Healthy citizens, healthy democracies? A review of the literature

A growing literature over the past 10 years on health and political behavior has established health status as an important source of political inequality. Poor health reduces psychological engagement with politics and discourages political activity. This lowers﻿ incentives for governments to respond to the needs of those experiencing ill health and thereby perpetuates health disparities. In this review article, we provide a critical synthesis of the state of knowledge on the links between different aspects of health and political behavior. We also discuss the challenges confronting this research agenda, particularly with respect to measurement, theory, and establishing causality, along with suggestions for advancing the field. With the COVID-19 pandemic casting health disparities into sharp focus, understanding the sources of health biases in the political process, as well as their implications, is an important task that can bring us closer to the ideals of inclusive democracy.


Introduction
Life expectancy at birth increased by 10 years between 1970 and 2017 (OECD, 2019), but longer lives often mean more illness.Currently, 58.1% of people aged 65 and older across 28 OECD countries have two or more chronic conditions.Even among those aged 15-64, the average is 25.4% (OECD, 2019).Moreover, an estimated one in two people experience a mental health problem during their lifetime (OECD, 2019).Yet, political scientists have been slow to recognize that ill health is politically consequential.This has changed over the past decade (see Figure S1 in the online appendix).By providing a critical synthesis of this rapidly developing literature 1 on the links between health and various aspects of political behavior, this article aims to be a catalyst for further research on the long-term implications of health disparities for political behavior.The COVID-19 pandemic has highlighted the importance of this research agenda: from vaccine hesitancy in populations whose experiences have taught them to distrust the healthcare system to higher infection rates in poor neighborhoods and distrust in politicians and experts, the consequences of health and health-related socioeconomic inequalities are evident.
We begin by discussing some of the institutional and individual-level factors that could affect the association between health and political behavior.Then we zoom in to focus on how health relates to different aspects of political behavior, including psychological predispositions, voter turnout, extra-electoral political activities, and political preferences.After this, we offer critical reflections on the state of the literature, highlighting some of the challenges in studying the relationship between health and political behavior.We conclude with suggestions for future studies.

Studying health and political behavior
We should not necessarily expect a straightforward relationship between health and political behavior.On the contrary, it is likely to be contingent on various institutional and individual-level characteristics.Importantly, we could expect generous welfare provision and universal health coverage to mitigate any negative effects of poor health.As policy feedback scholars have emphasized, government programs can help to compensate for resource deficits and convey important messages about the nature of politics and clients' worth as citizens (Béland and Schlager, 2019).When it comes to voter turnout, we could expect the relationship to vary depending on the institutional arrangements governing elections.The type of electoral system, the number of political parties, and voter facilitation policies (such as proxy voting and postal voting) can all affect the costs of voting, which are likely to be higher for people experiencing poor health (Mattila et al., 2017).The presence of a strong left-wing party may also be essential, given their record of mobilizing resource-poor citizens and voicing their needs.
Individual-level differences are likely to matter, too.First, people's general state of health may affect their political behavior by increasing the costs and reducing the perceived benefits of political engagement.Poor health may diminish the resources, mental and physical energy, social connections and associational involvements that encourage active citizenship.However, this may be too simplistic.The effects of ill health may differ, depending on whether the condition is temporary or chronic.Moreover, chronic illnesses are likely to differ in the extent to which they hamper-or encourage-engagement with politics (Gollust and Rahn, 2015;Sund et al., 2017).Chronic health conditions that are stigmatized, impede cognitive and/or physical functioning or require ongoing management may depress engagement, but conditions that foster a strong sense of collective identity and benefit from the mobilizing activities of disease advocacy groups may have the opposite effect.
Second, social background characteristics may moderate the effects of ill health.Poor health may compound the decline in political engagement that often accompanies old age.Higher socioeconomic status may translate into resources and social connections that counter any dampening effect of poor health.Meanwhile, sickness may reinforce the political disengagement of marginalized groups.It is possible, though, that poor health serves instead as a leveler by impeding the political engagement of the hitherto advantaged.
Reviewing the literature on health and political behavior provides an opportunity to assess these possibilities.While comprehensive, the review is not intended to be exhaustive.Table S1 in the online appendix provides systematic information about every study being referenced.

