The Effect of Issue Framing on Public Support for Obamacare

Currently, the United States Patient Protection and Affordable Care Act is one of the most contentious laws ever passed by Congress. For the first time in the nation’s history, thousands of individuals will be allowed to obtain health insurance under a government-controlled system. At the same time, in the field of political science, the framing of issues by both the media and politicians forms a highly controversial subject for scholars and students alike. Framing, which refers to a process by which a person develops a particular view on an issue based on its wording, has the ability to drastically affect public opinion. The goal of this paper is to examine the effects that framing has on this new healthcare law. Although prior studies have analyzed the effects of framing on other issues, none has yet evaluated the consequences of framing on Obamacare. I conduct an experiment manipulating three different equivalency frames and four different rationales. Ultimately, I find that issue rationales have the greatest effect on public support for healthcare. Specifically, what you call the law (i.e. Obamacare vs. Affordable Care Act) has little sway over public opinion. Instead, providing individuals with the rationale that the law will allow people with pre-existing conditions to obtain health insurance significantly increases support for this law. I end with a discussion of the implications of these findings. Introduction In the world of healthcare, one pressing concern prior to the implementation of Obamacare was the desire of individuals with pre-existing conditions to obtain health insurance. According to a 2009 CNN story by Elizabeth Cohen, individuals with pre-existing conditions that would normally obtain health insurance through employment are being dropped by health insurance companies or forced to pay thousands of dollars because their conditions leave them unable to work (Cohen 2009). For example, nineteen-year old Stuart Wald had not one, but three different pre-existing conditions, including schizophrenia and bipolar disorder. According to Cohen, Wald was so sick that he was unable to work, forcing his family into a financial crisis just to pay for his healthcare. However, the passage of the United States Patient Protection and Affordable Care Act, or “Obamacare” as it has become to be known, has sparked new hope for individuals like Stuart Wald. Meanwhile, public support towards the law itself is currently mixed. However, as we will see through this paper’s findings, if politicians were to slightly alter the way they frame the healthcare issue, then not only would support for the law increase, but individuals like Stuart Wald would be made aware of the fact that they are eligible for health insurance under this new law. According to political scientist James Druckman (2003), issue framing occurs when a speaker emphasizes certain aspects of an issue so that an individual may prioritize consideration of these aspects as they develop their own opinions. Imagine, briefly, how useful it would be for politicians or the media to know the best way to frame an issue so as to garner public support. Would this ability be more advantageous or harming to America’s form of democracy? In order to gain traction on this larger question, this project examines the effects of issue framing in the context of public opinion on the United States Affordable Care Act. Specifically, this paper seeks to answer the following question: How does issue framing affect public support for the recent healthcare law? In order to evaluate this question, I conducted an online survey-experiment testing two hypotheses. In short, I argue first that changing the issue frame (i.e. Obamacare v. Affordable Care Act) will not create a significant shift in public support for the healthcare law. Secondly, I argue that public support for the law will increase if respondents are told that the law will allow people with pre-existing conditions to obtain health insurance. However, before I elaborate further on these hypotheses, let us turn first to a review of the most relevant literature in the field of framing. Literature Review Defining Issue Framing and Related Concepts To begin, I will define each of the major terms related to framing. However, it is worth noting that there are multiple definitions of framing that each focus on a different actor in the framing process. For this paper, I use the Chong and Druckman (2007) definition, which refers to framing as a process by which an individual develops or redevelops their view on a specific issue. Furthermore, although the definition of framing used by political scientists is not consistent throughout the literature, two major sub-categories of framing emerge with some regularity. The first major subcategory, “equivalency framing,” refers to a situation in which individuals are presented with two or more identical options that differ only in the way that they are worded (Tversky and Kahneman 1981). The other major subcategory, “issue (emphasis) framing,” is referred to by Entman and colleagues (2009) as being, “...concerned with increasing or decreasing the salience of an issue or consideration when formulating an opinion.” Unlike equivalency framing, the options presented are not identical. In truth, depending on who is presenting the information, one option may be stressed more than another. Therefore, when choosing, individuals must weigh how relevant each option is to their own set of personal experiences and values. Another point worth emphasizing is the meaning of the term “framing effect.” Druckman (2001) and Entman (2009), define a framing effect as the process by which an individual forms an opinion toward a particular issue after exposure to frames presented by the media or politicians. However, one should be careful not to assume that this opinion will be in favor of the views supported by those doing the framing. In fact, even if an individual forms