A polling experiment on public opinion on the future expansion of Medicare and Medicaid

Abstract

Objective

To conduct a polling experiment to understand the possible framing effects that drive constituents’ views around Medicare For All (MFA) and Medicaid Buy‐In (MBI).

Data Sources and Study Setting

Five thousand and fifty‐one US adults aged 18 and older were recruited to participate in an online poll conducted between September 12, 2018, and September 26, 2018.

Study Design

Participants were randomized to receive one of four polls: (a) a poll measuring respondent approval for MFA, with the name of the proposal stated with a description; (b) a poll measuring approval for MFA, with only a description of the proposal; (c) a poll measuring approval for MBI, with the name stated with a description; or (d) a poll measuring approval for MBI, with only a description.

Principal Findings

Including the names “Medicare For All” and “Medicaid Buy‐In” increases approval by 3.4 (from 32.7 percent to 36.1 percent) and 5.0 (from 50.1 percent to 55.1 percent) percentage points, respectively. Support varies by age, where MBI is most strongly supported by Millennials, while Baby Boomers and those older than 65 are more likely to support MFA.

Conclusions

Constituents are more likely to support a proposal when given the names of the proposal. Approval is also higher for health policies that are framed as expansions of existing policies than as new programs.

What This Study Adds.

• Medicare For All (MFA) and Medicaid Buy‐In (MBI) are two policy proposals that would extend two popular and federally financed programs (Medicare and Medicaid, respectively) with the stated aim to achieve universal health insurance coverage. However, there is significant and persistent confusion over many of the key provisions that are stated as part of these proposals.

• Public support for or opposition to MFA and MBI may be driven more by (mis)perceptions and biases around the framing of these reforms than by informed opposition to the provisions that are established within these reforms themselves.

• Through our polling experiment, we find that constituents are more likely to support a proposal when the name of the program is explicitly used to describe the policy; similarly, disapproval for the proposal also declines when the name of the proposal is included.

• Reframing policy proposals as expansions of existing popular programs may serve to: (a) destigmatize both the proposal and the program, and (b) increase public support, and possibly even take‐up, even among enrollees who say that the program is important to them.

1. INTRODUCTION

Considerable polling has been conducted to better understand public opinion around the government's role in providing access to health care. Findings from these polls indicate that a majority of the public (60 percent) consistently supports the idea that the federal government has a responsibility to ensure that people have access to health care. ¹ , ² This perception changed during the passage of the Affordable Care Act (ACA), which faced opposition from a range of groups, including individuals who believed that the law did not go far enough to provide adequate coverage and believed that the government should act as the single payer. ³ Since the passage of the ACA, public favorability for government responsibility for the provision of health care has increased. ¹

Politicians have recently proposed plans that would extend the role of the government in the provision of health care. Two of these policy proposals, Medicare For All (MFA) and Medicaid Buy‐In (MBI), would extend two popular and largely federally financed programs, Medicare and Medicaid, respectively, with the stated aim to achieve universal health insurance coverage. MFA would replace the existing health financing care system with a federally run single‐payer plan, while MBI would allow uninsured individuals with moderate or high incomes who are currently ineligible for Medicaid to purchase Medicaid coverage while maintaining the current Medicare, employer, and Medicaid systems for those who are currently eligible for the program. ⁴

Several polls have been conducted to better understand public perception of Medicare and Medicaid and of proposals for their expansion. Results from these polls have found generally high support for both programs, with slightly more support for Medicare. The partisan divide in both of these health care plans is strong, and a 2017 poll by the Kaiser Family Foundation found the divide to be stronger on Medicare expansion proposals than on Medicaid expansion proposals despite the fact that there was greater partisan support for Medicare than for Medicaid. ⁵

Findings from national polls suggest that both MFA and MBI proposals hold relatively high levels of support; however, there is significant and persistent confusion over many of the key provisions of these plans, including how these program proposals would be rolled out, key beneficiaries from these proposals, how the government would be involved in their expansion, and how these expansions would be financed. ⁶ , ⁷ Recent polling has showed that nearly 40 percent of Medicare beneficiaries and 30 percent of Medicaid beneficiaries reported never having received a government benefit or using government program; this finding is reflective of similar confusion in public awareness of the ACA, where a recent poll found that more than one‐third of Americans did not know that Obamacare and the ACA referred to the same policy. ⁸ , ⁹ Public opinion has often been shaped by the strong association between names and their positive or negative perception, as has been illustrated through research on names ¹⁰ —in this case, the name of the Medicaid program is predominantly identified and recognized at the local or regional level, which may contribute to an underestimation of the level of support for the program. ¹¹

