## Abstract

Context: Although research has begun to examine perceptions of being on the losing side of politics, it has been confined to electoral politics. The context of health disparities, and particularly the opioid crisis, offers a case to explore whether frames that emphasize racial disadvantage activate loser perceptions and the political consequences of such beliefs.

Methods: White survey participants (N = 1,549) were randomized into three groups: a control which saw no news article, or one of two treatment groups which saw a news article about the opioid crisis framed to emphasize either the absolute rates of opioid mortality among whites or the comparative rates of opioid mortality among whites compared to blacks.

Findings: Among control group participants, perceiving oneself a political loser was unrelated to attitudes about addressing opioids, whereas those who perceived whites to be on the losing side of public health had a less empathetic response to the opioid crisis. The comparative frame led to greater beliefs that whites are on the losing side of public health, whereas the absolute frame led to more empathetic policy opinions.

Conclusions: Perceptions that one's racial group has lost ground in the public health context could have political consequences that future research should explore.

In a March 2018 survey by the Pew Research Center, a substantial majority of the public (67%; 78% of Democrats and 53% of Republicans) noted that “their side” has been losing more often than winning in politics. Given the high prevalence of this loser perception across both political parties (even among those whose party, at the time of the survey, controlled all three branches of the federal government), it is surprising that social scientists actually know very little about the antecedents or consequences of these perceptions. In particular, little research attention has been paid to the factors that influence perceptions of being on the losing side of politics (or the consequences of such beliefs), nor has existing work explored the causes or consequences of perceptions of being on the losing side in specific policy domains. In the domain of public health policy, the notion of winners and losers is often quite explicit. For instance, certain groups are commonly portrayed as more or less affected by some conditions over others or there may be more policy attention (or funding) to one disease over another (Armstrong, Carpenter, and Hojnacki 2006; Best 2012). The consequences of these loser messages may be significant if perceptions of being advantaged versus disadvantaged contribute toward support for policies that have genuine potential to help or harm people's health. Whereas a few studies have examined public opinion about health disparities (Benz et al. 2011; Booske, Robert, and Rohan 2011), this literature in public health has been disconnected from the literature in political psychology explaining the psychological predispositions of policy beliefs. The current study aims to bridge these literatures and offer new insights into the political consequences of believing oneself to be on the losing side of public health policy, focusing on opioid use disorder (OUD) as a case wherein these phenomena may be particularly salient. Elite messaging around OUD has often portrayed the problem as being particularly severe among white Americans, potentially activating the types of loser sentiments described above.

The objectives of this research are to examine the factors that predict whites' perceptions of being a loser both in politics and in health policy, assess whether such perceptions have an impact on support for policies to address OUD, and test whether these loser perceptions can be manipulated by elite messaging about the racialized demographics of the opioid crisis. There are two potential classes of policy responses to the opioid problem, one that is more empathetic of people with OUD (i.e., bolstering public health approaches of treatment and prevention while reducing stigma) and another that is more punitive (i.e., focusing on drug use and its correlates as crimes among users, dealers, and inappropriate prescribers). Identifying what factors predict the public's support for these positions thus provides important contributions to the evolving policy response to the crisis. The study also contributes to theory in two ways. First, we demonstrate that loser perceptions have consequences for policy attitudes in the domain of opioid policy, and they are not explained away by the usual predictors of political attitudes like demographics and partisanship. Second, we show that loser perceptions can be manipulated by elite messages. Although the magnitude of the findings are small, we anticipate that these propositions will foster additional scholarship into the political consequences of a type of elite messaging that situates one group as winning and another as losing—messaging that is extremely common in both current politics and public health, as we describe below.

## Background on the Political Psychology of the Loser

With the recent rise in populism, support for so-called populist candidates (such as Donald Trump), ethnocentrism, and anti-immigrant attitudes—especially among whites—has come an increase in scholarly attention to the role of concepts such as alienation, efficacy, and economic, cultural, and racial threat in affecting individuals' political attitudes and behaviors. For example, journalists and scholars alike have used individual- and community-level data to examine the causes of whites' support for Donald Trump, oftentimes in an attempt to disentangle the effects of racial, economic, and cultural threat (or to argue that one was more impactful than the others; see, e.g., Chokshi 2018). In White Identity Politics, Ashley Jardina (2019) argues that the perception that the United States is losing ground culturally, sparked by growing diversity in the country, is a primary determinant of support for policies and candidates that whites (especially those who strongly identify with their racial group) believe will “Make America Great Again.” In Dying of Whiteness, Jonathan Metzl (2019) argues that whites' investment in policies that contribute toward maintenance of racial superiority harms their own health and well-being. As he notes, “White backlash politics gave certain white populations the sensation of winning (emphasis added), particularly by upending the gains of minorities and liberals; yet the victories came at steep cost,” the cost of their own health (Metzl 2019: 8). At the root of these explanations is the notion that a dominant group's belief that they are losing ground is fundamental to our understanding of the rise of populist sentiments and ethnocentrism around the globe (Cramer 2016 makes a similar argument with regard to rural white Americans).

Research narrowly targeted on the consequences of the political psychology of losing is limited. Miller, Farhart, and Saunders (2018) have shown that people who perceive themselves to be on the losing side of politics are more likely to endorse conspiracy theories that impugn their political rivals and that the perceived threat of losing power is causally related to reduced support for political compromise (Barker, Bowler, Carman, and Wendelbo 2018). Other work shows that people who vote for the losing presidential candidate become less trusting of government (Anderson and LoTempio 2002; see also Nadeau and Blais 1993) and that electoral losers are more likely to believe in election fraud conspiracies than electoral winners (e.g., Edelson et al. 2017). Researchers have also shown that learned helplessness—a concept related to perceiving oneself a loser—is negatively correlated with conventional forms of political participation (e.g., voting) and positively correlated to unconventional participation (e.g., protesting), especially among traditionally disadvantaged racial minorities (Farhart 2017).

The perception that one, or one's group, is on the losing side (of politics, or of a particular policy domain) is also similar to one component of the psychological concept of relative deprivation. People naturally compare themselves, and their outcomes, with others (either as individuals comparing themselves to other individuals—egoistic relative deprivation—or as members of a group comparing themselves to members of other groups—fraternal or group relative deprivation; Runciman 1993). To the extent that people notice others possess something to which they feel entitled and think is feasible to obtain, and do not blame themselves for failure to possess it, resentment results (Crosby 1976). Our conceptualization of loser perceptions is akin to the first component of group level relative deprivation—the realization that another group possesses something that one's own group does not possess (e.g., the perception that one's political party is on the losing side of politics, or one's racial group is on the losing side of health policy).

The two domains in which the effects of relative deprivation have been most examined are worker satisfaction with their pay/income inequality (e.g., Sweeney, McFarlin, and Inderrieden 1990) and political protest and social movement participation (e.g., Pettigrew 2015; Smith and Huo 2014). Relative deprivation has been linked to a host of emotional and behavioral effects, such as objective and subjective (decreases in) happiness, depression, health, satisfaction with one's job and salary, aggression, and social movement participation (e.g., Subramanyam et al. 2009; but see Gurney and Tierney 1982 for a critique of the application of relative deprivation theory to social movements). Interestingly, although the theory of relative deprivation has been applied to make policy recommendations surrounding issues of pay equity, income inequality, social justice, and the like (Smith and Huo 2014), to our knowledge, research has not explored the policy attitudes of individuals who are experiencing relative deprivation. Moreover, little work has examined the causes and political consequences of perceptions that one's group is on the losing side of politics in general or within specific policy domains (although see Metzl 2019 for an example). As such, we focus in this study on the causes and policy attitude consequences of group-level loser perceptions.

