With status decline in sight, voters turn radical right: how do experience and expectation of status decline shape electoral behaviour?

Abstract We distinguish between the experience and expectation of subjective status decline in relation to electoral behaviour. Studies often link support for radical parties, especially radical right ones, to voters’ experience of status decline. A few other studies argue that voters’ expectation of status decline also triggers radical right support. Without precise measures of both perceptions, it has been difficult to distinguish which (or both) is most relevant for radical right support in Western Europe and the USA. Using survey data from 2018 (n = 4,076) and 2020 (n = 2,106) in Finland, we could precisely measure and distinguish between voters’ experience and expectation of status decline. Descriptively, voters who have experienced status decline have low income, whereas voters who expect status decline have (lower)middle income. Using multivariate analyses, we find that voters who expect status decline consistently prefer radical right parties more than voters who expect status improvement. However, there is no robust evidence of radical right support among voters who have experienced status decline. These findings suggest that the expectation, not experience, of status decline drives radical right support. If these expectations trigger radical right support in Nordic welfare states, they may be even more pertinent in less comprehensive welfare states.


Introduction
Political science research has historically paid much attention to economic voting, that is, voting based on voters' economic interests. Recently, it has begun focusing on the electoral consequences of subjective social status to understand the growth and persistence of radical right support in Western Europe (e.g., Gidron and Hall, 2019). Whereas some trace radical right support to economic disadvantage (Dehdari, 2021;Rodrik, 2021), a growing number link it to declining social status (e.g., Burgoon et al., 2019;Gidron and Hall, 2019;Im et al., 2019;Kurer, 2020). Specifically, recent economic and social transformations have dislodged voters' perceptions about their position in society. While some voters benefit from these transformations, others do not. Studies thus suggest that past experiences matter: those who feel that their social status has declined seek out parties that claim to restore their social status. In Western Europe today, radical right parties latch onto these voters' status grievances by offering nostalgic and populist appeals that seek to roll back these social and economic transformations (Gest et al., 2018). A recent analysis also suggests that some voters who experienced a status declinenamely, a feeling that their expected status was not realisedmay also favour radical left parties (Kurer and van Staalduinen, 2022).
Although these studies stress the experience of status decline, we contend that voters' future expectations are equally crucial for electoral behaviour. Specifically, we posit that voters who expect their status to decline also turn to radical right parties to prevent (rather than restore) their status decline. To date, fewer studies have focused on how expectations of status decline affect radical right support as compared to the burgeoning number of studies that examine how experiences of status decline affect it (e.g., Engler and Weisstanner, 2020;Hochschild, 2016;Mayer, 2015). In short, we contribute to this theoretical gap by distinguishing between experience and expectation of a status decline and then comparing how they influence electoral behaviour.
We conducted our analyses using two nationally representative survey data sets (n = 4,076; 2,106) collected in the Finnish multi-party context in 2018 and 2020. Like other advanced economies, Finland has undergone substantial societal and economic transformations that have shaken voters' perceptions of their societal position. Support for the radical right Finns Party has been significant and persistent since 2011. Finland, therefore, resembles other Western European countries, where radical right parties are an entrenched feature of the political landscape. Consequently, we consider our Finnish results as being relevant to other Western European countries. Further, we suggest that to understand political flux, it is just as important to identify those who feel that they are losers of past socioeconomic transformations, as it is essential to pay attention to those who fear worsening social status.
In the following section, we briefly review the growing literature on how experiences of status decline influence electoral behaviour. Next, we distinguish between experience and expectation of status decline and set out how they may influence electoral behaviour in subtly different ways. After introducing the data and methods, we present our results. We conclude by discussing the implications of our findings.
The decline in subjective social status and support for radical parties Leveraging psychology and sociological literature, subjective social status refers to voters' perceptions of their status vis-à-vis others living in the same society (Jackman and Jackman, 1973). It refers to how people 'rank their level of social esteem or respect that society is according to them' (Bolet, 2022, p. 2). Subjective social status is thus relational because voters form views about themselves based on a referent group within society (Pettigrew, 2008;Runciman, 1966;Smith et al., 2011). As such, changes in the socioeconomic status or life circumstances of individuals or their referent groups may alter the way individuals perceive their position and value within society. In terms of electoral behaviour, subjective social status (hereafter social status) can be as politically relevant as other factors, such as economic interests (e.g., income and unemployment), because individuals often care about their social position as much as their material position (Kuziemko et al., 2014;Ridgeway, 2013). In short, voters are reluctant to lose their status in society, which triggers political responses.
A burgeoning literature in political science shows that declines in social status (hereafter status decline) trigger support for radical right as well as radical left parties in Western Europe and the USA (e.g., Ballard-Rosa et al., 2022;Bolet, 2022;Burgoon et al., 2019;Gidron and Hall 2019;Hochschild, 2016;Kurer and van Staalduinen, 2022). Economic transformations like workplace automation, technological change, and job offshoring (Anelli et al., 2021;Im et al., 2019;Kurer, 2020) as well as social transformations like immigration, massification of higher education, and changing social roles (Ansell and Gingrich, 2018;Gest et al., 2018;Hochschild, 2016;Steenvoorden and Harteveld, 2018) alter the relative status of different groups within society. These studies find that voters who feel their social status has declinedcompared to their earlier status or their parents' social statusoften support radical right parties to address their status decline. Radical right parties attract these social groups by appealing to nostalgia, offering to roll back the transformations that have brought about their status decline or assuage their diminished status by marginalising social groups that have seen their social position improve like immigrants (Hochschild, 2016;Lamont et al., 2017). In contrast, social groups who feel that their social status has improved are unlikely to find appeals by radical right parties attractive. Thus, they do not favour these parties and have been shown to turn to centre-right parties instead Kurer, 2020;Kurer and van Staalduinen, 2022).