Research on support for single‐payer plans has shown that people are generally supportive of such plans until they are given additional details about how the plan will be financed and about the role of the government as a care provider. ¹² One key reason that Medicare has a positive public perception is that the program is available to everyone and is earned through employment ¹³ ; however, it is unclear as to which of these two features for eligibility are crucial to its popularity. Similarly, Medicaid has had public support, but it has also arguably been more politically divisive than other public programs. In addition, it is unclear as to how the means‐tested criteria for eligibility into the program, which would facilitate targeting of services to the poor, make it attractive. Taken together, these findings suggest that public support for or opposition to new reforms may be driven more by (mis)perceptions and biases around the framing of these reforms than by informed opposition to the provisions that are established within these reforms themselves. Of particular interest is the extent to which the inclusion of the words “Medicare” or “Medicaid” within the MFA and MBI proposal names and descriptions would contribute to public support independently from the content of the proposal descriptions.

1.1. Study objectives

In this spirit, we conduct a polling experiment to test how the framing of these proposals, specifically how the naming of the proposal as “Medicare For All” or “Medicaid Buy‐In,” could be driving constituents’ views around and reasons for opposition to both MFA and MBI. We explore the heterogeneity in favorability of these polls across a range of factors and study how support for Medicare and Medicaid also may drive support for the expansion of these programs. Our key hypotheses are as follows:

1. Given the public's level of familiarity with Medicare and Medicaid, perceptions of these programs will significantly shape their opinions on reforms, and aspects of these existing programs that are appealing in existing programs carry over to perceptions of the reforms.

2. Given the American public's lack of awareness around the specific proposals for Medicare and Medicaid expansion, framing (eg, naming) will significantly shape their perceptions of these reforms.

1.2. Medicare For All (MFA) policy

Following the 2016 Presidential campaign, the policies of health insurance coverage expansions through a single‐payer or publicly run health care program have regained national attention. ¹⁴ Medicare For All has become a political hot topic, with most Democratic presidential nominees supporting the idea or cosponsoring the bill in the US Congress. ¹⁵ Several states have also begun to closely examine reforms that they could make independently of the federal government in order to expand government‐run health care, coining the term Medicaid Buy‐In. ¹⁶

The Medicare program is and has historically been popular. ¹⁷ Polling over the last few decades has shown strong support for Medicare, with 80 percent in favor of the program and only 15 percent with negative opinions. ¹⁸ Moreover, Medicare has created a political constituency of beneficiaries who have a vested interest in preserving the program. ¹⁹ The program enjoys strong support among its beneficiaries, with 91 percent of beneficiaries satisfied with the program. ²⁰ In this regard, the Medicare voting block has made Medicare a “third rail of politics,” whereby any proposals to amend or reduce the scope of the program have been met with strong opposition.

MFA or other universal or single‐payer proposals have been in the political discourse for decades. ²¹ Over the years, the policy has gained traction but has failed to move forward either due to a lack of political support or an inability by policy makers to demonstrate the fiscal sustainability of the program. ²² The policy gained attention following the 2016 Democratic presidential primary, where it was a focal point in the Bernie Sanders campaign. ¹⁴ In the 2018 midterm election, MFA was a focal point of many congressional races and was even highlighted in an opinion editorial by President Trump. ²³ As the 2020 Democratic presidential primary begins, many of the candidates are running on the promise of an MFA policy, and details of the policy have been discussed by candidates in order to define eligibility and the cost of the program. ⁴

A key objective in current proposals that comprise the MFA is to establish a national single‐payer health insurance system that would eliminate private health insurance and provide all Americans with a single health insurance plan run by the federal government. ²⁴ The MFA program would have the name Medicare but would look very different from the current Medicare program. Proposals for the MFA have been less focused on defining the additional benefits that would be available to new enrollees and instead have proposed to work through the existing Medicare infrastructure. ²⁴ The overlap between the existing Medicare policy and the new proposals under the MFA has, as a result, created confusion within public opinion as to what “Medicare For All” entails.