## Background on Framing Health Disparities

Health disparities in the United States is one policy domain for which the concept of group-level loser perceptions may be particularly salient. Health disparities are defined as differences in health that “adversely affect socially disadvantaged groups; are systematic and plausibly avoidable” (Braveman et al. 2011). Research consistently shows that populations that face structural inequality (e.g., lower socioeconomic status, residential segregation, or racism) also experience disproportionate rates of illness (Braveman et al. 2011). Whereas the public health literature abounds with discussion of health disparities, understanding of health inequalities among the public is more limited. For instance, in 2010, only 45% of Americans were aware that African Americans have worse life expectancy than whites; 37% were aware that African Americans are more likely than whites to be diagnosed with diabetes (Benz et al. 2011). Low public awareness may result from a news media that comments infrequently about health disparities in typical health coverage (Gollust and Lantz 2009; Nagler et al. 2016).

Elite messaging—or framing—has the potential to shift public views on health disparities. Framing is the strategic emphasis communicators place on certain aspects of social or political issues, such as the causes, solutions, or target populations affected, in public or media discourse (Entman 1993). Research on framing, for instance, demonstrates that emphasizing the racial identity of a group affected by public policy can shape public attitudes about those issues (see, e.g., Kinder and Sanders 1996).1 Relatively few studies have sought to understand how framing group-level health disparities affects public understanding (see Niederdeppe et al. 2013 for a review). In one study that explicitly examines social comparison frames (i.e., presenting racial comparisons of illness risk as opposed to reporting a single group's risk), Bigman (2014) found that comparative frames tended to decrease perceptions of risk for the less at-risk group. In another experimental study, Nicholson and colleagues (2008) found that a frame describing blacks as having higher colon cancer rates led to black respondents having negative reactions and lowered colorectal cancer screening intentions. Neither of these studies included politically relevant outcome measures, however, such as support for policy. Gollust and Cappella (2014) tested various messages describing the causes of disparities between socioeconomic groups and found high levels of anger elicitation and counterarguing in response to the messages; conservative respondents counterargued the messages more than liberals did. This study suggests that disparity frames could produce politically consequential responses, either because they activate motivated reasoning (i.e., the motivation to challenge the very existence of disparities or the need for government to address them) or other types of racialized stereotypes (Kinder and Sanders 1996), such as minority racial groups being to blame for their disadvantaged status (Lynch and Gollust 2010). However, no work to our knowledge has examined whether framing disparities in health outcomes might contribute toward the political psychology of feeling like a loser or whether such a perception has broader political consequences.

## Opioid Use Disorder as a Case

The opioid crisis is a salient case in which to explore the political psychology of health disparities frames and the effect of loser perceptions on policy attitudes. Opioid use disorder (OUD) generally concerns problematic use of prescription pain medications and/or heroin or street versions of synthetic opiates like fentanyl. The opioid case offers a different perspective from how scholars traditionally understand health disparity. In national depictions of the opioid problem, the racial group typically perceived as more advantaged in US society—whites—is suffering (in terms of higher levels of addiction and overdose mortality based on national statistics), more than those groups typically perceived as disadvantaged (i.e., people of color; see Hansen and Netherland 2016). For instance, there were 37,113 total deaths from overdoses among whites in 2017 (a rate of 19.4 per 100,000) compared to 5,513 deaths among blacks (a rate of 12.9 per 100,000; KFF n.d.).2

This population comparative frame—emphasizing high rates among whites compared to other groups—is featured prominently in news coverage of OUD (Harbin 2019; Netherland and Hansen 2016; see also Shachar et al., this issue). For example, a May 2018 New York Times Magazine cover article reported on the higher prevalence of neonatal abstinence syndrome among whites and featured numerous anecdotes and photographs of white mothers and babies. The author even reflectively noted in her coverage, “Indeed, the perception of our opioid crisis as an epidemic, rather than a racial pathology, owes much to the fact that white Americans have been hard hit” (Egan 2018).

This novel disparity frame has also been the subject of scholarly interest. Health disparities researchers have questioned whether the dissemination of the concept of working-class white mortality from addiction and related mental health conditions will contribute toward public perceptions of a “new face of disadvantage,” despite the enduring health inequities for communities of color (Brown and Tucker-Seeley 2018: 124). At the core of our current research are similar questions: Do common racialized media frames of opioid use disorder promote whites' perception that they are losers in public health? And what are the consequences of such beliefs on attitudes about policy and resource allocation?

Understanding the consequences of loser perceptions surrounding the opioid crisis is significant for several reasons. First, a growing research narrative links whites' morbidity and mortality from overdose, suicide, and mental illness (see, e.g., Case and Deaton 2015) to voting in the 2016 election. Several recent studies identify correlations at the ecological level between poor community health (including addiction) and whites' turnout for Trump in 2016 (Bor 2017; Monnat and Brown 2017; Wasfy, Stewart III, and Bhambhani 2017). If racial comparison frames emphasizing white opioid threats activate perceptions that whites are losing, this could be politically consequential. For instance, emphasizing this losing status could boost whites' support of government spending on opioids—but it could potentially lower their support for spending on illnesses more common among nonwhites.

Second, framing the opioid crisis as a “white issue” could lead whites to be more sympathetic toward opioid users than they would otherwise be (in comparison to drug use problems that plague communities of color, as noted in the popular press; see, e.g., Lopez 2017). Policy responses to high rates of OUD might be categorized into an empathetic approach that favors public health policy strategies and a punitive approach that emphasizes law enforcement strategies (McGinty et al. 2016; see also Kim, Morgan, and Nyhan in this issue). The former includes public health education, treatment with effective pharmacological remedies (i.e., medically assisted treatment), guidelines to promote more appropriate opioid prescribing by providers, and harm reduction approaches (such as treatment of overdose with naloxone and needle exchange; Barry et al. 2016; Saloner and Barry 2018). Punitive approaches include arresting and prosecuting those who possess, use, or deal opioids, cracking down on providers and clinics that are unlawfully prescribing opioids, enforcing immigration laws to prevent the trafficking of drugs, and referring parents or pregnant women who use drugs to child welfare authorities (Kennedy-Hendricks et al. 2017). Loser perceptions could lead members of the public (and policy makers alike) to favor different types of empathetic versus punitive policies as appropriate responses to the opioid crisis (Kennedy-Hendricks et al. 2017).

## Research Objectives

This study has two major objectives. First, given the paucity of research on the political consequences of perceiving oneself to be a political loser, we examine the correlates, predictors of, and consequences of feeling like a loser in politics (among a sample of white Americans not exposed to news frames about the opioid crisis). We also develop a measure to assess the belief that whites are on the losing side of public health policy, and compare the correlates, predictors of, and consequences of health loser perceptions with political loser perceptions. Specifically, we assess whether loser perceptions are related to public attitudes about the appropriate approaches to deal with the opioid epidemic. Second, using an experiment, we examine whether frames that emphasize white mortality (in either an absolute or a comparative sense) affect whites' propensity to identify as a loser in the political context or the health context, as well as whether these frames affect public support for policy approaches.

Our study design is guided by a general conceptual model (see figure 1). The model proposes that loser perceptions—perceiving oneself on the losing side of politics or policy—may contribute to policy-relevant attitudes about an issue (i.e., empathetic vs. punitive policy approaches and attitudes about the target population). Examining the factors that are associated with loser perceptions (both predictors of and consequences on policy-relevant outcomes) is the first task of the study, relying on the control group from our experiment. Leveraging an embedded media framing experiment, we further anticipate that media frames describing white opioid mortality using a comparative frame (i.e., explicitly stating that whites have higher rates of opioid mortality than blacks) will activate perceptions of being on the losing side more than the same article with an absolute frame (i.e., just reporting opioid mortality rates among whites). We also explore whether these frames have effects on policy-relevant attitudes—directly or via loser perceptions.

## Data and Methods

Data were collected through Amazon Mechanical Turk (MTurk) on August 3, 2018. MTurk provides access to more demographically diverse samples of the US voting-age population than student-convenience and Internet samples (Berinsky, Huber, and Lenz 2012; Buhrmester, Kwang, and Gosling 2011; Levay, Freese, and Druckman 2016), as well as high-quality data (Crump, McDonnell, and Gureckis 2013; Goodman, Cryder, and Cheema 2013; Weinberg, Freese, and McElhattan 2014). After eliminating 124 respondents we suspected to be bots, there were 1953 participants in the study.3 Given our theoretical expectations that the messages would affect whites' beliefs specifically, we restrict all analyses to adults who identify as white (N = 1,549).

