Right-wing populist parties are surging in support across the Western world (Norris and Inglehart Reference Norris and Inglehart2019). Why do voters in developed democracies support such parties? A growing body of research has identified economic insecurity and cultural backlash as potential drivers of recent populist successes (Algan et al. Reference Algan, Guriev, Papaioannou and Passari2017; Hochschild Reference Hochschild2016; Inglehart and Norris Reference Inglehart and Norris2016; Mudde Reference Mudde2007; Oesch Reference Oesch2008; Rodrik Reference Rodrik2018; Smith and Hanley Reference Smith and Hanley2018). According to these explanations, once-dominant socioeconomic groups perceive an erosion of their economic opportunities or a threat to their privileged positions in society. These threats heighten voters’ perceived vulnerability, motivating them to support parties that promise to restore their socioeconomic standing through anti-multiculturalism, antiglobalism, and anti-immigration (Inglehart and Norris Reference Inglehart and Norris2016).
We argue that a voter’s perceived health may meaningfully contribute to populist support via a similar mechanism. The development of illness and disability often produces frustration with one’s physical and emotional limitations, and it prompts people to compare themselves with their healthier neighbors (Buunk, Gibbons, and Buunk Reference Buunk, Gibbons and Buunk2013; Martz and Livneh Reference Martz and Livneh2007). This experience may increase an individual’s sense of personal vulnerability regardless of their socioeconomic background. They may then blame their misfortunes on the political establishment as the architects of existing social, political, and economic structures (Laclau Reference Laclau2005; Nussbaum Reference Nussbaum2018). If true, individuals who suffer poorer health and more disability would be desirous of changing the political status quo. This desire for structural change would, in turn, draw them toward parties that campaign for a fundamental restructuring of a “biased and broken” system.
As such, health-related vulnerability may contribute to an antiestablishment sentiment alongside economic and cultural drivers of populism. Indeed, some research has associated declining population health with right-wing populist voting. U.S. counties that experienced the greatest rise in mortality over recent decades, especially among whites, were most likely to vote for President Trump and shift toward the Republican Party in 2016 (Bilal, Knapp, and Cooper Reference Bilal, Knapp and Cooper2018; Bor Reference Bor2017). Similar associations have been shown for rates of chronic opioid use (Goodwin et al. Reference Goodwin, Kuo, Brown, Juurlink and Raji2018) and other markers of poor public health (Wasfy, Stewart, and Bhambhani Reference Wasfy, Stewart and Bhambhani2017). In the U.K., localities that experienced greater rises in “deaths of despair” due to suicide or drug overdose in the previous decade were more likely to vote for Brexit (Koltai et al. Reference Koltai, Varchetta, McKee and Stuckler2019). However, the relationship between poor health and right-wing populist voting remains to be demonstrated at the individual level with appropriate controls for economic and cultural vulnerability.
Understanding how poor health influences right-wing populist support could have important implications for policy makers. It suggests that illness may counterintuitively increase support for political parties that seek to dismantle medical and public health services (Pavolini et al. Reference Pavolini, Kuhlmann, Agartan, Burau, Mannion and Speed2018). It could also enrich our understanding of the connection between health and political participation (Gollust and Haselswerdt Reference Gollust and Haselswerdt2019). Existing work has documented how, for example, people with disabilities, more depressive symptoms, and poorer self-reported health are less likely to vote (Landwehr and Ojeda Reference Landwehr and Ojeda2021; Pacheco and Fletcher Reference Pacheco and Fletcher2015; Schur et al. Reference Schur, Shields, Kruse and Schriner2002). We hypothesize that perceived health vulnerability may influence a voter’s decision in the polling booth itself. To test this argument, we explore whether an individual’s health vulnerability, as operationalized by their self-reported health, predicts their support for right-wing populist parties.
We pooled all available waves of the European Social Survey (ESS), collected from 2002 through early 2020. The survey measures social, political, and other attitudes and behaviors across Europe. Respondents were included in our main analyses if they stated the political party for which they voted in their country’s last national elections and their subjective general health. Cases were weighted using the ESS’s poststratification weights, which correct for sampling errors and nonresponse rates within each country to reduce selection bias. In total, our main analyses included 188,478 weighted cases across 24 countries, based on 192,896 unweighted cases.
Our analyses focus on right-wing populist support because most populist parties in Europe embrace this ideology. Respondents were coded as having voted for either a right-wing populist party or another party in their country’s last national elections. The list of right-wing populist parties was taken from Inglehart and Norris (Reference Inglehart and Norris2016) and Norris and Inglehart (Reference Norris and Inglehart2019). All other parties were coded as mainstream, including the two left-wing populist parties, Podemos of Spain, and Syriza of Greece. The ESS captured too few of their voters to empirically test whether self-reported health drives left-wing populist support. If, for a given round of the ESS, a party was grouped under the category “Other,” we coded it as mainstream because we could not distinguish it from other parties. In a few instances, this aggregation misclassified some right-wing populist voters as mainstream; however, any resulting classification errors should bias against finding support for our hypothesis.
