Implementing ObamaCare

Abstract

After President Barack Obama signed into law the Affordable Care Act (ACA) of 2010, 26 states filed lawsuits challenging key provisions. In 2012, the Supreme Court ruled in support of most provisions but held that the mandatory Medicaid expansion was unconstitutional, in effect making it optional for the states. Yet Medicaid expansion is critical if the ACA is to achieve its goal of substantially increasing coverage, especially for lower-income adults with health problems, many of whom are racial and ethnic minorities. This article examines the factors explaining state variation in Medicaid expansion from 2012 to 2014. Although partisan power in the states is a key predictor, other relevant factors include the state’s Medicaid policy legacy, providers’ influence, conservative ideology, and racial resentment. We demonstrate that the enactment of legislation is only a part of the policy formation process and that implementation is equally salient, with state dynamics playing a critical role.

Keywords

Medicaid Obamacare partisan politics policy legacy interest groups public opinion policy attitudes racial resentment

On March 23, 2010, President Barack Obama signed into law the Patient Protection and Affordable Care Act (ACA). For the first time in more than a century, the federal government made a commitment to provide near universal coverage through a complex mix of private incentives and public support. The ACA’s main features include state insurance exchanges, stringent regulations on insurance companies, fines on employers who do not offer coverage, a mandate that individuals purchase health insurance, subsidies to help low-income people with the costs, and a substantial expansion of Medicaid (Quadagno 2014).

The day that President Obama signed the ACA, Florida filed a lawsuit in federal district court challenging the constitutionality of the individual mandate and the Medicaid expansion. Florida was joined by 25 other states, all but two with Republican governors. On June 28, 2012, the Supreme Court held that the individual mandate was a constitutional exercise of Congress’ power to levy taxes but that the Medicaid expansion was unconstitutionally coercive on the states (Oberlander 2011). The majority decision meant that the federal government could not withhold existing federal Medicaid funds if a state did not comply with the Medicaid expansion. The practical effect has been to make Medicaid expansion optional for the states (Kaiser Family Foundation [KFF] 2012).

Medicaid expansion is a critical component of the ACA, one of its most important mechanisms for substantially increasing health insurance coverage. It is especially critical for lower-income adults with health problems (Sommers, Kenney, and Epstein 2014) and for racial and ethnic minorities, who have higher odds of being Medicaid-eligible compared to whites. Overall, it is estimated that because of states opting out of Medicaid, there will be 3.6 million fewer insured and more than $8 billion less in federal payments to states (Price and Eibner 2013).

Health policy innovations in the states are important for practical as well as theoretical reasons. In the past, state actions have signaled to elected officials what policy options are acceptable to voters and, thus, have provided a potential blueprint for federal legislation (Gray 1994). For example, a hospital reimbursement formula that was tested in New Jersey set the precedent for Medicare’s Prospective Payment System in 1983 (Morone and Dunham 1985). Similarly, state reforms in the early 1990s regulating the small group health insurance market were incorporated into the Health Insurance Portability and Accountability Act of 1996 (HIPAA).

A similar process occurred with the ACA, which was nearly identical to a health insurance plan implemented in Massachusetts in 2006 (Quadagno 2014). In this sense, the states have been the “laboratories of democracy” (Barrilleaux and Brace 2007). Yet when the states are asked to follow a federally mandated blueprint, they do not necessarily conform. After Medicaid was enacted under the Social Security Act of 1965, many states initially refused to implement it. Arizona notably did not create a Medicaid program until 1981. Thus, the states’ refusal to implement the ACA Medicaid expansion is not historically unique. The Medicaid controversy demonstrates that policy formation involves more than a simple evolution from state to federal legislation and that state dynamics have a substantial role in the entire process.

The objective of this article is to draw on welfare state theory to identify factors that help explain state variation in movement toward Medicaid expansion under the provisions of the ACA. We first discuss the crucial link between partisanship and expansion, then consider alternative factors that might influence a state’s decision about whether to comply, even where partisan control is consolidated. These factors include states’ Medicaid policy legacies prior to the ACA, the influence of provider groups, the effect of conservative political values, the level of racial resentment, and state capacity.

Factors Influencing Medicaid Expansion

Partisan Politics

Partisan politics have played a role in nearly every health policy debate since the 1940s at both the federal and state level, with Republicans generally opposed to expansion of public benefits and Democrats generally in favor. Republicans opposed a plan for national health insurance during Harry Truman’s Fair Deal in the 1940s, initially fought against Medicare in the 1960s, and blocked President Bill Clinton’s Health Security plan in the 1990s (Skocpol 1997). States with Republican legislatures and governors have also spent less on children’s health care after the State Children Health Insurance Program (SCHIP, later CHIP) was enacted in 1997 (Tope and Hickman 2012).

At first glance, the ACA appears to be a victory for the Democratic Party over Republican visions of limited government intervention. No Republicans in either the House or the Senate voted for the bill, and it was enacted by a Democratic Congress and signed into law by a Democratic president. Most health policy experts emphasize this partisan division (Beaussier 2012; Hacker 2011). As Lawrence D. Brown (2011:421, 425) argues, “Congress’ accomplishment was not merely but entirely a triumph for the Democratic Party . . . The ACA is a virtually pure Democratic product . . . .” The ACA also reflected the Clinton administration in important ways, notably with Medicaid expansion and insurance company regulation (Quadagno 2014). Thus, we would expect states with Democratic legislatures and Democratic governors to be more likely to implement the Medicaid expansion. In confirmation of this argument, Lawrence R. Jacobs and Timothy Callaghan (2013) found support for the effect of state partisan politics. Yet key provisions of the ACA, the employer mandate and the individual mandate, were Republican policy ideas, and its fundamental principles were nearly identical to a bill, the “Health Equity and Access Reform Today Act of 1993” (HEART), promoted by Republican senators to deflect support for President Clinton’s Health Security plan (Quadagno 2014). Further, some Republican governors have supported Medicaid expansion as have some states with Republican legislatures while some states with Democratic governors or legislatures have opted out. Thus, our first hypothesis is as follows:

Hypothesis 1: States with a strong Democrat presence in key government offices will be more likely to expand Medicaid than states with a strong Republican presence, but partisan domination alone will not explain state variation.

Our task is to identify key factors that help explain expansion net of partisanship.

Policy Legacies

The theory of policy legacies suggests that social welfare benefits are influenced by earlier policy choices. These initial policy choices give rise to widespread public expectations, institutional structures, and vast networks of vested interests that narrow the menu of future options by forming self-reinforcing paths that become increasingly difficult to alter (Hacker 2002; Pierson 2000). This process of path dependency is consistent with the historical trajectory of Medicaid, which, through a generous federal match and substantial state discretion, has fostered steady program expansion (Grogan and Andrews 2014).

