appeals have a positive impact, and when and why they can elicit a negative reaction. The context in which we encounter these asymmetric effects is in appeals to urban or rural voters. These appeals are of increasing substantive interest as urban-rural divides characterize many recent elections in the United States and across Europe. Indeed, ‘place’ is increasingly seen as a crucial element of people's social and political identities (Bornschier et al. 2021a; Cramer 2016; Munis 2022). Accordingly, place-based identities are an important potential target for group-based appeals. Rural (Dassonneville, Stubager, and Thau 2022; Jacobs and Munis 2019) and urban (Jacobs and Munis 2019) appeals have already been studied in the experimental literature on group-based appeals.

Moreover, the urban-rural divide is of particular interest since it increasingly stratifies Western societies, with urban people acquiring a socially dominant position while rural people are increasingly in a (objectively and subjectively) subordinate position. This conflict allows for positive in-group identification on both sides of the divide. While most rural citizens clearly have a positive identity as ‘rural’, a smaller but substantial number of urban citizens also hold a positive ‘urban’ identity (Lin and Trujillo 2023; Lyons and Utych 2023). Still, group-based appeals are typically studied in a context where only one group (the group with subordinate status; for example, the working class, or ethnic minorities) holds an in-group identity and sees the dominant, upper-status group as a clear out-group. By contrast, the members of the dominant group often hold a weaker group identity, at least as long as this identity is not being threatened. In the case of urban-rural divides, it is much more plausible that one person's out-group is another person's in-group. Hence, studying the power of appeals on both sides of the urban-rural divide provides a good context to study the scope conditions of group appeals.¹

Theoretically, we combine insights from the literature on cleavages and social positions with the group-appeals literature to theorize the scope conditions for when group appeals work. Specifically, we distinguish between dominant and subordinate groups based on their position and trajectory in a social divide and propose that group-based appeals will have asymmetric effects on these two groups. People on the dominant side of a social divide will typically have no interest in the framing of politics revolving around this divide and will thus resist antagonistic appeals. After all, such a framing necessarily emphasizes their dominant, privileged position, which often triggers negative emotions of self-directed anger and guilt (Dover 2022; Harth, Kessler, and Leach 2008; Leach, Iyer, and Pedersen 2006). Therefore, voters who perceive themselves to be on the winning side of a social divide reject appeals that emphasize this divide. Even though they might value a policy that would favour their in-group, they prefer appeals that depict this policy as serving a common interest of all groups and thereby downplay the divide. By contrast, the group that perceives itself as subordinate is more receptive to conflictual appeals as its position provides the moral grounds upon which to challenge the existing order and the out-group dominance. The effectiveness of group-based appeals among subordinate groups mirrors the existing literature's focus on studying appeals to disadvantaged social groups such as the working class, rural people, or people with lower education levels (for example, Dassonneville, Stubager, and Thau 2022, Robison et al. 2021).

We test this theoretical argument with a set of survey experiments that expose dominant (that is urban) or subordinate (that is rural) social groups to group-based appeals. We find that appeals to rural voters indeed have the expected effect. In both Germany and the UK, rural voters evaluate candidates who appeal to them as a group significantly better than generic candidates who make a general appeal to all voters. By contrast, we do not find a positive effect of appealing to urban voters. On the contrary, urban appeals reduce urban voters' evaluation of a candidate.

While we anticipated an asymmetry in the reaction of urban and rural voters, we were surprised by the intensity of urban voters' negative reaction to group-based appeals that have strong positive effects among rural voters. Urban voters' antipathy towards an in-group appeal is partly explained by the fact that urban citizens on average have weaker local identities and report lower place-based resentment. Nevertheless, there remains a substantial difference between urban and rural voters' reactions to the in-group appeal, even after controlling for these factors. Even people with a strong urban identity at best react neutrally to an urban appeal.

To gain a better understanding of this reaction, we ran a second survey targeted only at urban voters in Germany. Since our interest in this survey was specifically about the backlash among urban voters to in-group appeals, we recruited only urban respondents. Following our theoretical considerations, we adjusted the policy content and the antagonistic nature of the urban appeals to test whether this more advantaged group of voters prefers addressing their interests in a less conflictual way. In the second survey, we vary both the specific policy issues addressed in the appeal, as well as how the relationship between urban and rural interests is presented. We find that the specific issues do not make a difference: urban respondents dislike an innovation-focused appeal and a housing-focused appeal just as much as a generic economic appeal. However, how the appeal presents the relationship between urban and rural interests has a decisive impact; if the appeal presents politics as an antagonistic conflict between urban and rural citizens, urban voters reject the appeal. If the exact same policies are presented in a harmonious framework that emphasizes their benefits for all citizens, urban voters react positively.

These findings provide important scope conditions for the political effectiveness of group-based appeals. Antagonistic appeals to dominant groups can reduce a politician's support; this is a finding that has been suggested for appeals to higher-class voters (Robison et al. 2021), which we develop and explain systematically for urban residents. Conflictual appeals only mobilize support among subordinate groups, for whom such appeals present societal order in line with their own experiences. Our study suggests that successful group-based appeals require a consideration of the type of social group targeted by the politician with respect to perceptions of dominant vs. subordinate status.

This paper is organized as follows. In the next section, we develop our argument about the limits of group-based appeals and the importance of considering which side of a structural divide is the target of the appeal. We then explain why the urban-rural context is suitable for studying the differential reaction to group-based appeals. Thereafter, we explain our survey design. In the results section, we first present the results from our comparative survey, along with several subgroup analyses. We then dig deeper into the behaviour of urban voters by analyzing the results of the follow-up survey. In the conclusion, we discuss potential avenues for future research, with a focus on the limits of group-based appeals.