Polling on MFA indicates that the policy has support, but the support decreases once additional details of the plan are provided. A poll showed that support dropped by 19 percent once respondents were given additional details about the taxing structure that would be needed to finance the proposal and by 30 percent when it was implied that there would be delays in receiving care. ⁷ The details of MFA continue to be unclear in the public dialogue, and questions remain as to whether the policy would be a fundamental change to a single‐payer health care system or whether the policy would be an expansion of the Medicare program to individuals who would not otherwise qualify. ²⁵

1.3. Medicaid Buy‐In (MBI) policy

Individual states do not have the authority to regulate large employer health insurance plans, which is how a majority of Americans receive their health care services. ²⁶ States also do not have control over the decisions made by the federal Medicare program. However, states do have control over the private health insurance market for people who purchase health insurance coverage on their own, comprising of approximately 20 million people, and the Medicaid program, which serves approximately 72 million low‐income people and people with disabilities; together, these populations comprise approximately 25 percent of the health insurance market. ²⁷ , ²⁸

The Medicaid program gained prominence in the political discourse following the debate over the repeal of the ACA and potential cuts to the Medicaid program in the summer of 2017. Less polling has been conducted to determine the national popularity of the program; however, a 2005 poll found that 74 percent of the public had a favorable view of Medicaid, ¹⁸ , ²⁹ and polling during the 2009 and 2010 health care reform debate also indicated that there was strong support for expansion. ³⁰ Less polling has been conducted to identify Medicaid beneficiaries’ perceptions of the Medicaid program, though evidence has shown that beneficiaries are generally satisfied with their benefits. ³¹ A few recent studies using the 2014‐15 National Medicaid Consumer Assessment of Healthcare Providers and System (CAHPS) survey have shown that Medicaid enrollees were satisfied with their coverage, rating their overall health care to be 7.9 out of 10, and these findings were consistent across state expansion choices following the rollout of the ACA. ³² , ³³ With this said, Medicaid has received considerable criticism for its inefficiency in providing beneficiaries with adequate access to care and in its poverty‐based eligibility criteria, which critics have claimed would provide coverage to individuals who may be capable of covering their care through other means but instead earn program benefits that they do not deserve (ie, the “undeserving poor”). ³⁴

In a similar fashion to MFA, MBI is a policy that would utilize the existing infrastructure of the Medicaid program by either adopting payment rates, networks, or managed care contracts. ³⁵ The policy has attracted the attention of policy makers in several states, including Massachusetts, Oregon, and New Jersey, where legislation and working groups have been proposed to discuss MBI plans. ³⁶

1.4. The messaging of health reforms

Messaging around expanding health insurance coverage has typically been framed to the general public in two ways, either as an expansion of an existing program or as the creation of a new program. Framing a program as an expansion of coverage could have the effect of building approval in the program if the expansion would be building off of successful reforms. On the other hand, such framing could have the effect of reducing trust if it were implied that the expansion would come at the expense of coverage that the existing program provides. Similarly, framing a health reform as a new option could be an opportunity to build trust if it were believed that the reform would not adversely impact existing programs; however, a new reform may also introduce a level of uncertainty. To this end, the introduction of a health reform might be shaped by the extent of the government's involvement in operationalizing the reform, particularly if there exists a more general lack of trust in the government as an effective implementer of new programs.

Given how public support for or opposition to new health care reforms may be driven more by the framing of these reforms than by the content within these reforms themselves, it would be of particular importance to policy makers to identify the extent to which even simple framing around the reforms may impact public support.

2. METHODS

2.1. Source of participants and data

In collaboration with Civis Analytics, a consulting firm that specializes in person‐level data analytics, we implemented a polling experiment to test the impact of framing on public opinion and approval for MFA and MBI. A sample of 5051 adults agreed to participate in an opt‐in online poll that was conducted between September 12, 2018, and September 26, 2018. Participants were recruited through the Civis Analytics polling database of validated online poll respondents. The poll was posted on an online board along with other fielded polls, and respondents were given the option to choose which poll to take.