Medicaid was created as a joint federal-state program under Title XIX of the 1965 amendments to the Social Security Act (SSA), which gave states considerable leeway in deciding how generous a program to create or even whether to create a program at all. As a result, there was wide variation in terms of protected population groups, eligibility levels, and access to services (Olson 2010). Section 1115 of the SSA amendments furthered state variation by allowing states to apply for waivers to conduct research and demonstration projects that bypassed federal rules and regulations. Waivers were not utilized to a great extent until the 1990s when the Clinton administration actively encouraged states to use them to expand services and coverage. States responded rapidly, using waivers to increase Medicaid income eligibility levels, raise asset limits, and incorporate groups not categorically eligible (Schneider 1997). States also took advantage of the more generous federal match in CHIP to cover low-income children and uninsured parents (Kaiser Commission on Medicaid and the Uninsured 2005; Cunningham 2002). Thus, using Medicaid and CHIP resources, many states expanded coverage to populations otherwise ineligible, including parents with higher incomes and nonparents (Engquist and Burns 2002). Notably, the states did not engage in the open resistance that came to typify the response to the ACA. Still, in the years before the ACA, there was wide variation in state Medicaid programs (Kim and Jennings 2012). Since the passage of the ACA, these policy legacies have continued to influence Medicaid expansion (Jacobs and Callaghan 2013).

What has yet to be determined is whether a state’s policy legacy has an effect on Medicaid expansion beyond state partisanship. Because the variation in program implementation is so considerable, we expect that policy legacies may be influential enough to break through the partisan gridlock surrounding the ACA. Thus, our second hypothesis is as follows:

Hypothesis 2: States that have already instituted more generous coverage will be more likely to take advantage of the Medicaid expansion opportunity, and this policy legacy will help explain variability in expansion that runs counter to partisan tendencies.

Provider Influence

Provider groups have lobbied against threatened cuts to Medicaid in the past and have been identified as advocates in many cases currently pushing for expanded benefits (Grogan 2014; Quadagno 2004; Sommers et al. 2013). The hospital industry, in particular, has a considerable stake in the Medicaid expansion. Charitable and safety-net hospitals, especially the larger hospitals, have long provided uncompensated care to low-income and underserved individuals. If a state expands Medicaid, there will be fewer uninsured individuals and a guaranteed source of payment for medical services. An equally compelling incentive is the potential loss of federal funds. Many hospitals receive funds through the Medicaid and Medicare Disproportionate Share Hospital (DSH) programs to help offset the cost of providing care to the uninsured. The ACA decreases the amount of DSH payments under both programs because it is presumed that the number of uninsured and underinsured people will fall precipitously (Davis 2012). This reduction in DSH payments will dramatically impact the financial situation of some hospitals. In states that do not expand Medicaid, the demand for uncompensated care may remain relatively stable, while the amount of DSH funds that were previously used to subsidize some of that care will decline. As a result, hospitals are likely to be a primary pressure group lobbying for Medicaid expansion.

Physician organizations have also been central players in health care debates since the progressive era (Starr 1983). From the New Deal until the 1960s, the American Medical Association (AMA) opposed every effort to enact national health insurance. Physicians lost a major battle with the enactment of Medicare in 1965 and then saw their influence dwindle further in the 1970s as corporate purchasers began seeking ways to rein in health care costs. Physician organizations were less a factor in 2010 than in previous health care reform debates (Quadagno 2010). Nonetheless, the ACA will have an enormous impact on the way many physicians provide services, as it will play a major role in shaping the composition and size of the pool of patients who can afford doctor visits.

It is difficult to predict the direction of physician influence on the Medicaid expansion. One issue is that there is no equivalent to DSH programs for physicians, so the incentive to support Medicaid expansion may be less than for hospitals. Another issue is that many low-income persons—especially those enrolled in Medicaid or who have no insurance—have difficulty finding physicians who are willing to accept them as patients (Casalino 2013). One study found a gradual decline in the percent of physicians serving Medicaid patients and an increase in the number of physicians who are closing their practices to new Medicaid patients due to low Medicaid reimbursement rates (Cunningham 2002). In addition, constraints on payments from managed care and other sources have reduced physicians’ ability to cross-subsidize the care they provide to uninsured or Medicaid patients (Cunningham and May 2006). Yet the expansion of Medicaid will dramatically reduce the number of the uninsured and increase physician practices. Our third hypothesis is as follows:

Hypothesis 3: Hospital and physician influence will have a positive effect on movement toward Medicaid expansion, and this effect will persist even after accounting for partisan control in state governments.

Political Values

Although there is a debate about the extent to which public attitudes influence policy decisions, there is considerable evidence that on highly salient issues, state politicians are attuned to what the public favors (Erikson, Wright, and McIver 1993). On the issue of Medicaid expansion, a state population’s degree of conservatism should be particularly salient, because key elements of modern conservatism include opposition to government activism in general and to government spending to redress social ills in particular. Conservative pundits and political leaders regularly describe the ACA as a huge expansion of federal authority that will be a financial drain on the economy (Cannon 2013) even though the Congressional Budget Office estimated that the ACA would reduce actual budget deficits (Elmendorf 2011). If the conservative position is absorbed by the public and translated into political opposition, Medicaid expansion is likely to be delayed. Thus, our fourth hypothesis is as follows:

Hypothesis 4: States with higher average levels of conservatism will be less likely to expand Medicaid, and this will remain the case net of partisanship among government leaders as those leaders seek public support for reelection.

Racial Politics

Historically, race has been a fundamental influence on American politics, and racial resentment has proven to be a robust predictor of public attitudes toward race-targeted policies such as affirmative action (Sears, Sidanius, and Bobo 2000) as well as seemingly nonracial policies such as welfare, crime, and Social Security (Gilens 1999; Hurwitz and Peffley 2005; Winter 2006). In the debate over the ACA, racial resentment played a substantial role in shaping public opinion toward various policy proposals (Tesler 2012). One example concerned the controversial “public option,” which would have allowed the federal government to operate a health insurance plan in the exchanges to compete with private plans (Hacker 2010). Compared with party identification, political ideology, or demographic factors, racial resentment was the strongest predictor of negative attitudes toward the public option (Tesler 2012). Thus, our fifth hypothesis is as follows:

Hypothesis 5: States with greater levels of racial resentment will be laggards in Medicaid expansion, and this will be the case net of partisan control of government.