Argument

Group-Based Mobilization

The literature on group-based appeals starts from the theoretical premise that politics is fundamentally characterized by a conflict between social groups (Thau 2021). Therefore, these studies argue that group-based appeals improve people's evaluation of the appealing political candidate or party if people perceive the group that is the target of the appeal as an important part of their own social identity.

Some studies in this literature use observational data to study parties' actual appeals (Dolinsky 2023; Horn et al. 2021; Stuckelberger and Tresch 2024; Thau 2019). They show the continuing importance of group-based appeals across countries and types of groups. While the most important group membership analyzed is class (Evans and Tilley 2017; Robison et al. 2021; Thau 2021), other studies focused on gender (Kam, Archer, and Geer 2017), or a wider range of groups (Dolinsky 2023; Huber 2022; Stuckelberger and Tresch 2024). Moreover, these studies show that parties target these appeals strategically (Huber 2022).

In addition to these observational studies, a growing number of studies use survey experiments to provide causal evidence for the effectiveness of group-based appeals. This literature typically asks survey participants to evaluate fictitious candidates for political office based on short statements or visual cues (Dassonneville, Stubager, and Thau 2022; Jacobs and Munis 2019; Robison et al. 2021). Here, it is usually argued that signalling attention to and tuning of policies towards a specific social group will be rewarded by group members as this enables a linking of political candidates to the group and its attributes (Robison et al. 2021). Indeed, these experimental studies generally show that group-based appeals elicit a positive reaction from the targeted group.

Based on the theoretical understanding of politics as being structured by group conflict, we thus derive our first hypothesis about the effectiveness of group-based appeals.

H1: Group-based appeals improve candidate assessment.

That said, this literature indicates two scope conditions for the success of group-based appeals.² Firstly, voters must identify with the group targeted by the appeal, and the appeal must make this identity salient and imbue it with political meaning (Huddy 2013). Secondly, group members need to see themselves in conflict with the out-group of the appeal and harbour resentment against this out-group. Hence, as depicted in the framed part of Fig. 1, group-based appeals only work when there is a certain identification with the in-group as well as a certain resentment towards the out-group of the social divide.

Figure 1. The general model of group-based appeals and its potential variation across social groups.

These considerations lead to a second hypothesis about the scope conditions for the effectivity of group appeals:

H2: The effect of group-based appeals is moderated by the strength of in-group identification and out-group resentment.

However, the existing literature typically studies appeals to socially subordinate groups. Robison et al. (2021), for example, focus on the effectiveness of working-class appeals, while Dassonneville, Stubager, and Thau (2022) include appeals to four groups: the working class, people with lower education levels, rural people, and the young. In all these cases (except, perhaps, for the young), the in-group of the appeal is in a socially subordinate position while the out-group is in a dominant position. An exception to the focus on the disadvantaged side of societal divides is the study by Jacobs and Munis (2019), who studied both urban and rural appeals in the American context. Here, the authors find clear positive effects among rural voters but less effectiveness of appeals among urban voters.

Asymmetry of Group-Based Conflict

Political cleavages typically emerge in reaction to the discontent of groups who perceive themselves as subordinate or threatened. Thus, the focus on subordinate groups in the political appeals literature makes sense from a broader theoretical perspective. As Hooghe, Kamphorst, and Marks (2022, 2) put it: ‘In the Lipset and Rokkan model, a major shock to the fabric of society motivates opposition to the status quo on the part of groups whose lives are affected in ways that are difficult or impossible for them to escape.’ Hence, parties representing a specific cleavage – be it Socialists, Catholic parties, or Farmers' parties – typically emerged as representatives of those who perceived themselves on the losing side of the industrial or national revolution, where losing could mean either a generally low status or a relatively high but declining status.³

By contrast, few parties actively seek to mobilize the other side of a cleavage. Whereas working-class parties historically mobilized class-based divides, liberal or Christian-Democratic parties often sought to deflect the importance of these divides, rather seeking to emphasize the commonalities between bourgeois and working-class voters – for example, their common nationality or faith.

Thus, the losing side typically drives the political mobilization of societal divides with clear winners and losers. In 1959, Ralf Dahrendorf had already argued that groups of lower social status were much more likely to see society in terms of conflict (Dahrendorf 1959). As he suggested, ‘the dominant groups of society express their comparative gratification with existing conditions inter alia by visualizing and describing these conditions as ordered and reasonable; subjected groups, on the other hand, tend to emphasize the cleavages that, in their opinion, account for the deprivations they feel’ (Dahrendorf 1959, 293). While Dahrendorf focused on the class divide, contemporary accounts emphasize that dominant groups such as white people or men are reluctant to see society as structured by racial or gender divides, unless they start to see their position being threatened (Friedman, O'Brien, and McDonald 2021; Jardina 2019; Phillips and Lowery 2018).

This suggests an important scope condition for the effectiveness of group-based appeals. As the literature amply demonstrates, such appeals generally work for the subordinate side of a political divide, as presented in the framed part of Fig. 1 (for example, Robison et al. 2021). For groups who perceive their status as low and/or declining, these appeals ‘emphasize the cleavages that in their opinion account for the deprivations they feel’ (Dahrendorf 1959, 293). Thus, these appeals present a group's self-interest as morally justified. Moreover, they may increase the political entrepreneur's credibility as an advocate of this group.