## Study Design

Participants in the treatment conditions were asked three questions to assess their opinions about the article (how interesting/informative they perceived it to be, and how likely they would be to share it, measured on Likert scales coded 1 = not at all, 2 = slightly, 3 = somewhat, 4 = very, and 5 = extremely) before answering additional survey items that the no-exposure control group also answered.

#### Manipulation Check

To test whether the two articles were effective at communicating that the rates of opioid addiction are higher among whites, all respondents were asked a question adapted from previous research assessing public awareness of health disparities (Benz et al. 2011): “Compared to the average white person, do you think the average African American person is more likely, just as likely, or less likely to be affected by . . . ” [opioid addiction]. The response options were coded such that 1 = less likely, 2 = just as likely, and 3 = more likely. This question was asked following the news article and the loser perception questions.

#### Loser Status

The general loser perception question, based on a question adapted from the Pew Research Center (Fingerhut 2015) by Miller, Farhart, and Saunders (2018) asked, “Thinking about the way things are going in politics today, on the issues that matter to you, would you say that your side has been winning more often than it's been losing, or losing more often than it's been winning?” Response options were “winning more often than losing” and “losing more often than winning” which branched to a second question. The second question asked, “Would you say that your side is winning [losing]?” and response options included “a great deal more often than losing [winning],” “somewhat more often than losing [winning],” or “a little bit more than losing [winning].” These questions were combined to create a 6-point scale, with higher numbers representing greater loser perceptions.

We also created a health policy-specific loser perception measure.4 Given that one of our goals was to examine whether the correlates and consequences of health loser perceptions are similar to general political loser perceptions, we constructed this question to be as similar as possible to the Pew question. However, rather than using the language of “your side,” we chose to highlight a racial group comparison, to tap the kinds of comparisons that are frequently made in the health policy domain. Respondents were asked, “Thinking about the way things are going with the public's health, do you think that white Americans are on the winning side or the losing side of public policies aimed at improving health outcomes compared to black Americans?” As with the general political loser question, the winning/losing question was then branched to a follow-up question, which asked, “Would you say that white Americans are winning [losing]” “a great deal more often than losing [winning],” “somewhat more often than losing [winning],” or “a little bit more than losing [winning].” These questions were combined to create a 6-point scale, with higher numbers representing greater loser perceptions. We chose not to include a middle option because neither the original Pew item nor Miller et al.’s (2018) branched version did so.

#### Policy Opinions

We assessed attitudes toward a set of 12 policy proposals, adapted from previous work (Barry et al. 2016; see  appendix 2 for the full set, as well as the means from the control group). The items were selected based on whether they would be considered empathetic or punitive responses toward people with OUD (and/or their health care providers). Response options were coded such that 1 = strongly oppose, 2 = oppose, 3 = somewhat oppose, 4 = neither favor nor oppose, 5 = somewhat favor, 6 = favor, and 7 = strongly favor. These items, and all other policy items described below, were displayed in random order. We constructed scales from these items based on whether they were public health–oriented (empathetic) measures or law-enforcement (punitive) measures. The empathetic policy opinion scale (alpha = .84) is an average of the first six items listed in  appendix 2. The punitive policy opinion scale (alpha = .80) is an average of the second six items.

#### Government Spending

Respondents were asked whether they thought the government should spend more or less on the following three empathetic domains: treatment for opioid addiction, drug overdose prevention programs, and education campaigns about the dangers of opioid addiction; and the following two punitive domains: law enforcement to arrest opioid dealers and opioid users, respectively. All responses were measured on a 5-point scale such that 1 = spend much less, 2 = spend less, 3 = spend about the same, 4 = spend more, and 5 = spend a lot more.

#### Responsibility

Respondents were asked how much responsibility they thought “the following groups should have for addressing the problem of opioid (prescription pain medication and heroin) abuse in the United States” (adapted from Barry et al. 2016): pharmacies and pharmacists, pharmaceutical companies, doctors, and health insurance companies (averaged to comprise a health care industry responsibility scale with an alpha of .81), the federal government (we treated these two groups as empathetic), dealers, addicts, and law enforcement (we treated these latter three groups as punitive). Responses were coded such that 1 = none, 2 = a little, 3 = a moderate amount, 4 = a lot, and 5 = a great deal.

#### Attitudes toward People with Opioid Addiction

We used a standard 0–100 feeling thermometer measure to assess attitudes toward people with opioid addiction (McGinty et al. 2018). In addition to people who abuse opioids (mean in control condition = 45.6), participants were asked about people who have been diagnosed with depression (mean = 75.0), people who abuse alcohol (mean = 47.3), people who have been diagnosed with HIV (mean = 65.0), and people who have been diagnosed with cancer (mean = 80.8).5 Not surprisingly, respondents reported feeling the coldest toward people whose afflictions are perceived to be more in their control (those who abuse opioids and alcohol, as well as people diagnosed with HIV), with those who abuse opioids garnering the most negative rating (Weiner 2006).

#### Spending on Other Health Disparities

To assess whether the frames might have a negative effect on people's support for government spending to prevent health problems more prevalent among black Americans, respondents were asked, “Do you think the federal government spends too much money, spends the right amount of money, or spends too little money on programs aimed at preventing health problems that are more prevalent among black Americans compared to white Americans?” Responses were coded such that 1 = spends too much money, 2 = spends enough money, and 3 = spends too little money.

#### Control Variables

We measured the following political and demographic variables: party identification, gender, annual income, age, education, Hispanic/Latinx ethnicity, strength of white identity, and respondents' subjective perceptions of their relative socioeconomic status (see  appendix 3 for the question wordings and coding for these variables).

## Analysis

In the first set of analyses, we focused only on the white respondents in the control condition to examine the distributions of the two loser perception variables (political and health policy), as well as their correlations. Next, we examined the factors that predict these loser perceptions, by regressing each of the two loser perception variables on the set of sociodemographic and political characteristics using ordinary least squares (OLS) regression. Third, to assess the political consequences of these perceptions, we regressed the three sets of responses to the opioid crisis as dependent variables (empathetic policy attitudes, punitive policy attitudes, and stigma held toward users) on the two loser perception variables and included the same set of sociodemographic and political controls.

For the experimental analyses, we first verified that respondents were balanced in observable characteristics (the control variables described above) across the three experimental groups, using linear regression and chi-square tests. We observed no significant differences across the randomly assigned groups in these characteristics. We used t-tests and chi-square tests to assess whether there were any differences between the two treatment groups (absolute vs. comparative racial framing) in responses to the knowledge question about rates of addiction by racial group (e.g., our manipulation check) and in evaluations of the news vignettes (how interesting, shareable, and informative they were).

To assess whether there were differences in loser perceptions and in participants' responses to the opioid epidemic by the experimental frame to which they were exposed, we regressed the three sets of dependent variables (loser perceptions, policy attitudes, feeling thermometer) on the treatment groups. We estimated the regression models with and without covariates to assess whether inclusion of control variables affected the precision of our experimental estimates. We used postestimation Wald tests of the coefficients on the experimental conditions to test for differences between conditions.

## Results

### Nonexperimental Results (Control Condition Only)

Table 1 displays the characteristics of the group without exposure to any opioid news story (white respondents only). The sample is reasonably varied in terms of socioeconomic status and partisanship, but as is typical of MTurk samples, it is more Democratic, younger, and more educated than the US adult population as a whole.