2.2. Study design

As part of the polling experiment, all respondents first received brief descriptions of Medicare and Medicaid in order to ensure that there would be minimal confusion between the two programs as well as to control for baseline awareness of these programs. Participants were subsequently asked whether they agreed with various statements about Medicare and Medicaid, and the extent to which those statements were instrumental in their support for, or opposition to, those programs. In particular, they were asked about their beliefs about the generosity of benefits from receiving those programs (“Medicare/aid covers most medical needs”), availability of care through the programs (“Medicare/aid would allow me to see most doctors”), eligibility into the program (“Medicare is available to all people that have paid into it, Medicaid is available to me if I need it”), their perceptions of program cost (“Medicare/aid keeps my cost low”), and the impact of the program on the current health insurance system (“Medicare/aid does not disrupt the health care system”).

Participants were then randomized into one of four groups: (a) a group “MBI with Name” in which respondents were asked about their approval for MBI, with the name of the proposal stated along with a description (N = 1221); (b) a group “MBI without Name” in which respondents were asked about their approval for MBI, but with only a proposal description (N = 1282); (c) a group “MFA with Name” in which respondents were asked about their approval for MFA, with the name of the proposal stated along with a description (N = 1280); or 4) a group “MFA without Name” in which respondents were asked about their approval for MFA, but with only a proposal description (N = 1268). Figure S2 in the Appendix illustrates the experimental design, and Table 1 presents the framing across the four polls that respondents within each experimental arm received. Finally, detailed sociodemographic data, including age, race, gender, income, educational attainment, insurance status, and political preference, were collected from participants at the conclusion of the poll. The complete poll questionnaire for each of the groups is presented in Appendix S2.

TABLE 1.

Experimental group item descriptions

Experimental group	Poll wording
1: MBI with name	Under current law, a person can buy a private health care plan if they do not get health coverage at work. At present, politicians are considering a proposal to have a state health plan, Medicaid Buy‐In, which would be an expansion of the Medicaid program (the state‐run health program for low‐income residents and certain people with disabilities). The program would let people buy health insurance through existing state Medicaid programs instead of purchasing a private health insurance plan.
2: MBI without name	Under current law, a person can buy a private health care plan if they do not get health coverage at work. At present, politicians are considering a proposal to have a state health plan that would let people buy health insurance through existing state programs instead of purchasing a private health insurance plan.
3: MFA with name	At present, politicians are considering a proposal to have a national health plan known as Medicare For All. This plan would include an expansion of the Medicare program (the federal health insurance program for people over 65 and certain people with disabilities), in which all Americans would get their health insurance from a single government plan run by the federal government.
4: MFA without name	At present, politicians are considering a proposal to have a national health plan in which all Americans would get their health insurance from a single government plan run by the federal government.

2.3. Outcomes

Our key outcome variable is a 5‐point scale measure of a participant's approval for the stated proposal, where a score of 1 indicates that the participant is strongly opposed to the proposal, and a score of 5 indicates that the participant is strongly in favor of the proposal. In addition, we generate a binary measure of approval, which takes on a value of 1 if the participant reported an approval rating of 4 or 5 (either somewhat in favor or strongly in favor) for the proposal, and 0 otherwise. Similarly, we generate a binary measure of disapproval that takes on a value of 1 if the participant reported an approval rating of 1 or 2 (either very opposed or somewhat opposed) for the proposal, and 0 otherwise.

2.4. Data analysis

Our first analysis compares the level of approval and disapproval between poll groups 1 and 2 that received information about MBI, and we infer the causal effect of framing by estimating the differences in rates between respondents who were assigned to the “MBI with Name” poll (poll group 1) and those who were assigned to the “MBI without Name” poll (poll group 2). In our second analysis, we similarly infer the causal effect of framing in the MFA proposal by calculating the differences in rates between respondents who were assigned to the “MFA with Name” poll (poll group 3) and those assigned to the “MFA without Name” poll (poll group 4). Finally, we compare the relative public approval and disapproval for MBI against MFA, for either with or without the name included, respectively, through a comparison of poll groups 1 and 2 against poll groups 3 and 4. Our analytic models are calculated using multivariate linear regressions that control for respondent age, gender, race, education, income level, party affiliation, political ideology, and insurance status. All analytic models include state fixed effects, and standard errors are clustered at the state level.