State Fiscal Capacity

State fiscal capacity refers to the state’s economic ability to cover the costs of social programs on its own or along with federal funds. Throughout Medicaid’s history, federal matching funds have provided expansion incentives to states, and states have been adept at designing strategies to maximize federal dollars. For example, CHIP authorized a federal match 15 percent higher than the regular Medicaid match, and the states responded by covering more low-income children (Olson 2010).

The federal share of the Medicaid match is significantly more generous for the ACA expansion than for the states’ existing Medicaid programs and CHIP. Under the ACA, the federal government will cover 100 percent of the states’ costs in years 2014 through 2016, gradually decreasing to 90 percent in 2020 and thereafter. The federal government will also assume some of the costs for segments of the uninsured population that were previously funded using state budgets. In many cases, the effect of expansion will be to reduce both state and local government costs for uncompensated care (Holahan et al. 2012). These financial incentives will be difficult for states to resist (Kim and Jennings 2012). Yet states with lower fiscal capacities may hesitate to accept even the relatively small costs of Medicaid expansion not covered directly by the federal government or covered by the federal government through grants and subsidies in other areas. Furthermore, a strong theoretical tradition suggests that states with greater fiscal capacity are more likely to have infrastructure and expertise in place to facilitate expansion (DeViney 1983; Flora and Alber 1981). Accordingly, our final hypothesis is as follows:

Hypothesis 6: States with greater fiscal capacity will be further along in expanding Medicaid, net of partisan control of government.

Method and Data

Testing whether the theorized factors influence states’ movements toward Medicaid expansion is fairly straightforward in statistical terms. Yet determining the independent influence of one factor net of other factors with only 50 observations (N = number of states) can be challenging, especially where predictors are associated among themselves. We address this challenge with two different approaches. Each has its own benefits and drawbacks, but each also informs results obtained using the other. Our first approach is to analyze bivariate models and then examine split-bivariate models, which decompose the analyses by subgroups for states controlled by Democrats, states with split control, and states controlled by Republicans. This approach allows us not only to identify which factors are associated with Medicaid expansion in the direct sense but also to identify which factors remain influential even when state governments are dominated by the politics of a single party. In our second approach, we proceed with conventional multivariate regression analyses.

Bivariate and Split-bivariate Analyses

Movement toward Medicaid

For the bivariate and split-bivariate analysis, our dependent variable is a continuous measure of states’ movements toward Medicaid expansion. This measure was developed by Jacobs and Callaghan (2013). Their objective was to capture important behind-the-scenes variation in states’ recent activities as well as states’ potential for future change regarding Medicaid expansion under the ACA. The measure is constructed from a combination of three factors in an additive scale. The first factor is an assessment of gubernatorial statements, budgets, or collective decisions of the legislature and governor regarding Medicaid expansion (weighted using scores from −3 through +3 points). The second factor captures state planning for Medicaid expansion in terms of preparatory grants received from the federal government (1 point each, up to 4). The third and final component reflects recent shifts in state Medicaid programming made in response to the ACA (1 point positive or negative). Scores range from the least movement toward Medicaid expansion at −4 to the most movement toward Medicaid expansion at 7. More details on the construction of the measure can be found in Jacobs and Callaghan (2013).

The continuous measure has a key advantage in that it captures more information than a binary or trinary variable. For an illustration, consider that some states classified as rejecting expansion or undecided about expansion according to the KFF’s (2013a) three-level scale are almost certainly further along toward expansion than other states in the same broad categories (Jacobs and Callaghan 2013). This information is lost with a binomial or trinomial variable. While a straightforward, up-to-the-minute binary variable indicating expansion or rejection has its own advantages and will be useful for supplementing the continuous measure as demonstrated below, the more sophisticated continuous measure also bears on the process of Medicaid expansion under the ACA and contains added information that should not be ignored. The score for each state is presented in Table 1.

Table 1.

Jacobs and Callaghan’s (2013) Measure of States’ Movements toward Medicaid Expansion.

Least advanced					Most advanced
−4	−3	−2	−1	0	1	2	3	4	5	6	7
ME WI	AK LA OK TX WY	GA IN NC PA SC SD	AL ID IA MS NE	UT VA	KS	IL NH TN WV	FL KY OH	AZ CA CO DE MD MN MO MT NJ ND	AR CT HI MI NM OR	MA RI VT WA	NV NY

Predictors

Prior research suggests several factors may influence states’ decisions regarding Medicaid expansion. These include partisan power, policy legacies, provider influence, public political values, racial political attitudes, and state capacity. We draw on a number of data sources to capture the effects of each factor.

We measure partisan power in three ways. First, we use the percentage of Democrats and Republicans in a state’s House of Representatives. Second, we consider partisan control of the governorship. Third, we note where partisan control of the governorship aligns with that of each chamber of the respective state’s Congress. The latter element identifies states controlled by Democrats, split states, and states controlled by Republicans. The data we use come from the National Conference of State Legislatures (NCSL) and is updated through November 6, 2013, thereby reflecting 2012 and 2013 elections (National Conference of State Legislatures 2013).

Policy legacy is measured with data on legacy Medicaid eligibility cutoffs reported as a percentage of the federal poverty line (FPL) defined each year by the Department of Health and Human Services. Eligibility data, current through 2013, has been compiled for each state in a joint effort between the KFF and the Georgetown Center for Children and Families (KFF 2013b). If states are to receive additional Medicaid funding offered by the federal government through the ACA, eligibility limits must be at least 138 percent of the FPL. Yet, as noted above, legacy cutoffs have historically varied widely between states. We use the cutoffs for working parents and working nonparents. Income cutoffs for jobless parents and nonparents are typically the same, or similar to, the benefits for their working counterparts (KFF 2013b).

Provider influence is captured with several measures. First, we adopt the conventional method of estimating the influence of hospitals by counting the number of hospital beds per 1,000 people in each state. Second, we supplement the “beds per 1,000” measure with the number of physicians per 10,000 people. Data for these two measures is obtained from the National Center for Health Statistics and is current as of 2010. While these measures, particularly the measure of hospital beds, are commonly used to represent the power of medical providers in comparative research, it is important to assess providers’ actions as directly as possible. Ideally, data on lobbying expenditures by hospital and physician groups at the state level would be available. Unfortunately, such data are not systematically collected and reported (King 2008). State data on campaign contributions, however, are readily available. Medicaid expansion has been a central concern in recent elections, and support for Democrat candidates should roughly indicate support for expansion, while support for Republican candidates should indicate opposition.