Participants more readily acknowledged feeling like they are on the losing side with regard to politics than they acknowledged that white Americans are on the losing side with regard to health policy. Specifically, the mean of the political loser measure (6-point scale ranging from winning a lot more to losing a lot more) was 3.64 (SD = 1.73), whereas the mean of the racialized health policy loser measure was 2.47 (SD = 1.4). Over half (51.9%) of respondents reported that they believed they were losing a little, somewhat, or a lot more than winning in politics. In contrast, only 18.6% of respondents reported that white Americans were losing a little, somewhat, or a lot more than they were winning in health policy.

#### Relationship between Political and Health Policy Loser Status

The political loser and the health policy loser variables were uncorrelated (r = −.05, ns). The fact that the political loser question asks respondents to indicate whether they think their side is losing [winning] more often than winning [losing] (likely activating partisanship and/or political ideology), whereas the health policy loser question explicitly asks respondents whether they think white Americans are losing [winning] more often than winning [losing] in the health policy domain relative to black Americans, may be one of the reasons for a lack of association between the two. Consistent with this reasoning (and unsurprisingly given that at the time of the survey Republicans controlled both the executive and legislative branches of the federal government), Democrats scored higher on the political loser variable than Republicans (means = 4.46 and 2.51, respectively, t = 14.49, p < .001). In contrast, Democrats scored lower on the explicitly racialized health policy loser variable than Republicans (means = 2.11 and 2.76, respectively, t = −5.24, p < .001).

#### Predictors of Loser Status

Before running OLS regressions to examine the predictors of political and health loser perceptions, we recoded all variables (independent and dependent) to range from 0–1. Table 2 displays the factors that predict loser perceptions. Men, Republicans, political conservatives, those with a bachelor's degree or some college (compared to more than a college education), and those reporting their social standing was high were all significantly less likely to report being a loser in the domain of politics, while those over age 50 were more likely to report being a political loser. Those with a stronger white racial identity were also significantly less likely to report being a political loser. These findings have strong face validity, illuminating the groups who arguably were achieving more wins than losses in politics during the summer of 2018, before the fall electoral victories of women, people of color, and Democrats.

The pattern of predictors was different for the racialized health policy loser variable. Whites who were older, Independent, and politically conservative were more likely to perceive whites to be on the losing side of health policy. Those who reported a higher social standing were less likely to perceive whites to be on the losing side of health policy. In fact, those who reported a higher social standing were significantly less likely to endorse being on the losing side of both politics and health policy, reinforcing the validity of these measures. Interestingly, despite the racially explicit nature of the health policy loser variable, strength of white identity had no relationship to respondents' belief that whites are health policy losers.

#### Consequences of Loser Perceptions on Opioid Policy Attitudes

For these loser perceptions to be politically consequential, they need to be related to policy attitudes. Table 3 reports the results of separate OLS regression models predicting the empathetic and punitive attitudes, as well as the stigma measure, with both the political and health loser perception variables in models that include the controls (all independent and dependent variables were recoded to range from 0–1; full models are available from the authors upon request). Political loser status was associated with three of the empathetic attitudes and two of the punitive policy attitudes (and inconsistently so); political loser perception was negatively associated with the punitive policy index (b = −.11, p < .001) and positively associated with the belief that dealers are responsible (b = .15, p < .001). It was not associated with the stigma measure.

Results for the health policy loser status question are more consistent. The perception that white Americans are on the losing side in health policy is negatively associated with every one of the empathetic attitudes, and is not associated with any of the punitive attitudes or the feeling thermometer. Whites who perceive that they are on the losing side of public policies aimed at improving health outcomes compared to blacks are less supportive of policies and government spending programs that are empathetic to opioid users, and believe that the health care sector and the federal government are less responsible for addressing the problem compared to whites who perceive that they are on the winning side.

### Experimental Results

#### Manipulation Check

To examine whether the articles were effective at communicating to respondents that opioid addiction rates are higher among whites, we examined the percentage of respondents in each condition who said that African Americans are less likely to be affected by opioid addiction. Compared to the control condition (16%), more respondents in the absolute (27%) and comparative (49%) conditions indicated that African Americans are less affected (from logistic regression, b = .71, se = .15 and b = 1.64, se = .15 for the absolute and comparative conditions respectively compared to the no-exposure control condition, p < .05). Although the effect of the manipulation was not very strong, those exposed to the treatment conditions were more accurate in their assessments that African Americans were less likely to be affected by opioid addiction than respondents in the control condition, and they were most accurate when provided the explicit comparative frame in the article.

In addition, respondents in the absolute (mean = 2.31) and comparative (mean = 2.37) conditions were equally likely to indicate that they would share the article with friends or family if they were to see the article online (t = .78, ns). Respondents in the comparative condition were significantly more likely to indicate that they found the article to be interesting (means = 3.49 and 3.37, respectively, t = 1.98, p < .05) and marginally significantly more likely to indicate that they found the article to be informative (means = 3.67 and 3.57, respectively, t = 1.74, p < .10).

#### Effect of the Frames on Loser Perceptions

We tested our hypothesis that the comparative frame would be more likely to activate loser perceptions than the absolute frame (compared to the control) by regressing the political loser question, and then the health policy loser question, on the two treatment dummy variables. As table 4 shows, consistent with expectations, the respondents in the absolute condition were no more likely to perceive themselves to be on the losing side of politics, nor that whites are on the losing side of health policies, than respondents in the no-exposure condition. However, respondents in the comparative frame condition were more likely to perceive whites to be on the losing side in the health domain compared to respondents in the no-exposure condition (b = .03, p < .10 in the model without control variables; b = .04, p < .05 in the model with control variables). The effect of the comparative condition on perceptions of being a political loser was not statistically significant.

Thus far, we have shown the perception that whites are losing ground to blacks in the health policy domain is powerfully, and negatively, associated with empathetic opioid attitudes, and that the comparative frame is predictive of this belief that whites are losing ground. Next, we turn to examining the impact of the frames on opioid attitudes.

#### Effect of the Frames on Opioid Attitudes

Results of the effects of the frames relative to the control on the punitive policy measure and the empathetic policy measure are shown in table 5. Specifically, participants in the absolute condition were significantly more likely to support empathetic policies (b = .03, p < .01 in the model without controls; b = .04, p < .001 in the model with controls) compared to those in the no-exposure condition. Participants in the comparative condition were marginally significantly more likely to support empathetic policies (b = 02, p < .10) compared to the no-exposure condition, but only in the model without controls. However, we could not reject the null hypothesis of no difference between the absolute condition and comparative condition in their impact on support for empathetic policies (Wald test F = 1.20, p = .274). The treatment conditions had no impact on support for punitive policies, relative to the no-exposure condition.

Neither of the treatment conditions had statistically significant effects (at p < 0.05) on any of the other policy attitudes examined (responsibility attributions or government spending), nor on the feeling thermometer measure of stigma (results available upon request).

To review the key experimental findings, we identified that the comparative frame led to greater perceptions of whites' being on the losing side of health policy, and that in the control condition perceptions of being on the losing side of health policy were associated with lower support for empathetic policies. In contrast, we found that the absolute frame led to greater support for empathetic policies, and minimal evidence that the comparative frame had any impact on policy attitudes at all. Given these conflicting findings, we have no conceptual evidence for the mediation path depicted in figure 1, that any effect of the comparative frame on policy-relevant attitudes is indirect, via loser perceptions.6

#### Effect of the Frames on Health Disparity Spending

The final goal of the analysis was to evaluate whether there may be any negative consequences of the emphasis of white Americans as having a high rate of opioid mortality on perceptions of other racial health disparities. Collapsing across the three conditions, 9.6% of respondents indicated that they thought that the government spends too much money on health problems more prevalent among black Americans compared to white Americans; 47.3% of respondents indicated that the government spent enough money, and 43.2% indicated that the government spends too little money. There were no significant differences across the treatment groups in the belief that the government spends too much money on health problems among blacks (χ2 = 4.61, ns), indicating that the treatment conditions did not cause a backlash on support for spending on other health disparities.

## Discussion

This study aimed to identify whether perceptions of feeling like a loser in health politics might be politically consequential, and whether certain ways of framing racial disparities for one issue—the opioid epidemic—might affect such perceptions. The results are mixed but offer new insights into the politics of the opioid epidemic as well as suggest new lines of inquiry within the broad intersection of political psychology and health disparity research in public health.