More recently, Kurer and van Staalduinen (2022) clarify that voters in Germany who have suffered status declinenamely that their realized social status fell short of their parents' (thus their expected) social statusfavour either radical right or radical left parties. However, whether they favour radical right or radical left parties depends on their family and their own social backgrounds. Specifically, higher educated voters who feel unable to keep up with the status of their higher educated parents favour radical left parties. These parties blame the current socioeconomic model for much of the social and economic malaise observed in advanced economies today and view it as a system that benefits the elites rather than the masses (March, 2011;Rooduijn et al., 2017). They therefore call for the radical transformation and wholesale overhaul of this current socioeconomic model rather than refining or improving it. By contrast, lower educated workers from middle-or working-class families who cannot keep up with the status of their parents favour radical right parties. With this clarification, the authors demonstrate that feelings of status decline are not limited to social groups who have lost out from recent socioeconomic transformations like routine workers. Instead, they suggest that such feelings of status decline may be far more pervasive and prevalent, even among socially advantaged groups (see Häusermann et al., 2021).

Experience of status decline and expectation of status decline
Although this burgeoning scholarship demonstrates the political implications of voters' subjective social status, there is still ambiguity as to whether the relationship between status decline and support for radical right and radical left parties is led by the experience of status decline, or expectation of status decline or both. Within this scholarship, most studies argue that experience of status decline influences support for radical parties (e.g., Anelli et al., 2021;Ballard-Rosa et al., 2022;Bolet, 2022;Burgoon et al., 2019;Gest et al., 2018;Kurer and van Staalduinen 2022). A smaller number of studies raise an equally pertinent point: expectations of status decline also trigger support for radical right parties (Engler and Weisstanner, 2020;Häusermann et al., 2021;Im et al., 2019;Iversen and Soskice, 2020;Iversen, 2021;Mayer, 2015).
Yet, few studies compare the experience of status decline against the expectation of status decline within this burgeoning scholarship (for an exception, see Engler and Weisstanner, 2020;Rebechi and Rohde, 2022). One of the main reasons for the lack of this comparison is the paucity of direct measures of both types of status decline which, in turn, increases a reliance on imperfect proxies. For instance, the experience of status decline is often measured in terms of changes in the objective circumstances of particular social groupsexposure to industrial robots (Anelli et al., 2021), differences in growth rates across income deciles (Burgoon et al., 2019), different occupational trajectories among routine workers (Kurer, 2020), the difference in voters' actual and predicted status based on their parents and their own socioeconomic background (Kurer and van Staalduinen, 2022), and the difference in occupational vulnerability to workplace automation between men and women (Gingrich and Kuo, 2022). While a handful of studies use subjective proxies to capture the experience of status decline, such as societal pessimism (Steenvoorden and Harteveld, 2018) and perceived current social status (Gidron and Hall, 2019), they remain indirect measures. Notably, Gest and colleagues (2018, p. 9) came closest to a direct subjective measure when they asked respondents to rate how central and important people like them were in society thirty years ago.
On the side of expectation of a status decline, similar issues arise among studies relying on survey analyses instead of ethnographic methods (for ethnographic methods, see Mayer, 2015). For instance, Engler and Weisstanner (2020) projected expectations of status decline based on the interaction of objective measures of long-run income inequality trends and respondents' current subjective social status, while Mutz (2018, p. E4336) identified social groups that were expected to suffer status threat. Likewise, Iversen (2021) reflected on how housing price inequality may reduce intergenerational social mobility, which raises the prospect of a status decline and its consequent impact on radical right support. A few other studies rely on subjective indicators, including respondents' perceptions of their future income insecurity (Im et al., 2019), respondents' perceptions of their employment and social opportunities (i.e., safe, and fulfilled life over the life course) for themselves and their children . However, these are still indirect proxies of the expectation of a status decline. The study that comes closest to measuring subjective perceptions of both experiences and expectations of status decline was conducted by Rebechi and Rohde (2022). As their research focuses on the electoral impact of economic insecurity, the authors used two variables based on respondents' assessment of how their financial situation had changed from the past and how their lives might change in the future. Thus, the indicators used in Rebechi and Rohde's study still do not fully or precisely capture respondents' perceptions of their subjective social status. Therefore, to our knowledge, no study within this burgeoning literature directly measures voters' subjective perceptions of both experiences and expectations of status decline.
Electoral responses to experience and expectation of status decline Consequently, we know little about the similarities and differences in the electoral responses of voters who experienced status decline and voters who expect status decline. Both perspectives relate to how voters respond to their views about themselves in relation to others. However, the mechanisms by which they influence electoral behaviour may differ subtly. When voters experienced status decline, their electoral responses may be triggered by resentment and grievance over their diminished social status. When voters expect a status decline, their electoral responses may be triggered by a fear of falling down the social ladder. Correspondingly, these two groups of voters may have somewhat different demands. Voters who have experienced status decline may seek to restore their lost status. Voters who expect a status decline may instead seek to prevent a loss of status. In turn, voters who have experienced and voters who expect status decline are perhaps drawn to radical right and radical left parties to varying degrees.