When comparing our sample to the US population, we note that our respondents are not nationally representative; respondents in our sample are more likely to be white and have completed college than the average US adult, although the proportion of respondents with health insurance is slightly less than the US population. ³⁷ For this reason, all descriptive statistics are weighted using estimated population‐based weights that are calculated by Civis Analytics to obtain representative estimates. The weights use proprietary data from their National Consumer Database and previously fielded Civis polls and are calculated by calibrating respondents’ key demographic characteristics (gender, age, race, income, and education) and answers to a set of proprietary questions to aggregated population estimates for each characteristic and question response such that they are then representative of the US general population. Following Wooldridge (2002) and, more recently, Solon, Haider, and Wooldridge (2015), our regressions are not weighted by the estimated population weights so as to not introduce inconsistency in the estimates in the presence of heteroscedasticity. ³⁸ , ³⁹

2.5. Ethical considerations

Civis Analytics provided only de‐identified, validated data on polled participants. The study therefore received a nonhuman subjects research determination from the institutional review board at Boston University (protocol number 4991X).

3. RESULTS

Table 2 presents weighted descriptive statistics for the sample of 5051 poll participants. When asked to rate their respective proposals from 1 (strongly opposed) to 5 (strongly in favor), participants responded with an average favorability rating of 3.24. When examining approval and disapproval for their respective proposals, 43.7 percent of participants reported that they were either strongly or somewhat in favor of their respective proposals. 23.9 percent of participants reported that they were either strongly or somewhat opposed to their respective proposals. The average age of participants was 47.3 years, with the distribution of participants by age presented in Figure S1 in the Appendix, and slightly more than half of all participants (51.6 percent) were female; moreover, 71.9 percent of respondents were white, 62 percent earned more than the estimated US median income of $50 000, and 26.2 percent attended some years of college. When asked about party leaning, more participants identified with the Democratic party than with the Republican party (37.1 percent vs 32.7 percent, respectively); however, when asked about political ideology, more participants identified as having conservative political leanings than liberal political leanings (37.6 percent vs 25.6 percent, respectively). A large proportion of participants (91.6 percent) reported having insurance of some type, and almost 3 in 10 participants (29.1 percent) had Medicare, while 4.4 percent of participants had Medicaid. Tables S1‐S6 present additional detailed tabulations of descriptive variables for the sample.

TABLE 2.

Descriptive statistics, population weighted

	Mean	SD	No. of cases
Outcomes (1‐5 scale, 5 = strongly in favor/agree, 1 = strongly opposed/disagree)
Level of support for proposal (1‐5 Scale)	3.242	1.260
Approval of proposal (1 = yes)	0.437		2841
Disapproval of proposal (1 = yes)	0.239		1552
Treatment groups
1: MBI with Name	0.241		1221
2: MBI without Name (Words Only)	0.254		1282
3: MFA with Name	0.253		1280
4: MFA without Name (Words Only)	0.251		1268
Covariates
Age, years	47.331	17.407
Female (1 = yes)	0.516		3356
White (1 = yes)	0.719		4676
No college education (1 = yes)	0.262		1703
Above median income (>$50K) (1 = yes)	0.620		4033
Democrat (1 = yes)	0.371		2411
Republican (1 = yes)	0.327		2126
Liberal (1 = yes)	0.256		1666
Conservative (1 = yes)	0.376		2447
Has insurance (1 = yes)	0.864		5620
Insured by Medicare (1 = yes)	0.236		1538
Insured by Medicaid (1 = yes)	0.087		564
Insured through private insurance (1 = yes)	0.497		2511
Insured by different plan (1 = yes)	0.044		221
Insured, but does not know insurance type (1 = yes)	0.052		260
N	5051

When comparing unadjusted levels of participant approval for each of the four proposals (Table 3), approval for MBI is higher than approval for MFA, regardless of whether the name of the proposal is included in the description. Slightly more than 50 percent of participants reported to be in favor of the MBI proposals, while between 32.7 and 39.5 percent of participants were in favor of MFA proposals, with favorability varying depending on the framing of the proposal. On the other hand, between 34.1 and 41.8 percent of participants disapproved of the MFA proposals, compared to significantly lower disapproval rates (between 8.7 and 10.4 percent) for MBI proposals.

TABLE 3.

