Symbolic politics and self-interest in post-Affordable Care Act health Insurance coverage

Abstract

Social scientists have long shown that self-interest rarely has a large effect on citizens’ sociopolitical attitudes. Instead, Americans’ opinions about public policy are driven more by symbolic predispositions. Yet those opinions about policy may just be cheap talk. When Americans are faced with costly behaviors that pit their interests against their values, perhaps self-interest will play a larger role. Post-Affordable Care Act (ACA) health insurance coverage provides an important test for self-interest and symbolic politics. Faced with costly decisions to pay fines or purchase insurance, did Americans put symbolic politics over material self-interests? Using a monthly tracking poll, we first show that symbolic attitudes overwhelm self-interest in shaping public opinion about health policy. Marshaling data from over one million respondents from Gallup’s Tracking Poll, however, we show that both self-interest and symbolic predispositions are associated with decisions to purchase health insurance under the ACA.

Introduction

In March of 2010, President Barack Obama signed the Affordable Care Act (ACA) into law, the largest expansion of the social safety net since the 1965 enactments of Medicare and Medicaid. Republicans spent the remainder of Barack Obama’s tenure in office attempting to repeal the law—unsuccessfully voting over 60 times to fully or partially repeal the ACA.

In seeking to explain the loss, journalists proposed that expanded access to health benefits cultivated new constituencies with vested interests who would loudly advocate to keep the ACA in place and that the program would become gradually more popular over time. Similar arguments have been made with regards to other social programs including pensions (Pierson, 1992), veterans’ benefits (Mettler, 2002), welfare (Soss, 1999), and more recently Medicaid expansion under the ACA (Clinton and Sances, 2017).

Decades of political science research finds little evidence that self-interest plays a large role in most sociopolitical attitudes and behaviors, however. Instead this research points to symbolic attitudes (Sears, 1993) as more usually the primary determinants. Extant research has shown that self-interest plays a central role primarily when the political issue has clear and potentially large material effects on voters (Green and Gerken, 1989; Sears and Citrin, 1985; Sears and Funk, 1991). The unusual scope, salience, and politically polarized nature of the ACA provide social scientists with a strong test case for this latter contention.

In this article, we pool tens of thousands of respondents from the Kaiser Family Foundation Health Tracking Poll and over one million respondents from Gallup’s Daily Tracking polls spanning from 2008 to 2016, to test the relative importance of self-interest and symbolic predispositions in health care attitudes and enrollment both before and after the passage of the ACA. Consistent with the extant literature, we find clear and consistent evidence that symbolic predispositions are much stronger predictors of ACA attitudes than self-interest, but that self-interest plays an important role in enrollment decisions, drawing an important distinction between political attitudes and behaviors.

Self-interest or symbolic motivations?

Self-interest

Theories of rational choice, which for decades formed the theoretical core of many or most models of political attitudes and behavior (Arrow, 1951; Downs, 1957; Olson, 1965), are predicated on the idea that individuals efficiently pursue self-interested goals. These self-interested goals are often objective and material in nature—wealth, power, health, well-being, and re-election (Sears and Funk, 1991). ¹ It is difficult to overstate the impact that rational choice models have had on the study of politics. In one review, Ordeshook noted that no accomplishment has been more important than the “reintegration of politics and economics under a common paradigm and deductive system” (Ordeshook, 1993: 76). Riker insists that rational choice theory accounts for one of the only general advances to occur in political science (Riker, 1990: 177).

While intuitively plausible, and a strong starting point for theorizing about social behavior, a review of the literature finds little empirical evidence for many rational choice models (Green and Shapiro, 1994). Similarly, material self-interest only rarely influences political behaviors and attitudes (Citrin and Green, 1990; Lau and Heldman, 2009; Sears and Citrin, 1985; Sears and Funk, 1991). ² Shifts in life circumstances and objective self-interest—losing a job, having children, or becoming a victim of a violent crime—are not usually associated with drastic shifts in attitudes towards the size of the social safety net, childcare policy, or the carceral state, but are rooted instead in social values, group identifications, and crystallized predispositions. Self-interest tends to play an important role in socio-political attitude formation or change only when the stakes are high, clear, and certain, as is the case with tax policy and smoking (Chong et al., 2001; Sears and Citrin, 1985; Sears and Funk, 1991).

Symbolic predispositions

Alternatively, a substantial volume of empirical evidence has been amassed for a symbolic theory of politics. Theories of symbolic politics generally posit that political attitudes and behaviors are motivated primarily by emotion-laden enduring predispositions such as partisanship, ideology, and racial attitudes. These predispositions are acquired through a process of classical conditioning and often are stable through adult life (Sears et al., 1986). As adults are exposed to political objects, their attitudes towards that object are a product of its symbolic elements and the affect attached to those symbols (Sears, 1993). As such, many attitudes are formed reflexively, without calculations of personal cost or benefit.