On the one hand, voters who have experienced status decline may favour both radical right and radical left parties, as both parties offer solutions to restore their lost status but in different ways. Radical right parties propose to return to the past and marginalise certain social groups, especially minorities (e.g., Gest et al., 2018;Gidron and Hall, 2019), whereas radical left parties seek to usher in a new socioeconomic order, such as post-growth economic models (Reitz and Jörke, 2021; see also Rooduijn et al., 2017). In this regard, if voters who have experienced status decline consider going back to the past and marginalizing certain social groups (see Hall and Lamont, 2013;Lamont, 2018) or going forward with a radically different system as viable ways to restore their lost status, they would vote for radical right or radical left parties, respectively. Kurer and van Staalduinen (2022) support this when they find that German voters who have experienced a status decline (with respect to their expected status) favour radical right or radical left parties based on their own and their parents' sociodemographic background.
On the other hand, we propose that voters who expect status decline may favour radical right parties more than radical left parties. If these voters fear falling down the social ladder and seek to prevent status decline, they may be less inclined towards solutions that are framed as radical as it conveys uncertainty. Instead, they may prefer solutions that they feel familiar with such as solutions which rely on nostalgic frames, even if these solutions are radical in practice. Related literature on risk aversion shows that individuals who are exposed to both economic and noneconomic risks are unwilling to make risky decisions (Courbage et al., 2018;Guiso and Paiella, 2008). Furthermore, Fernandez and Rodrik (1991) demonstrated that individuals favour the status quo over reforms when the distributive outcomes from these reforms are uncertain, even if these individuals are risk-neutral, forward-looking, and rational. Thus, these voters who are under status threat may turn away from solutions which they feel less certain about in favour of solutions which they feel more familiar with.
To be clear, the solutions proposed by radical right and radical left parties are both radical and potentially transformative when compared with the proposals of mainstream parties. Research from party politics has consistently demonstrated a general convergence of positions on economic and sociocultural issues among mainstream parties (e.g., Kurella and Rosset, 2017;Lefkofrifdi et al., 2014). By contrast, the radical right often proposes to impose strict restrictions on immigration, exit the European Union, enforce strict trade tariffs and protectionist measures, and reform the principles that underpin the welfare state by restricting social policy access only to 'deserving' citizens. Likewise, the radical left often proposes to depart from current capitalist economic orthodoxy and pursue new economic models such as de-growth, reduce European economic integration, and impose far-reaching redistribution funded by higher taxes (March, 2011;Rooduijn et al., 2017;Rooduijn and Akkerman, 2015). However, radical right and radical left parties differ in how they frame their solutions. When compared with the frames of the radical left (Agustín and Briziarelli, 2017;Fassin, 2017;Marlière, 2019), the frames of radical right parties rely much more on nostalgia (Gest et al., 2018;Fenger, 2018;Schreurs, 2021). Betz and Johnson (2004, p. 324) described radical right parties' frames as backward-looking, reactionary and reflects 'a deep sense of nostalgia for the good old days', and Taggart (2004, p. 274) characterized them as relying on 'an ideal world : : : constructed retrospectively from the past'. Nostalgia enables radical right parties to frame their solution in terms of the 'pleasure of familiarity, the comfort of tradition and feelings of reassurance' (Kenny, 2017, p. 263) which evokes familiarity rather than uncertainty. In short, radical right parties use frames that convey familiarity more than radical left parties do. If voters who expect status decline are more risk-averse, they may find the frames of radical left parties to be less appealing than the ones of radical right parties. To summarize, voters who have experienced status decline and voters who expect status decline may both support radical right parties, but they may differ in their support for radical left parties.
Lastly, voters who have experienced status improvement or expect status improvement are unlikely to worry about their status. Therefore, they are unlikely to seek to restore or prevent their status decline. Recent research finds that voters who have experienced status improvement (Kurer, 2020;Kurer and van Staaalduinen, 2022) or voters who expect status improvement  are unlikely to turn to radical right or radical left parties. Instead, they prefer centreright parties.
Our expectations are summarized below. However, it is pertinent to point out that Finland lacks an electorally successful radical left party. The previously hard-left, the Left Alliance is now, by most accords, a new-left party (Dunphy, 2007). Therefore, we cannot test the relationship between experience and expectations of status decline and support for radical left parties.
• Hypothesis 1: Voters who have experienced status decline support radical right parties more than voters who have experienced status improvement. • Hypothesis 2: Voters who have experienced status decline support centre right parties less than voters who have experienced status improvement. • Hypothesis 3: Voters who expect status decline support radical right parties more than voters who expect status improvement. • Hypothesis 4: Voters who expect status decline support centre right parties less than voters who expect status improvement.