We measured two constructs, the status of whites' perceiving themselves to be a loser in politics (endorsed by a majority of the sample, as with the Pew 2018 survey) and in a racialized health policy domain (endorsed by only 1 in 5 respondents). We found that these beliefs are uncorrelated with one another, yet are systematically related to other sociodemographic and political characteristics in ways that support the validity of these items. Namely, whites in 2018 who were conservative, Republican, and identified more strongly with their white identity were less likely to feel like a political loser. In contrast, conservatives were more likely to feel whites were losing in the health policy domain. Importantly, the higher respondents perceived themselves on the ladder of subjective social status, the less likely they were to perceive themselves a loser in either politics or in health policy, as one would expect.

These two beliefs predicted attitudes about opioids differently. The belief of being a loser in politics was largely unrelated to any opioid attitudes—on policy, spending, responsibility attributions, or stigma. In contrast, white respondents' perception that whites are on the losing side of public health policy was consistently and negatively related to empathetic approaches to deal with the opioid epidemic. Even when adjusting for demographic factors, partisanship, and ideology, this belief of whites being a loser in health policy was related to opposition toward a set of policies considered by public health authorities to be evidence-based ways to deal with the epidemic, including increasing access to treatment, naloxone availability, and education campaigns (Barry et al. 2016; Saloner and Barry 2018). Our findings suggest that this belief of whites being on the losing side—often discussed in terms of the 2016 election results—could be an important contributor to public health politics as well. As noted by others (Goodwin et al. 2018; Monnat and Brown 2017; Wasfy, Stewart III, and Bhambhani 2017), there may be a relationship between the extent of the opioid crisis and county-level voting patterns for Trump in 2016. If whites living in these areas also feel higher levels of being a loser in public health, they may be less supportive of the very policies that might best ameliorate the opioid crisis in their areas. And, if media frames continue to emphasize this perception of whites losing ground—as journalists have done in the past (Netherland and Hansen 2016), the media could contribute to reifying these beliefs—and consequently policy opinions—more strongly.

We found limited evidence to support the experimental expectations of the study: neither framing the opioid crisis to emphasize whites as the group affected, nor framing it to compare whites to blacks, had strong effects on either perceptions of whites' being a disadvantaged group in politics (null results only) or in health policy (small effects for the comparative condition). We did find that the absolute condition, which emphasized whites as the dominant group suffering from the opioid epidemic, led to stronger support for an empathic policy approach to deal with the epidemic but had very little effect on other beliefs about the epidemic.7 The fact that the absolute framing of whites as the major population affected by opioids, relative to no exposure to any media depiction, did shape a more empathic response to the epidemic is broadly consistent with media commentary on the epidemic (see, e.g., Lopez 2017; also see Kim, Morgan, and Nyhan in this issue). In other words, emphasizing the white target population may contribute to a more empathetic policy approach to deal with the problem among a white audience. Finally, concerns of an unintended effect of framing white mortality in the opioid context—that it could drive down support for spending on conditions that are more common in nonwhite racial groups—were not borne out in these data.

## Limitations and Future Directions

This study offers an initial exploration into the political consequences of losing ground in public health. Future work should continue to unpack the relationships between health loser perceptions and policy attitudes, recognizing that cross-sectional analyses have limitations. Our experimental and nonexperimental results together raise some questions we cannot completely unpack, such as why the comparative frame contributed toward whites' heightened perceptions of feeling like a loser but also increased support for empathetic policy responses, whereas whites' perception that their racial group is losing was negatively correlated with empathetic policy responses. There may be selection issues, confounding, and/or omitted variables that the current models cannot address.8 We cannot, for instance, distinguish health loser perception (as measured here, with its explicit whites versus blacks comparison) from various types of racial resentment (Feldman and Huddy 2005; Jardina 2019; Kinder and Sanders 1996). The health loser perception question may be tapping the general belief that blacks get more than they deserve rather than the perception that whites are on the losing side in a specific public policy domain. Although we did show that strength of white identity was unrelated to health policy loser perceptions, future research should measure and explicitly control for racial resentment.

Our measure of whites' perceptions of being on the losing side of health policy was also limited in a few other ways. First, we did not assess other components of relative deprivation that might moderate the effects of loser perceptions (e.g., Crosby 1976): whether respondents felt that they were entitled to better health policies, whether they thought it was feasible to be on the winning side of health policies, and whether they made internal or external attributions for why they were on the losing side. Although, in this context, it is likely that whites who perceived themselves to be on the losing side of health policies felt that they were unjustly so, future research should explicitly measure perceptions of entitlement, feasibility, and attributions of responsibility. Second, to be consistent with the general political loser question, we chose not to include a middle option (i.e., neither winning nor losing). Whereas some research has found no differences in the univariate distribution of responses to a question that includes versus does not include a middle option (Schuman and Presser 1996), other research has found this not to be the case (Bishop 1987). Future research might experiment with using a middle response option to assess the distribution of loser perceptions. Finally, our measure is agnostic as to whether whites who perceive themselves to be on the losing side of health policy view this as a recent (and possibly transient) development, or whether they view it as a more chronic state. Such temporal perceptions could moderate the range, intensity, and direction of the effects on policy attitudes.

Future research also should examine these phenomena in more representative samples and examine the duration of the (admittedly weak) effect of the comparative frame on whites' loser perceptions (see, e.g., Lecheler and de Vreese 2016). It is also important to examine whether frames for the opioid epidemic (e.g., emphasizing white mortality) have different types of effects than do frames for other conditions that exhibit health disparities in outcomes but for the opposite group comparisons, such as diabetes or heart disease.

Scholars have recently been paying much more attention to examining the connections among health and political behavior, examining, for instance, how community health factors related to the 2016 election (e.g., Bor 2017) or how health insurance gains relate to voter turnout (e.g., Haselswerdt 2017). Whereas these studies at the community or aggregate level are important, our study suggests that future studies at the intersection of health and politics should also engage more deeply at the individual level, to consider the underlying political psychological factors that may be politically consequential in how groups determine the appropriate scope of policy in combating public health challenges.

## Acknowledgments

We thank the participants at the Brown University workshop on the Politics of the Opioid Epidemic for their feedback. We also thank the Grand Challenges Research Initiative at the University of Minnesota for funding and Emma Klinger for her research assistance. We received helpful feedback on earlier versions of this article from participants at the American Political Science Association 2018 annual meeting as well as from participants at the Media and Politics Research Group at the University of Minnesota, particularly Benjamin Toff.

## Notes

1.

Another broad class of literature on framing in health communication concerns loss or gain frames (Rothman et al. 2006). Whereas the idea of a loss frame is conceptually related to the idea of being a loser, loss-framed messages in the health communication context are typically messages that identify the costs of not taking some protective action to reduce one's health risk versus a gain-framed message, which identifies the health benefits of taking the action. Since these frames are typically focused on individualized health behavior choices (and not group issues or policy preferences), these frames are not our focus here.

2.

Recent health data document accelerating rates of OUD among all population groups, with particularly steep increases among Native American and urban African American populations, especially low-income African Americans in certain metropolitan areas such as Washington, DC (CDC 2018; Jamison 2018).

3.

A concern has arisen among scholars who use MTurk that there has been a substantial uptick (since spring 2018) in the number of responses submitted from identical GPS coordinates (possibly completed by bots). Anecdotal evidence from scholars' social media conversations indicates that between 5% and 50% of recently collected MTurk survey responses come from repeating GPS coordinates. Some repeating GPS coordinates are to be expected (www.qualtrics.com/support/survey-platform/data-and-analysis-module/data/download-data/understanding-your-dataset/), and the bot problem may be not as pervasive as the social media discussions among academics make it out to be (blog.turkprime.com/2018/08/concerns-about-bots-on-mechanical-turk.html?m=1). We erred on the side of caution by dropping from our data any response that had the same GPS coordinates as at least nine other responses (29 with GPS coordinates that put them in a reservoir in Kansas, 39 from a park in Buffalo, NY, and 13 from the same GPS coordinates in Venezuela).