By contrast, for dominant/ascending groups, these appeals clash with their preference for ‘visualizing and describing these conditions as ordered and reasonable’. On the contrary, the antagonistic nature of these appeals has quite the opposite effect. It taps into distributive conflicts between a person's in-group and out-group, and thereby emphasizes the existing inequalities that favour the dominant group. Group-based appeals to dominant social groups thus run the risk of activating privilege and concerns for out-group disadvantage. Indeed, inequalities across social groups induce feelings of guilt and unfairness that play a key role in producing out-group concern among individuals belonging to dominant groups (Chudy, Piston, and Shipper 2019; Powell, Branscombe, and Schmitt 2005). In line with that, dominant groups have been shown to bias political party agendas by exempting divisive issues from political competition (Traber et al. 2022; Weber 2020). We therefore expect that in-group appeals will be less effective on the dominant side of a structural divide. Instead, we expect that groups who perceive their status as high and ascending prefer harmonious appeals, which address the interests of a social group but avoid opening up a conflict with other groups and, rather, emphasize that the group's interests are aligned with those of other groups. Based on these considerations, we thus hypothesize:

H3: Group-based appeals work better for subordinate than for dominant groups, even after controlling for in-group identification and out-group resentment.

A priori, it is hard to predict how strong the difference between dominant and subordinate groups will be. In the preregistration of our first survey experiment,⁴ we pre-registered a strong positive effect of rural appeals and a weaker, yet still positive, effect of urban appeals. Theoretically, however, we can mainly predict an asymmetry but not its magnitude.

That said, within the dominant group, the reaction to group-based appeals will again depend on the factors identified above: in-group identification and out-group resentment. Among group members without in-group identification or out-group resentment, these appeals might even elicit a negative reaction as these groups prefer to avoid and diffuse social conflicts over inequalities from which they are benefitting. Trying to mobilize people as ‘wealthy’ or ‘West German’ can backfire since few individuals strongly identify with these groups. This was probably also the case for ‘whites’ in the United States but this may be changing since whites increasingly perceive their position as eroding (Jardina 2019). Similarly, if the dominant group does not harbour resentment towards the out-group, group-based mobilization attempts could trigger negative reactions. However, this becomes less clear if membership in the dominant group is an important social identity. For example, highly-educated people are clearly privileged yet they hold a positive in-group identity (Stubager 2009). In these situations, the reaction to in-group appeals among the dominant group may be more neutral. Still, it will not have the same positive effect as for subordinate groups.

Research Design

Case Selection: The Urban-Rural Divide

To provide empirical evidence on the asymmetry of group-based appeals, we ideally would want to study a societal divide in which dominance and subordination are clearly distributed yet allow people on both sides of the divide to hold positive in-group identities. Many divides, such as class, only fulfil the first criterion. While dominance and subordination are clearly assigned, those on the dominant side usually do not develop strong in-group identities. Other divides, such as age, only fulfil the second criterion. While ‘young’ and ‘old’ may be relevant identities, it is less clear which group is dominant and which is subordinate. We propose that the urban-rural divide is well-suited to fulfilling both criteria. There is a widely shared perception of which group is winning and losing, yet it still offers a positive identification to both groups.

We studied this divide in England and Germany, following a most-different case design. In England, urban places (in particular, London) clearly dominate the country in economic and cultural terms. Moreover, British (English) centralism reinforces centre-periphery dynamics. In Germany, the economic model is centred on a strong industrial sector, whose ‘hidden champions’ are often based in rural areas. Moreover, strong federalism and the proportional electoral system incentivize parties to represent voters from all parts of the country. We then ran the follow-up survey only in Germany, given that our initial results were very similar in both countries.

Despite these objective country differences, urban dwellers are typically perceived as economic and cultural winners in both countries. In economic terms, the countryside is often characterized as being ‘left behind’, while urban areas benefit from the transformation to a knowledge economy (Rodríguez-Pose 2017; Iversen and Soskice 2019). At the same time, there is a growing cultural divergence between ‘cosmopolitan’ cities and ‘nationalist’ rural areas (Cramer 2016; Huijsmans et al. 2021; Maxwell 2019, 2020). Indeed, in our survey, those who classified their place of residence as ‘very urban’ reported a significantly better economic and cultural trajectory of their community over the last decade compared to all other respondents (Appendix B, Table B4–B7). Finally, urban dwellers have better chances to make their political voices heard (Gimpel, Lee, and Thorpe 2011; Haffert 2024). Accordingly, rural voters perceive systematically lower levels of political efficacy across European countries (García del Horno, Rico, and Hernández 2023).

Most empirical analyses focus on rural voters and argue that their strong local attachments can lead them to develop ‘rural resentment’ (Cramer 2016; Munis 2022) because they feel that they do not get their ‘fair share’ and that their way of life is not being respected by urban voters. Both types of grievances are much less common among urban voters. This line of research has established that, especially in the US context, rural appeals are a very common and highly effective feature of political mobilization (Munis 2020).

While this characterization of rural places is relatively uncontested, there are two potential objections to the characterization of urban places as being advantaged. Firstly, not all urban places are on the winning side. Rustbelt cities such as Sunderland or Newcastle in England or Gelsenkirchen or Chemnitz in Germany often have higher unemployment and lower incomes than rural counties. Secondly, not all inhabitants of booming cities are winners. Processes of gentrification often push out longstanding dwellers who cannot afford rising rents and costs of living.

The important point, however, is that these relative losers are unlikely to see their situation as being related to an urban-rural divide. Even if they identify strongly with their place, they will not see rural people as the out-group responsible for their plight. Rustbelt inhabitants are more likely to blame globalization or immigration, whereas gentrification losers may blame wealthier urban residents. At the same time, they are unlikely to feel strong commonalities with rural losers. For example, they will often struggle with the consequences of population growth (in the form of rising rents) rather than population decline. Still, our empirical design includes appeals that tap into both the strength of booming cities (‘innovation centres’) as well as the difficulties of some urban dwellers (‘affordable housing’).