The Finnish case
Like other Western European countries, Finland has been substantially affected by economic and social transformations that have weakened previous linkages between voters and parties. For the latter half of the 20th century, large class-based parties like the centre-left workers' party (the Social Democratic Party), the centrist agrarian party (the Centre Party), and the centre-right bourgeois party (the National Coalition Party) dominated the Finnish political landscape. Recently, however, the landscape has fragmented, and electoral outcomes have become less predictable. There seems to be a shift in support from these parties towards smaller parties like the new-left Green League, the previously hard-left but now new-left Left Alliance, and the radical right Finns Party. By the new-left, we mean parties that adopt liberal positions on sociocultural issues (Busch, 2016;Oesch, 2012), support social investment, and back climate-friendly policies (Abou-Chadi et al., 2021). Notably, Finland does not have an electorally successful radical left party. 1 There are several plausible reasons for the rise and persistence of the Finns Party. First, the Finnish electoral system is based on proportional representation and mandatory preferential voting which may lower the initial barrier of entry for challenger parties. Second, because of the electoral system, voters may be more willing to consider smaller parties as viable parties to vote for. In the initial years prior to 2011, the Finns Party may have benefitted from these direct and indirect effects of the Finnish electoral system. Third, it participated in the previous government from 2015 to 2017 until they split into two separate parties. The new Blue Reform party continued in the government, while the rest of the party kept its original name and went into the opposition. This stint in office may have further enhanced the credibility of the Finns Party as a governing party and thus raised voters' willingness to vote for it.
The rise and persistence of the Finns Party thus bring Finland in line with other Western European countries, where radical right parties are a mainstay in their electoral landscapes (Oesch and Rennwald, 2018;Rooduijn et al., 2017). Yet Finland and its Nordic counterparts differ from other Western European countries regarding their comprehensive Nordic universal welfare state (Esping-Andersen, 1990). However, the strength and persistence of radical right parties in Denmark, Norway, Sweden, and Finland suggest that universal welfare may not insulate against radical right support (see Gingrich, 2019, contrast Vlandas andHalikiopoulou, 2019). Overall, we consider the results from Finland relevant for other Western European countries because it has undergone similar societal and economic transformations and has a radical right party that is regarded as a credible and popular party within the electoral landscape.

Data
Our study is based on two nationally representative cross-sectional surveys conducted in 2018 and 2020. 2 Thus, the two surveys allowed us to compare results before and during the COVID-19 pandemic. For the first survey, data were collected in three stages. The first stage used an online survey with 20,000 randomly sampled from a national population register, with a response rate of 12.1%. As the higher educated and pensioners were overrepresented among these respondents compared to the national population, the second stage targeted 13,197 persons who belonged to underrepresented sociodemographic groups from the initial 20,000 people for in-person or phone interviews (response rate 7.6%). By the second stage, the sample was still lower than the targeted size of 4,000 respondents, and lower educated working-age people remained underrepresented. Therefore, we conducted a third step focusing on lower educated, working-aged 1 See Table A1 and Figure A1 in the supplementary material. 2 The questionnaire and description of data collection (in Finnish) can be found at https://bibu.fi/wp/wp-content/uploads/ 2020/12/BIBU-kansalaissurveyn-kysymyslomake.pdf. The data set is available for replication purposes upon request from authors.
people. Altogether, 4,214 participants were contacted from an internet panel (n = 40,000) maintained by the research company in charge of the data collection (response rate 16.2%). The final data set comprised 4,076 respondents. Finally, we checked our sample against the national population for representativeness and generated weights that adjusted our sample for gender, age, and education. After dropping observations with missing values on any of the variables used for the analyses, the final study sample comprised 2,680 respondents. 3 The second survey was collected from August 31 to early September 2020. It included 2,106 respondents drawn from a large, ongoing online panel administrated by Taloustutkimus. The respondents were asked to share their views on a wide range of policy measures to boost the economy and employment, organise social security, and mitigate climate change amid the COVID-19 crisis. To ensure the sample's representativeness against the national population, Taloustutkimus generated weights that adjusted for gender, age, and place of residence. After dropping observations with missing values on any of the variables used for the analyses, the final sample comprised 1,190 observations. 4

Variables
In both data sets, our dependent variable measures respondents' preferred party choice if national parliamentary elections were held now. We focused on the six parties that received the largest vote shares in the last parliamentary elections (2017) preceding our data collection: the Left Alliance, the Green League, the Social Democratic Party, the Centre Party, the National Coalition Party, and the Finns Party. Our dependent variable was therefore categorical, indicating whether respondents would vote for one of these parties. Unfortunately, the survey did not include an option asking if respondents would not vote if elections were held now. As such, vote abstention could not be included in the dependent variable.
Our explanatory variables in both data sets were designed to capture respondents' subjective social status and distinguish between the experience of status decline and expectation of status decline. Using a five-point Likert scale, respondents were asked to rate how their status (1) had changed during the past five years and (2) would change during the upcoming five years. The possible responses ranged from 'improved a lot' to 'worsened a lot', and 'will improve a lot' to 'will worsen a lot' respectively. However, the number of observations with responses at the tail ends of the scale was small; 2.24% (2018) and 4.37% (2020) indicated that their status had improved a lot, and 4.37% (2018) and 5.38% (2020) said their status had worsened. Likewise, 2.28% (2018) and 4.29% (2020) of respondents said that their status would improve a lot, and 2.72% (2018) and 3.19% (2020) said that their status would worsen a lot. 3 Most observations with missing values come from our dependent variable, respondents' preferred party choice (911 observations), and household income (436 observations). A number of respondents opted for parties that are not part of the six major parties we are interested in or did not give a response. Likewise, a number of respondents did not indicate their income. The spearman correlation between missing observations on these two variables is very weak (0.087). For robustness, we reestimated Model 3 for the 2018 dataset (see Figure 7 in this text and Table A6 in the supplementary material) using an alternate sample without dropping respondents with missing values on this income variable and without controlling for income. We summarise the results in Figure A2 in the supplementary material. The main findings from our original sample, which we present later below, are insensitive to these two different samples and model specifications. 4 Like in the 2018 data set, most observations with missing values come from our dependent variable, respondents' preferred party choice (551 observations), and household income (574 observations). A number of respondents opted for parties that are not part of the six major parties we are interested in or did not give a response. Likewise, a number of respondents did not indicate their income. The spearman correlation between missing observations on these two variables is higher than in the 2018 data set, but still very weak (0.152). For robustness, we re-estimated the Model 5 for the 2020 data set (see Figure 9 in this text and Table A8 in the supplementary material) using an alternate sample without dropping respondents with missing values on this income variable and without controlling for income. We summarize the results in Figure A3 in the supplementary material. The main findings from our original sample, which we present later below, are insensitive to these two different samples and model specifications.