4.

Given the limited research examining loser status in health or politics, we first fielded a pilot study from September 27 to October 5, 2017, also using an MTurk sample of whites only. In this survey we piloted a few other items assessing dimensions of being a loser in public health, including a dichotomous version of the measure described above (winning vs. losing) along with the political loser status item and a smaller number of opioid policy opinions. To assess the validity of the key measures, we compared the predictors of the political loser and public health loser items in both the 2017 and 2018 surveys. A comparison of the 2017 pilot with the 2018 survey provides substantial evidence that the items are measuring similar underlying concepts despite different samples and different times of data collection (results available from authors upon request).

5.

We note that our survey wording used the language of abuse in response to the problem and in assessing stigma toward people who use drugs and alcohol, which we acknowledge is problematic and can itself perpetuate stigma. Thus, in this article we used the word abuse in reference to the survey language respondents saw and otherwise used opioid use disorder and other nonstigmatizing language elsewhere. See, for example, “Words Matter: How Language Choice Can Reduce Stigma,” Substance Abuse and Mental Health Services Administration (SAMHSA), mnprc.org/2017/02/04/words-matter-how-language-choice-can-reduce-stigma/ (accessed November 12, 2019).

6.

We also tested a simple mediation model with regression analysis and found no statistical support that loser perceptions mediate the effects of the experimental frames on policy attitudes. However, we acknowledge that analyses attempting to apply a causal interpretation to cross-sectional mediation analyses have challenges related to confounding (see, e.g., MacKinnon and Pirlott 2015).

7.

Per a suggestion by a workshop participant for this special issue, we also examined whether there was any heterogeneity in the effects of the frames across whites based on their measured strength of white identity. This variable was measured postrandomization so our post hoc assessment of interaction effects was exploratory (and we note that the frames had no effect on strength of white identity). We did not find any evidence that the frames had a significantly different effect on perceptions of whites being on the losing side of health policy or on policy attitudes among respondents who perceived their white identity to be weaker or stronger.

8.

We attempted to examine endogeneity concerns in the relationship between loser status and policy attitudes with instrumental variables analysis, where random assignment to the comparative frame was an instrument for health loser perceptions (see, e.g., MacKinnon and Pirlott 2015). However, we do not believe the frames affected the outcome exclusively through loser perceptions, so the assumption for this analysis was not met. The first-stage F statistic was 4.11, much lower than conventional cut-offs to be considered a strong instrument, so these analyses were not conclusive.