Campaign contribution data are less desirable compared with lobbying data, since they are expected to be collinear with electoral victory and because donor groups often give to incumbents or those they expect to win regardless of their political party preferences because they want access to politicians in any case (Hacker and Pierson 2011). Thus, we may see no effect of contributions or only an effect that is subsumed by partisan control. Yet an independent effect of partisan campaign contributions by provider groups would be a highly suggestive indicator of a relationship between not just the preferences but also the actions of specific donor groups and Medicaid expansion in a state. We accordingly assess the contributions of hospitals, physicians, and branches of the American Hospital Association (AHA). These data are publicly available from the Federal Election Commission, processed and standardized by the National Institute for Money in State Politics, and made available by the Sunlight Foundation on influenceexplorer.com. Because providers are less likely to have taken a clear stand on Medicaid expansion before the June 28, 2012 Supreme Court decision entrenched the ACA and rendered Medicaid expansion optional, data in the current analysis are limited to contain information on contributions made between June 28, 2012, through the 2012 election cycle to December 31, 2012.

Public political values in each state are measured using averaged individual-level survey responses from the Cooperative Campaign Election Study (CCES) of 2012, which is designed to include reliable representative data for each state. Test results from this survey closely match results from state-level representative samples (Ansolabehere 2013). The CCES contains a measure of respondents’ self-reported liberal-conservative ideology. Responses range from 1 = very liberal to 7 = very conservative. Drawing on these responses, we assembled states’ average levels of liberal-conservative ideology.

Research on racialized politics suggests racial political attitudes may influence the political climate beyond political ideologies that are not explicitly racial. Our analysis therefore includes a second item from the CCES that captures racialized political attitudes with questions regarding racial resentment. Derived from symbolic racism theory, racial resentment is a measure of contemporary racial attitudes developed to assess the influence of racial animosity on policy attitudes in an age when “old-fashioned” blatant racism is no longer commonly expressed in public or in surveys (Kinder and Sanders 1996). Racial resentment is typically measured with a 4-item scale. The CCES contains only two questions from the standard scale, yet these two items form a highly reliable measure nonetheless with an alpha reliability score of .76. The scale derived from these two questions (exact wording can be found in the Supplementary Appendix A) ranges from 1 = most racially sympathetic to 9 = most racially resentful. We take the averaged responses for each state as our measure of racial resentment.

Our final factor, state capacity, is measured in two ways. First, we use the average per capita income in each state as reported by the Bureau of Economic Analysis for 2010. Second, we use the total of tax revenues, in millions of dollars, for each state in fiscal year 2010 as reported by the U.S. Census Bureau’s Survey of State Government Finances (Snyder et al. 2012). Summary statistics are presented in Table 2.

Table 2.

Summary Statistics.

Variable	M	SD	Minimum	Maximum
DV
Movement Toward Medicaid Expansion	1.72	3.36	−4	7
Expanded Medicaid	0.54	0.50	0	1
IVs
Political Structure
Percentage of Democrats in the House	46.96	17.73	13	92
Percentage of Republicans in the House	52.35	17.95	8	87
Democrat Governor	0.4	0.49	0	1
Republican Governor	0.6	0.49	0	1
Democrat Control	0.28	0.45	0	1
Split Control	0.24	0.43	0	1
Republican Control	0.48	0.50	0	1
Policy Legacy
Parental Eligibility Cutoff	84.28	55.01	16	215
Nonparent Eligibility Cutoff	15.36	38.99	0	160
Provider Groups
Beds Per 1,000 People	2.73	0.75	1.7	5
Physicians Per 10,000 People	24.7	5.08	17	40
Hospital Contributions, $ Per Capita, D-R	0.0045	0.02	−0.02	0.06
Physician Contributions, $ Per Capita, D-R	0.0081	0.03	−0.06	0.08
AHA Contributions, $ Per Capita, D-R	0.0011	0.01	−0.01	0.02
State Political Values and Racial Resentment
Average Liberal-Conservative Ideology	4.28	0.31	3.71	4.92
Average Racial Resentment	6.20	0.35	5.39	7.09
State Fiscal Capacity
State Revenues	14.12	17.50	1	105
Income Per Capita, $	39,312	6,970	30,401	68,843

Note. AHA = American Hospital Association; DV = dependent variable; IVs = independent variables; D-R = Democrat-Republican.

Partisan politics

What factors influence Medicaid expansion? As noted above, a growing body of research suggests that partisan politics are a key predictor of movement toward Medicaid expansion by state governments (Jacobs and Callaghan 2013; Oberlander 2011). Consistent with these studies, we find that the partisan environment in state governments is a central factor for Medicaid expansion. Each of our measures of partisanship is highly correlated with movement toward Medicaid expansion statistically (p < .001) and visually when the relationships are displayed in two-way scatter plots (Figure 1). In each case, Democratic power has a strong positive influence on movement toward Medicaid expansion, and Republican power has a strong negative influence.

Although highly influential, partisan presence is not wholly determinative. For example, in Figure 1E, we see that Illinois and West Virginia both lagged in movement toward Medicaid expansion even though the House and the governorship were held by Democrats. Conversely, Figure 1F shows that states such as Michigan, Arizona, and North Dakota moved toward expanding Medicaid despite being controlled by Republicans. This is consistent with prior research that suggests there may be factors that “cross pressure” politicians to act contrary to party lines (Jacobs and Callaghan 2013). Because it is these cross pressures that have bedeviled research on the Medicaid expansion so far, the remainder of the analysis is devoted not only to identifying which factors beyond party power are most influential but also to identifying which factors remain influential even in states where partisan control of Congress and the governorship are consolidated.

Policy legacies

Do policy legacies influence the adoption of future policies? We find that states with a more generous legacy of eligibility requirements have moved further toward Medicaid expansion than states with less generous policy legacies. This applies when we look at eligibility requirements for parents (Figure 2A) and when we look at eligibility requirements for nonparents (Figure 2B). In each case, the correlation is significant with a p value of less than .01 (solid lines). The results also suggest that policy legacy may be influential even when states are controlled by a single party. Whenever Democrats or Republicans control the government, the plots visually indicate a positive relationship between prior generosity and movement toward Medicaid expansion (dashed lines). Although this finding is suggestive of a cross-pressure influence, the positive relationship receives statistical significance with a p value of less than .10 only with the nonparent measure among Republican-controlled states (r = .36), and this effect is due to expansion in a single Republican-dominated state, Arizona (Figure 2B), which is the only Republican-controlled state to have offered Medicaid assistance to nonparents prior to the ACA. Overall then, policy legacy does have a strong association with advancement toward Medicaid, but we see only modest evidence that existing institutions have the capacity to work counter to the powerful influence of consolidated partisan power on expansion. We therefore withhold strong conclusions for the moment.