Health and psychological predispositions
We begin with the psychological predispositions that are widely seen as key influences on political activity.Intuitively, we would expect health problems to encourage more negative views of politics (Peterson, 1987).Poor health may feed a perceived lack of system responsiveness (external political efficacy) and erode a sense of political competence (internal political efficacy) as a result of status quo bias and a diminished sense of agency.Individuals experiencing health problems may question whether positive social change is possible and doubt their ability to influence political matters, especially given the time and energy sickness can consume (Mattila et al., 2017;Pacheco and Fletcher, 2015;Shore et al., 2019).Health problems, particularly if accompanied by struggles to overcome bureaucratic hurdles to proper treatment (Christensen et al., 2020), can lead to the 'perception that a person's needs are not being met and their preferences ignored' (Shore et al., 2019: 439).This may affect citizens' political self-confidence by diminishing their self-esteem (Shore et al., 2019).
Studies have reported a consistent connection between health problems and political efficacy, which can diminish satisfaction with democracy (Mattila et al., 2017).We could expect this connection to be weaker in countries with generous welfare provision.Access to social benefits has interpretive effects that encourage a perception that the state responds to citizens' needs (Shore, 2019) and resource effects that decrease the association between poor health and internal efficacy by reducing impediments to political engagement. 2 Yet, while both types of political efficacy tend to be higher in more encompassing welfare regimes, the health bias persists (Shore et al., 2019).This could reflect the 'Scandinavian paradox': welfare regimes are more successful in addressing socioeconomic inequalities than health disparities.However, the weaker link with external political efficacy in Finland may reflect the accessibility and quality of the country's healthcare (Mattila et al., 2017), Mental health conditions may have particularly negative effects on people's sense of political competence and perceptions of system responsiveness (Bernardi, 2021b;Bernardi and Johns, 2021;Bernardi et al., 2022;Landwehr and Ojeda, 2021), which may well contribute to the depressionvoting gap (Bernardi et al., 2022).Hong Kong is an interesting exception, where users of psychiatric services reported higher-than-average levels of external efficacy (Chiu and Chan, 2007).
Studies of political interest and political knowledge tell a similar story.Two competing hypotheses have been proposed.The political motivation hypothesis (Landwehr and Ojeda, 2021) suggests that health problems leave people preoccupied with everyday coping and concerns related to their personal well-being, weakening their motivation to follow political news and gather information.The self-interest hypothesis (Mattila, 2020;Mattila et al., 2017Mattila et al., , 2020)), by contrast, suggests that persons suffering from ill health have higher stakes in public health and social welfare policies, and should therefore be more motivated to pay attention (see also Söderlund and Rapeli, 2015).Empirical evidence favors the former (Mattila et al., 2017), although panel data from the UK shows that a decline in health led to a small increase in interest (Mattila et al., 2020).Self-reported depressive symptoms also have a negative effect on political interest (Landwehr and Ojeda, 2021).This may reflect the tendency of those with depression to focus more on negative cues and underestimate benefits (Bernardi, 2021a), again supporting the political motivation hypothesis.
The mechanisms linking health to political trust are more complex and vary depending on the institutional context.The psychological-democratic contract theory perceives political trust as a contract between citizens and the political system based on expectation management (Mattila andRapeli, 2018, citing Wroe, 2004).Citizens expect certain benefits, and a government's failure to deliver is seen as a breach of contract.When people fall ill, they may hold governments at least implicitly responsible, especially if the system is unresponsive to their needs (Mattila and Rapeli, 2018).However, the relationship between health and political trust may be modified by a person's evaluation of health services or their political orientation.If services seem inadequate, the association between poor health and low trust may intensify.Similarly, it may be stronger among those with a leftist political ideology, who expect more support, and in universalist welfare states that have made 'a strong commitment to ensure the well-being of their citizens, which may, in turn, raise citizens' expectations to a higher level than those in states with less extensive welfare systems' (Mattila and Rapeli, 2018: 122).