## References

References
Adler, Nancy, Epel, Elissa S., Castellazzo, Grace, and Ickovics, Jeannette R.
2000
. “
Relationship of Subjective and Objective Social Status with Psychological and Physiological Functioning: Preliminary Data in Healthy White Women
.”
Health Psychology
19
, no.
6
:
586
92
.
Anderson, Christopher J., and LoTempio, Andrew J.
2002
. “
Winning, Losing, and Political Trust in America
.”
British Journal of Political Science
32
, no.
2
:
335
51
.
Armstrong, Elizabeth M., Carpenter, Daniel P., and Hojnacki, Marie.
2006
. “
Whose Deaths Matter? Mortality, Advocacy, and Attention to Disease in the Mass Media
.”
Journal of Health Politics, Policy and Law
31
, no.
4
:
729
72
.
Barker, David C., Bowler, Shaun, Carman, Christopher Jan, and Wendelbo, Morten.
2019
. “
Compromise Is for Losers? Socio-Political Power, Threat, and Public Resistance to Political Compromise.
” Paper presented at the annual meeting of the American Political Science Association,
Washington, DC
,
August 29–September 1
.
Barry, Colleen L., Kennedy-Hendricks, Alene, Gollust, Sarah E., Niederdeppe, Jeff, Bachhuber, Marcus A., Webster, Daniel W., and McGinty, Emma E.
2016
. “
Understanding Americans' Views on Opioid Pain Reliever Abuse
.”
111
, no.
1
:
85
93
.
Benz, Jennifer K., Espinosa, Oscar, Welsh, Valerie, and Fontes, Angela.
2011
. “
Awareness of Racial and Ethnic Health Disparities Has Improved Only Modestly over a Decade
.”
Health Affairs
30
, no.
10
:
1860
67
.
Berinsky, Adam J., Huber, Gregory A., Lenz, Gabriel S., and Alvarez, R. Michael.
2012
. “
Evaluating Online Labor Markets for Experimental Research: Amazon.com's Mechanical Turk
.”
Political Analysis
20
, no.
3
:
351
68
.
Best, Rachel Kahn.
2012
. “
Disease Politics and Medical Research Funding: Three Ways Advocacy Shapes Policy
.”
American Sociological Review
77
, no.
5
:
780
803
.
Bigman, Cabral A.
2014
. “
Social Comparison Framing in Health News and Its Effect on Perceptions of Group Risk
.”
Health Communication
29
, no.
3
:
267
80
.
Bishop, George F.
1987
. “
Experiments with the Middle Response Alternative in Survey Questions
.”
Public Opinion Quarterly
51
, no.
2
:
220
32
.
Booske, Bridget C., Robert, Stephanie, and Rohan, Angela M. K.
2011
. “
Awareness of Racial and Socioeconomic Health Disparities in the United States: The National Opinion Survey on Health and Health Disparities, 2008–2009
.”
Preventing Chronic Disease
8
, no.
4
:
A73
.
Bor, Jacob.
2017
. “
Diverging Life Expectancies and Voting Patterns in the 2016 US Presidential Election
.”
American Journal of Public Health
107
, no.
10
:
1560
62
.
Braveman, Paula A., Kumanyika, Shiriki, Fielding, Jonathan E., Laveist, Thomas, Borrell, Luisa N., Manderscheid, Ron, and Troutman, Adewale.
2011
. “
Health Disparities and Health Equity: The Issue Is Justice
.”
American Journal of Public Health
101
,
suppl. 1
:
S149
55
.
Brown, Lauren, and Tucker-Seeley, Reginald.
2018
. “
Will ‘Deaths of Despair’ among Whites Change How We Talk about Racial/Ethnic Health Disparities?
Ethnicity and Disease
28
, no.
2
:
123
28
.
Buhrmester, Michael, Kwang, Tracy, and Gosling, Samuel D.
2011
. “
Amazon's Mechanical Turk: A New Source of Inexpensive, Yet High-Quality, Data?
Perspectives on Psychological Science
6
, no.
1
:
3
5
.
Case, Anne, and Deaton, Angus.
2015
. “
Rising Morbidity and Mortality in Midlife among White Non-Hispanic Americans in the 21st Century
.”
Proceedings of the National Academy of Sciences
112
, no.
49
:
15078
83
.
CDC (Centers for Disease Control and Prevention)
.
2018
. “
Opioid Overdose.
www.cdc.gov/drugoverdose/index.html (accessed
April
8
,
2019
).
Chokshi, Niraj.
2018
. “
Trump Voters Driven by Fear of Losing Status, Not Economic Anxiety, Study Finds
.”
New York Times
,
April
24
. www.nytimes.com/2018/04/24/us/politics/trump-economic-anxiety.html.
Cramer, Katherine J.
2016
.
The Politics of Resentment: Rural Consciousness in Wisconsin and the Rise of Scott Walker
.
Chicago
:
University of Chicago Press
.
Crosby, Faye.
1976
. “
A Model of Egoistical Relative Deprivation
.”
Psychological Review
83
, no.
2
:
85
.
Crump, Matthew J. C., McDonnell, John V., and Gureckis, Todd M.
2013
. “
Evaluating Amazon's Mechanical Turk as a Tool for Experimental Behavioral Research
.”
PloS ONE
8
, no.
3
:
e57410
.
Dawson, Michael C.
1994
.
Behind the Mule: Race and Class in African-American Politics
.
Princeton, NJ
:
Princeton University Press
.
Edelson, Jack, Alduncin, Alexander, Krewson, Christopher, Sieja, James A., and Uscinski, Joseph E.
2017
. “
The Effect of Conspiratorial Thinking and Motivated Reasoning on Belief in Election Fraud
.”
Political Research Quarterly
70
, no.
4
:
933
46
.
Egan, Jennifer.
2018
. “
Children of the Opioid Epidemic.
New York Times
,
May
9
. www.nytimes.com/2018/05/09/magazine/children-of-the-opioid-epidemic.html.
Entman, Robert M.
1993
. “
Framing: Toward Clarification of a Fractured Paradigm
.”
Journal of Communication
43
, no.
4
:
51
58
.
Farhart, Christina.
2017
. “
Look Who Is Disaffected Now: Political Causes and Consequences of Learned Helplessness in the US
.” PhD diss.,
University of Minnesota
.
Feldman, Stanley, and Huddy, Leonie.
2005
. “
Racial Resentment and White Opposition to Race-Conscious Programs: Principles or Prejudice?
American Journal of Political Science
49
, no.
1
:
168
83
.
Fingerhut, Hannah.
2015
. “
In Politics, Most Americans Feel They're on the Losing Side
.”
Pew Research Center
,
November
25
. www.pewresearch.org/fact-tank/2015/11/25/winners-and-losers-in-politics/.
Gay, Claudine, Hochschild, Jennifer, and White, Ariel.
2016
. “
Americans' Belief in Linked Fate: Does the Measure Capture the Concept?
Journal of Race, Ethnicity, and Politics
1
, no.
1
:
117
44
.
Gollust, Sarah E., and Cappella, Joseph N.
2014
. “
Understanding Public Resistance to Messages about Health Disparities
.”
Journal of Health Communication
19
, no.
4
:
493
510
.
Gollust, Sarah E., and Lantz, Paula.
2009
. “
Communicating Population Health: Print News Media Coverage of Type 2 Diabetes
.”
Social Science and Medicine
69
, no.
7
:
1091
98
.
Goodman, Joseph K., Cryder, Cynthia E., and Cheema, Amar.
2013
. “
Data Collection in a Flat World: The Strengths and Weaknesses of Mechanical Turk Samples
.”
Journal of Behavioral Decision Making
26
, no.
3
:
213
24
.
Goodwin, James S., Kuo, Yong-Fang, Brown, David, Juurlink, David, and Raji, Mukaila.
2018
. “
Association of Chronic Opioid Use with Presidential Voting Patterns in US Counties in 2016
.”
JAMA Network Open
1
, no.
2
:
e180450
.
Gurney, Joan Neff, and Tierney, Kathleen J.
1982
. “
Relative Deprivation and Social Movements: A Critical Look at Twenty Years of Theory and Research
.”
Sociological Quarterly
23
, no.
1
:
33
47
.
Hansen, Helena, and Netherland, Julie.
2016
. “
Is the Prescription Opioid Epidemic a White Problem?
American Journal of Public Health
106
, no.
12
:
2127
29
.
Harbin, M. Brielle.
2019
. “
The Contingency of Compassion: Media Depictions of Drug Addiction.
” Unpublished paper.
Haselswerdt, Jake.
2017
. “
Expanding Medicaid, Expanding the Electorate: The Affordable Care Act's Short-Term Impact on Political Participation
.”
Journal of Health Politics, Policy and Law
42
, no.
4
:
667
95
.
Jamison, Peter.
2018
. “
Falling Out: A Generation of African American Heroin Users Is Dying in the Opioid Epidemic Nobody Talks About. The Nation's Capital Is Ground Zero
.”
Washington Post
,
December
18
. www.washingtonpost.com/graphics/2018/local/opioid-epidemic-and-its-effects-on-african-americans.
Jardina, Ashley.
2019
.
White Identity Politics
.
Cambridge
:
Cambridge University Press
.
Kennedy-Hendricks, Alene, Barry, Colleen L., Gollust, Sarah E., Ensminger, Margaret E., Chisolm, Margaret S., and McGinty, Emma E.
2017
. “
Social Stigma toward Persons with Prescription Opioid Use Disorder: Associations with Public Support for Punitive and Public Health–Oriented Policies
.”
Psychiatric Services
68
, no.
5
:
462
69
.
KFF (Kaiser Family Foundation)
. n.d. “
Opioid Overdose Deaths by Race/Ethnicity
.” State Health Facts. kff.org/other/state-indicator/opioid-overdose-deaths-by-raceethnicity/ (accessed
November
12
,
2019
).
Kinder, Donald, and Sanders, Lynn.
1996
.
Divided by Color: Racial Politics and Democratic Ideals
.
Chicago
:
University of Chicago Press
.
Lecheler, Sophie, and de Vreese, Claus H.
2016
. “
How Long Do News Framing Effects Last? A Systematic Review of Longitudinal Studies
.”
Annals of the International Communication Association
40
, no.
1
:
3
30
.
Levay, Kevin E., Freese, Jeremy, and Druckman, James N.
2016
. “
The Demographic and Political Composition of Mechanical Turk Samples
.”
Sage Open
January-March
:
1
17
. doi.org/10.1177/2158244016636433.
Lopez, German.
2017
. “
When a Drug Epidemic's Victims Are White.
Vox
,
April
4
. www.vox.com/identities/2017/4/4/15098746/opioid-heroin-epidemic-race.
Luhtanen, Riia, and Crocker, Jennifer.
1992
. “
A Collective Self-Esteem Scale: Self-Evaluation of One's Social Identity
.”
Personality and Social Psychology Bulletin
18
, no.
3
:
302
18
.
Lynch, Julia, and Gollust, Sarah E.
2010
. “
Playing Fair: Fairness Beliefs and Health Policy Preferences in the United States
.”
Journal of Health Politics Policy Law
35
, no.
6
:
849
87
.
MacKinnon, David P., and Pirlott, Angela G.
2015
. “
Statistical Approaches for Enhancing Causal Interpretation of the M to Y Relation in Mediation Analysis
.”
Personality and Social Psychology Review
19
, no.
1
:
30
43
.
McGinty, Emma Beth., Kennedy-Hendricks, Alene, Baller, Julia, Niederdeppe, Jeff, Gollust, Sarah, and Barry, Colleen L.
2016
. “
Criminal Activity or Treatable Health Condition? News Media Framing of Opioid Analgesic Abuse in the United States, 1998–2012
.”
Psychiatric Services
67
, no.
4
:
405
11
.
McGinty, Emma E., Barry, Colleen L., Stone, Elizabeth M., Niederdeppe, Jeff, Kennedy-Hendricks, Alene, Linden, Sarah, and Sherman, Susan G.
2018
. “
Public Support for Safe Consumption Sites and Syringe Services Programs to Combat the Opioid Epidemic
.”
Preventive Medicine
111
:
73
77
.
Metzl, Jonathan.
2019
.
Dying of Whiteness: How the Politics of Racial Resentment Is Killing America's Heartland.
New York
:
Basic Books
.
Miller, Joanne M., Saunders, Kyle, and Farhart, Christina.
2018
. “
The Relationship between Losing an Election and Conspiracy Theory Endorsement.
” Paper presented at the annual conference of the Midwest Political Science Association,
Chicago, IL
,
April 5–9
.
Monnat, Shannon M., and Brown, David L.
2017
. “
More than a Rural Revolt: Landscapes of Despair and the 2016 Presidential Election
.”
Journal of Rural Studies
55
:
227
36
.
1993
. “
Accepting the Election Outcome: The Effect of Participation on Losers' Consent
.”
British Journal of Political Science
23
, no.
4
:
553
63
.
Nagler, Rebekah H., Bigman, Cabral A., Ramanadhan, Shoba, Ramamurthi, Divya, and Viswanath, K.
2016
. “
Prevalence and Framing of Health Disparities in Local Print News: Implications for Multilevel Interventions to Address Cancer Inequalities
.”
Cancer Epidemiology Biomarkers and Prevention
25
, no.
4
:
603
12
.
Netherland, Julie, and Hansen, Helena B.
2016
. “
The War on Drugs That Wasn't: Wasted Whiteness, ‘Dirty Doctors,’ and Race in Media Coverage of Prescription Opioid Misuse
.”
Culture, Medicine, and Psychiatry
40
, no.
4
:
664
86
.
Nicholson, Robert A., Kreuter, Matthew W., Lapka, Christina, Wellborn, Rachel, Clark, Eddie M., Sanders-Thompson, Vetta, Jacobsen, Heather M., and Casey, Chris.
2008
. “
Unintended Effects of Emphasizing Disparities in Cancer Communication to African-Americans
.”
Cancer Epidemiology, Biomarkers, and Prevention
17
, no.
11
:
2946
53
.
Niederdeppe, Jeff, Bigman, Cabral A., Gonzales, Amy L., and Gollust, Sarah E.
2013
. “
Communication about Health Disparities in the Mass Media
.”
Journal of Communication
63
, no.
1
:
8
30
.
Pettigrew, Thomas F.
2015
. “
Samuel Stouffer and Relative Deprivation
.”
Social Psychology Quarterly
78
, no.
1
:
7
24
.
Pew Research Center
.
2018
. “
Key Findings on Americans' Views of the US Political System and Democracy
.” Fact Tank,
April
26
. www.pewresearch.org/fact-tank/2018/04/26/key-findings-on-americans-views-of-the-u-s-political-system-and-democracy/.
Rothman, Alexander J., Bartels, Roger D., Wlaschin, Jhon, and Salovey, Peter.
2006
. “
The Strategic Use of Gain- and Loss-Framed Messages to Promote Healthy Behavior: How Theory Can Inform Practice
.”
Journal of Communication
56
,
suppl. 1
:
S202
20
.
Runciman, Walter Garrison.
1993
.
Relative Deprivation and Social Justice: A Study of Attitudes to Social Inequality in Twentieth-Century England
.
Berkeley
:
University of California Press
.
Saloner, Brendon, and Barry, Colleen L.
2018
. “
Ending the Opioid Epidemic Requires a Historic Investment in Medication-Assisted Treatment
.”
Journal of Policy Analysis and Management
37
, no.
2
:
431
38
.
Schildkraut, Deborah J.
2017
. “
White Attitudes about Descriptive Representation in the US: The Roles of Identity, Discrimination, and Linked Fate
.”
Politics, Groups, and Identities
5
, no.
1
:
84
106
.
Schuman, Howard, and Presser, Stanley.
1996
.
Questions and Answers in Attitude Surveys: Experiments on Question Form, Wording, and Context
.
Thousand Oaks, CA
:
Sage
.
Smith, Heather J., and Huo, Yuen J.
2014
. “
Relative Deprivation: How Subjective Experiences of Inequality Influence Social Behavior and Health
.”
Policy Insights from the Behavioral and Brain Sciences
1
, no.
1
:
231
38
.
Sweeney, Paul D., McFarlin, Dean B., and Inderrieden, Edward J.
1990
. “
Using Relative Deprivation Theory to Explain Satisfaction with Income and Pay Level: A Multistudy Examination
.”
33
, no.
2
:
423
36
.
Subramanyam, Malavika, Kawachi, Ichiro, Berkman, Lisa, and Subramanian, S. V.
2009
. “
Relative Deprivation in Income and Self-Rated Health in the United States
.”
Social Science and Medicine
69
, no.
3
:
327
34
.
Wasfy, Jason H., Stewart, CharlesIII, and Bhambhani, Vijeta.
2017
. “
County Community Health Associations of Net Voting Shift in the 2016 US Presidential Election
.”
PloS ONE
12
, no.
10
:
e0185051
.
Weinberg, Jill D., Freese, Jeremy, and McElhattan, David.
2014
. “
Comparing Data Characteristics and Results of an Online Factorial Survey between a Population-Based and a Crowdsource-Recruited Sample
.”
Sociological Science
,
August
4
. www.sociologicalscience.com/articles-vol1-19-292/.
Weiner, Bernard.
2006
.
Social Motivation, Justice, and the Moral Emotions: An Attributional Approach
.
Mahwah, NJ
:
Lawrence Erlbaum Associates
.

