Health, Disability and Political Attitudes: Advancing Theory and Methodology

• In Event: Health, Health Policy Design, and the Effects on Political Attitudes & Behavior

Abstract

Objectives

1. We develop a novel theoretical framework for understanding and testing how experiences with chronic illness and disability may shape political attitudes and preferences.
2. Drawing from the literature on racial inequities, we propose a set of best practices for selecting covariates when exploring the relationship between health/disability and political attitudes.
3. We apply these theoretical and methodological advancements to characterize the political attitudes and preferences of Europeans in poor health and disability.

Motivation/Theory

As populations age and recover from COVID-19, chronic illness and disability may become increasingly important in electoral politics. While illness and disability have been shown to suppress political participation, how they shape political attitudes remains unclear. This population benefits from redistributive social policy, so they may support government action. However, they are also more likely to face isolation and administrative burden, driving lower public trust and animosity towards out-groups.

Health has been understudied as an independent variable in politics and important methodological issues remain unaddressed. For example, given the bidirectional relationship between health/disability and socioeconomic status, what control variables are appropriate without inducing post-treatment bias? Should models permit these predictors to vary by membership in other groups or identities? How can researchers responsibly explore causality given that health and disability are usually not manipulable under experimental and quasi-experimental study designs? And when health and disability status are self-reported (as in our data), how should researchers manage the risk that self-reported health is endogenous (i.e. a function of observed and unobserved variables that may also impact our outcome)?

In response, we develop a theoretical framework based on political trust and social deprivation models to understand these relationships, and propose best practices for methodological concerns.

Empirical Design

We use the European Social Survey, a serial cross-sectional survey of social and political attitudes from 2002 to 2020 across 38 countries. With over 400,000 respondents, we use fixed effects OLS models to explore the relationship between illness or disability and a variety of attitudes on economic policy (about redistribution and caring for the poor) and social policy (about immigrants, queer rights, and women’s rights). We also explore the empirical implications of different sets of control variables with mediation analyses to both articulate promising methodological practices and hypothesize about potential causal relationships. Lastly, we make use of random slope models to permit these relationships to vary by intersectional social strata, enabling us to efficiently explore potential confounders.

Results/Implications

We find that Europeans with worse health and disabilities are more economically progressive and socially conservative. After adjusting for demographic characteristics, they are modestly more supportive of redistributive policies yet considerably more opposed to immigrant and gender minority rights. Thus, poor health and disability appear to meaningfully predict political attitudes. These findings may help explain why this population has been shown to support far-right populist parties in prior work.

More generally, we situate these findings in a broader theoretical framework built on political trust and social deprivation models. We also discuss a set of unanswered methodological questions and outline promising practices for the space. Although our design is not causal, we consider potential avenues for addressing concerns of endogeneity when examining the effects of poor health and disability, drawing on the literature on measuring the causal effects of race/ethnicity and inequities. Lastly, by using random slope models, we can assess the extent to which our findings hold across a wide range of population subgroups, offering an efficient way to examine many potential demographic confounders.

Authors

• Nolan M Kavanagh, Harvard University

• Ari Neeman, Harvard University