Strategic Messaging to Promote Policies that Advance Racial Equity: What Do We Know, and What Do We Need to Learn?

Abstract

Policy Points.

• Many studies have explored the impact of message strategies to build support for policies that advance racial equity, but few studies examine the effects of richer stories of lived experience and detailed accounts of the ways racism is embedded in policy design and implementation.

• Longer messages framed to emphasize social and structural causes of racial inequity hold significant potential to enhance support for policies to advance racial equity.

• There is an urgent need to develop, test, and disseminate communication interventions that center perspectives from historically marginalized people and promote policy advocacy, community mobilization, and collective action to advance racial equity.

Context

Long‐standing racial inequities in health and well‐being are shaped by racialized public policies that perpetuate disadvantage among Black, Brown, Indigenous, and people of color. Strategic messaging can accelerate public and policymaker support for public policies that advance population health. We lack a comprehensive understanding of lessons learned from work on policy messaging to advance racial equity and the gaps in knowledge it reveals.

Methods

A scoping review of peer‐reviewed studies from communication, psychology, political science, sociology, public health, and health policy that have tested how various message strategies influence support and mobilization for racial equity policy domains across a wide variety of social systems. We used keyword database searches, author bibliographic searches, and reviews of reference lists from relevant sources to compile 55 peer‐reviewed papers with 80 studies that used experiments to test the effects of one or more message strategies in shaping support for racial equity–related policies, as well as the cognitive/emotional factors that predict their support.

Findings

Most studies report on the short‐term effects of very short message manipulations. Although many of these studies find evidence that reference to race or use of racial cues tend to undermine support for racial equity–related policies, the accumulated body of evidence has generally not explored the effects of richer, more nuanced stories of lived experience and/or detailed historical and contemporary accounts of the ways racism is embedded in public policy design and implementation. A few well‐designed studies offer evidence that longer‐form messages framed to emphasize social and structural causes of racial inequity can enhance support for policies to advance racial equity, though many questions require further research.

Conclusions

We conclude by laying out a research agenda to fill numerous wide gaps in the evidentiary base related to building support for racial equity policy across sectors.

Introduction

Long‐standing racial inequities in health and well‐being are largely driven by racialized public policies that perpetuate disadvantage among Black, Brown, Indigenous, and people of color. ¹ Successful efforts to dismantle the structures that perpetuate existing power imbalances will require broad societal changes in how public policies are designed and implemented. Although by no means a panacea, there is also evidence that strategic messaging can help to accelerate public and policymaker support for public policies that advance population health in general. ² , ³ Although scholars across many disciplines have studied the role of communication in shaping attitudes toward specific policies with racial health equity implications, we lack a comprehensive understanding of lessons learned from that work and the gaps in knowledge it reveals.

This has not stopped a variety of scholars, however, from drawing sweeping conclusions about the impact of various message strategies related to racial equity. For instance, on April 26th, 2021, English and Kalla published a preprint, with accompanying commentary on Twitter, reporting on a study that compared “race frames” with “class frames” in shaping public support for policies with evidence of equity‐promoting effects. ⁴ Although the study involved >5,000 participants and examined six different policies, the messages in question were only three sentences long, and the effects of the messages were only gauged immediately after exposure. Yet the authors interpreted their results in very broad terms: “Despite increasing awareness of racial inequities and a greater use of progressive race framing by Democratic elites, linking public policies to race is detrimental for support of these policies.” ^{4 p13} Subsequent debates in social and legacy media on the merits of the study highlight sharp divisions in the interpretation and application of these and other studies that purport to offer a definitive word on the merits or pitfalls of messaging about race and social policy. ⁵

This and other debates highlight the need for a reckoning of the evidence and knowledge claims related to strategic messaging to promote policies that advance racial equity. What do we know from the scientific literature on the subject, and what do we need to learn? This paper reports on a scoping review of peer‐reviewed studies from communication, psychology, political science, sociology, public health, and health policy, which have tested how various message strategies influence support and mobilization for racial equity policy domains, as well as their cognitive and social antecedents, across a wide variety of social systems. In the sections that follow, we first define the scope of racial equity policy domains and describe relevant health and social policies that prior evidence suggests have implications for racial equity (or, in some cases, are associated with increased racial inequity). We then describe a variety of modifiable cognitive and social antecedents to support and advocate for racially equitable policies and/or to oppose and advocate against policies likely to increase racial inequities. We next review the volume and strength of evidence for various message design considerations for strategic efforts to promote racial equity policy domains and/or oppose racial inequity policies. We conclude by offering some messaging recommendations based on the available state of the evidence while highlighting the need to address large gaps in the evidence base.