Domination and subordination in the urban-rural divide are thus clearly recognized. Moreover, ‘place’ is increasingly seen as an important element of people's social and political identities (Bornschier et al. 2021a; Bornschier et al. 2021b). Many studies emphasize the conflict between identities based in rural communities (Cramer 2016) and cosmopolitan cities (Cunningham and Savage 2015) when seeking to explain the rise of anti-establishment parties (Lunz Trujillo 2022; Lyons and Utych 2023). Indeed, place-based identities are becoming more important to people as other sources of identity such as membership in collective organizations become less salient (Fitzgerald 2018). In the European context, Hegewald and Schraff (2022) demonstrate the existence of an identity-based affective polarization between urban and rural voters while Schulte-Cloos and Bauer (2023) show that voters have an identity-based preference for candidates from their own municipality.

As with the objective position of both groups, it is widely accepted that ‘rural’ is an increasingly important social identity. By contrast, it is less clear whether ‘urban’ is a meaningful identity.⁵ We argue that it is meaningful for an important subgroup of urban dwellers. One way to demonstrate this is by asking about people's perceived closeness to ‘urban’ and ‘rural’ people (for a validation of this question, see Bornschier et al. 2021b). In Appendix A (Figures A7 and A8), we show that perceived closeness to urban people varies just as much with local population density as perceived closeness to rural people. However, identification is weaker for both urban and rural voters who grew up in a different type of place (Hegewald and Schraff 2022; Lunz Trujillo 2022).

While there seem to be distinct identities on both sides of the urban-rural divide, this is less clear for the existence of place-based resentment. Indeed, the literature has a clear focus on ‘rural resentment’ (Cramer 2016). Using a place resentment scale, Munis (2022) finds that place-based resentment is much higher among rural than among urban Americans. However, the growing salience of urban-rural conflicts in politics may mean that urban dwellers start to develop anti-rural resentment for political reasons. In the US context, this is epitomized in the sentence ‘land does not vote, people vote’ in response to choropleth maps of electoral results. More generally, rural voters may be seen as blocking important reforms in areas from climate policies to social liberalization.

In terms of parties' actual behaviour, finally, existing observational studies show that place-based appeals are an important element of parties' repertoire. This is particularly true of rural appeals, which parties regularly employ in both Austria (Huber 2022) and the UK (Thau 2019). Urban appeals are much less common in Austria, at least between 2013 and 2019 (Huber 2022). In the UK, by contrast, they have been employed about as often as rural appeals between 1964 and 2015 (Thau 2019).

Data and Operationalization

To study group appeals across the urban-rural divide, we fielded two online survey experiments with the survey company Bilendi. The first experiment was conducted in Germany and England in September 2021. The second experiment was conducted in Germany in June and July 2022, serving as a follow-up study for questions arising from the first study. In all analyses, we exclude speeders and those (roughly 10 per cent) respondents who failed to pass a simple attention check. In the following, we describe the design of the first survey in detail. When introducing the follow-up survey below, we only explain where the design deviated from the first survey.

The design of both experiments follows established approaches from the literature on group-based appeals (Dassonneville, Stubager, and Thau 2022; Robison et al. 2021). In both surveys, we asked respondents to evaluate a candidate for parliament at the next national election.⁶ Within each survey, we hold constant the overall candidate description. In the first survey, we presented the candidate as follows, ‘David Williams is 46 years old. He is an office clerk and has two kids. He has been living in your constituency for 20 years and volunteers as a youth coach in a local sports club.’ In the second survey, we changed the candidate's gender, name, and occupation to make her a more authentic representative of urban interests (see below).

The treatment conditions then attach a certain appeal to this candidate and test whether this affects the respondents' candidate evaluation. Our main outcome is the question: ‘How would you rate a candidate with political views like those of David Williams?’ The outcome measure then varies between ‘0 – I think very poorly of him’ to ‘10 – I think very highly of him’.

We measure subjective place of residence pre-treatment to establish whether people understand themselves as residing in a rural or urban context. For this, we use a four-category scale, which we then dichotomize into ‘urban’ and ‘rural’. In Germany, 62 per cent of respondents describe their residence as ‘very urban’ or ‘somewhat urban’; this number is 70 per cent in the UK.⁷ To maximize statistical power, we then expose all participants to a treatment that fits their subjective place of residence. In other words, there is in principle no mistreatment (Hersh and Schaffner 2013).

We measure the strength of place-based identity by showing the respondents five pairs of photos, one of an urban and one of a rural environment, and have them choose in which of the two places people are more like themselves ‘in terms of their lifestyle and their opinions’. These photos seek to illustrate both the advantages and disadvantages of an urban or rural lifestyle (see the photos in Appendix B).⁸ Respondents who chose four or five photos matching their perceived place of residence are coded as having a strong place-based identity. We prefer this measure to a survey question about attachment to one's place of residence since the reaction to the photos better captures an abstract identification with urban or rural places, not the very concrete identification with one's specific place of residence.⁹ This is important since we are interested in the mobilizing power of appeals to urban/rural places as such, not in the power of specific appeals to respondents' local places. Furthermore, our visual-based measure helps to mitigate concerns that especially the dominant group may hesitate to convey a strong identity through conventional language-based items.

In Germany, 62 per cent of those who described their place as ‘very urban’ picked four or five urban photos while 92 per cent of those who described their place as ‘very rural’ picked four or five rural photos. In the UK, the respective numbers were 53 per cent and 90 per cent, respectively. While rural identities are more widespread than urban identities, more than half of urban respondents seem to have some form of urban identity.

We measure people's place-based resentment using a five-question battery based on Munis (2022). These questions ask whether people feel that their place-based in-group is disadvantaged compared to the out-group in terms of their economic, cultural, or political situation (see the full list of questions in Appendix B). Answers for all items range from 1 to 5, and we code everyone with a mean across all items above 3 as having place-based resentment.

This setup of the survey was basically identical in the first and the second surveys. However, the surveys differed regarding the target population, the description of the candidate, and the content of the presented appeals. We first present the specific design and the results of the comparative survey before we move on to the German survey of urban respondents.