Mean approval and disapproval by treatment group, weighted

Variables	(1)	(2)
	Approve	Disapprove
1: MBI with Name	0.528	0.087
2: MBI without Name	0.501	0.104
3: MFA with Name	0.395	0.341
4: MFA without Name	0.327	0.418
N	5051	5051

The top panel of Table 4 presents results for favorability, approval, and disapproval for the MBI proposal from regressions that compare MBI with the name of the proposal in the description against MBI with only the proposal description. Findings from the adjusted analyses indicate that including the name “MBI” in the proposal significantly increases favorability by 0.12 points on the 5‐point approval scale (column 2), increases the likelihood of approval by 5.0 percentage points (column 4), and decreases the likelihood of disapproval by 2.3 percentage points (column 6). Similarly, our findings from the bottom panel of Table 4 demonstrate that including the name “MFA” in the proposal description for MFA significantly increases favorability by 0.115 points on the 5‐point approval scale (column 2), increases the likelihood of approval by 3.4 percentage points (column 4), and decreases the likelihood of disapproval by 4.8 percentage points (column 6). When considering respondents’ insurance status, we find that respondents who are covered by Medicare and private insurance are relatively more likely to support MBI by 9.0 and 8.3 percentage points, respectively; however, the same is not true for respondents’ support of MFA. Appendix Tables S8 and S9 present the fully adjusted analysis with covariate estimates.

TABLE 4.

Average treatment effect analysis of MBI and MFA proposals, with and without including the proposal name

Top Panel: Average treatment effect analysis of MBI with name vs MBI with description only
	(1)	(2)	(3)	(4)	(5)	(6)
	Favor proposal (1‐5 scale)	Favor proposal (1‐5 scale)	Approve (Y/N)	Approve (Y/N)	Disapprove (Y/N)	Disapprove (Y/N)
MBI with name	0.107**	0.119***	0.044**	0.050**	−0.021**	−0.023**
	0.023 to 0.191	0.032 to 0.205	0.005 to 0.083	0.011 to 0.090	−0.042 to −0.001	−0.045 to −0.001
Controls	N	Y	N	Y	N	Y
State fixed effects	Y	Y	Y	Y	Y	Y
Observations	2503	2503	2503	2503	2503	2503
R‐squared	.020	.096	.021	.096	.026	.062

Bottom Panel: Average treatment effect analysis of MFA with name vs MFA with description only
	(1)	(2)	(3)	(4)	(5)	(6)
	Favor proposal (1‐5 scale)	Favor proposal (1‐5 scale)	Approve (Y/N)	Approve (Y/N)	Disapprove (Y/N)	Disapprove (Y/N)
MFA with name	0.132***	0.115***	0.041***	0.034**	−0.052***	−0.048***
	0.047 to 0.217	0.038 to 0.192	0.014 to 0.068	0.008 to 0.060	−0.086 to −0.019	−0.080 to −0.016
Controls	N	Y	N	Y	N	Y
State fixed effects	Y	Y	Y	Y	Y	Y
Observations	2548	2548	2548	2548	2548	2548
R‐squared	.037	.121	.034	.123	.030	.093

All models are estimated using ordinary least squares, with 95 percent confidence intervals presented in parentheses. Covariates in adjusted models include age, sex, race, education, income level, party affiliation, political ideology, and insurance status. Standard errors are clustered at the state level.

^*** P < .01, **P < .05, *P < .1.

In Figure 1, we present estimates of public favorability for MFA relative to MBI and find that favorability for MBI, regardless of whether the name is included or not in the proposal, is 0.6 points higher, at 4.2 on the 5‐point favorability scale, when compared to favorability for MFA, which is 3.6 on the scale. This finding corroborates our descriptive evidence to show that public support for MBI is generally higher than public support for MFA. Regression estimates from this analysis are presented in Appendix Table S10.

FIGURE 1.

Level of support for Medicaid Buy‐In (MBI) vs Medicare For All (MFA) [Color figure can be viewed at wileyonlinelibrary.com]

Note: Models to estimate predicted levels of support are estimated using ordinary least squares. Covariates in adjusted models include age, sex, race, education, income level, party affiliation, political ideology, and insurance status. Standard errors are clustered at the state level

We also run stratified analyses by age groups, which we divide into either under 65 or over 65, as well as by categorical generational bands (Silent Generation, Baby Boomers, Gen X, Millennials, and Gen Z). Our results (Appendix Tables S9‐S11) show that public support for the new proposals varies considerably by age. We find MBI to have modestly stronger support among younger age groups, particularly Gen X and Millennial participants. Our results show that respondents under the age of 65 have a 7.0 percentage point higher likelihood of supporting MBI compared to respondents older than 65. In contrast, support (or, rather, lack of disapproval) for the MFA proposal is relatively stronger among Baby Boomers and participants older than 65. In particular, we find that respondents older than 65 have a 14.1 percentage point lower likelihood of disapproval for MFA compared to respondents under the age of 65.