Provider influence

Do medical providers influence states’ ACA decisions? The measure of hospital beds per 1,000 people (Figure 3A) is significantly related to movement toward expansion (r = −.30; p < .05), but the negative direction of the effect is contrary to our expectations. Even with the exclusion of the outlier, South Dakota, the contrary finding does not go away (r = −.26; p < .10).

Perhaps this unexpected result appears because the number of hospital beds in states does not adequately represent influence from the medical profession. Further investigation reveals that our measure of hospital beds is inversely related to our measure of physicians per 10,000 people (r = −.32; p < .05), which casts doubt on the idea that hospital beds are an effective indicator of provider power in the context of U.S. states. Looking at our analysis of physician power (Figure 3B), we see the expected evidence that physicians have pushed states further toward Medicaid expansion (r = .48; p < .001). Moreover, the effect remains statistically significant within states controlled by Democrats (r = .60; p < .05) and is visible within Republican-controlled states.

It is beyond the scope of this study to determine the precise cause of the inconsistent findings produced by conventional measures of provider power, but we can perhaps move past any impasse by looking beyond these relatively abstract measures and examining what hospitals and providers are actually doing. Looking at Table 2, we gain a preliminary notion of why the mere presence of providers shows no consistent linkage with expansion. Contrary to our expectations, the descriptive statistics for provider contributions show that Republicans receive more support than Democrats from hospitals, physicians, and AHA groups. This suggests there is no clear tendency among these groups to support Medicaid expansion as suggested in prior studies. There is certainly no consensus.

We can now ask whether provider support, if the balance lies with any one party in any particular state, has an influence on expansion. In Figure 2, panels C to E, we see that contributions by hospitals and physicians to political campaigns are associated with Medicaid expansion in the expected direction (r = −.35; r = −.45), and these associations are statistically significant (p < .05; p < .01). However, to determine that this is not merely the result of the relationship between contributions and electoral success or a result of donors hedging contributions by contributing to the likely victor, these effects would need to be significant even among states controlled by a single party. Judging by the split-bivariate models presented in Figure 2, this is not the case. Visibly, the split-bivariate slopes have mixed signs, and there is no statistical significance in any case. While we withhold strong conclusions until after the remaining analyses, the split-bivariate regression offers little evidence of a statistically significant influence of providers on expansion net of party power.

Political values and racial politics

Does public ideology influence Medicaid expansion? Like partisanship among government elites, aggregate liberal-conservative ideology among the public is closely related to movement toward Medicaid expansion (Figure 4A). States where people identify more as liberals have moved closer to Medicaid expansion than states where people identify more as conservatives (r = −.58; p < .001). Looking at the dashed lines in Figure 4A, this appears to remain the case within state governments that are dominated by Democrats, split, or dominated by Republicans. Within states controlled by Democrats, the effect remains statistically significant (r = −.68; p < .01).

Racial resentment is also closely linked to Medicaid expansion decisions (Figure 4B). States with lower racial sympathy and higher racial resentment on average show a strong resistance to Medicaid expansion (r = −.40; p < .01). When we break down the analysis by partisan domination of the government, we see that Democrat-controlled states generally range lower in racial resentment scores overall (West Virginia is a notable exception). However, Democrat-controlled states are also more sensitive to varying levels of racial resentment in their Medicaid expansion decisions (r = −.62; p < .05) when compared with Republican-controlled states or split-control states, which show no statistically significant within-group difference in regard to racial resentment. Thus, as with political values that are not explicitly racial, racial resentment has a strong negative influence on expansion, and while racial resentment acts as a cross pressure against expansion among Democrat-controlled states, racial sympathy does not overpower opposition to expansion among Republican-controlled states.¹

State fiscal capacity

Does a state’s fiscal capacity influence decisions to expand Medicaid? Personal income in the states (Figure 5A) has an overall positive effect on movement toward Medicaid expansion (r = .28; p < .05), and this effect remains when the outlier Delaware is dropped from the analysis. Yet state revenue has no effect at all, and neither personal income nor state revenue has an effect when analyses are broken down into subgroups of states dominated by a specific party. This suggests only a modest effect of state capacity on expansion.

Multivariate Analysis

The bivariate and split-bivariate models offer helpful insight into the process of Medicaid expansion. Specifically, they indicate which factors matter for Medicaid expansion beyond the obvious influence of party power in the states. Although these methods are illuminating, we could place greater confidence in our findings if they were supported by analyses employing more traditional multivariate regression analyses.

This section proceeds in three phases. First, we take predictor variables with significant bivariate associations revealed above and enter them together into a multivariate ordinary least squares (OLS) regression model using the same outcome measure discussed above.² We then examine these same predictors in a multivariate logistic regression model using a dichotomous outcome measure indicating whether or not states have expanded. This model includes only the most up-to-the-minute information on state expansion. Finally, to make the best possible use of all available information, we assemble and pool two waves of time-series panel data that capture the time-varying status of the predictors and dichotomous outcome. These data are used to fit a Pooled Panel Time Series (PPTS) model that contains twice the observations of a simple cross-sectional regression and therefore overcomes some of the problems associated with having only 50 states to examine at any one point in time. Jackknifed standard errors are reported in each analysis, as discussed in the Supplementary Appendix B.

Modeling movement toward Medicaid with multivariate OLS regression

Based on our theoretical foundation concerning the direction of the predictors’ effects, we report one-tailed tests of statistical significance and accordingly employ a cutoff of p < .10. Judging from the multivariate OLS model results displayed in Column A of Table 3, there is little evidence that any factor besides public ideology has an independent influence on movement toward Medicaid expansion, including elite partisanship. The net effect of public ideology, however, is substantial. A one-unit increase in the seven-level political ideology scale results in a 4.2-unit drop in the scale measuring movement toward Medicaid expansion, net of other factors. It is well known that partisan politics have played a major role in Medicaid expansion, yet this model suggests that it is public ideology, not elite partisan power, that has the clearest independent impact on expansion.

Table 3.

Regressions Predicting Medicaid Expansion under the Affordable Care Act.

	Column A	Column B	Column C
Model	OLS, multivariate	Logit, multivariate	Logit, PPTS with RE
Outcome variable	Movement toward Medicaid expansion	Expanded Medicaid	Expanded Medicaid
Predictors
Democrat Control	1.57	—	1.75^†
	(1.23)		(1.01)
Strong Medicaid Legacy	1.04	—	1.97^†
	(0.73)		(1.10)
Provider Influence	−14.10	−44.82*	−31.59
Pro-anti	(11.73)	(19.47)	(20.74)
Public Ideology	−4.20*	−4.64	−7.85^†
Liberal-Conservative	(1.89)	(4.23)	(4.00)
Public Racial Resentment	−0.53	−0.29	−1.04
Sympathetic-Resentful	(1.52)	(2.66)	(1.64)
Income Per Capita	−0.00	−0.00	−0.00
	(0.00)	(0.00)	(0.00)
R ²	.39 (Adjusted)	.39 (Pseudo)
N	50	34	100

Note. Jackknifed standard errors are in parentheses. PPTS = Pooled Panel Time Series; RE = random effects.

