Ten years after its enactment, public support for the Affordable Care Act (ACA) still only reaches a scant majority. Candidates for the presidency—and the sitting president—have endorsed health reforms that would radically transition US health care away from the current system upon which the ACA was built. Few opinion surveys to date have captured dominant preferences among alternative health reform policies or characterized attitudes and experiences that might be associated with policy preferences. Using a 2019 nationally representative telephone survey, this article considers how variations in political values, attitudes toward government, and experiences with the health care system relate to competing health reform preferences. Differences between those who favor Medicare for All over building on the ACA largely reflect different levels of satisfaction with the status quo and views of private health insurance. By contrast, differences between ACA supporters and those who would favor replacing it with a state-based alternative reflect sharply different political values and attitudes. Key differences remain significant after controlling for demographic, health, and political characteristics. Overwhelming public support still eludes the ACA, and reaching consensus on future directions for health reform will remain challenging given differences in underlying beliefs.
In the decade since the Affordable Care Act (ACA)’s passage, approval of the law has climbed modestly, but consensus levels of public support remain elusive. Whereas the decades following the enactment of Medicare and Medicaid were characterized by “politics of consensus” (Oberlander 2003), the politics of President Barack Obama's signature legislative achievement have been relentlessly turbulent (Oberlander 2018; McIntyre and Song 2019).
Congressional Republicans have voted dozens of times to repeal the ACA in whole or in part and have taken other actions to weaken its implementation, believing the law to be an unwarranted intrusion of the federal government into the health care system (Redhead and Kinzer 2017; Pear and Kaplan 2017). The most thorough replacement plan put forward would replace the ACA with block grants to the states; this proposal was deeply unpopular and never came to a vote (Nelson 2017). In addition to legislative repeal, since 2016, President Donald Trump and many of his allies have sought to weaken the law through regulatory avenues and a lawsuit challenging the ACA's constitutionality (Sanger-Katz 2020). That case, which has the potential to strike down the ACA in its entirety, will be heard by the Supreme Court in the fall of 2020 (Liptak and Goodnough 2020).
Across the political aisle, some hold that the law did not go far enough and that the health insurance system demands reinvention, not incremental reforms. Ten years after the ACA's enactment—and just two years after their party put it center stage in a successful bid to recapture the US House of Representatives (Fowler, Franz, and Ridout 2020)—several Democratic presidential candidates offered Medicare for All as an ambitious replacement. Senator Bernie Sanders (D-VT) released legislation in 2019 that would replace the current patchwork of insurance schemes with a comprehensive national health plan; Senator Elizabeth Warren (D-MA) also endorsed this approach. More moderate candidates, including former vice president and (as of this writing) presumptive Democratic nominee Joe Biden, hewed to policies that would build on the ACA by making coverage more generous and introducing a “public option” that people could choose instead of private insurance (Scott 2019).
Public Opinion and the Affordable Care Act
The ACA's resilience in this environment has come despite underwhelming public support. According to the Kaiser Family Foundation's monthly tracking poll, the ACA's overall favorability vacillated between 36 and 45% between 2010 and 2016 (KFF 2020). In recent years, approval ratings have risen above the halfway mark, peaking at 55%. In late 2017, Pew reported that, for the first time in their experience fielding the question, more Americans thought that the ACA had a mostly positive effect on the nation than a mostly negative effect; at the same time, most Americans reported that they were not directly affected by the ACA (Pew Research Center 2017).
The ACA has experienced its highest sustained approval ratings since the 2016 election, likely buoyed by the repeal efforts that followed. On average, 49% of Americans have reported favorable views of the law since November 2016, compared to an average favorability rating of 39% between October 2013 and October 2016 (Brodie et al. 2020).
National ACA favorability ratings obscure sharp partisan divisions that have widened over time (Brodie et al. 2020). Whereas 84% of Democrats approved of the law in January 2020, only 55% of independents and 19% of Republicans expressed approval (KFF 2020). Understanding public opinion is further complicated by the fact that these divided views on the law as a whole belie wide, bipartisan support for many of its central provisions. Although the individual mandate was a contentious feature, majorities of Americans across the political divide favor protections for individuals with preexisting conditions, eliminating out-of-pocket costs for preventive care, and the creation of health insurance exchanges (Kirzinger, Wu, and Brodie 2018; Blendon and Benson 2017).
Elite messaging and partisan motivated reasoning likely explain some of this discrepancy (Dancey and Goren 2010; Druckman, Peterson, and Slothuus 2013; Fowler et al. 2017), but it is important not to overlook the fact that most Americans believe the law did little or nothing to improve their own lives. The share of Americans reporting that the law has made things somewhat or much better for themselves nearly doubled between 2010 and 2018, but at 20% this remains a minority opinion (Jacobs, Mettler, and Zhu 2019).