the opposite opinion, it is still said to be a framing effect. Thus, it is exposure to a frame and the subsequent formation of an opinion that characterizes a framing effect, and not the nature of the opinion itself. One final concept that needs defining is the term “considerations.” Two terms that are often used interchangeably with considerations are “implications” and “rationales.” Unlike framing, “consideration” is welldefined in the existing literature in the field and little disagreement exists on what it entails (Chong and Druckman 2007; Druckman 2001; Tversky and Kahneman 1981; Tversky and Kahneman 1986). Briefly put, a consideration is any effect or outcome that will occur as a consequence of a particular frame. For instance, using the example of a hate speech rally, a news story using a public safety frame might emphasize the possibility of violent crime that could occur as a result of the rally. Different Types of Framing in Use In terms of practical usage, there have been several important studies that have examined the effects of equivalency framing. It is worth noting though that these studies often disagree as to the merits of using equivalency framing to affect public opinion. For example, Tversky and Kahneman’s 1981 Asian disease experiment presented individuals with two different problems (Tversky and Kahneman 1981). Although the answers to these problems were identical apart from slightly altered wording, Tversky and Kahneman found that individuals chose the option that they believed would avoid the most risk. However, Druckman (2001) disagreed with this conclusion, arguing instead that equivalency framing was nothing more than a method of deception used to trick individuals. On the other hand, issue (emphasis) framing has received less criticism from contemporary scholars. In fact, there have been a number of studies that have analyzed the effect of issue framing on support for political policies (Druckman 2001; Nelson et al. 1997). Overall, most of these studies agree that issue framing can have a large effect on public opinion. However, there is some disagreement as to how often framing effects occur. For instance, Druckman (2001) finds that framing effects are less likely to occur if an individual already has strong beliefs about the issue in question. Furthermore, he also finds that instead of blindly supporting one frame over another, individuals tend to carefully weigh the benefits and risks of each frame before deciding which to support. Another group of scholars finds the opposite (Nelson et al. 1997). Nelson and colleagues suggest that individuals who are already familiar with a particular issue are, in fact, more likely to go along with a particular frame. In making their argument, they also carefully contrasted issue framing with persuasion. Unlike persuasion, they insist that framing activates information already at an individual’s disposal. Here, we can see that the concepts that separate priming and framing sometimes bleed into one another. Additionally, a number of studies suggest that the inclusion of considerations greatly influences support for a particular frame. As stated above, Tversky and Kahneman (1981) found that individuals were more likely to choose the option that they believed would benefit them the most. What is more, individuals also tended to choose the option that they believed avoided unnecessary risk. However, Druckman (2001) concluded that framing effects were less likely to occur if a rationale or consideration was provided to an individual. So, although including rationales or considerations seems to create framing effects on an experimental level, it is still unclear how effective such considerations would be if put into actual use by politicians or the media. Where’s the Gap? Now, with the above in mind, it is only natural to ask how this analysis will attempt to fill a gap in current literature on issue framing. First, little has been written on the effects that both issue framing and equivalency framing may have on support for the healthcare system in the United States. Similar studies have focused on political issues as diverse as immigration (Merolla et.al 2013), government spending (Jacoby 2000), and international relations (Mintz and Redd 2003), but healthcare has been all but absent from such discussions. It is worth briefly noting that late-night television host Jimmy Kimmel did reveal in one of his skits that some Americans believed that Obamacare and the Affordable Care Act were two entirely different laws. Unfortunately, he was far less scholarly in his methods than was desirable for such a potentially important issue. This study attempts to use both equivalency and issue framing in order to analyze support for the United States Affordable Care Act. Although Druckman (2001) suggests that equivalency framing is rarely used by politicians or the media, by switching the frame presented to respondents (i.e. Obamacare v. Affordable Care Act), I am doing just that. Essentially, these frames are identical to one another (both refer to the same law), but by changing what you call law, I am attempting to gauge whether or not the law’s name factors in to support for the law itself. Furthermore, this study will also examine how the inclusion of considerations affects public support for the recent healthcare law. This is where the issue (emphasis) framing aspect of this experiment comes into play. In accordance with Druckman’s concept of issue framing (2003), I am testing to see if these considerations are at all relevant as the public forms its preference on a particular law. Thus, there are several ways in which this study will hopefully contribute to filling the current gaps in the literature on issue framing. Hypotheses Hypothesis 1 H1: Varying the issue frame (i.e. Obamacare vs. Affordable Care Act) will not affect the level of support for the underlying health care policy. In discussing my first hypothesis, I would like to note that many of the places where typical