These hypotheses are only partially confirmed.Poor health is associated with lower levels of political trust in Europe (Mattila, 2020;Mattila and Rapeli, 2018;Mattila et al., 2017) and the US (Peterson, 1991).However, while poor health plays a more important role for those identifying with the left, evaluations of health services do not have the predicted effect (Mattila and Rapeli, 2018).Moreover, the health gap in political trust is most evident in Northern European welfare states, even though overall trust is relatively high, even among those in poor health.By committing to universalism and egalitarianism, Scandinavian countries increase citizens' expectations, and if these are not met, the resulting erosion of trust may further dampen internal and external political efficacy (Shore, 2019).Little research has examined the link between mental health and political trust, although Chiu and Chan (2007) found a negative relationship in Hong Kong.
We might expect these negative psychological orientations to be reflected in low levels of identification with established political parties.Using German panel data, Papageorgiou et al. (2019) found that deteriorating health was indeed associated with weaker party identification and that recovery from poor health did not restore previous levels in the short term.However, Bernardi (2021a) argued that a desire to regain control of their own lives might lead people vulnerable to depression to identify with a political party, particularly if that party makes explicit representational claims (cf.Saward, 2010) for marginalized groups.Contrary to expectations, though, the probability of identifying with a political party decreased when self-reported depression symptoms increased, although the association was weak (Bernardi, 2021a).
In summary, the story is fairly grim: poor physical or mental health tends to discourage and demotivate rather than empower.Even generous welfare state provision is unable to overcome the negative political predispositions associated with health problems.

Health and voting
Much of the research on health and political behavior has focused on voter turnout.This is easy to understand: if there is a health bias in turnout, governments may have less incentive to respond to the needs of those in poor health.Multiple studies, spanning different welfare regimes and electoral rules, have confirmed the association between health and turnout in the UK (Denny and Doyle, 2007a;Rapeli et al., 2020), Canada (Couture and Breux, 2017), Ireland (Denny and Doyle, 2007b), the US (Burden et al., 2017;Pacheco and Fletcher, 2015;Peterson, 1991), the Nordic countries (Adman, 2020;Lahtinen et al., 2017;Mattila et al., 2017Mattila et al., , 2018;;Söderlund and Rapeli, 2015), and across Europe (Mattila et al., 2013(Mattila et al., , 2017;;Stockemer and Rapp, 2019;Wass et al., 2017).Some studies have even found that the effect rivals that of education (Burden et al., 2017) and social class (Gagné et al., 2020).
Implicitly or explicitly, most studies have been informed by the civic voluntarism model (Verba et al., 1995), which encompasses three broad explanatory factors: resources, psychological engagement, and recruitment.Health is likely to affect all three: 'People with poor health are . . .less likely to participate since they either can't, don't want to, or nobody asked them to' (Söderlund and Rapeli, 2015: 31).
Voting may be more demanding, physically and cognitively, than typically assumed (Burden et al., 2017).Poor health can sap the energy, time, and initiative needed to register and attend a polling place (Denny and Doyle, 2007b;Gollust and Rahn, 2015;Landwehr and Ojeda, 2021;Mattila et al., 2013).It can also affect voting indirectly through other resource deficits: if sickness compromises educational attainment or employment opportunities, citizens may acquire fewer civic skills that promote political engagement (Stockemer and Rapp, 2019).These factors may also constrain financial resources (Gollust and Rahn, 2015;Mattila, 2020).However, resource-based explanations raise a tricky question: Do those in poor health vote less because they have fewer resources, or do they have poor health because they have fewer resources?
Poor health can also reduce turnout by diminishing psychological engagement with politics.People with health problems may have little time or energy to follow politics (Burden et al., 2017;Mattila et al., 2013), a topic that may seem irrelevant when 'holding body and soul together' (Denny and Doyle, 2007b: 55) is the priority.As a result, they may know little about politics.Their sense of personal political competence may be eroded by a perceived lack of control over their lives, while frustrating experiences with the healthcare system and government failure to address their needs may decrease their political trust.If citizens think that voting will not improve their situation, the costs of voting may outweigh any conceivable benefits (Mattila, 2020;Stockemer and Rapp, 2019).Put bluntly, 'casting a vote might not seem imperative if you don't expect to live through the next term' (Reitan, 2003: 55).Finally, ill health may weaken people's sense of civic duty (Denny and Doyle, 2007b;Peterson, 1991).