Based on these small case numbers, we collapsed the five-point scale to a three-point scale. Thus, respondents who indicated that their status improved a lot or worsened a lot were combined with respondents who said that their status improved or worsened, respectively. Similarly, respondents who indicated that their status would improve a lot or would worsen a lot were combined with respondents who said that their status would improve or would worsen, respectively. We treated both explanatory variables as categorical to account for potential nonlinear associations between experience and expectations of status decline and party choice. In short, the experience of status decline was coded as follows: status (1) improved, (2) stayed the same, or (3) declined. Likewise, the expectation of status decline was coded as follows: status (1) will improve, (2) will stay the same, or (3) will decline. In sum, our operationalization allowed us to precisely identify and distinguish voters' experiences and expectations of status decline, which contrasts with the analytical approach taken in most of the burgeoning scholarship.
We also controlled for a battery of sociodemographic variables, including age, education, gender, marital status, current labour market status, gross monthly household income, native language (as a proxy for respondents' ethnic background), and household economic condition when they were growing up (see Bolet, 2022;Kurer and van Staalduinen, 2022). We also controlled for respondents' self-reported left-right and liberal-conservative ideologies. Additionally, we controlled for respondents' issue positions on redistribution and immigration as proxies for their issue positions on economic and cultural issues, respectively (e.g., Kurella and Rosset, 2017;Lefkofridi et al., 2014).

Modelling strategy
We conducted our analyses twice: once using the 2018 survey and another using the 2020 survey. In both analyses, we employed multinomial logit models with region-clustered standard errors. We applied region fixed effects and weights that adjusted for age, gender, and education. In our first set of analyses conducted with the 2018 survey data set, we first regressed respondents' party choice on respondents' experience and expectation of status decline and the battery of sociodemographic controls. In the second set, we added respondents' left-right economic ideology and liberal-conservative ideology. In the third set, we added respondents' issue positions on redistribution and immigration.
In our second set of analyses, we used the 2020 survey data set; we first regressed respondents' party choices on respondents' experience and expectation of status decline and the battery of sociodemographic controls. However, we did not control for respondents' economic situation when growing up or their mother tongue because these variables were absent in this data set. Second, we added respondents' left-right economic ideology and liberal-conservative ideology. Unlike the first set of analyses, we did not perform an additional step controlling for respondents' issue positions on redistribution and immigration because these variables were absent from the data set.

Descriptive results
We begin by describing the sociodemographic background of respondents relating to their experience and expectation of status decline based on the 2018 (Figures 1 and 2) and 2020 surveys (Figures 3 and 4). The figures present the distribution of respondents regarding their experience or expectation of status decline based on five sociodemographic variables. For each specified category (response level) of each sociodemographic variable, we calculated the distribution of respondents in terms of their experience or expectation of status decline. Therefore, any overlap between categories of each sociodemographic variable represents the difference in percentages between these categories. For example, the yellow shaded area of the right-most bar (label:

124
Zhen Jie Im et al.  'declined') in Panel A of Figure 1 shows the difference in the percentage of respondents with comprehensive education and university education who have experienced status decline. In Figure 1, Panel A shows that the experience of status decline is less prevalent among higher educated respondents. It is most prevalent among respondents with comprehensive school education and vocational education, whereas the experience of status improvement is most prevalent among respondents with university education. Panel B shows that the experience of status decline does not vary substantially between men and women. Panel C shows that the experience of status decline is most prevalent among older age groups, specifically among respondents who are close to the retirement age (60-65 years old) and among respondents who are older than the retirement age. Panel D shows that the experience of status decline is most prevalent among respondents residing in northern and eastern parts of Finland, where economic activity lagged behind other regions of Finland. Furthermore, Panel E illustrates that the experience of status decline is most prevalent among respondents who are in the lowest income quartile. Figure 2 shows the distribution of expectations of status decline based on the same sociodemographic variables as Figure 1. Panel A shows that expectation of status decline is most prevalent among respondents with medium levels of educationi.e., respondents with vocational education. Panel B shows a little variation in the expectation of status decline between men and women. Panel C illustrates that the expectation of status decline is most prevalent among respondents near retirement age (60-65 years old). Panel D shows that the expectation of status decline is most prevalent among respondents residing in the northern and eastern parts of Finland. Furthermore, Panel E shows that the expectation of status decline is most prevalent among respondents in the third income quartile (3,000-4,999 euros).