American citizens get their news about new laws such as this one have used each of the issue frames (Obamacare, Affordable Care Act, and Health Care Reform) interchangeably. Perhaps the label “Obamacare” was originally used as a derogatory term for the law by its detractors. However, because the law has been referred to by this name so frequently since 2008 and 2009, I believe that the connotations associated with the law have shifted from being wholly negative to more neutral in nature. Merolla and colleagues (2013) have found a similar trend with the effectiveness of the term “amnesty” on support for a pathway to citizenship. Moreover, as Druckman and Nelson (2003) note, framing effects tend to be short-lived phenomena. Hence, this hypothesis was constructed with the belief that what you call the recent healthcare law would have similar short-lived effects on public opinion. Hypothesis 2 H2: Mentioning that individuals with pre-existing conditions will be guaranteed access to health insurance will increase support for the health care law (compared to those not receiving this rationale). Here, my hypothesis was based on a logical assumption. Those citizens who have been denied insurance under a private healthcare plan because of a pre-existing condition will be overjoyed if they are told that the new healthcare law will finally allow them to be insured. What is more, those individuals who know or are related to someone with a pre-existing condition will be sympathetic to the fact that the new law will allow their friends and relatives to get insurance. Clearly, as evidenced by the 2009 story related above, insurance and health issues can take a large toll on a family, both economically and emotionally. Therefore, my reasoning for this expectation is based on self-interest. Methods To assess these hypotheses, I conducted a survey experiment in which the unit of analysis was the individual. Using the Qualtrics survey building software, I created a survey and ran it online using Amazon’s Mechanical Turk marketplace. In order to recruit individuals for this survey, I created a “hit” on Amazon’s Mechanical Turk marketplace, offering $0.35 as an incentive to take the survey. When the potential respondent clicked on this “hit,” they were brought to a page stating the terms and conditions of the survey. In order to obtain payment, potential respondents were required to consent to the terms and conditions, to complete the survey in its entirety, and to provide their Mechanical Turk ID. Also, potential respondents were required to be both 18 years of age and have a United States IP address. Once the individual agreed to these terms, they were forwarded to the survey itself. The survey consisted of ninety-one questions querying respondents on questions dealing with demographics, SES, political predispositions, and policy preferences. Of particular importance, participants were presented with a question assessing support for the new healthcare law that included manipulations for both issue frame and rationale. The question read: “Some people say Figure 1: Difference in Mean Support for the Pre-Existing Condition Rationale across 3 Equivalency Healthcare Frames that (insert manipulation 1 here) will (insert manipulation 2 here). To what extent do you support or oppose this policy?” This question was based off of a 4x3 cell experimental design (2 manipulations) in which participants were randomly distributed across 12 cells. Participants were randomly exposed to one of the following “issue frame” manipulations: “Obamacare,” “The Affordable Care Act,” or “Health Care Reform.” Additionally, participants randomly received one of the following four rationales regarding the health care law: “drive up the cost of healthcare premiums,” “drive down the cost of healthcare premiums,” “allow individuals with pre-existing conditions to receive obtain health insurance,” or a control in which no rationale was provided. Finally, I used difference of means t-tests and ordinary least squares regression analyses with balance checks to assess each hypothesis. Results I now turn to the results of the statistical analyses (T-tests, bivariate regressions, and multiple linear regressions) assessing each hypothesis. Hypothesis 1 To assess my first hypothesis, which argued that changing the equivalency frame would not affect support for the healthcare law, the policy questions were grouped into four sets of three pairings, for a total of twelve pairings in all. These four sets were created according to the rationale provided to the respondent. This step was taken to ensure that the rationale was being held constant, while the equivalency frame was the manipulation being analyzed. The first set of pairings compared each equivalency frame and its respective control group to one another. The first pairing compared “Obamacare” without any rationale to “The Affordable Care Act” without any rationale. T-test results indicate that there is no statistically significant difference in mean support for these two groups (p=.499). Furthermore, a multiple linear regression including relevant balance checks returns the same result (p=.779). The second pairing from this set, comparing “Obamacare” without any rationale to “Health Care Reform” without any rationale, is statistically insignificant as well. The third pairing from this set compared “The Affordable Care Act” without any rationale to “Health Care Reform” without any rationale. After running a T-test, evidence shows that there was a marginally significant statistical difference between policy support for these groups (p=.077). Substantively, evidence shows that when moving from “Health Care Reform” without any rationale to “The Affordable Care Act” without any rationale there is a .243 shift in policy support. However, a multiple linear regression with balance checks reports that the difference between groups is even less significant (p=.204). The second set of pairings compared each equivalency frame grouped with the “premium