People in poor health often have fewer social connections and take part in fewer social activities.The resulting social isolation means that they are also less likely to be mobilized to vote by friends and acquaintances (Burden et al., 2017;Denny and Doyle, 2007b;Gollust and Rahn, 2015;Mattila et al., 2013).
While several possible mechanisms have been identified, few mediation analyses have been conducted.Evidence from 30 European countries suggests that social connectedness partly mediates the relationship between health and turnout (Mattila et al., 2013) and a Finnish study showed that political trust mediated the relationship between health and institutional participation (Mattila, 2020), although voting was only one of the activities included in the participation measure.
Earlier, we raised the possibility that poor health could offset the advantages associated with higher socioeconomic status, but ill health only appears to be a leveler of class differences in turnout for those in the worst health (Lahtinen et al., 2017).In the US, poor health does not affect the turnout of high-income citizens, but depresses the turnout of low-income citizens, perhaps because they lack the resources to offset the effects of ill health (Lyon, 2021).However, low-income workers who were not obese, did not smoke, and had health insurance were more likely to vote (McGuire et al., 2021a).The findings for education are mixed: cancer patients in the US with less schooling were more likely to vote than those with more education (Gollust and Rahn, 2015), but a study of 20 European countries found no evidence that education moderated the relationship between poor health and turnout (Stockemer and Rapp, 2019).
Little attention has been paid to race as a potential moderator.Far from reinforcing the effects of marginalization, health problems appear to boost the turnout of racial minorities in the US.African Americans with cancer proved more likely to vote than whites with cancer (Gollust and Rahn, 2015) and a particularly strong diabetes-voting relationship was found among those who identified as Hispanic and multiracial (McGuire et al., 2021b).Clearly, more studies are needed to understand how race, class, and health interact to affect voter turnout (Gollust and Rahn, 2015;Ojeda and Pacheco, 2019).The same applies to sex.
Poor health seems to disrupt the formation of a voting habit for young adults.Young Americans who experience poor health during adolescence are less likely to vote in their first elections (Ojeda and Pacheco, 2019).Similarly, in the UK, poor health is associated with lower turnout in a young person's first and second eligible elections (Gagné et al., 2020).At the opposite end of the life cycle, declining health can disrupt the voting habit (Rapeli et al., 2020).Administrative data on hospital admissions in Denmark confirm that poor health is one reason turnout declines among the elderly (Bhatti and Hansen, 2012; see also Engelman et al., 2022;Goerres, 2007;Mattila et al., 2013).
Associations between health and turnout generally hold cross-nationally, but effect sizes vary.Institutional arrangements matter.Mattila et al. (2017) found that compulsory voting and partyversus candidate-centered electoral systems help explain the variation in Europe.Left-wing parties' vote share and trade union density were important, too.Community type may also matter.In the US, health has less effect on rural voters because religious attendance offsets the effects of poor health (Cahill and Ojeda, 2021).The relationship between health and turnout may also vary by level of government.Couture and Breux (2017) found that poor general health mattered more at the national level in Canada, whereas poor mental health was more important at the municipal level, perhaps because voting is costlier.The theory of second-order elections could be useful in theorizing the differential effects of poor health in national, supra-national, and local elections, given differences in stakes (Reif and Schmitt, 1980).
The diagnosis matters.A US study revealed that people with heart disease were less likely to vote, whereas those with cancer were more likely, reflecting possible differences in disease identity and the advocacy efforts of disease-focused organizations (Gollust and Rahn, 2015; see also McGuire et al., 2021b).A Finnish study of 17 chronic health conditions similarly found that cancer, like COPD/asthma, was positively associated with voting, but turnout was unaffected by heart disease, arthrosis, multiple sclerosis, or kidney disease (Sund et al., 2017).Neurodegenerative diseases, especially dementia, had the strongest negative associations.