Evidence for symbolic theories of politics has emerged across a variety of domains. With a series of papers from the 1970s through the 1990s, David Sears and colleagues found that: White Los Angeles residents who felt personally threatened by possible neighborhood segregation were not more likely to oppose Black city councilman Tom Bradley in the 1969 and 1973 mayoral elections (Kinder and Sears, 1981); Los Angeles residents personally affected by the energy crisis of 1974 were no more likely to support or oppose Richard Nixon, the president at the time (Sears et al., 1978); White parents of school-aged children in areas undergoing desegregation were no more opposed to busing than other Whites (Sears et al., 1979); working women were no more likely to support the Equal Rights Amendment than homemakers (Sears and Huddy, 1990); the unemployed do not disproportionately support job programs (Lau and Sears, 1981); and most importantly for our purposes, those with poor health insurance were no more likely to support government health care than those fully insured (Sears et al., 1980). Instead, these attitudes were strongly associated with symbolic predispositions such as partisanship, ideology, and racial attitudes. More recent work (Lau and Heldman, 2009) has largely reinforced these findings from the 1970s and 1980s.

Self-interest, symbolic politics, and health care attitudes

Despite this evidence, it is still plausible that a piece of legislation as large in scope and as consequential as the ACA could generate enduring self-interested support. Indeed, the ACA substantially affected millions of Americans, given the current cost of health insurance and health care. Over time, its impact became increasingly clear as Americans purchased health insurance under ACA regulations—the individual mandate and the uninsured rate were cut nearly in half from 15% in 2010 to 8.6% in 2016. Those conditions, as noted above, have been shown to generate self-interested support for policies.

Policies can also generate support and activism among interest groups and voters to maintain and expand benefits. Campbell (2003) shows how the development of Social Security energized senior citizens in the U.S., turning them from one of the least active to the most politically active age groups in American politics. Pierson (1996) argues that the expansion of the welfare state has generated enduring support from dense interest-group networks and strong popular attachments from recipients (but see Soss (1999) who argues that the design of the governmental program shapes how respondents come to view their own political efficacy and subsequent participation). Mettler (2002) shows how the GI Bill increased membership in civic organizations and spurred political engagement. Finally, recent research inds that Medicaid expansion under the ACA led to short-term bursts in participation among those in counties that benefited the most from the bill (Clinton and Sances, 2017) and an increase in support favorability toward the ACA among low-income Americans (Hopkins and Parish, 2019). Underlying these policy feedback studies is the idea that beneficiaries of certain policies will become politically engaged to protect their benefits. If this is true, we would expect support for such policies to correlate with objective self-interest after controlling for competing explanations.

Thus, ACA health care attitudes and health insurance coverage offer a particularly difficult test for the theory of symbolic politics. Indeed, many of the conditions thought to facilitate self-interested effects are present. And in the case of health insurance coverage, Americans could even be risking their health and financial well-being to voice their opposition to the ACA.

Data and methods

We begin by modeling support for the ACA. Because survey respondents can engage in “expressive responding” to attitudinal survey questions, in which they intentionally provide inaccurate responses as a way of showing support for their political viewpoints (Bullock et al., 2015; Schaffner and Luks, 2018; though see Berinsky, 2017), we also measure enrollment behavior, as a distinguishably different test of symbolic theories of politics.

To measure attitudes, we cleaned and pooled every publicly available Kaiser Family Foundation Health Tracking Poll from Cornell’s Roper Center (n = 80,969) that included all variables needed for our analysis. Our primary attitudinal dependent variable is favorability toward the ACA (1 = very favorable, 0.66 = somewhat favorable, 0.33 = somewhat unfavorable, and 0 = very unfavorable). For behavioral outcomes we cleaned and merged Gallup’s Daily Tracking Polls from 2008 to 2016 (n = 3,189,666) and use the variable measuring whether the respondent had health insurance or not (1 = insured and 0 = not insured). These large datasets allow us to estimate small effects with statistical precision. Each Kaiser poll that contained questions about support for the ACA, self-reported health, partisanship, ideology, and demographic controls was included in the final dataset. Full poll methodology for both the Kaiser and Gallup polls are available in Online Appendix 1.

Our primary independent variables in both datasets include self-reported health (self-interest) and partisanship and ideology (symbolic predispositions). ³ Control variables include income, age, education, gender, race, and marital status. All analyses include state fixed effects to control for state-level differences. We include a number of additional robustness checks using a broad variety of independent and dependent variables that tap into several different operationalizations of self-interest and support for the ACA. All models are restricted to respondents under 65, and thus ineligible for Medicare. Details on all variables used in this study can be found in Online Appendix 2.