increase” rationale to one another. The first pairing in this set compared “Obamacare” with the “premium increase” rationale to “The Affordable Care Act” with the “premium increase” rationale. T-test results show that there is no statistically significant difference in average support for these groups (p=.495). Moreover, a multiple linear regression with appropriate balance checks reflects a similar outcome (p=.940). Also, the following two pairings, which compared “Obamacare” with the “premium increase” rationale to “Health Care Reform” with the “premium increase” rationale and “The Affordable Care Act” with the “premium increase” rationale to “Health Care Reform” with the “premium increase” rationale respectively, also return non-significant results. The third set of pairings compared each equivalency frame grouped with the “premium decrease” rationale to one another. The first pairing in this set compared “Obamacare” with the “premium decrease” rationale to “The Affordable Care Act” with the “premium decrease” rationale. T-test results demonstrate that there is a marginally significant statistical difference between average support for these groups (p=.102). Substantively, evidence shows that when moving from “The Affordable Care Act” with the “premium decrease” rationale to “Obamacare” with the “premium decrease” rationale, there is a -.247 shift in policy support (measured on a four point scale where 0=strongly opposed to 4=strongly supportive). However, a multiple linear regression with balance checks reports that the difference between groups is not significant (p=.404). The second pairing in this set compared “Obamacare” with the “premium decrease” rationale to “Heath Care Reform” with the “premium decrease” rationale. T-test results show that there is no statistically significant difference in average support for these groups (p=.977). Also, a multiple linear regression with balance checks returns a similar, non-significant result (p=.576). The third pairing in this set compared “The Affordable Care Act” with the “premium decrease” rationale to “Health Care Reform” with the “premium decrease” rationale. T-test results indicate that there is a marginally significant statistical difference between policy support for these groups (p=.091). Substantively, evidence demonstrates that when moving from “Health Care Reform” with the “premium decrease” rationale to “Obamacare” with the “premium decrease” rationale, there is a +.252 shift in policy support. However, a multiple linear regression with balance checks reports that the difference between groups remains marginally significant (p=.092). The final set of pairings in in my first hypothesis compared each equivalency frame grouped with the “preexisting condition” rationale to one another. The first pairing in this set compared “Obamacare” with the “preexisting condition” rationale to “The Affordable Care Act” with the “pre-existing condition” rationale. T-test results show that there is no statistically significance difference between average support for these groups (p=.309). What is more, a multiple linear regression with relevant balance checks returns the same, statistically insignificant result (p=.434). The following pairing, which compared “Obamacare” with the “pre-existing condition” rationale to “Health Care Reform” with the “pre-existing condition,” is also statistically insignificant. The third pairing in this set compared “The Affordable Care Act” with the “pre-existing condition” rationale to “Health Care Reform” with the “pre-existing condition” rationale. T-test results indicate that there is a statistically significant difference between average policy support for these groups (p=.037). Substantively, evidence shows that when moving from “Health Care Reform” with the “pre-existing condition” rationale to “The Affordable Care Act” with the “pre-existing condition” rationale, there is a -.285 shift in policy support. However, a multiple linear regression with balance checks reports that the difference between groups is not significant (p=.156). In short, the evidence suggests that what you call the recent healthcare law really has no bearing on average public support. Although several pairings were originally marginally significant and one pairing was statistically significant, the inclusion of balance checks indicates that these findings may only be due to small sample sizes. Thus, with balance checks taken into account, none of the pairings analyzed in my first hypothesis were statistically significant. Hypothesis 2 In order to assess the second hypothesis, I compared the groups that received the “pre-existing condition” rationale to the respective control group for each equivalency frame. This enabled me to measure the robustness of the pre-existing rationale’s framing effect. The first pairing in my second hypothesis compared “Obamacare” with the “pre-existing condition” rationale to “Obamacare” without any rationale. T-test results demonstrate that there is a highly significant statistical difference between average policy support for these groups (p=.000). Substantively, evidence shows that when moving from “Obamacare” without any rationale to “Obamacare” with the “pre-existing condition” rationale, there is a +.669 shift in policy support. Furthermore, a multiple linear regression with the proper balance checks does not alter the statistical significance of this pairing. The other two pairings, which compared “The Affordable Care Act” with the “pre-existing condition” rationale to “The Affordable Care Act” without any rationale and “Health Care Reform” with the “pre-existing condition” rationale to “Health Care Reform” without any rationale respectively, also have a p-value of .000 both before and after balance checks. In summation, the highly significant p-value returned by each of these three T-tests indicates a strong correlation between inclusion of the “pre-existing condition” rationale and average policy support for the healthcare law. What is more, the