Overall, Figures 1 and 2 provide a descriptive overview of the similarities and differences in the sociodemographic backgrounds of respondents who felt that they had experienced status decline and respondents who expect status decline. Regarding similarities, both experiences and expectations of status decline are more prevalent among older respondents without higher education and who reside in less economically prosperous areas in northern and eastern parts of Finland. Additionally, both experiences and expectations of status decline do not vary substantially between men and women. However, some crucial differences exist between respondents who have experienced status decline and respondents who expect status decline. The experience of status decline is most prevalent among respondents who have the lowest education levels (comprehensive education), are oldest (above retirement age of 65 years old), and whose household income is lowest (first income quartile). By contrast, expectations of status decline are most prevalent among respondents who are medium-educated (or medium-skilled), of the age group that is just before retirement (60-65 years old), and who have reasonably comfortable household income (third income quartile). If education and income reflect objective socioeconomic status, then it seems (descriptively at least) that respondents who have experienced status decline have low socioeconomic status, whereas respondents who expect status decline have (lower) middle socioeconomic status.
For comparison, Figures 3 and 4 provide similar descriptive overviews using the 2020 data set. Unlike respondents in 2018, both experience and expectation of status decline are most prevalent among lower educated respondents (comprehensive school). However, there are some differences regarding income and age. Respondents who have experienced status decline are often in the lowest income quartile, whereas respondents who expect status decline are in the second income quartile. There are also more respondents between the ages of 60 and 65 years old who expect status decline than those who have experienced status decline.
When comparing these descriptive findings from 2018 and 2020, we can observe broad similarities in the social background of respondents who have experienced status decline and respondents who expect status decline. Respondents who have experienced status decline are often in lowincome households, whereas respondents who expect status decline are often in middle (2018) or lower middle-income (2020) households. If income reflects (at least partly) citizens' objective social status, then it seems (descriptively at least) that respondents who have experienced status decline have low socioeconomic status, whereas respondents who expect status decline have (lower) middle socioeconomic status. Crucially, these similarities persist despite contextual exceptionalities from the COVID-19 pandemic in 2020. In short, it appears that respondents who have experienced status decline are somewhat sociodemographically different from respondents who expect status decline.

Multivariate analyses results
Next, we focus on the results of our multivariate analyses of the associations between experience and expectation of status decline and party choice. We first present results from analyses conducted on data from our 2018 survey, using average marginal effects based on our multinomial logit regression estimates. The average marginal effects reflect differences in group means with respect to 'status has improved' and 'status will improve', respectively. Figure 5 presents the results of our first regression model, which controls for sociodemographic variables. It shows that the experience of status decline is significantly associated with support for the new left Left Alliance and opposition to the bourgeois centre-right National Coalition Party. However, it illustrates that expectation of status decline, not experience of status decline, is significantly associated with support for the radical right Finns Party. Respondents who reported that their status had declined support the Left Alliance significantly more than respondents who reported that their status had improved (P < 0.05). Next, respondents whose status had stayed the same or had declined support the National Coalition Party less than respondents whose status had improved (P < 0.005; P < 0.01, respectively). Further, respondents who responded that their Notes: The regression model includes sociodemographic controls (see Table A4 in the supplementary material). Whiskers represent 95% confidence intervals. When they intersect the red dotted line, the difference in group means is not statistically significant (P < 0.05).
status will decline are significantly more likely to support the Finns Party than respondents who responded that their status will improve (P < 0.005). Figure 6 shows the results when both sociodemographic variables and respondents' selfreported left-right and liberal-conservative political ideologies were controlled for. When compared with Figure 5, Figure 6 does not show any difference in results. The sole difference relates to respondents who expect their status to stay the same. They are now significantly more likely to vote for the Finns Party than respondents who expect their status to improve (P < 0.05).
Next, Figure 7 presents the results when sociodemographic controls, left-right and liberalconservative political ideologies, and issue positions on redistribution and immigration were controlled for. The results are broadly similar to the results presented in Figures 5 and 6. Respondents who reported that their status has stayed the same or has declined support the National Coalition Party less than those who reported that their status had improved. Respondents who expect their status to decline significantly support the Finns Party more than respondents who expect their status to improve. However, there is no longer a significant association between experience of status decline and support for Left Alliance. Likewise, respondents who expect their status to stay the same are no longer significantly more likely to support the Finns Party than those who expect it to improve. On a probability scale from 0 to 1, respondents whose status will decline support the Finns Party 0.063 points more than respondents who expect their status to improve. When the predicted intercept for support for the Finns Party starts from a low baseline of 0.124 points, this difference of 0.063 points is substantial.
Next are the results based on analyses conducted on data from our 2020 survey. Figure 8 shows some differences from the results based on data from our 2018 survey when sociodemographic variables were controlled. Significant differences in group means are only reflected for the Notes: The regression model includes sociodemographic controls and respondents' left-right and liberal-conservative political ideologies (see Table A5 in supplementary material). Whiskers represent 95% confidence intervals. When they intersect the red dotted line, the difference in group means is not statistically significant (P < 0.05).
National Coalition Party and the Finns Party. Respondents whose status had declined oppose the National Coalition Party significantly more than respondents whose status had improved (P < 0.005). Additionally, respondents who expect their status to stay the same oppose the National Coalition Party significantly more than do respondents who expect their status to improve (P < 0.05). For the Finns Party, only respondents who expect their status to decline support it significantly more than respondents who expect their status to improve (P < 0.05).