†

p < .1. *p < .05.

Modeling Medicaid expansion with multivariate logistic regression in cross-section

Here, we shift from the continuous measure of movement toward Medicaid expansion to a simpler but straightforward and more up-to-the-minute dichotomous outcome measure indicating whether or not a state has expanded Medicaid under the ACA. As of December 2014, 27 states have expanded Medicaid, and 23 have rejected the expansion (KFF 2014).³

The same predictors from the multivariate OLS model are now entered into a logistic regression model with the dichotomous outcome (1 = expanded, 0 = rejected). The results are seen in Column B of Table 3. Immediately apparent in this model are the missing coefficients for Democrat control and Medicaid legacy. They are missing because every state under Democrat control or with a strong Medicaid legacy has expanded or accepted expansion as of this writing. In other words, this model suggests either factor is, itself, sufficient for expansion.⁴ These states are automatically dropped from the analysis.⁵

Looking at the remaining predictors, we see that only the logit coefficient for provider influence is statistically significant. This contrasts with the OLS model where public ideology had the only statistically significant independent effect. Thus, although the results from each model are suggestive, they must be interpreted with caution due to their apparent instability. This is particularly the case with the logit model, as the already-small N has been reduced even further from 50 states to 34, and results from the maximum likelihood estimation procedure involved with logistic regression should be viewed with caution with an N this size. We would be more confident with a model providing more reliable coefficients.

Medicaid expansion with a PPTS analysis

To transcend the limitations imposed by the presence of only 50 states, we employ a PPTS model that uses multiple waves of data from the same units over time. Limitations in available data remain important, as it was only possible to collect two waves of data. Yet this alone doubles the N available for analysis. All data for the first wave were collected from before the 2012 Supreme Court decision and elections (see Supplementary Appendix C). Data from the logistic regression above compose the second wave.

There are two commonly employed types of PPTS analysis, fixed effects (FE) models and random effects (RE) models. FE models are considered more consistent, while RE models are considered more efficient (Hausman 1978). The Hausman test implemented in Stata 12.0 was used to determine which model is most appropriate in our case. The resulting coefficient was nonsignificant, which indicates the RE model is not only more efficient but is also consistent and therefore preferable to the FE model.

Logit coefficients from the RE model are reported in Column C of Table 3. While Democrat control or a strong policy legacy is sufficient for Medicaid expansion according to the cross-sectional logit model, this is not evident in the RE model because the latter includes state-time observations from a time before all Democrat-controlled states and states with strong Medicaid legacies committed to expanding. Nonetheless, the results for party dominance and policy legacy are still informative because each still influences expansion net of other factors.⁴ Public ideology also has an independent effect on expansion, which suggests the inconsistent findings for this factor between the cross-sectional OLS and logit models were due to their lesser statistical power.

At first, one might be surprised that the coefficient for provider influence is not significant in the RE model, as this model could be considered more powerful. However, we must remember that provider data for the first wave in the RE model were collected before the June 2012 Supreme Court decision entrenched the ACA and rendered Medicaid optional. There was much less of a reason for providers to advocate for or against Medicaid expansion specifically. Accordingly, the first wave data wash out the effect seen in the cross-sectional logit model. In this case, we should rely on the cross-sectional logit model as it incorporates up-to-the-minute data on provider support from after the Supreme Court decision. Its relatively strong coefficient for provider influence allows us to maintain our assertion that even after accounting for other factors, in the time since Medicaid expansion was deemed optional, providers have swayed Medicaid expansion decisions in their respective states, whether toward expansion or rejection.

Discussion

Each of the approaches above provide unique information. It is therefore useful to summarize the findings before making general conclusions. Our bivariate and split-bivariate analyses confirm prior research suggesting partisan control in a state is a key predictor of a state’s progress with Medicaid expansion. However, these models also show that there are cross pressures on politicians that cause them to reject their party’s official position. Policy legacy was linked with expansion even within the set of states dominated by Republicans, as demonstrated by Arizona, and ideology and racial resentment remained significant within the set of states dominated by Democrats.

The traditional measures of provider power proved inconsistent. To overcome this issue, we examined the actions of providers directly. Contrary to the expectations of researchers thus far, there is no unified effort to expand Medicaid on the part of health care providers. Hospitals, physicians, and AHA organizations did not show strong support for Medicaid expansion through their contributions to Democrats in the 2012 election, even after they were aware of the June 28 Supreme Court Decision. In fact, each of these groups tended to support Republicans, which suggests they largely opposed Medicaid expansion, perhaps preferring alternatives.

Although providers did not speak with one voice, the prevailing preferences of providers in the respective states were heard. Results from the split-bivariate models were not decisive, but these results were strengthened by findings from the more up-to-the-minute multivariate logistic regression model, which provided strong evidence that providers had an independent influence in their respective states’ Medicaid decisions since the Supreme Court rendered expansion optional. Greater provider support tended to produce expansion, while greater provider resistance tended to obstruct expansion.

On balance, the multivariate regression results confirm that partisan control, policy legacy, provider influence, and public ideology have statistically significant independent effects on expansion. The cross-sectional logit model also suggests that Democrat dominance and policy legacy are each themselves sufficient for expansion.⁴

Future research investigating the varied political activism displayed by providers in the different states is clearly warranted. Future research could also advance our understanding of Medicaid expansion by examining key states in-depth, particularly those that bucked the general trends during the rollout of the ACA. Looking at the split-bivariate figures, we might not expect Ohio and Iowa to have expanded, as they tend to group with states that did not expand by most measures. Yet both states have expanded Medicaid under the ACA. Michigan and North Dakota also chose expansion despite being dominated by Republicans. Case studies on such exceptional states could reveal important but as-yet unidentified factors influencing Medicaid expansion idiosyncratically or generally.

Conclusion

The ACA represents the single largest expansion of the American welfare state since the enactment of Medicare and Medicaid in 1965. Moreover, it occurred during an era when the primary trend in welfare state development appeared to be either retrenchment or modest restructuring of existing programs (Huber and Stephens 2010).