While the ACA precipitated a historic drop in the nation's uninsured rate, an overwhelming majority (84%) of Americans already had health insurance coverage when it passed (DeNavas-Walt, Proctor, and Smith 2011). Polls consistently show that Americans' top priority is lowering their own health care expenses (Jones and Reinhart 2018; Blendon, Benson, and McMurtry 2019; Kirzinger et al. 2019). Individuals with employer-based health insurance have seen cost-sharing obligations grow over the past decade, a trend that predates the implementation of health reform (Claxton et al. 2019). Despite the new coverage expansions, consumer protections, and sources of financial assistance introduced by the ACA, Americans may not generally associate the law with lower personal health costs.
As we reach the close of the ACA's first decade, important questions remain: why do vocal calls for sweeping—though sharply different—alternatives to the ACA remain, even as the law has embedded itself into the fabric and function of the United States health care system? And what factors might explain continued divisions in public preferences for health reform?
Overall approval of the ACA and its component policies is crucial to understanding the current moment in health policy but is only one window into understanding the current debates in health policy: half of Americans report favorably viewing the ACA, but some more strongly favor alternative approaches to health reform. This article uses a probability-based national survey requiring respondents to make a choice about which policy option they most prefer: building on the Affordable Care Act, moving the United States toward Medicare for All, or replacing the ACA with state health plans. This survey is the first, to our knowledge, to elucidate preferences among these three options and also to ask about political values and experiences with the health care system.
Study Design and Sample
Data were obtained from an original telephone survey designed collaboratively between the Harvard T. H. Chan School of Public Health, the Commonwealth Fund, and the New York Times. The nationally representative, probability-based survey was administered by SSRS from July 10, 2019, through August 10, 2019. The survey reached respondents on both landlines (500 respondents) and cell phones (1,505 respondents).
Up to five follow-up attempts were made for nonresponsive numbers, varying the day of the week and time of day to maximize the survey response. The overall response rate for this survey was 8%, as calculated using the American Association for Public Opinion Research's (AAPOR) RR3 formula (AAPOR 2016). Because data from this study were drawn from a probability sample and used the best available sampling and weighting practices in polling methods, they are expected to be reliably generalizable to the broader US population, with a margin of error of ±2.5 percentage points at the 95% confidence interval.
Sampling weights were constructed to account for variations in selection probability (both across and within households), oversampling, and nonresponse. Weights accounted for demographics (age, gender, educational attainment, race and Hispanic origin, and census region), political affiliation, cell phone and landline use, and household size.
The full research sample included 2,005 American adults 18 or older. The analyses in this article focus on the subset of 1,779 respondents who expressed a preference among three health reform approaches at the beginning of the survey. The remaining 226 respondents were excluded from analysis if they did not support any of the three approaches or did not know which approach they preferred most.
Our poll asked respondents to identify which policy they most preferred among three options: keeping the existing Affordable Care Act, also known as Obamacare, and passing additional legislation to improve how it works (shorthand: keep/improve ACA); changing the health care system so that all Americans receive health insurance from Medicare, a plan often called Medicare for All (shorthand: Medicare for All); or, replacing the Affordable Care Act with a new law that would give taxpayer funding to states to design their own health insurance systems with fewer federal rules (shorthand: state-based option). This last alternative was intended to reflect legislation promoted by congressional Republicans during the 2017 repeal effort (Antos and Capretta 2017). The order of these three policy options was rotated at random to mitigate against order effects bias.
The poll also elicited information across the following four domains: (1) general political values, (2) attitudes toward government and health care, (3) general assessment of the current health care system, and (4) satisfaction with personal experiences with the health care system. Full question wording is available in the online appendix.
We classified respondents into three groups based on their preferred approach to health reform: keep/improve ACA, Medicare for All, or state-based option. After calculating survey-weighted descriptive statistics for the sample, we compared the responses of the three “policy preference” groups on items across the four “values, attitudes, and experiences” domains (see above). We used pairwise t-tests of differences in proportions to determine when keep/improve ACA respondents differed significantly from Medicare for All or state-based option respondents.