The specifications of our models align with those of previous analyses of right-wing populist voting (Inglehart and Norris Reference Inglehart and Norris2016; Oesch Reference Oesch2008). We included controls for demographics (age, gender, years of formal education, religiosity, and being an ethnic minority), socioeconomics (Goldthorpe occupation scheme [Goldthorpe, Llewellyn, and Payne Reference Goldthorpe, Llewellyn and Payne1980], income decile within one’s country, self-reported economic insecurity, being on government benefits, having been unemployed for 3+ months, and rurality of domicile), and cultural attitudes (negative attitudes about immigrants, mistrust in national institutions, mistrust in international institutions, espousing authoritarian values, and degree of right-wing self-placement), as well as fixed effects for country (to adjust for differences in standard of living and evaluation of health across the countries in our sample) and survey year (to account for differences in the general political climate within Europe across years). To ensure that health status did not simply reflect general quality of life, we controlled for satisfaction with life as a whole. To account for negative health care experiences, we controlled for satisfaction with a country’s health system. The inclusion of potential posttreatment variables—that is, variables that might be affected by health status—could bias the observed association between health and right-wing populist voting. To address this concern, we provide results from the sequential introduction of variables to our models as a robustness check in the appendix.
We operationalized health vulnerability using two continuous self-reported health measures: (1) subjective general health, which the ESS elicited from respondents on a five-point scale ranging from “very good” to “very bad,” and (2) being hampered in daily activities by illness, disability, mental health problems, or other infirmities, measured on a three-point scale from “none” to “a lot.” Full details on all variables and reliability analyses of scales are provided in the appendix.
Missing observations were multiply imputed for variables other than voting and subjective general health using a bootstrapping-based algorithm with five imputations. We tested the relationship between health and voting using binomial logistic regressions on poststratification-weighted cases. Full details, tables, and robustness analyses are provided in the appendix.
Between 2002 and 2020, 13% of respondents voted for right-wing populist parties. With respect to subjective general health, 22% of voters reported “very good” health, 44% “good,” 26% “fair,” 6% “bad,” and 1% “very bad.” With respect to being hampered by health problems, 73% of voters reported “no” limitations, 21% were limited “to some extent,” and 6% “a lot.”
We find that Europeans with worse self-reported health were more likely to vote for right-wing populist parties between 2002 and 2020 (Figure 1). We examined this relationship using both measures of self-reported health. After controlling for demographics, socioeconomics, cultural attitudes, and more, a voter who rated their health as “very bad” had 26% greater odds of voting for a right-wing populist party than someone who responded as “very good” (Model 1; odds ratio [OR] for each unit: 1.059, 95% CI, 1.034–1.084, p < 0.001; Table 1, Column 1). Meanwhile, a voter whose daily activities were hampered “a lot” by health problems had 14% greater odds of voting for a right-wing populist party than someone who was not constrained by such conditions (Model 2; OR = 1.067, 95% CI, 1.033–1.103, p < 0.001; Table 1, Column 2). When the measures are tested together, a voter with “very bad” health had 20% greater odds of voting for a right-wing populist party, while the other health measure no longer meets conventional levels of statistical significance (Model 3; for general health, OR = 1.046, 95% CI, 1.018–1.075, p = 0.001; for hampered by disability, etc., OR = 1.034, 95% CI, 0.996–1.074, p = 0.08; Table 1, Column 3).
Note: Coefficients indicate the influence of a one-unit change in the self-reported health measures on the odds of right-wing populist voting in binomial logistic regressions. Please refer to Methods for the controls. Robust 95% confidence intervals (in parentheses) and standardized odds ratios are provided.
Consistent with existing research, we find that many demographic, socioeconomic, and cultural characteristics also predict populist voting. Right-wing populist voters were more likely to be white, male, younger, and less educated than other voters. Manual (skilled and nonskilled) workers voted for right-wing populists more often than professionals and managers. On average, these voters lived in more rural places, were more likely to have been unemployed for long periods, and were more likely to be receiving government benefits. They felt more negatively about immigrants, were more mistrustful of national and international institutions, espoused more authoritarian values, and identified more as right-wing. However, they did not feel less economically secure nor had lower incomes than mainstream voters. The standardized magnitudes of these predictors suggest that health had a greater influence on voters than income, self-reported economic insecurity, and living in rural areas, among other measures (Figures A.1–A.3). Meanwhile, they suggest that health had a smaller influence on voters than gender, education, and attitudes toward immigrants.