Finally, Figure 9 shows the results when both sociodemographic controls and left-right and liberal-conservative political ideologies were controlled for. Compared to the results in Figure 8, the difference in support for the National Coalition Party between respondents whose status had declined and those whose status had improved is no longer statistically significant (P = 0.063). However, the difference in support for this party between respondents who expect their status to be the same and respondents who expect their status to improve remains significant (P < 0.05). Additionally, the difference in support for the Finns Party between respondents who expect their status to decline and respondents who expect their status to improve remains statistically significant (P < 0.05). On a probability scale from 0 to 1, respondents who expect their status to decline support the Finns Party by 0.113 points more than respondents who expect their status to improve. When the predicted intercept for support for the Finns Party starts from a low baseline of 0.193 points, this difference of 0.113 points is substantial.
Based on data from the 2018 and 2020 surveys, the overall results do not show that respondents who experienced a status decline favour the radical right Finns Party more than respondents who experienced status improvement as stated in Hypothesis 1. Hypothesis 1 is thus rejected.
Next, respondents who experienced a status decline support the bourgeois centre-right National Coalition Party less than those who experienced status improvement in 2018, as stated Notes: The regression model includes sociodemographic controls, respondents' left-right and liberal-conservative political ideologies, and their issue positions on redistribution and immigration (see Table A6 in supplementary material). Whiskers represent 95% confidence intervals. When they intersect the red dotted line, the difference in group means is not statistically significant (P < 0.05).
in Hypothesis 2. However, it is not significant when respondents' political ideologies were controlled for in the 2020 data set. These findings suggest that experience of status decline significantly reduces the likelihood of voting for the National Coalition Party in 2018, but it matters less for the centre-right vote in 2020. Hypothesis 2 can therefore be accepted in 2018 but rejected in 2020.
Regarding expectations of status decline, there is consistent and robust evidence showing that respondents who expect a status decline consistently support the Finns Party more than respondents who expect a status improvement. This result was found in both the 2018 and 2020 data sets and across all models. Therefore, we can accept Hypothesis 3, which states that respondents who expect status decline support radical right parties more than their counterparts who expect status improvement.
Finally, we do not find that respondents who expect a status decline support centre-right parties less than those who expect a status improvement, as stated in Hypothesis 4. Additionally, respondents who expect their status to stay the same are significantly less likely to support centre-right parties than those who expect a status improvement only in 2020, but not in 2018. Hence, Hypothesis 4 can be rejected.
In summary, we find a consistent relationship between expectation of status decline and support for radical right parties despite using different data sets and different estimation strategies. 5 Namely, when respondents expect their status to decline, they favour radical right parties. Notes: The regression model includes sociodemographic controls (see Table A7 in the supplementary material). Whiskers represent 95% confidence intervals. When they intersect the red dotted line, the difference in group means is not statistically significant (P < 0.05).

5
These main findings are insensitive to alternate specifications of our sample and estimation strategies (see footnotes 3 and 4 and Figures A2 and A3 in the supplementary material).

Discussion and conclusion
Our study contributes to a burgeoning literature on the political consequences of status decline in advanced economies (Burgoon et al., 2019;Engler and Weisstanner, 2020;Gest et al., 2018;Gidron and Hall, 2019;Häusermann et al., 2021;Im et al., 2019;Iversen and Soskice, 2020;Kurer, 2020;Kurer and van Staalduinen, 2022) by distinguishing between voters' experience and expectation of status decline and their corresponding relationship with electoral behaviour. To do so, we went beyond the indirect proxies used by studies in this literature and relied instead on voters' assessments of whether their social status has declined and if they expect it to decline.
On one level, we descriptively explored the sociodemographic background of voters who had experienced and who expect status decline. Both groups of voters share some similarities in their sociodemographic backgrounds; they are often older, not higher educated, and live in the less economically prosperous northern and eastern parts of Finland. However, they differ in terms of their household income. Voters who have experienced status decline have low income, whereas voters who expect status decline have (lower)middle income. Crucially, this sociodemographic difference was observed both during normal times in 2018 as well as during exceptional times of the COVID-19 pandemic in 2020. Therefore, we find some descriptive evidence showing that voters who expect status decline differ somewhat from voters who have experienced status decline. Echoing recent research (Engler and Weisstanner, 2020;Mayer, 2015), voters who expect and are threatened by status decline are not at the bottom of the social ladder but are instead located somewhat higher up and fear falling. Put differently, it seems that voters who experienced status decline are often located among the lower classes, whereas voters who expect status decline are Notes: The regression model includes sociodemographic controls and respondents' left-right and liberal-conservative political ideologies (see Table A8 in supplementary material). Whiskers represent 95% confidence intervals. When they intersect the red dotted line, the difference in group means is not statistically significant (P < 0.05).
often located among the 'sandwiched' (lower)middle class. These differences indicate that it would be useful for researchers to conceptually distinguish between experience and expectation of status decline.