Mental health has attracted less attention than physical health (Landwehr and Ojeda, 2021).Several studies have hypothesized that depressive symptoms drain physical energy, diminish interest in politics, and impair personal political competence (Landwehr and Ojeda, 2021).Alcoholism and mental disorders both have strong negative effects on voter turnout in Finland (Sund et al., 2017), and in the US, severe symptoms lower turnout (Ojeda, 2015).Adolescent depression makes voting less likely in young adulthood, which is partly explained by its negative effect on educational attainment and the formation of party ties (Ojeda, 2015; see also Wray-Lake et al., 2019).Moreover, the turnout of young American sufferers declines across time (Ojeda and Pacheco, 2019).The negative association between depression and turnout generally applies across gender and race in the US (Ojeda and Slaughter, 2019), and to varying degrees, across 25 of the 30 countries in the European Social Survey (Landwehr and Ojeda, 2021) as well as Canada (Couture and Breux, 2017).However, mental well-being appears unrelated to turnout in Ireland (Denny and Doyle, 2007b). 3 In sum, with few exceptions, the evidence suggests that poor health depresses turnout.However, a more concerted effort to systematically test causal mechanisms and theorize potential moderating factors is needed.

Health and other forms of political participation
Less attention has been paid to other forms of political participation, partly because suitable data are lacking.Studies have typically focused on a single country or small group of countries, raising questions about generalizability.Results are mixed, with positive, negative, and null relationships, which is not altogether surprising as political activity can take many forms.
According to the convenience hypothesis, the health gap should be smallest for easier activities with fewer obstacles to participation (Söderlund and Rapeli, 2015).Impaired physical functioning is unlikely to impede donating to a campaign, for example, as donors need only provide credit card details (Burden et al., 2017).Similarly, wearing a campaign button or signing a petition is less demanding than attending a demonstration or contacting a government official.
While the convenience hypothesis focuses on costs, the increased activism hypothesis emphasizes expected benefits.Health is expected to make less difference to activities with a stronger impact on policymaking, such as working for a political party, participating in demonstrations, or contacting public officials (Söderlund and Rapeli, 2015).This hypothesis is founded on the assumption of self-interest: people in poor health may be motivated to participate because they have high stakes in policy outcomes.Mattila (2020) similarly argues that citizens in poor health will focus on political activities that could impact policy, either from self-interest or from a sense of grievance related to a decline in their social status.However, he predicts they will be less likely than healthy people to contact public officials or be active in a political party because they cannot control the agenda.Instead, health problems will motivate them to engage in non-institutional forms of political action, such as signing petitions, participating in demonstrations or boycotts, and posting on social media.Stockemer and Rapp (2019) likewise note that these activities allow people in poor health to have their voices heard.
None of these hypotheses have been consistently confirmed.An early American study reported that poor health depressed engagement in campaign activities and in communal activities such as working with others to solve community problems or get the government to act, but was unrelated to particularistic contacting (Peterson, 1991).Scandinavian studies have produced conflicting results.In Finland and Sweden, older adults in poor health proved less likely to engage in protest activities, such as signing petitions, boycotting, and demonstrating, but no less likely to undertake 'influencing activities,' such as contacting politicians, appealing decisions, and writing letters to the press (Nygård and Jakobsson, 2013).However, Mattila (2020) showed that people in poor health were more likely to engage in non-institutional forms of participation in Finland, and Adman (2020) concluded that poor health stimulates contact and protest activities in Sweden.
Studies of specific activities have also failed to produce clear verdicts (see Table S1 in the online appendix).Support for the convenience hypothesis is mixed.Whether poor health depresses, fosters, or makes no difference to the propensity to boycott products, donate to campaigns, sign petitions, or wear a campaign button depends on how health is measured and the countries or regions studied.Similarly, for demanding activities such as contacting officials, demonstrating or working for political organizations, the evidence does not consistently support the increased activism hypothesis.Consistent support is similarly lacking for Mattila's (2020) argument that citizens with health problems are less likely to participate in activities that do not allow them to control the agenda.People in poor health are not necessarily less likely to contact public officials, be involved in political parties or engage in non-institutional activities such as signing petitions or expressing their opinions on social media.