In the months leading up to the enactment of the ACA, bitter partisan debates were waged even though the ACA’s main provisions were modeled after a Republican plan that initially had broad bipartisan support. These partisan conflicts spilled over into the ACA implementation process in the form of numerous delaying tactics and lawsuits against the individual mandate and the mandatory Medicaid expansion. When the Supreme Court ruled in favor of the mandate but made Medicaid expansion optional for the states, this issue became the new battleground. Blocking Medicaid expansion is the main vehicle opponents of the ACA have to delay implementation. As the conservative Cato Institute advised, “(T)he Supreme Court empowered states to block the Medicaid expansion . . . States should exercise that freedom” (Cannon 2013:24).

Although partisan politics did have a significant effect on state decisions, other factors including policy legacies, provider interests, political ideologies, and racial attitudes have had an important role in states’ decisions to expand Medicaid under the ACA. These results indicate that states are not merely laboratories of democracy in the sense that they demonstrate what is possible at the federal level. Factors within states are relevant to both the policy formation process and the implementation process. Theorists seeking to understand the development of the American welfare state should expand their focus beyond the legislative process to include factors involved in putting policy into action. For practitioners, the case of Medicaid implementation demonstrates that policies will be more robust where they provide mechanisms through which states can overcome their varied inhibitions.

Footnotes

Acknowledgements

We thank Dan Tope for valuable feedback on a prior version of this manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biographies

Daniel Lanford is a doctoral candidate in the Department of Sociology at Florida State University. He is currently conducting research at the Pepper Institute on Aging and Public Policy, where his interests include health policy, the social psychology of politics, politicized understandings of race and gender, and social movements. His current works investigate the social dynamics of policy relating to the Patient Protection and Affordable Care Act and elite social movements. His dissertation focuses on the causes and consequences of stability and change in mass opinion as it relates to the politics of salient welfare policy debates.

Jill Quadagno is a professor of sociology at Florida State University, where she holds the Mildred and Claude Pepper Eminent Scholar Chair in Social Gerontology. She is the author of more than 75 articles and 12 books. Her research interests include social welfare policy, health policy, aging, and political sociology. Her most recent book is One Nation, Uninsured: Why the U.S. Has No National Health Insurance (Oxford University Press). A past president of the American Sociological Association, in 2010, she was elected to the Institute of Medicine.

References

Ansolabehere

Stephen.

2013. Cooperative Congressional Election Study, 2012: Common Content. Cambridge, MA: Harvard University. Retrieved May 6, 2015 (http://cces.gov.harvard.edu).

Barrilleaux

Charles

Brace

Paul

. 2007. “Notes from the Laboratories of Democracy: State Government Enactments of Market and State-based Health Insurance Reforms in the 1990s.” Journal of Health Politics, Policy and Law 32(4):637–65.

Beaussier

Anne L.

2012. “The Patient Protection and Affordable Care Act: The Victory of Unorthodox Lawmaking.” Journal of Health Politics, Policy and Law 37(5):741–78.

Blumer

Herbert

. 1958. “Race Prejudice as a Sense of Group Position.” The Pacific Sociological Review 1(1):3–7.

Bobo

Lawrence D.

1999. “Prejudice as Group Position: Microfoundations of a Sociological Approach to Racism and Race Relations.” Journal of Social Issues 55(3):445–72.

Brown

Lawrence D.

2011. “The Elements of Surprise: How Health Reform Happened.” Journal of Health Politics, Policy and Law 36(3):419–27.

Cannon

Michael

. 2013. 50 Vetoes: How States Can Stop the Obama Health Care Law. Washington, DC: Cato Institute.

Casalino

Lawrence

. 2013. “Professionalism and Caring for Medicaid Patients—The 5% Commitment?” New England Journal of Medicine 369:1775–77.

Cunningham

Peter J.

2002. “Mounting Pressures: Physicians Serving Medicaid Patients and the Uninsured, 1997–2001.” Center for Studying Health System Change 2002(6):1–4.

10.

Cunningham

Peter J.

May

Jessica

. 2006. “Medicaid Patients Increasingly Concentrated among Physicians.” Center for Studying Health System Change 2006(16):1–4.

11.

Davis

Corey

. 2012. “Disproportionate Share Hospitals and the Medicaid Expansion.” National Health Law Program. Retrieved May 6, 2015 (http://www.healthlaw.org/images/stories/DSH_QA_final.pdf).

12.

DeViney

Stanley

. 1983. “Characteristics of the State and the Expansion of Public Social Expenditures.” Comparative Social Research 6:151–74.

13.

Dixon

Jeffrey C.

2006. “The Ties That Bind and Those That Don’t: Toward Reconciling Group Threat and Contact Theories of Prejudice.” Social Forces 84(4):2179–204.

14.

Elmendorf

Douglas W.

2011. CBO’s Analysis of the Major Health Care Legislation Enacted in March 2010. Testimony by Douglas Elmendorf, Director of the Congressional Budget Office, before the Subcommittee on Health, Committee on Energy and Commerce, U.S. House of Representatives. March 30.

15.

Engquist

Gretchen

Burns

Peter

. 2002. “Health Insurance Flexibility and Accountability Initiative: Opportunities and Issues for States.” State Coverage Initiatives Issue Brief 3(2):1–6.

16.

Erikson

Robert S.

Wright

Gerald C.

McIver

John P.

1993. Statehouse Democracy: Public Opinion and Policy in the American States. Cambridge, MA: Cambridge University Press.

17.

Flora

Peter

Alber

Jens

. 1981. “Modernization, Democratization, and the Development of Welfare States in Western Europe.” Pp. 37–80 in The Development of Welfare States in Europe and America, edited by Flora

Heidenheimer

A. J.

New Brunswick, NJ: Transaction.

18.

Gilens

Martin

. 1999. Why Americans Hate Welfare: Race, Media and the Politics of Antipoverty Policy. Princeton, NJ: Princeton University Press.

19.

Gray

Virginia

. 1994. “Federalism and Health Care.” PS: Political Science & Politics 27(2):217–20.

20.

Grogan

Colleen

. 2014. “Medicaid: Designed to Grow.” Pp.142-163 in Health Politics and Policy. 5th ed., edited by Morone

Ehlke

Stamford, CT: Cengage Learning.

21.

Grogan

Colleen

Andrews

Christian

. 2014. “Medicaid.” Pp.337-354 in The Oxford Handbook of U.S. Social Policy,edited by Beland

Morgan

New York: Oxford University Press.

22.

Hacker

Jacob

. 2002. The Divided Welfare State. New York: Cambridge University Press.

23.

Hacker

Jacob

. 2010. “The Road to Somewhere: Why Health Reform Happened, or Why Political Scientists Who Write about Public Policy Shouldn’t Assume They Know How to Shape It.” Perspectives in Politics 8(3):861–76.

24.

Hacker

Jacob

. 2011. “Why Reform Happened.” Journal of Health Politics, Policy and Law 36(3):439–11.