mean difference returned by each of these pairings indicates that there is a very large shift in average policy support when moving from the groups without any rationale to the groups with the “preexisting condition” rationale. Discussion and Conclusion With these results in mind, we can now discuss how they pertain to my hypotheses. My first hypothesis, which argued that changing the equivalency frame would not affect public support for the recent healthcare law was largely supported by the results. Initially, however, several marginally significant findings (the pairings comparing support levels for “The Affordable Care Act” without any rationale to “Health Care Reform” without any rationale, “The Affordable Care Act” with the “premium decrease” rationale to “Health Care Reform” with the “premium decrease” rationale, and “Obamacare” with the “premium decrease” rationale to “The Affordable Care Act” with the “premium decrease” rationale) casted some doubt on the veracity of this expectation. Interestingly though, after including balance checks, these marginally significant pairings all fell into agreement with my hypothesis. Specifically, these pairings saw a shift in p-value from around p=.100 (marginally significant) to well above p=.100 (not significant) after doing balance checks. Furthermore, difference of means tests reveal a statistically significant difference in support between the group receiving “The Affordable Care Act” with the “pre-existing condition” rationale and the group receiving “Health Care Reform” with the “pre-existing condition” rationale (p=.037). However, once balance checks were included, this outlier also fell into line with my hypothesis (p=.156). In short, these non-findings lend strong support to my first hypothesis. This finding indicates that changing what you call the recent healthcare law (equivalency frame) has little effect on public opinion. It is likely that these nonfindings result from the American public’s high level of familiarity with each of the equivalency frames presented. Although at the time of passage in 2010 FOX News and MSNBC’s discussion of the healthcare law was dominated by use of a single term (FOX News = Obamacare; MSNBC = Affordable Care Act), it is now common for all news organizations and politicians to use these issue frames interchangeably. Therefore, since this survey was conducted a few months after the rollout of the new Affordable Care Act website and several years after the passage of the law, any equivalency framing effect that may have been there in 2010 has long since disappeared. Unlike my first hypothesis, my second hypothesis found strong support. Briefly put, the second hypothesis argued that if individuals were provided with the “preexisting condition” rationale as opposed to receiving no rationale at all, support levels would be much higher. Results from each of the three pairings used to test this hypothesis showed a highly significant difference in support between conditions (p=.000) both before and after balance checks. In fact, the most significant of these pairings, which compared “Health Care Reform” with the “pre-existing condition” rationale to “Health Care Reform” without any rationale, had a mean difference of +.671 (on a four point scale where 0=strongly opposed to 4=strongly supportive). Here, the mean support shifted from 2.301 for “Health Care Reform” without any rationale to 2.972 when the “preexisting condition” rationale was included instead (See Figure 1). Moreover, these findings show that the “preexisting condition” rationale is of the utmost concern to the individuals who took my survey. Out of the three possible frames that a respondent might have received, “Health CareReform” with the “pre-existing condition” rationale had thehighest mean support level at 2.972. It is likely that thisframe garnered the most support of the three because itprimed individuals to think about improving the Americanhealthcare system. Since allowing individuals with pre-existing conditions to obtain health insurance is one of thetop concerns of health reformers, it makes sense that thosewho received the “Health Care Reform” frame had thehighest level of support. Hence, future attempts at framingthe recent healthcare law would benefit from providing a“pre-existing condition” rationale to the general public.In conclusion, we can now get a good idea of theimplications of this paper. In terms of practical politicalimplications, proponents of the law who read this papermight discover that they should restructure their efforts byfocusing on the benefits that the Affordable Care Act has forthose with pre-existing conditions. Here, one possibleoption might be to invest in a media campaign in order tospread word about this rationale quickly and efficiently. Interms of the implications for policy creation, all politiciansmight learn to focus less on what future health laws arecalled, and focus more on what these laws actually do forthe citizens that will be affected by it. Lastly, in terms ofresearch implications, political scientists benefit in twomajor ways from this paper. First, this paper serves asevidence of the first major attempt to analyze issue andequivalency framing in terms of healthcare. Furthermore,the results of this paper suggest that, more so thanequivalency or emphasis frames, the rationales that areprovided to an individual are what is important in thedevelopment of personal opinions about an issue. So, interms of future research, not only should more work be doneon rationales, but on the connection between rationales andspecific policies. Hence, this paper will undoubtedly be ofgreat use to politicians, the media, and political scientistsalike. ReferencesChong, Dennis, and James N. Druckman. 2007. “FramingTheory.” Annual Review of Political Science 10:103-126. Cohen, Elizabeth. 2009. “Tips for getting insurance whenyou have a pre-existing condition.” May 14.http://www.cnn.com/2009/HEALTH/05/14/preexisting.condition.insurance/index.html?iref=24hours