Neither dissatisfaction with life as a whole nor dissatisfaction with the country’s health system predicted voting for right-wing populist parties. These findings suggest that voters distinguished their health status from general life satisfaction, which minimizes the concern that self-reported health might merely act as a proxy for a voter’s satisfaction with life. Similarly, perceptions of health status appear to influence political decision making independently of experiences with and attitudes about a country’s medical and public health infrastructure.
The relationship between self-reported health and right-wing populist voting was robust to a variety of sensitivity analyses. The associations are retrieved when we respecify the health measures as dichotomous (Model 1, OR = 1.072 [1.027–1.118], p = 0.002; Model 2, OR = 1.077 [1.032–1.124], p < 0.001; and Model 3, ORgen. = 1.048 [0.999–1.098], p = 0.05, and ORhamp. = 1.055 [1.006–1.106], p = 0.03) (Table A.8). Given the well-established association between poor health and nonvoting, we estimated a multinomial probit model with a polytomous outcome (mainstream voting, right-wing populist voting, and nonvoting). The association between poorer self-reported health and right-wing populist voting is robust to the inclusion of nonvoters. We also retrieve the relationship between poorer health and nonvoting identified in the literature (Table A.9). Next, we tested the robustness of our findings by sequentially introducing the controls in our fully specified models (Tables A.10–A.12). The direction and statistical significance of the association remain stable across these specifications, mitigating concerns about posttreatment bias.
To address survivor bias—whereby mainstream voters may die younger than right-wing populist voters and bias the observed association—we stratify our sample into three age groups and find that the association is preserved among younger voters (Tables A.13–A.15). The association also manifests in both pre- and post-2008 periods (Tables A.16–A.17), suggesting that the relationship was not an artifact of the Great Recession. Similarly, it persists when county-year controls for GDP growth, life expectancy, health inequality, percentage of government expenditure on health, and out-of-pocket spending on health are included (Table A.18). Our findings are also robust to the inclusion of population-size weights and countries that lacked populist parties during the study period, such as Ireland and Portugal (Tables A.19–A.20). Last, the continent-wide association remains unchanged after a one-by-one exclusion of each country, mitigating the concern that the observed relationship is driven by any single country (Table A.21).
Our findings suggest that a voter’s perception of health-related vulnerability—beyond economic and cultural vulnerabilities—may be fueling right-wing populist support. This finding builds on research that has identified economic and cultural vulnerability as contributing to recent right-wing populist successes. Our models tested two distinct measures of health as proxies for health vulnerability: subjective general health and being hampered in daily activities by health problems. Their association with right-wing populist voting was robust to the inclusion of demographic, socioeconomic, and cultural controls. The results also suggest that the magnitude of this association may be comparable to that of some socioeconomic factors.
We had hypothesized that health vulnerability contributes to frustrations with the political status quo, not unlike how other vulnerabilities of a voter’s economic and cultural position could feed an antiestablishment sentiment (Algan et al. Reference Algan, Guriev, Papaioannou and Passari2017; Hochschild Reference Hochschild2016; Inglehart and Norris Reference Inglehart and Norris2016; Laclau Reference Laclau2005; Mudde Reference Mudde2007; Rodrik Reference Rodrik2018; Smith and Hanley Reference Smith and Hanley2018). That is, illness inspires grievances with societal structures, driving voters to seek out political parties that promise to restructure them. Our findings complement work that has documented the association of declining community health with populist voting but at the individual level (Bilal, Knapp, and Cooper Reference Bilal, Knapp and Cooper2018; Bor Reference Bor2017; Goodwin et al. Reference Goodwin, Kuo, Brown, Juurlink and Raji2018; Koltai et al. Reference Koltai, Varchetta, McKee and Stuckler2019; Wasfy, Stewart, and Bhambhani Reference Wasfy, Stewart and Bhambhani2017).
Self-reported health measures reliably and cost-effectively estimate “true” health status (Kaplan and Baron-Epel Reference Kaplan and Baron-Epel2003). They have been shown to predict mortality in cohort studies with up to 25 years of follow-up (Idler and Benyamini Reference Idler and Benyamini1997; Lorem et al. Reference Lorem, Cook, Leon, Emaus and Schirmer2020); correlate with symptoms, disease diagnoses, and physicians’ ratings of health status (Larue et al. Reference Larue, Bank, Jarvik and Hetland1979); and, to some degree, capture health as a positional good relative to a person’s healthier neighbors (Eriksson, Undén, and Elofsson Reference Eriksson, Undén and Elofsson2001). Meanwhile, research has shown that health is intimately linked to a person’s social, economic, and political context (Woolf and Braveman Reference Woolf and Braveman2011) and that it is associated with their perception of their social standing (Ostrove et al. Reference Ostrove, Adler, Kuppermann and Eugene Washington2000). It is, therefore, conceivable that economic and cultural factors could mediate the relationship between health status and right-wing populist voting. Because including mediators in a model can introduce bias, we showed that our results were robust to the inclusion of socioeconomic and cultural measures as controls as well as to their exclusion as potential posttreatment variables. As a result, our findings suggest that health vulnerability shapes right-wing populist sentiment beyond the influence of other individual-level factors already identified in the literature.