To determine whether these associations remained significant after adjustment for other factors, we estimated a multinomial logit model for each of the four domains, with policy preference as the dependent variable, and each variable of interest as an independent variable. Multinomial logit specifications yield relative risk ratios, which are analogous to odds ratios in logistic regressions: values greater that 1 indicate that a variable is associated with higher likelihood of falling into an alternate categorical outcome relative to the base outcome; values lower than 1 indicate a lower likelihood. We set keep/improve ACA as the reference group; the resulting relative risk ratios estimate the likelihood of favoring each alternative policy option (Medicare for All or state-based option) over the keep/improve ACA option.
These models adjusted for variation due to self-reported demographic and health characteristics, insurance status, and political characteristics. Demographic covariates were gender, age (18–29, 30–49, 50–64, ≥ 65), race/ethnicity (Hispanic, non-Hispanic black, non-Hispanic white), household income (< $25,000; $25,000–$49,999; $50,000–$74,999; $75,000–$99,999; ≥ $100,000), education (less than college degree or college graduate), metropolitan status (urban, suburban, rural), and geographic region. Using self-reported health insurance information, respondents were characterized as having coverage primarily through a private insurer, primarily through a public insurer (Medicare or Medicaid), or being uninsured. Political variables were party affiliation (Republican, Democrat, Independent) and ideology (very conservative, somewhat conservative, moderate, somewhat liberal, very liberal). To ease interpretation of the resulting relative risk ratios, we dichotomized categorical variables with four levels into two (e.g., very/somewhat satisfied vs. somewhat/very dissatisfied).
Some questions on values, attitudes, and experiences were only fielded to random half samples of respondents; as a result, adjusted models reflect smaller sample sizes than unadjusted results. Sampling weights were used in all statistical procedures to generate nationally representative estimates. All analyses were conducted in Stata 16.0 (Statacorp) and statistical significance was determined at p < .05.
General Sociodemographic and Health Characteristics
Demographic, health, and political characteristics of this nationally representative sample are displayed in table 1, by favored approach to health reform. Respondents' most-favored approaches to health reform were split among those who most favored keep/improve ACA (28%), Medicare for All (32%), or a state-based option (29%). Compared to Americans who favored keeping/improving the ACA, those who preferred Medicare for All were less likely to be black (14% vs. 19%, p < .05), less likely to have attained a college degree (31% vs. 39%, p < .01), and more likely to report being uninsured 11% vs. 6%, p < .05). Compared to those favoring the ACA, people favoring the state-based option were more likely to be male (57% vs. 43%, p < .01) and white (75% vs. 54%, p < .01), and less likely to have attained a college degree (39% vs. 31%, p = .01). Those favoring the state option were also less likely to report a chronic condition (41% vs. 49%, p = .01), disability (83% vs. 77%, p < .05), or fair/poor health (17% vs. 23%, p < .05).
General Political Values
In uncontrolled comparisons across groups for general attitudes toward government, there were some notable differences between respondents who favor keep/improve ACA and other groups (table 2). While there were similarities in general attitudes toward government between respondents favoring keep/improve the ACA and those favoring Medicare for All, compared with respondents who favor keeping the ACA, those who prefer Medicare for All were more likely to report having positive images of socialism (57% vs. 40%, p < 0.001) and less likely to report positive images of capitalism (41% vs. 54%, p < 0.001). Compared with respondents favoring keep/improve ACA, those favoring a state-based option held more negative attitudes toward government on every measure, including saying things run by the federal government are not run too well or not well at all (69% vs. 48%, p < 0.001).
After we controlled for relevant sociodemographic, health insurance, health status, and political characteristics in a multinomial logistic regression model, some group differences in general attitudes toward government persisted (fig. 1). Compared with keeping/improving the ACA, holding positive views of socialism were associated with a higher likelihood of favoring Medicare for All and lower likelihood of favoring the state-based option in the adjusted model, and believing that government regulation usually does more harm than good was associated with a higher likelihood of favoring the state-based option.
Assessment of the Current Health Care System
Unadjusted differences in views on the current health system are presented in table 3. While respondents who favored Medicare for All and keeping/improving the ACA both had strongly favorable attitudes toward Medicare and Medicaid, those favoring Medicare for All were more likely to rate the US health care system as fair or poor (75% vs. 66%, p = .01) and more likely to have an unfavorable opinion of employer-based insurance (41% vs. 27%, p < .001). In contrast, people favoring the state-based option over the ACA were more likely to rate the health care system as good or excellent (48% vs. 33%, p < .001), less likely to have unfavorable views of employer-based insurance (14% vs. 27%, p < .001), and more likely to have unfavorable views of Medicaid (32% vs. 18%, p < .001). Both groups favoring alternative reforms were significantly more likely to have unfavorable views of the ACA than the keep/improve ACA group were, but this was a minority view for those who favored Medicare for All (24%) and a majority view for those who favored the state-based option (75%).