Results from Comparative Surveys in Germany and the UK

The survey experiment follows a multi-arm vignette design with a control group and three treatment groups. This leads to a set of seven groups receiving separate appeals, as presented in Table 1. Following the appeals literature (Robison et al. 2021), we differentiate between symbolic and policy appeals. Symbolic appeals portray the political elite as an ally of some social group but are void of any specific policy statements (Hersh and Schaffner 2013). Policy appeals, by contrast, try to appeal to a social group by elaborating on a substantive political issue that might be of concern to this group.

Table 1. Structure of the vignette experiment

The control group received the following generic appeal that does not engage with the urban-rural divide: ‘I want to be a voice for everyone in our country. It is time to put the people at the centre of politics.’ All other appeals start with an identical frame, which presents politics as a conflict between urban and rural people: ‘Too much attention has been given to big cities/rural areas in recent political debates.’¹⁰ After this, we add the appeals. The symbolic appeal is formulated in broad terms, appealing to the urban or rural groups but not mentioning any substantive societal issues: ‘I want to be a voice for rural areas/our cities. It is time for politicians to prioritize rural areas/our cities.’

Our policy appeals are derived from the place-based resentment literature, as outlined in the theory part. The economic appeal implies competition regarding investment between urban and rural areas: ‘I want to support economic development in rural areas/our cities. The government needs to invest more in infrastructure and the creation of jobs in rural areas/our cities.’ The cultural appeal, by contrast, forwards a competition in lifestyles across the urban-rural divide: ‘I want to fight for the lifestyle of people in rural areas/our cities. The government needs to support social and cultural activities in rural areas/our cities.’

We collected nationally representative samples (according to gender, age [18–74], education, and region/state) from around 2500 respondents. Our study is well-powered to uncover effect sizes of 0.6 or more, assuming a standard deviation of 2.5 on the standard normally distributed outcome and splitting the sample into two equally sized groups (for example, urban and rural residents).¹¹ We can capture even smaller effects if we pool across treatment conditions or people's places of residence. Appendix C presents descriptive statistics for the control and treatment groups. The means indicate that samples are balanced and randomization was successful.

The empirical analysis presents the results from ordinary least squares (OLS) regressions of candidate evaluation on the treatments. Figure 2 presents predicted values of the candidate evaluation with 95 per cent confidence intervals over the type of place and the experimental conditions. The complete regression results are presented in Appendix D. Overall, we see similar patterns in both countries. There is no statistical difference between urban and rural residents' candidate evaluation in the control group, suggesting that our neutral description of the candidate did not trigger an urban or rural frame. Looking at the treatments, place-based appeals tend to increase candidate evaluation among rural voters but consistently decrease the rating among urban respondents. Among respondents with subjective rural residency, we find that cultural appeal increases candidate evaluations by 0.6 points in Germany and 0.5 points in the UK. These average treatment effects (ATEs) are statistically significant at the one and five per cent level, respectively. Economic and symbolic appeals do not affect candidate evaluations among rural voters in England. In Germany, both appeals improve candidate evaluations among rural respondents. The symbolic appeal increases evaluations by 0.5 points, statistically significant at the 5 per cent level. The economic appeal has an ATE of 0.3 and is only marginally significant at the 10 per cent level.

Figure 2. Main results – candidate evaluation: (a) Candidate evaluation Germany, (b) Candidate evaluation UK.

We thus find clear, although not unequivocal, evidence for Hypothesis 1 in rural areas. Among those with subjective urban residency, by contrast, all appeals strongly decrease candidate evaluations. This negative response is significant at the 1 per cent level in all treatment conditions. The cultural and economic appeals decrease candidate assessment by about 0.7 points compared to the control group while the negative effect of the symbolic appeal is even bigger at around 1.1 points. These results are nearly identical across both countries. While rural voters seem to be most susceptible to appeals that emphasize a cultural conflict, urban voters dislike all three appeals to a similar extent.

In line with Hypothesis 3, these findings suggest that the effects of place-based appeals are highly asymmetric: place-based appeals hurt candidates in an urban setting but they can benefit their evaluation in a rural context. However, the strength of this asymmetry and the negative reaction of urbanites to the urban appeal is remarkable and deserves further attention.

In the next step, we test whether different reactions of urban and rural voters can be explained by a weaker place-based identity among urban voters (cf. Hypothesis 2). To do so, we collapse all three treatments into a single category ‘treated’ and compare all treated respondents to the non-treated for each country. We interact the treatment variable with our measures of place-based identity. Figure 3 shows how respondents with weak and strong place-based identities react to the treatment. Panel a) shows the results from Germany and panel b) the findings from the UK. Respondents with a weak urban identity react negatively to the treatment in both countries. In the UK, this effect disappears among high identifiers. That said, high identifiers still do not rate the candidate better than the control candidate, meaning that the appeal may not hurt, but neither does it benefit the candidate. In Germany, even strong urban identifiers react negatively to the place-based appeal, although the effect is considerably smaller than among low identifiers.

Figure 3. Strength of place-based identity and candidate evaluation. (a) Candidate evaluation Germany. (b) Candidate evaluation UK.

Unfortunately, in this survey, we only measured the respondents’ place-based resentment after the treatment since we had pre-registered a hypothesis on the effect of treatment on place-based resentment. Hence, this measure might be confounded by the treatment. We thus only look into the effect of resentment in the second survey. Moreover, we also pre-registered a range of hypotheses about the different types of appeals (symbolic/economic/cultural) and subgroup effects (by age or political ideology). These were meant to explore variation within the group of urban and rural respondents. However, our results indicate that the most important question is not the variation within these groups but the massive difference between these groups. While our results indicate that place-based appeals are a promising strategy for rural candidates, we see no evidence for positive reactions even among high-identifying urban voters to any of our appeals.