These results may generalize to other stable democracies. For example, in the U.S., perceptions of health status and health policy are even more tightly tied to socioeconomic inequities and political identity than in Europe (Hero, Zaslavsky, and Blendon Reference Hero, Zaslavsky and Blendon2017; Kirzinger et al. Reference Kirzinger, DiJulio, Wu and Brodie2017). Indeed, as mentioned earlier, other researchers have observed a relationship between health and voting in the U.S. at the population level. However, the lack of universal health insurance and a robust safety net may mean that mainstream parties can better differentiate themselves along the health dimension, limiting the ability of right-wing populists to appeal to voters in poorer health. Moreover, detecting this relationship at the individual level in the U.S. might be more challenging. In a parliamentary system, voters can select parties that closely reflect their own ideologies, less so in a two-party, candidate-driven system.
Last, we turn our attention to implications for policy makers and health professionals. Many right-wing populists in Europe and the U.S. propose to dismantle public health systems (Pavolini et al. Reference Pavolini, Kuhlmann, Agartan, Burau, Mannion and Speed2018). If mainstream parties accommodate these proposals and voters attribute the resulting deterioration in their health to the mainstream, the relationship we identified might be bolstered. If, however, voters attribute a deterioration in their health to the right-wing populist parties, as might happen if right-wing populists become mainstream, our proposed relationship might weaken (Meguid Reference Meguid2005). While the mechanism proposed in this paper remains too preliminary for specific policy recommendations, any relationship between health and political activity should train our attention on health inequities and their consequences. Support for policies designed to enhance public health infrastructure, improve access to medical care, and reduce socioeconomic inequality could affect not only well-being but also the political landscape.
Meanwhile, we discourage dismissing populist voters as irrational or irresponsible. To them, support for parties outside the mainstream reflects democratic accountability at work, whereby parties that don’t serve their needs are punished for their poor performance. Thoughtful proposals that serve disaffected voters could allow mainstream parties to reengage their attention.
Limitations and Future Directions
We must consider the potential for reverse causality in this setting. That is, could voting for a right-wing populist party worsen a voter’s self-reported health? We believe that this concern is minimized for a few reasons. If a voter’s preferred candidate had come into office, it is unlikely that sufficient time would have passed between the election and the fielding of the ESS for structural changes that damage health to have been implemented. Alternatively, if their preferred candidate lost, dissatisfaction with the outcome of the election might negatively affect their self-perceptions. However, if it influenced their perceived health, we might also expect it to influence their satisfaction with life as a whole, and right-wing populist voters were no less satisfied with their lives than mainstream voters in our models. Lastly, community-level analyses indicate that changes in health status have predated surges in populist support (Bilal, Knapp, and Cooper Reference Bilal, Knapp and Cooper2018; Bor Reference Bor2017; Koltai et al. Reference Koltai, Varchetta, McKee and Stuckler2019).
The serial cross-sectional design of this study cannot establish causality. Moreover, the ESS’s limited measures cannot parse the intricacies of the experiences of illness and disability, nor can they distinguish whether absolute or positional changes in health are driving right-wing populist support. Also, self-evaluations of health may vary within countries over time. To test the mechanism more precisely, future research might correlate longitudinal changes in health status with political attitudes. For example, we might capture voters just after major injuries or chronic disease diagnoses and follow their voting behavior over time. At present, few large-N datasets include both political and health outcomes, complicating the study of their relationship at the individual level. Given the influence of political and social structures on health—and vice versa—we encourage more surveys to track both together.
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0003055421000265.
DATA AVAILABILITY STATEMENT
Replication files are available at the American Political Science Review Dataverse: https://doi.org/10.7910/DVN/LIXQUW.
We thank Mark Dincecco, Sarah E. Gollust, Scott L. Greer, Ronald Inglehart, Pauline Jones, Paula M. Lantz, Walter Mebane, and Atheendar S. Venkataramani for their helpful feedback on the project. We also extend our gratitude to Juli Highfill for overseeing the thesis from which this paper grew.
CONFLICT OF INTEREST
The authors declare no ethical issues or conflict of interest in this research.
The authors affirm this research did not directly involve human participants.