Health Care Attitudes

We begin by modeling support for the ACA in the Kaiser polls using multivariate regression (ordinary least squares). ⁴ In Figure 1 we display the effect of moving each variable from its in-sample minimum to maximum on support for the ACA in the full pooled dataset, before the ACA rolled out, and after the ACA rolled out, holding all other variables at their means, and relegate the regression tables to Online Appendix 3.

Figure 1.

Effects of self-interest and symbolic politics on attitudinal opposition to the Affordable Care Act (ACA).

Consistent with symbolic theories of politics, we find clear evidence that the two central symbolic predispositions, party identification and ideology, are far more predictive of attitudes toward the ACA than is self-interest, operationalized as an individual’s overall health. While moving from strong Democrat to strong Republican is associated with a 0.45 point increase in opposition to the ACA (95% confidence interval (CI): [0.43, 0.46]), and from strong liberal to strong conservative with a 0.19 point increase (95% CI: [0.17, 0.20]), moving from best to worst health is associated with a substantively tiny but positive increase in opposition to the ACA (0.02, 95% CI: [0.01,0.04]). Furthermore, the associations did not change after the ACA was enacted and benefits began. In Online Appendix 4 we show that our results hold when we operationalize self-interest using objective health and economic self-interest measures. ⁵

Health care enrollment

While we find strong evidence for symbolic theories of politics over self-interest in public opinion about health care, we next subject the theory to a different test, examining the relationship between symbolic politics and self-interest with actual enrollment in the ACA in daily Gallup Tracking poll data. We use a logistic regression to test whether respondents have health insurance (1) or not (0) as a function of their partisanship, general health, a host of demographic controls, and state fixed effects. We interact all covariates with a dummy indicator for post-January 2014, to estimate the average change in insurance rates moving from pre- to post-enactment at the state level. The full regression can be found in Online Appendix Table 3.4.

In Figure 2 we plot the change in enrollment, as measured in the Gallup polls, from before the enactment of the ACA (January 1, 2014) to after enactment as a function of the full range of partisan categories (symbolic predispositions) and personal health categories (self-interest), holding all other variables at their means. Looking first at the top panel, we see that those in worse health were far more likely to enroll post-ACA enactment than those in excellent health. On average, coverage increased about 3.4 percentage points for those who rated themselves in “poor health” compared to just 1.6 percentage points for those in “excellent health,” a difference of 1.8 percentage points (p < 0.001). This suggests that self-interest, operationalized as the state of one’s current health, does matter more in deciding to get insurance post-ACA than it does for attitudes. Looking next at the second panel, partisanship, we see that coverage increases about 2.7 percentage points among strong Democrats compared to just 1.3 percentage points for strong Republicans, a 1.4 percentage point difference. ⁶ In Online Appendix 4 we replicate these findings using objective rather than subjective well-being.

Figure 2.

Self-interest and symbolic politics in the Affordable Care Act enrollments pre–post January 1, 2014.

Conclusion

In seeking to foster opposition to the establishment and adoption of large government programs, politicians will frequently argue that social programs can generate new constituencies with vested interests in the programs’ success (Campbell, 2003). Indeed, political journalists frequently argued that failure to repeal and replace the ACA was due to the fact that over 20 million Americans who had gained health insurance under the law grew to favor its provisions and it served as a public opinion check on repeal efforts. This self-interest model of health care attitudes and politics is frequently featured in journalism on the issue.

While some researchers have found that the ACA may have led to a short-term increase in political participation (Clinton and Sances, 2017), and a small bump in support for the law among lower-income Americans who gained Medicaid coverage (Hopkins and Parish, 2019), we find little evidence that objective self-interest is a primary driver of health care policy attitudes in the U.S. Indeed, consistent with past research on health care attitudes, we find that Americans rely strongly on their symbolic predispositions such as ideology and partisanship to shape their attitudes towards health care and health care behaviors (Lerman et al., 2017).

When put to a harder test, however, we also find that self-interest also plays an important role in health-related behavior. In this case, we find that self-interest is associated with acquisition of health insurance after the passage of the ACA. While these effects sizes appear small, with almost 200 million adults under the age of 65 in the U.S., very small percentage point shifts correspond to very large numbers of individuals.

Supplemental Material

SUPPLEMENTAL_INDEX – Supplemental material for Symbolic politics and self-interest in post-Affordable Care Act health Insurance coverage

Supplemental material, SUPPLEMENTAL_INDEX for Symbolic politics and self-interest in post-Affordable Care Act health Insurance coverage by Tyler T. Reny and David O. Sears in Research & Politics