Few studies have focused on mental health.In the US, Ojeda (2015) concluded that adolescent depression can have lasting negative effects, but among adults, depressive symptoms were unrelated to donating to political parties, attending rallies, and contacting officials.Landwehr and Ojeda's (2021) 30-country study found that depressive symptoms were unrelated to physically undemanding activities like contacting a politician, signing a petition, or displaying campaign material but had a negative effect on activities requiring physical energy such as demonstrating or working for a political group.However, other studies suggest that people experiencing poor mental health may be more politically active.In Hong Kong, mental health service users were more likely to have participated in seven 'protest activities,' such as attending a demonstration and lodging a complaint (Chan and Chiu, 2007), though they were less likely to have signed a petition or belonged to a civic group.
In Canada, those reporting poor mental health were more likely to have signed petitions and searched for information about political issues (Couture and Breux, 2017).
Evidently, assuming that health problems necessarily adversely affect political participation would be mistaken.Under certain conditions, they can motivate citizens to be politically active.

Health and political preferences
Unequal participation in politics is especially concerning if poor health is systematically associated with political preferences.Given that people in poor health are more likely to attribute health problems to income inequality, healthcare costs, and lack of health insurance (Robert and Booske, 2011), we might expect their political preferences to be more leftist.However, empirical findings are somewhat mixed.
From a self-interest perspective, people with health problems should be deeply concerned about healthcare (Bernardi, 2021b;Denny and Doyle, 2007b;Rapeli et al., 2020).Their personal stake should increase the appeal of leftist parties, which are traditionally more supportive of public healthcare than conservative parties (see Rapeli et al., 2020).This would explain why Americans with health problems were more supportive of the 2010 Healthcare Reform, although surprisingly, not having personal health insurance had the opposite effect (Richardson and Konisky, 2013).Health can even modify the influence of partisan identity: Republicans who were worried about healthcare costs were as likely to support universal healthcare as Democrats (Henderson and Hillygus, 2011).Difficulties in accessing healthcare also pushed preferences toward Democrat candidates (Ziegenfuss et al., 2008).
People in poor health are similarly left-leaning in Finland (Mattila et al., 2017), Germany (Bullenkamp and Voges, 2004), Ireland (Kelleher et al., 2002), Japan (Subramanian et al., 2010), the UK (Bernardi, 2021a;Rapeli et al., 2020), the US (Pacheco and Fletcher, 2015;Peterson, 1991;Subramanian and Perkins, 2009), and Europe (Bernardi, 2021a;Huijts et al., 2010;Subramanian et al., 2009).Ecological studies using mortality as a health indicator (Davey Smith and Dorling, 1996;Dorling et al., 2001;Goldman et al., 2019;Kannan et al., 2019;Kondrichin and Lester, 1998) point in the same direction.However, it has been suggested that the causal arrow runs the other way (Subramanian et al., 2010): right-wing voters may be healthier because they invest more in health-promoting lifestyles, reflecting their individualist outlook and dislike of the 'nanny state.' The rise of right-wing populists who are skilled at addressing voters' grievances has challenged the assumption that poor health increases support for the left.Support for Trump has been linked to 'deaths of despair' related to alcohol, drugs, suicide (Goldman et al., 2019), and opioid use (Goodwin et al., 2018).Trump's vote share in 2016 was higher in counties with stagnating or declining life expectancy (Bor, 2017).Greater health inequalities and worsening premature mortality also shifted votes toward Trump, with the Republican Party being more popular in counties with more unequal life expectancy (Bilal et al., 2018).In Europe, poor health predicted voting for rightwing populist parties (Kavanagh et al., 2021), right-wing populist voters reported worse health than conservatives (Backhaus et al., 2019), and areas with poorer health were more likely to favor the far-right presidential candidate in France (Zeitoun et al., 2019).

Critical reflections
Health clearly affects political behavior, but for this research agenda to advance, researchers must wrestle with issues relating to reliance on self-reports, the meaning and measurement of health, theorizing the effects of health, and establishing causality Reliance on self-reports may be especially problematic for the association between turnout and health status since both are subject to social desirability bias.People may be reluctant to admit failing to perform their civic duties, and may report that their health is better than it is, especially if they have stigmatized conditions, such as diabetes, obesity, drug addiction, and mental illness (Gollust and Rahn, 2015;Stockemer and Rapp, 2019).In Nordic countries, researchers have obviated this by using register-based administrative data (Bhatti and Hansen, 2012;Bryngelson, 2009;Lahtinen et al., 2017;Mattila et al., 2018;Sund et al., 2017), but these are unavailable in many countries, and have their own limitations (Mattila et al., 2018).Importantly, register data cannot provide insight into the underlying cognitive and psychological mechanisms or other forms of political participation.