Hence, rather than studying within-group variation, we choose to focus on the big difference between urban and rural residents. To understand the reactions of the dominant side of the urban-rural divide to place-based appeals better and to discern why a majority of urbanites reject urban appeals, we fielded a second survey in Germany that focused on urban voters, testing our argument on the relevance of harmonious appeals for dominant groups (see Hypothesis 3). As negative reactions among urban voters were very similar in Germany and England, we think that just focusing on Germany still speaks to the broader puzzle.

Results From Urban Survey in Germany

While repeating the fundamental setup of the first survey among urban voters in Germany, we introduce two major types of variation in the design of the follow-up survey. Firstly, we increase the specificity of the policy proposal in the appeals, since one reason for the failure of urban treatments in Study 1 could be that they do not pick up relevant policy concerns. Secondly, we vary whether urban and rural interests are presented as antagonistic or harmonious, following our theoretical argument on the reluctance of dominant groups to endorse antagonistic appeals. In addition, we also changed the demographic attributes of the fictitious candidate and presented the candidate as follows: ‘Carolin Reuss is 42 years old. She is a graphic designer and a single mother of a daughter. She has been living in your constituency for 20 years and volunteers in the admin team of a local climbing hall.’ This allows us to see whether urban voters may have considered the original candidate as an inauthentic advocate of urban issues because of his gender, occupation, or lifestyle.

For the follow-up survey, we recruited a total of 2850 inhabitants from the 150 biggest German cities (that is, all cities with more than 60,000 inhabitants). Of these respondents, 54 per cent described their place of residence as ‘very urban’, 39 per cent as ‘rather urban’, 6 per cent as ‘somewhat urban’, and 1.5 per cent as ‘not urban’ (see the descriptive statistics in Appendix E). Our results are robust to excluding the two latter groups. This sampling strategy allows us to focus on the main puzzle emerging from Study 1, namely the strong negative reactions to political appeals among urban respondents.

As a control condition, we used the same generic appeal as in Survey 1. As our first treatment, we repeat the urban economic appeal from Study 1 to see whether we could replicate our initial findings. As Fig. 4 demonstrates, urban voters again react negatively to this appeal (complete regression results can be found in Appendix F). Even if we describe the candidate in a way that more closely fits the image of an urban lifestyle, respondents dislike the antagonistic group-based appeal just as much. Thus, it seems that a lack of authenticity was not the problem with the original appeal.

Figure 4. Effect of treatment on candidate evaluation, Study 2.

We use the rural economic appeal from Study 1 as our second treatment to test whether urban respondents may prefer a pro-rural (out-group) appeal. Exposing a subgroup of respondents to the supposed out-group appeal is similar to the design of Robison et al. (2021), who administered a working-class appeal not just to the working class but also to middle and upper-middle-class respondents. Similarly, Dassonneville, Stubager, and Thau (2022) expose all respondents to the same working class/rural/non-graduate/young appeal. Hence, they also administer the same appeal as an in-group- and an out-group appeal.

Figure 4 shows that urban voters also dislike the rural appeal. That is, they do not just reject a specific type of place-based appeal for targeting the wrong group but, rather, reject any type of place-based appeal. This negative reaction demonstrates that the original finding is not due to the fact that people perceive rural people as their in-group or have strong out-group concerns. Even low-educated urbanites do not show any sign of class-based solidarity with (potentially assumed, similarly lower-educated) rural voters (see Appendix Figure F4). Such a negative reaction to appeals to a deserving out-group is a relatively uncommon finding. Dassonneville, Stubager, and Thau (2022) find that people react neutrally to out-group appeals in most cases. The exception is appeals to the young, which are actively disliked by old respondents. Yet young people are significantly more positive towards this appeal. In our case, however, urban respondents dislike pro-urban and pro-rural appeals just as much. This suggests that urban voters have a more general preference for attenuating the urban-rural divide, no matter from which side it is polarized. This is in line with our argument on the behaviour of dominant groups and with broader insights on the political interest representation of advantaged groups (Weber 2020).

In addition to these replications and extensions, we introduce four new appeals. Here, the most important change concerns how the appeal presents the relationship between urban and rural citizens. In Study 1, all three appeals clearly depicted an antagonistic us-them dichotomy, a conflict between urban and rural areas. As demonstrated above, the lower level of place-based resentment among urbanites may partly explain their negative reaction to this framing. However, even among the resentful urbanites, the original treatments did not generate positive effects. We thus vary whether respondents are confronted with a similarly antagonistic appeal or whether the appeal emphasized the harmonious interests of all voters while still catering to urban interests.

Secondly, we also vary the specificity of the candidate's proposal and seek to address it to different aspects of the urban-rural divide. On the one hand, we introduced an appeal that addresses cities as beneficiaries of the transition to the knowledge economy.

The antagonistic version of this appeal read as follows:

Germany is falling behind with regard to innovations and technologies, since policymakers have given too much attention to the economic interests of rural areas. The government needs to invest more into the generation of future-proof jobs and the creation of innovation centers in our cities.

By contrast, the harmonious appeal reads as follows:

Germany is falling behind with regard to innovations and technologies, since policymakers have given too little attention to the future. The government needs to invest more into the generation of future-proof jobs and the creation of innovation centers in the whole country.

Moreover, we introduce two appeals on the issue of housing, which addresses problems arising from the growth of cities. Housing has been one of the most discussed political issues in German cities in recent years. Hence, rent should be a topic of high salience to urban voters and one where they see a shared group interest with other urbanites. In the antagonistic version, the rise in costs of housing is attributed to the influx of rural people. Investment in housing in this case is geared towards alleviating the worries of urban residents and their rent costs. The appeal is formulated as follows:

More and more people in our cities suffer from rising rents. Because of the influx from the countryside, people who have been living there for many years cannot afford the rent in their district anymore. I want to invest in affordable housing for the people in our cities. The government needs to stop the influx from the countryside and to improve the housing situation in the cities.