### Support for Policies to Address the Opioid Epidemic, Control Group Only

How much do you favor or oppose the following policy to address the problem of opioid (prescription pain medication and heroin) abuse? (7-point scale, higher values = more favorability)

### Question Wording and Coding for Control Variables

###### Party identification

Respondents were first asked, “Generally speaking, do you usually think of yourself as a Democrat, a Republican, an Independent, or what?” Respondents who responded Democrat or Republican were then asked, “Would you call yourself a strong Democrat [Republican] or a not very strong Democrat [Republican]?” Respondents who responded Independent or other were then asked, “Do you think of yourself as closer to the Democratic Party or the Republican Party?” (with “neither” included as a response option). Responses were coded as two dummy variables representing Republicans/Republican leaners and Democrats/Democratic leaners, with pure Independents as the comparison group.

###### Ideology

Respondents were asked, “We hear a lot of talk these days about liberals and conservatives. Here is a 7-point scale on which the political views that people might hold are arranged from extremely liberal to extremely conservative. Where would you place yourself on this scale?” The 7-point scale was coded to range from 0–1; higher numbers = more conservative.

###### Gender

Respondents' self-reported gender was coded such that 1 = male and 0 = female.

###### Income

We assessed income with the following question: “The next question is about the total income of your household for the past 12 months. Please include your income plus the income of all members living in your household (including cohabiting partners and armed forces members living at home). Please count income before taxes, including income from all sources (such as wages, salaries, tips, net income from a business, interest, dividends, child support, alimony, Social Security, public assistance, pensions, and retirement benefits).” Respondents were asked to choose 1 of 28 income groupings, which were then recoded into 3 dummy variables: less than or equal to $27,599;$27,500–$49,999; and$50,000–$74,999, with$75,000 and up as the comparison group.

###### Age

Respondents' reported age was coded into three dummy variables: 30–39 years; 40–49 years; and 50 years or older, with younger than 30 years as the comparison group.

###### Education

Respondents were asked, “What is the highest level of school you have completed or the highest degree you have received?” They were given 10 categories from which to choose. Responses were recoded into three dummy variables: high school or less; some college; and bachelor's degree, with more than a bachelor's degree serving as the comparison group.

###### Latinx ethnicity

Respondents were asked, “Are you Spanish, Hispanic, or Latino?” (coded such that 1 = Hispanic/Latinx identity and 0 = not Hispanic/Latinx identity).

###### Strength of white identity

Strength of white identity was assessed with four items developed by Luhtanen and Crocker 1992 (each measured on a 7-point agree-disagree Likert scale): “Being White has very little to do with how I feel about myself” (reverse-coded), “Being White is an important reflection of who I am”, “Being White is unimportant to my sense of what kind of person I am” (reverse-coded), and “Being White is an important part of my self-image.” Responses were recoded to range from 0–1 and then averaged to form an index.

###### Subjective perception of relative socioeconomic status

We used the MacArthur Scale of Subjective Social Status, Adult Version (see Adler et al. 2000), to assess subjective perceptions of socioeconomic status. Respondents were shown a picture of a ladder with 10 rungs numbered 1–10 and were given the following instructions (responses were coded to range from 0–1):

Think of this ladder as representing where people stand in the United States. At the top of the ladder are the people who are the best off—those who have the most money, the most education, and the most respected jobs. At the bottom are the people who are the worst off—those who have the least money, least education, and the least respected jobs or no job. The higher up you are on this ladder, the closer you are to the people at the very top; the lower you are, the closer you are to the people at the very bottom. Where would you place yourself on this ladder? Please type the whole number that represents the rung where you think you stand at this time in your life, relative to other people in the United States.”