Typically measured with a single item, such as, 'In general, how is your health?' self-reported health is widely used in health-related fields and correlates with objective indicators, such as chronic illness, use of health services, and early mortality.However, Bollen et al. (2021) estimated that this measure's reliability was modest, and if left uncorrected, reduced correlations with other health measures by 20-40%.They recommended structural equation modeling, which treats health as a latent construct.Differences in how respondents evaluate their health are another possible source of bias, as 'self-ratings of health are produced in a cognitive process that is inherently subjective and contextual' (Jylhä, 2009: 314).This is problematic when differences are systematic, and their sources (e.g., race, ethnicity, language, sex, age, and socioeconomic status) correlate with political behavior (Pacheco, 2019).When Pacheco (2019) adjusted for interpersonal incomparability by rescaling responses to a common scale, the association between self-rated health and turnout was attenuated, and the association with partisanship disappeared.Relying on a single measure of self-reported health is often the only feasible option, but scholars should be cautious about treating it as an accurate and reliable measure of health, especially when making intergroup or crossnational comparisons.
Given that health is 'a broad and extremely multifaceted concept' (Sund et al., 2017: 475), there is no 'direct objective measure of "true health"' (Jylhä, 2009: 313).Instead, different measures tap into different aspects of health.Greater attention is needed to theorizing how these different aspects relate to political behavior.The association between health and turnout generally holds, but the effects on other forms of political activity vary depending on how health is measured.Chronic illness is very different from temporary poor health.Moreover, as we have seen, the effects of different chronic diseases vary (Gollust and Rahn, 2015): some are easily managed with medication and lifestyle; others may cause severe pain or breathing difficulties.The effects of mental health also vary from study to study.We need to develop a deeper understanding of the different facets of health and political behavior and how they might interact, including whether self-rated health moderates the effects of chronic conditions.
Establishing causality is another major challenge.Most studies use cross-sectional data.Researchers assume that the causal arrow runs from health to political behavior, but participating in politics could promote well-being (Reitan, 2003).A British cohort study found that not voting increased the subsequent likelihood of poor self-reported health (Arah, 2008).It is unclear, though, whether this is because participating in the democratic process enhances well-being or because voting is an indicator of social capital.It is doubtful that elections occur frequently enough to have an enduring effect on health.Controlling for self-selection bias, another British longitudinal study found no association between voting and young people's subsequent physical health (Ballard et al., 2019).
In principle, panel data should show whether changes in health precede changes in political behavior.Longitudinal studies in the UK and Sweden have shown that declining health reduces the propensity to vote (Adman, 2020;Rapeli et al., 2020), but using panel data to address possible reverse causality comes with its own methodological challenges (Leszczensky and Wolbring, 2022).Another solution is to employ instrumental variables by selecting an exogenous variable that is strongly correlated with health, but does not directly influence political behavior and is unrelated to unobservables that could affect the latter.The challenge is finding valid instruments.A Canadian study used religious attendance as their instrument in assessing whether voting enhances self-reported health (Habibov and Weaver, 2014), but attending religious services has health benefits (Cahill and Ojeda, 2021) and is associated with higher turnout.Pacheco and Ojeda (2020) suggested that contextual variables, such as poor water quality or exposure to lead, could be valid instruments.
Selection bias also threatens causal inferences (Mattila et al., 2018;Söderlund and Rapeli, 2015).People with serious health problems may not have time or energy to respond to surveys and people who are uninterested in politics are unlikely to spend time answering questions about politics.The gold standard for combating selection bias and maximizing confidence in causal inferences is the randomized experiment.We obviously cannot randomize health, but piggybacking on randomized drug trials may be possible (Pacheco and Ojeda, 2020).In observational studies, matching healthy and unhealthy respondents on potential confounding factors could mitigate selection bias (see Ballard et al., 2019).The challenge is identifying-and having measures of-all the key variables that could affect both health and political behavior.