4. DISCUSSION

We conduct a polling experiment to test how the framing, specifically how the naming, for two health care reform proposals, Medicare For All and Medicaid Buy‐In, drives constituents’ views around and reasons for opposition to these proposals. Our findings suggest that the public is divided on support for MFA proposals but have stronger support for MBI. In total, 36.1 percent of poll respondents were in favor of a MFA plan, 38 percent were opposed, and 25.9 percent have no opinion. In contrast, 51.3 percent of poll respondents are in favor of a MBI plan, and 9.6 percent are opposed, while 39.1 percent have no opinion. The fact that a significant portion of respondents was neutral in their opinion of both proposals may reflect a lack of awareness of these policies, which would serve as an opening for policy makers to shape public opinion around these proposals. To date, health care expansion, particularly MFA, has been a central issue of discussion and debate in the 2020 Democratic presidential primary election. ⁴⁰ We also find that constituents were more likely to support a proposal, both for MFA and MBI, when given the names of the proposal. Similarly, disapproval for these two proposals significantly declined when the name of the proposal is included. We also find that support for these two proposals varied by age, where MBI demonstrated the strongest support among Millennials, while Baby Boomers and those older than 65 were more likely to support MFA. Finally, our findings indicate that people are more comfortable with policies that are framed as expansions of current policies than as new programs. For example, people older than 65 are more likely to support, and less likely to oppose, MFA when the program is described as an expansion of Medicare.

Our experiment draws comparisons with prior studies that have been conducted around measuring support of the ACA. In those studies, people were supportive of the ACA when receiving a description of the policy, but opposition grew when the program was referred to as “Obamacare”. ³ , ⁸ Like polling on the ACA, we find that support for the components of MFA and MBI was stronger than overall support for either proposal individually. We find that respondents believed that these proposals will achieve goals that are important to the public, like reducing health care costs, improving access to care, and providing good coverage. In contrast to framing studies of the ACA, however, we see that support increases when the names of the programs are used to describe the policies.

Medicare expansion and MFA have dominated the media over the last few years, and it is evident that the increased attention has contributed to more strongly held opinions of the proposal. We find that people support Medicare, and despite a common misconception that voters would not support an expansion of the program, support among all groups increased when Medicare was used to describe a single‐payer reform. In contrast, MBI, a proposal that has recently begun to be discussed in several states and that would allow people to purchase Medicaid plans, received significantly higher support. Many respondents did not have an initial opinion of the proposal, but those who supported the proposal outnumbered those who opposed it by a factor of five to one. Given that more than half of Americans are connected to the Medicaid program—either through their own coverage or that of a family member or close friend—we hypothesize that MBI likely enjoys strong support because the proposal opens a popular program to the larger public. Our findings also reinforce prior hypotheses that speculate that renaming programs within Medicaid, which may serve to destigmatize the program in general, would likely increase support and possibly even take‐up by reducing program confusion around Medicaid, even among enrollees who say that the program is important to them. ⁴¹ Given that the proposal also has not been in the spotlight for as long, our findings may suggest that political polarization over the proposal has not yet sorted public support based on party affiliation. In general, our results reinforce the findings of previous studies that highlight the relative favorability and increasingly positive opinion of both proposals among the American public.

CONFLICT OF INTEREST

All authors declare that they have no competing interests.

AUTHOR CONTRIBUTIONS

All authors participated equally in the conception, analysis, design, and writing of the article. All authors have read and approved the final manuscript and are aware that the manuscript is being submitted to the journal.

Supporting information

Author Matrix

Appendix S1 and S2

Karra M, Sandoe E. A polling experiment on public opinion on the future expansion of Medicare and Medicaid. Health Serv Res. 2020;55:578–586. 10.1111/1475-6773.13295