On the second level, we explored these two groups of voters' electoral behaviour. In 2018 and 2020, voters who expect their status to decline are consistently more likely to support the radical right Finns Party than voters who expect their status to improve. In contrast, there is no consistent evidence that voters who had experienced status decline support the Finns Party more than voters who had experienced status improvement in 2018 and 2020. However, we did find evidence that voters who had experienced status decline opposed the centre-right and bourgeois National Coalition Party more than those who experienced status improvement, especially in 2017. In short, it seems that expectations of status decline, not experiences of decline, explain persistent support for the radical right in Finland.
With regard to the electoral behaviour of voters whose status had declined, our findings depart somewhat from those of recent studies, which showed that these voters favour radical right parties, or radical left parties (if present, and if they and their parents have higher socioeconomic status) (e.g., Anelli et al., 2021;Ballard-Rosa et al., 2022;Burgoon et al., 2019;Gidron and Hall, 2019;Kurer, 2020;Kurer and Staalduinen, 2022). There are several reasons for these different findings. First and foremost, these differences may be related to how experiences of status decline are measured. While we asked respondents directly about whether they had experienced status decline, other studies relied on indirect proxies such as objective indicators and, to a lesser extent, subjective indicators. There are also a few theoretical explanations for these differences in findings. These voters' party preferences may be dispersed across all six major parties that were included in our analyses. If some voters who had experienced status decline favour radical left parties (Kurer and van Staalduinen, 2022), the absence of a credible radical left party in Finland (Dunphy, 2007) may disperse these voters' party preferences. Alternatively, these voters may abstain from voting, which we could not capture in our analyses. Descriptive evidence shows that voters whose status had declined are often the most socioeconomically vulnerable. Prior research shows that the most socioeconomically vulnerable do not vote more for radical right parties, but instead vote less (e.g., Aytaç et al., 2020;Gallego, 2007;Kurer, 2020;Mayer, 2015;Rovny and Rovny, 2017). Because experiences of status decline intersect with socioeconomic vulnerability in Finland, this may explain why these voters do not favour radical right parties.
In short, expectations of status decline, not experiences of status decline, drive support for the radical right in Finland. Expectations of status decline can strongly influence radical right support because voters react strongly to fears of losing status (Engler and Weisstanner, 2020;Kuziemko et al., 2014;Mayer, 2015). Recent socioeconomic transformations have disrupted upward social mobility and ushered downward social mobility (Kurer and van Staalduinen, 2022;Lahtinen et al., 2017). Social groups that enjoyed respectable status during the post-war years, like blue-collared workers in industrial jobs and farmers residing in areas specializing in agriculture, now fret about losing their place in the national pecking order. As Lamont and colleagues (2017) described, radical right parties destigmatize these social groups, attribute their fraying social status to structural transformations, and marginalize social groups that were previously status-inferior. In doing so, radical right parties buttress the moral worth of voters who fear status decline and thus consolidate their fraying social position. This, in turn, makes radical right parties attractive to these voters.
It is also important to highlight the success of the Finns Party in attracting voters who expect status decline in 2020 during the COVID-19 pandemic. Since concerns about public health and unemployment were often more salient than other issues during the COVID-19 pandemic, the significant association between expectations of status decline and radical right support in 2020 as well as in 2018 would suggest that such expectations are highly pertinent to understanding radical right support. Overall, this persistent link between expectation of status decline and radical right support suggests that other parties will struggle to dislodge this segment of the Finns Party's electorate.
Conversely, the absence of a relationship between experiences of status decline and radical right support spurs a reflection on what radical right parties' appeals can do for voters who have lost their status. With 'sticky floors' that prevent upward social mobility (OECD, 2018), radical right parties' appeal may do little to restore status that has already been lost, especially without radical and profound political change. Except for two-party systems, where radical right parties are one of the dominant parties, most radical right parties struggle to pursue these profound changes even if they are in power because they are often in coalition governments. In this regard, radical right parties' appeals are perhaps more successful at reasserting symbolic boundaries to bolster the status of social groups whose position in the national pecking order is threatened but is not yet lost.
Although our study is based on Finnish respondents, we anticipate that expectations of status decline will also influence radical right support in other advanced economies where similar socioeconomic transformations are taking place. If there are voters with middling income and middling education who worry about their status, even in a country with a comprehensive Nordic welfare state such as Finland, there will likely be more of such voters in countries with less comprehensive welfare states. This may also explain why support for Donald Trump is driven not only by white working-class Americans but also by middle-class Americans (Lamont et al., 2017, p. S155). In other words, the relationship between the expectation of status decline and radical right support could be even more pronounced in countries with more residual welfare states. Additionally, we suspect that the strength of this relationship depends on the success of radical right parties and the policies they implement or support when in government. The more successful the radical right parties are, the more voters with expectations of status decline may perceive them as viable avenues to channel and address their status worries (see Kreiss et al., 2017). However, if they pass or tacitly support policies (especially if they are in a coalition government) that maintain or entrench socioeconomic transformations that have given rise to these voters' expectations of status decline in the first place, this relationship between expectations of status decline and support for radical right parties may diminish.
In this regard, we propose two further lines of research. First, future studies could directly measure voters' experience and expectations of status decline in other countries. In doing so, they may examine whether the lack of an association between the experience of status decline and radical right support is an artefact of measurement or specific to the Finnish context. Crucially, a crossnational comparison would allow researchers to assess how strong radical left parties, which Finland lacks, may influence the relationship between experiences and expectations of status decline and radical right support. Next, future studies could disaggregate voters who expect status decline to examine how they may vary in their electoral responses according to their sociodemographic backgrounds.