The final threat to causal inference is omitted variable bias.Health and political behavior may be associated because both are affected by genetic predispositions or childhood influences.By comparing the political behavior of monozygotic twins whose health status differs, twin studies could control for both shared genetic inheritance and shared environmental influences (Johnson et al., 2009).However, potential confounders include unshared environmental influences, such as peer groups and life experiences, and even monozygotic twins can differ genetically and epigenetically.Moreover, twin studies require access to a twin registry.An alternative is a discordant sibling design.Comparing the political behavior of siblings whose health differs mimics the logic of a controlled experiment; in effect, siblings from the same family serve as each other's control group (Burden et al., 2017;Gidengil et al., 2019;Lahtinen et al., 2019).Sibling designs control for unobservable characteristics that are shared by siblings in the same family, but unlike twin studies, only control for about 50% of genes that are shared by non-twin siblings.
These designs cannot prove causality, but they can enhance confidence in causal inferences.Using causal mediation analysis to test hypothesized causal mechanisms can also increase confidence, provided underlying assumptions are satisfied (VanderWeele, 2016).Focus groups and semi-structured interviews could be used to identify possible causal mechanisms based on people's insights into their own health and political behavior.

Concluding discussion
Despite these challenges, there is ample evidence that health can affect political behavior.Poor health can deprive people of the mental energy and physical ability to be politically active, diminish their psychological engagement with politics, and shrink their social networks.However, it may also mobilize people, as we have seen, especially if health conditions serve as a basis for a collective identity, and disease advocacy organizations encourage members to be politically active.
There are many possible future research directions.The association between poor health and support for radical right populist parties (see Kavanagh et al., 2021) highlights the need for further investigation of how health affects voting preferences and partisanship.In the US, partisan differences in evaluating the Affordable Care Act were attenuated when personal health concerns were primed (James and Van Ryzin, 2017), suggesting that health can cut across the deep partisan divides created by affective polarization.We need to know more about the impact of people's experiences with public healthcare.Policy feedback research highlights how people's encounters with government agencies and service providers can influence their perceptions of government responsiveness, sense of political agency, and political participation.Experiences with the healthcare system may have similar 'interpretive effects' (Pierson, 1993).Register data could be used to explore whether the intergenerational transmission of health inequality (Wilson and Shuey, 2019) contributes to the reproduction of political inequality across generations (Brady et al., 2015).The consequences of health inequalities for representation and responsiveness also warrant much more attention (Bernardi, 2021b), given evidence that 'political clout is not just about income, but also health' (Pacheco andOjeda, 2020: 1245).
We clearly cannot assume invariant relationships between health and political behavior across social divisions, yet research on intersections between health and social background characteristics, such as ethnicity, social class, and age, is in its infancy.Context matters, but how and why findings vary across countries needs attention, both theoretically and empirically.The Scandinavian experience indicates that even countries with generous welfare provision can exhibit health biases in political behavior.Very little is known about health and political behavior beyond the context of established postindustrial democracies.Finally, we located only one study on the effects of communicable diseases: turnout rates were lower in Finland and the US in regions where influenza was prevalent (Urbatsch, 2017).
More generally, the field could benefit from greater interdisciplinarity.Political scientists should collaborate with sociologists, health economists, psychologists, epidemiologists, and researchers in other health-related fields.Developing collaborations with scholars in the health sciences can be challenging, given differences in disciplinary norms, theoretical and methodological approaches, and work practices, but potential payoffs include better understanding of the physiology of different health conditions, data from randomized trials, and access to funding.Their input could prove vital for voter facilitation and ensuring the participation rights of those in poor health (see Brown et al., 2020).
The COVID-19 pandemic highlighted the importance of this research field, casting health disparities into sharp relief.Now, more than ever, we need to understand how unequal health translates into political inequality and how political inequality can result in a lack of responsiveness to the needs of those whose health is suffering, thereby perpetuating health disparities (Brown et al., 2020;Gollust and Haselswerdt, 2019;Ojeda, 2015;Pacheco and Ojeda, 2020;Rodriguez, 2018).Poor health is not just a personal challenge, but affects the functioning of the entire political system.A fuller understanding of the sources of health biases in the political process can bring us one step closer to the ideals of inclusive democracy.