25.

Hacker

Jacob

Pierson

Paul

. 2011. Winner Take All Politics: How Washington Made the Rich Richer and Turned Its Back on the Middle Class. New York, NY: Simon & Schuster.

26.

Hausman

Jerry A.

1978. “Specification Tests in Econometrics.” Econometrica 46(6):1251–71.

27.

Holahan

John

Buettgens

Matthew

Carroll

Caitlin

Dorn

Stan

. 2012. “The Cost and Coverage Implications of the ACA Medicaid Expansion: National and State-by-State Analysis.” Kaiser Commission on Medicaid and the Uninsured, The Kaiser Family Foundation, Washington, DC.

28.

Huber

Evelyn

Stephens

John

. 2010. Development and Crisis of the Welfare State. Chicago, IL: University of Chicago Press.

29.

Hurwitz

Jon

Peffley

Mark

. 2005. “Playing the Race Card in the Post-Willie Horton Era: The Impact of Racialized Code Words on Support for Punitive Crime Policy.” Public Opinion Quarterly 69(1):99–112.

30.

Jacobs

Lawrence R.

Callaghan

Timothy

. 2013. “Why States Expand Medicaid: Party, Resources, and History.” Journal of Health Politics, Policy and Law 38(5):1023–50.

31.

Kaiser Commission on Medicaid and the Uninsured. 2005. “Medicaid Section 1115 Waivers: Current Issues.” Key Facts, The Kaiser Family Foundation, Washington, DC.

32.

Kaiser Family Foundation. 2012. A Guide to the Supreme Court’s Affordable Care Act Decision. Retrieved May 6, 2015 (http://kaiserfamilyfoundation.files.wordpress.com/2013/01/8332.pdf).

33.

Kaiser Family Foundation. 2013a. State Decisions for Creating Health Insurance Exchanges and Expanding Medicaid. Washington, DC. Retrieved December 2, 2013 (www.kff.org).

34.

Kaiser Family Foundation. 2013b. Where Are States Today? Medicaid and CHIP Eligibility Levels for Children and Non-disabled Adults. Washington, DC. Retrieved December 2, 2013 (www.kff.org).

35.

Kaiser Family Foundation. 2014. Status of State Action on the Medicaid Expansion Decision. Retrieved April 1, 2015 (http://kff.org/health-reform/state-indicator/state-activity-around-expanding-medicaid-under-the-affordable-care-act/).

36.

Kim

Ae-sook

Jennings

Edward

. 2012. “The Evolution of an Innovation: Variations in Medicaid Managed Care Program Extensiveness.” Journal of Health Politics, Policy and Law 37(5):815–49.

37.

Kinder

Donald R.

Sanders

Lynn M.

1996. Divided by Color: Racial Politics and Democratic Ideals. Chicago, IL: University of Chicago Press.

38.

King

Linda

. 2008. 50-State Assessment of Lobbying Expenditure Data. National Institute on Money in State Politics. Retrieved May 6, 2015 (http://classic.followthemoney.org/press/PrintReportView.phtml?r=452).

39.

Morone

James

Dunham

Andrew

. 1985. “Slouching towards National Health Insurance: The New Health Care Politics.” Yale Journal on Regulation 2:263–90.

40.

National Conference of State Legislatures. 2013. Post-election 2013 State and Legislative Partisan Composition. Washington, DC. Retrieved May 6, 2015 (www.ncls.org).

41.

Oberlander

Jonathan

. 2011. “Under Siege: The Individual Mandate for Health Insurance and Its Alternatives.” New England Journal of Medicine 364(12):1085–87.

42.

Olson

Laura Katz

. 2010. The Politics of Medicaid. New York: Columbia University Press.

43.

Pierson

Paul

. 2000. “Increasing Returns, Path Dependence and the Study of Politics.” American Political Science Review 94(2):252–67.

44.

Price

Carter

Eibner

Christine

. 2013. “For States That Opt Out of Medicaid Expansion: 3.6 Million Fewer Insured and $8.4 Billion Less in Federal Payments.” Health Affairs 32(6):1030–36.

45.

Quadagno

Jill

. 2004. “Why the United States Has No National Health Insurance: Stakeholder Mobilization against the Welfare State, 1945–1996.” Journal of Health and Social Behavior 45:25–44.

46.

Quadagno

Jill

. 2010. “Institutions, Interests and Ideology: An Agenda for the Sociology of Health Care Reform.” Journal of Health and Social Behavior 51(2):125–36.

47.

Quadagno

Jill

. 2014. “Right Wing Conspiracy? Socialist Plot? The Origins of the Patient Protection and Affordable Care Act.” Journal of Health Politics, Policy and Law 39(1):35–56.

48.

Schneider

Saundra K.

1997. “Medicaid Section 1115 Waivers: Shifting Health Care Reform to the States.” Publius 27(2):89–109.

49.

Sears

David O.

Sidanius

Jim

Bobo

Lawrence

eds. 2000. Racialized Politics: The Debate About Racism in America. Chicago, IL: University of Chicago Press.

50.

Skocpol

Theda

. 1997. Boomerang. New York: W.W. Norton.

51.

Snyder

Laura

Rudowitz

Robin

Garfield

Rachel

Gordon

Tracy

. 2012. Why Does Medicaid Spending Vary Across States: A Chart Book of Factors Driving State Spending. Kaiser Commission on Medicaid and the Uninsured, The Kaiser Family Foundation, Washington, DC. Retrieved May 6, 2015 (www.kff.org).

52.

Sommers

Benjamin

Arntson

Emily

Kenney

Genevieve

Epstein

Arnold

. 2013. “Lessons from Early Medicaid Expansion under Health Reform: Interviews with Medicaid Officials.” Medicare & Medicaid Research Review 3(4):1–19.

53.

Sommers

Benjamin

Kenney

Genevieve

Epstein

Arnold

. 2014. “New Evidence on the Affordable Care Act: Coverage Impacts of Early Medicaid Expansions.” Health Affairs 33(1):78–87.

54.

Starr

Paul

. 1983. The Social Transformation of American Medicine. New York: Basic Books.

55.

Tesler

Michael

. 2012. “The Spillover of Racialization into Health Care: How President Obama Polarized Public Opinion by Racial Attitudes and Race.” American Journal of Political Science 56(3):690–704.

56.

Tope

Daniel

Hickman

Lisa

. 2012. “The Politics of Children’s Health Insurance Policy.” Sociological Perspectives 55(1):43–66.

57.

Winter

Nicholas J. G.

2006. “Beyond Welfare: Framing and the Racialization of White Opinion on Social Security.” American Journal of Political Science 50(2):400–20.