After adjusting for sociodemographic, health, health insurance, and political characteristics, rating the health system as fair or poor remained associated with higher likelihood of supporting Medicare for All over keep/improve ACA, as was viewing employer-based insurance unfavorably and viewing the ACA unfavorably (fig. 2). After adjustment, some differences between the ACA group and the state-based option group related to ratings of the US health system and favorability were no longer significant. However, viewing the ACA unfavorably remained associated with a higher likelihood of favoring a state-based option over keep/improve the ACA.
Personal Experiences with the Health Care System
In unadjusted comparisons, levels of dissatisfaction with health care experiences were consistently higher among those favoring Medicare for All compared with those who would rather keep/improve the ACA (table 4). They were more likely to report being dissatisfied with their ability to get health care when they need it (34% vs. 18%, p < .001), quality of care (24% vs. 15%, p < .001), costs of care (56% vs. 44%, p < .001) and, among the insured, their current insurance coverage (20% vs. 9%, p < .001). They were also more likely to report being worried about paying medical bills (63% vs. 51%, p < .001). Compared the ACA group, those favoring a state-based option were more likely to be dissatisfied with their current insurance (15% vs. 9%, p = .01), but less likely to worry about medical bills (33% vs. 51%, p < .001).
Some differences remained significant in an adjusted multinomial logit model (fig. 3). Dissatisfaction with health care access and with one's own insurance coverage (among the insured) were associated with higher likelihood of favoring Medicare for All over keeping/improving ACA. Worrying about medical bills was associated with a lower likelihood of favoring a state-based option.
Attitudes toward Government and Health Care
In unadjusted comparisons, marked differences in attitudes toward government and health care emerged between repsondents favoring keep/improve the ACA and the other groups (table 5). Those favoring Medicare for All were more likely than those of the keep/improve the ACA group to believe it is the government's responsibility to guarantee coverage (85% vs. 73%, p < .001), more likely to be willing to pay more in taxes to support health coverage expansion (79% vs. 68%, p < .001), and more likely prefer a health insurance system run mostly by the government (73% vs. 55%, p < .001), but these views were held by more than half of each group.
Differences were wider between the keep/improve ACA and state-based option group. Those favoring a state-based option were less likely to believe it is the government's responsibility to guarantee health coverage (20% vs. 73%, p < .001), and less likely to express willingness to pay higher taxes to guarantee coverage expansion (23% vs. 68%, p < .001). Differences also emerged on questions that were not notable when comparing the ACA and Medicare for All groups: compared to the keep/improve ACA group, respondents favoring a state-based option were less likely to say all Americans have a right to health care, regardless of ability to pay (60% vs. 92%, p < .001), and less likely to say government health care programs should be available to anyone living in the US, regardless of immigration status (13% vs. 49%, p < .001).
Some differences persisted in an adjusted model controlling for demographic, health, health insurance, and political characteristics (fig. 4). Respondents who preferred a system based mostly on private insurance had higher likelihoods of preferring a state-based option (and lower likelihoods of favoring Medicare for All) compared to the keep/improve the ACA group. Those who would rather pay higher taxes or premiums in exchange for lower cost-sharing when seeking care had higher likelihoods of preferring Medicare for All (and lower likelihoods of preferring the state-based option) over keep/improve the ACA. Respondents believing it is not the government's responsibility to guarantee health coverage, those less willing to pay taxes to expand health coverage, and those who believed that government programs should only cover legally present US residents had higher likelihoods of preferring the state-based option over keeping and improving the ACA.
In adjusted models, political and ideological covariates remained significantly associated with reform preferences across most domains. Identifying as somewhat or very conservative (compared to moderate) was associated with higher likelihood of preferring the state-based option over the ACA in models for three of the four domains. Identifying as very liberal (compared to moderate) was associated with preferring Medicare for All over the ACA across models for all four domains, while identifying as somewhat liberal (compared to moderate) was associated with a higher likelihood of preferring Medicare for All over the ACA across two models. Compared to identifying as a Democrat, Republican identification was associated with higher likelihood of favoring the state-based option over the ACA in all models. Refer to the online appendix for full model results across all four domains.
Ten years after its enactment, keeping and improving the Affordable Care Act is the most-preferred health reform direction for only one-third of Americans. The remaining two-thirds of Americans are equally divided between wanting to replace the ACA with more sweeping Medicare for All reforms and preferring to give taxpayer funding and authority to the states. Respondents tend to be satisfied with their own health insurance across all groups despite shared dissatisfaction with health costs, but differences become apparent when examining other measures of satisfaction with the health care system and political values and attitudes.