In the other condition, we refrain from pitting urban against rural constituencies and present the suggested housing policies as benefitting people across the country. This of course hides the fact that high rent prices are primarily an issue for urban residents. The appeal, therefore, pitches a policy that is very much tuned towards the urban in-group but sells it as a beneficial policy for everyone. This appeal reads as follows:

More and more people in our country suffer from rising rents. Increasingly, people who have been renting for many years cannot afford their flats anymore. I want to invest in affordable housing. The government needs to stop rent increases across the country.

We expect that, compared to the appeals presented to urban voters in the first survey, increasing the specificity of the appeal and reducing how antagonistic it is should result in higher approval of the candidate by urban voters. Especially, the non-antagonistic framing of pro-housing policies should bolster the approval of the candidate amongst urban constituencies as it presents a clear in-group interest in a way that hides the group's dominance.

The mid-part of Fig. 4 shows that the specificity of the policy has, at most, minor effects. The antagonistic innovation appeal is evaluated just as negatively as the original economic appeal while the antagonistic housing appeal is evaluated slightly but is not significantly better. Thus, all four appeals that present the urban-rural divide as an antagonistic, group-based conflict are negatively evaluated by urban respondents – no matter whom politicians seek to benefit in this conflict and which policy they propose to adopt.

Yet our findings show that urban voters react positively to appeals that use a more harmonious language. As soon as the same policies are presented as benefitting the whole country, respondents evaluate them positively. This is most clearly the case for housing. The promise to tackle an issue that is of primary concern in urban areas receives a very positive reaction among urban voters when it is presented as helping people ‘across the country’. However, even the somewhat less specific innovation appeal is evaluated positively when presented in a harmonious framework.

To assess how urban respondents evaluate the candidate in more depth, we draw on post-treatment questions about the candidate's credibility, competence, and local understanding. We find that all these more specific candidate evaluations exhibit similar patterns and align with the overall assessment presented in the main analysis (see Appendix Figure F5). Urbanites find a candidate making harmonious appeals more credible, while antagonistic appeals lead to lower ratings in terms of credibility, competence, and the candidate's understanding of local issues.

This is in line with our argument on the asymmetric reaction of dominant and subordinate groups to group-based appeals. While subordinate groups support the antagonistic framing of appeals as they reference a legitimate fight against social inequalities, dominant groups prefer de-emphasizing inequalities and conflicts. This finding is in line with insights from social psychology that demonstrate the strong role of guilt and fairness concerns in understanding privileged individuals' reactions to inequality (Chudy, Piston, and Shipper 2019; Powell, Branscombe, and Schmitt 2005), as well as with Dahrendorfian conflict sociology. Our findings indicate that dominant groups such as urban dwellers actively punish politicians for polarizing societal conflict. This shows that individuals do not just adjust their own behaviour in reaction to dominance, as the psychological literature indicates, but that dominant groups also adjust their expectations towards the behaviour of political elites accordingly.

In this survey, we measured both identity and resentment pre-treatment. This allows us to observe to what extent a combination of identity and resentment affects the results – although the small number of respondents in this group leads to imprecise estimates. We therefore pool the innovation and rent treatment.¹² As expected, those respondents who combine urban identity and urban resentment are indeed considerably more positive towards the antagonistic appeals. That said, they still do not rate the antagonistic appeals better than the control appeal. Hence, the antagonistic appeal does not even work in the most likely condition (Fig. 5).

Figure 5. Identity and resentment among urbanites, Study 2.

Alternative Explanations

An alternative interpretation of these results is that they are less about the implicit degree of conflict but rather about the implicit justifications provided in the appeals. After all, an important difference between antagonistic and harmonious appeals is that antagonistic appeals provide ‘pluralistic’ justifications, aimed at advancing the interests of a group, while harmonious appeals provide ‘republican’ justifications, appealing to the public good (Rehfeld 2009; Wolkenstein and Wratil 2021).

Urban citizens might have a specific self-conception of championing universal values and policies that are advancing the greater good of the entire society, not just their own places. Hence, even if these respondents would not reject the ‘us-them’ setting of the antagonistic appeals in general, they might reject these appeals because of their group-centred justifications.

To test this interpretation, we asked respondents about their representation ideals; that is, whether they preferred politicians to focus their work on improving the situation of their constituents or on the public good. While there was a quite substantial variation in respondents' representation ideals, they contributed relatively little to the explanation of candidate evaluations. As Fig. 6 shows, urban respondents with a universalist representation ideal always evaluated the candidate slightly but insignificantly more positively. However, this is also the case for antagonistic appeals, which universalists should dislike more strongly than particularists (full regression in Appendix G).

Figure 6. Effect of treatment on candidate evaluation by representation ideal, Study 2.

Our results could also be driven by urbanites who perceive the out-group (rural people) as more deserving than the in-group.¹³ Negative reactions to the antagonistic appeals would then be due to out-group concern and not hinge on the framing of the societal divide as either antagonistic or harmonious.

Such an explanation is already put into question by the fact that urbanites actively dislike the rural appeal as well. To further probe this explanation, we distinguish the reactions of urbanites who rate rural dwellers as highly deserving from those who attribute less deservingness to the out-group.¹⁴ Figure 7 shows that both urbanites who perceive the out-group as deserving and those who are less out-group oriented react similarly negatively to the antagonistic frame. Moreover, the reactions of both respondent groups are similar for the harmonious appeals, to which both react more positively (see the full regression in Appendix G).

Figure 7. Effect of treatment on candidate evaluation by out-group deservingness, Study 2.