Beyond differences in political ideology, our results showed respondents who want to build on the ACA are more content with the status quo and their own experiences with the health care system when compared to those who favor Medicare for All, and their reluctance to endorse more far-reaching federal reforms may be related to loss aversion (Mound 2018; Klein 2019). For example, respondents favoring keeping the ACA are more likely than Medicare for All proponents to rate the US health system highly, to view private and employer-sponsored insurance favorably, and to prefer a system based mostly on private insurance, even after adjusting for demographic characteristics, health status, health insurance, and political characteristics. Although there are discernible differences in some attitudes toward government and health care, a majority of ACA supporters believe it is the government's responsibility to make sure all Americans have health insurance coverage and would be willing to pay more in taxes so that everyone can have health insurance—these majorities are just larger among those who favor Medicare for All.
In terms of satisfaction with their own experiences with the health care system, people who want to keep the ACA are more similar to people who favor replacing the ACA with a state-based option than to those who prefer Medicare for All. However, the state-based option and keep ACA groups diverge sharply not only on political partisan affiliation but also on views about whether health care is a right, the appropriate role and scope of government in health care, willingness to pay higher taxes to finance a more generous health care system, and whether government health care programs should be available to anyone living in the United States, regardless of immigration status. Large differences between these groups remained significant in adjusted models.
In a political age where some progressive Democratic political leaders are agitating for Medicare for All (Johnson, Kishore, and Berwick 2020), our results suggest that current ACA supporters may shift to greater support for Medicare for All if they become disenchanted with their own coverage, or if other circumstances or policies erode enthusiasm for private, employer-sponsored coverage. Views differentiating the third of Americans who favor devolving health care to the states are of a different flavor entirely: they are much more staunchly opposed to growing governmental power and paying taxes to support universal coverage. These attitudes will likely prove more intractable.
Several limitations should be considered when interpreting our results. Self-reported data are subject to recall bias and other human error. Respondents may have been subject to social-desirability bias in responding to questions. Question order effects could be a concern; asking respondents about their health reform preferences before other survey questions could affect responses. Nonresponse bias is another concern in public opinion surveys, though evidence suggests that low response rates do not bias results if the survey sample is representative of the study population (Kohut et al. 2012; Keeter et al. 2017). Recent research has shown that such surveys, when based on probability samples and weighted using census parameters, yield accurate estimates in most cases when compared with both objective measures and higher-response-rate surveys (Keeter et al. 2006; Yeager et al. 2011; Kohut et al. 2012; Keeter et al. 2017). However, it is possible some selection bias may remain related to the experiences and views measured. Sample size also limited our ability to estimate adjusted models, particularly where questions were only fielded to a subset of respondents; coefficients should be cautiously interpreted.
Due to limitations in survey length, we did not elaborate on policy options presented to respondents. Respondents may have found options more or less favorable depending on policy framing (Grande, Gollust, and Asch 2011), ordering, or provision of additional information (Hamel, Wu, and Brodie 2017). Questions about attitudes toward government did not indicate a level of government; specifying federal, state, or local government could affect reported attitudes. Lastly, because this was designed as a cross-sectional study, results should be carefully interpreted and readers should not infer causality from these results.
This survey was fielded in mid-2019. Accordingly, it is unclear whether or how the COVID-19 pandemic, which has had widespread consequences for the US economy and health care system, may have changed public attitudes and opinions evaluated in the present study.
Despite these limitations, this study was strengthened by its probability sampling design and the breadth of questions asked on values and experiences related to favored approaches for health reform.
The Affordable Care Act has become a fundamental part of the US health care system in its first ten years. However, public support remains middling, with approval at 50% in April 2020 (KFF 2020). Given a choice, only about one-third of Americans most favor keeping and improving on the law; the other two-thirds are split between supporting more expansive Medicare for All reforms and letting states take the lead. Results from this survey make evident that the views, values, and attitudes that comprise each of these groups different from those who support maintaining and strengthening the ACA are not the same. Respondents who favor Medicare for All share many common values and attitudes with respondents who want to build on the ACA—and most have a favorable view of the ACA itself—but tend to be more dissatisfied with their own experiences with the current health care system, including their health care costs. In contrast, those favoring a state-based option tend to have similar levels of satisfaction as those who want to build on the ACA, but starkly different opinions on taxation and the broader role of government in health care. A broad consensus on future directions for health reform will likely continue to be difficult to achieve given these differences in underlying beliefs. However, the one key thing those who support improving the ACA and those favoring Medicare for All agree on is the need to move toward universal coverage.