Moreover, the negative reaction to antagonistic appeals by urbanites could be driven by those residing in objectively well-off places, where the proposed policies are less relevant. We thus measure the objective economic conditions in the cities of our urban respondents and show the different reactions of those who live in places with low/high welfare quotas.¹⁵ As with the other subgroup analyses, we find relatively little variation in terms of economic conditions or hardship in the respective place of residence (see the full regression in Appendix G) (Fig. 8 ).

Figure 8. Effect of treatment on candidate evaluation by objective economic conditions, Study 2.

Moreover, in Appendix G we distinguish urbanites according to their length of residence in the city in which they live. Overall, the results show that there is almost no difference between urbanites who have lived in the respective city their entire life and those who have lived there less than five years.

Last, we analyze whether the effect of our group appeals is driven by respondents' assumptions about the candidates' partisanship. Whereas we did not present a party affiliation of our fictitious candidate, we asked respondents to indicate how likely they thought that the candidate was affiliated with each of the parties represented in the German parliament. Figure 9 shows the average association of each appeal (and appealing candidate) with one of the six parties (the scale ranged from ‘not at all likely’ [1] to ‘very likely’ [4]).

Figure 9. The average likelihood of candidate party affiliation by condition, Study 2

The innovation appeal is more closely associated with the policy programme of centre-right parties (CDU and FDP), while the rent appeal is perceived to come from a left party (SPD or LINKE). Interestingly, there is very little variation for the Greens, and the AfD is most strongly associated with antagonistic appeals. Since urban respondents dislike the rent and the innovation appeals to a similar extent, it does not seem to play a major role whether the appeal is more strongly perceived to come from the right or the left.

Conclusion

In this paper, we study the conditions under which the mobilization of social groups by politicians appealing to them works or backfires. Using the case of place-based appeals across the urban-rural divide, we argue that the success of group-based appeals hinges on the targeted groups' relative position in a social divide. While subordinate social groups buy into antagonistic appeals that mobilize against the dominant group, dominant groups react negatively to attempts to appeal to them by polarizing existing inequalities.

We test this argument with two survey experiments on how urban and rural voters react to political appeals that target people`s subjective places of residence. Our findings show that it is indeed possible to garner support from rural voters with place-based appeals; candidates are evaluated significantly more positively when they address rural voters with such appeals, especially when these appeals focus on cultural issues. By contrast, urban voters actively oppose appeals that pitch urban and rural places against each other.

In a follow-up survey of urban voters, we find that the antagonistic framing of place-based appeals produces a strong negative reaction among urban residents. However, more harmonious appeals to urban interests that emphasize the common societal good are successful and substantially increase candidate approval. This demonstrates that the success of conflictual appeals hinges on the type of social group to which they are directed.

Compared to the existing survey-based literature on group-based appeals, we thus find an important new pattern. While it is not uncommon to find that group-based appeals may have null effects among certain groups of voters, it is rare that groups react negatively to such appeals (the only example is Robison et al. [2021], who report a negative reaction among upper-middle-class voters in their Appendix). Introducing new evidence on the urban-rural divide, we uncover a broader puzzle around the negative reaction of advantaged groups to group-based political appeals. Our findings show that urban respondents react decisively negatively to appeals that cater to them as a group, be they economic, cultural, or symbolic appeals. Importantly, these voters also react negatively to group-based appeals that emphasize the other side of the divide, that is, rural voters. Thus, urban voters reject any appeal that presents an urban-rural divide as a group-based conflict. At the same time, these voters are not critical of the policies proposed in these appeals – they react positively to the same policy proposals when they are presented in a harmonious framework.

A few tentative conclusions can be drawn from our study with respect to the conditions under which group-based appeals might work or backfire. While our study is empirically restricted to the case of urban-rural divides, the findings may hold broader importance for the analysis of group-based appeals. After all, the fundamental setup of this divide is quite common – it is very clear which of the two groups is seen as being on the winning side of the conflict and which of the two groups is perceived as being on the losing side. This also holds for other group-based divides such as class and education, which are often studied in the literature on group appeals. Hence, we may speculate that group-based appeals would also have an asymmetric effect on those divides (as suggested by Robison et al.'s [2021] finding). Indeed, it may be common that the dominant group actively rejects a framing that presents them as a group that is in conflict with another group. However, the situation is less clear for divides based on race or gender, where the historically dominant social groups may perceive their position as threatened, such as whites in the United States. Here, antagonistic appeals may potentially be able to tap into some group members’ perception of status threat. At the same time, they may still trigger negative emotions of self-directed anger and guilt among other group members, making the total effect indeterminate.

Potentially, another scope condition could concern the reaction of disadvantaged groups to harmonious appeals. These groups could oppose a harmonious framing of a divide that is perceived to exist at their cost. Since our analysis focused on dominant groups, we have no evidence of this. However, because our results show similarly positive responses among rural voters across a range of substantively different appeals (see also the findings of Dassonneville, Stubager, and Thau 2022; Robison et al. 2021), we suspect that disadvantaged groups are likely to approve of a wide range of appeals (including harmonious ones). Overall, this would suggest that harmonious appeals do not face similar scope conditions as antagonistic appeals.

For future research on the scope conditions under which group-based appeals work, we thus think that it would be most pertinent to focus on antagonistic appeals and to further disentangle the specific contribution of status level and status threat to voters' perceptions of group dominance and their reactions to group-based appeals. Do they have an additive or an interactive impact on the reaction to such appeals? Moreover, such research should also investigate the formation of status and conflict perceptions across major societal divides and should be open to the possibility of country-specific differences. It might very well be possible that perceptions of group dominance differ systematically among group members; for example, between different regions in bigger countries. This promises to generate a more general model for the effectiveness of political appeals based on perceived group status.