Racial (In)Equity, Health Outcomes, and Public Policy

There are widespread and long‐standing disparities in health outcomes among Black, Brown, Indigenous, and White populations in the United States. ⁶ , ⁷ , ⁸ These disparities are driven by systemic differences in social determinants of health, the conditions of the environments in which people are born, live, learn, work, play, worship, and age that affect a wide range of health, functioning, and quality‐of‐life outcomes and risks. ⁹ , ¹⁰ These disparities are inequitable because they are driven by long‐standing, systemic patterns of racism and oppression (i.e., they are unjust) and can be addressed through the design, passage, and implementation of public policies that remove racial barriers to upward mobility (i.e., they are modifiable). ¹¹ , ¹² , ¹³ Although health behaviors also matter, decisions about diet, exercise, health screenings, etc. are also shaped by the social, economic, and environmental factors that constrain opportunities and access. ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ As a result, interventions targeting individual behaviors and individual provision of clinical care are unlikely to change long‐standing patterns of racial health inequity. Interventions must target larger social systems that shape income, education, housing, employment, policing, etc. ¹⁸ , ¹⁹ , ²⁰ , ²¹

Changing social systems to support racial health equity can be achieved by changing public policies in at least two ways: through the passage and implementation of policies that are known or likely to reduce racial inequities in social conditions and/or by opposing and repealing policies that are known or likely to increase racial inequities. Although the specific racial equity impact of all public policies may not be fully known a priori, a robust evidence base surrounding the racial equity implications of many public policies exists, ²² , ²³ , ²⁴ and there are frameworks to assess the plausible racial equity implications of novel policy ideas and proposals. ¹⁹ , ²⁵ This evidence and these frameworks can inform health and racial equity advocates to set public policy goals and priorities that maximize the impact of their advocacy.

Media Messaging and Support for Public Policies to Advance Racial Equity

We also know that the broader communication environment (e.g., overall patterns of messages seen in news, advertising, and entertainment media) and strategic messaging conducted alongside broader policy advocacy efforts can play an important role in shaping how both the general US public and policymakers think about health‐related public policies. There are robust theories and literatures on the effects of message framing ²⁶ , ²⁷ (the idea that “[often small] changes in the presentation of an issue or event produce [sometimes large] changes of opinion” by altering a belief or the weight of that belief in shaping an overall opinion) ^{26 p104} and media priming ²⁸ , ²⁹ (the idea that exposure to messages make some concepts more easily accessible from memory and thus exert disproportionate influence on thoughts and opinions) that describe the ways mediated messages can shape interpretation of issues and policies. Informed by these literatures, studies have documented effects of messaging about health and social policy in shaping public and policymaker views in numerous policy domains that include climate change, ³⁰ , ³¹ tobacco control, ³² , ³³ obesity, ³⁴ , ³⁵ substance use disorder, ³⁶ health insurance, ³⁷ early childhood care and education, ³ , ³⁸ and social determinants of health more broadly. ² , ³⁹ However, although often impactful, even well‐crafted, theory‐informed messaging strategies designed to promote policy change can fail or backfire among some audiences, undermining support for the very policies the messages purport to promote. ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ Scholars and advocates have also long debated whether policies that are poised to reduce racial inequities are most likely to gain support and passage when framed explicitly to address specific racial groups’ needs or when these policies are described in more race‐neutral terms. ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ One position argues that racial specificity allows for direct acknowledgment of the inequity and appropriate redress to the harmed groups. Another position suggests that taking a universal stance reduces concerns of exclusion and reactionary (i.e., politically opposing) responses while still addressing the harm at hand. Further complicating matters is the fact that the choice between race‐neutral frames and race‐conscious frames may depend on the specific policy domain, audience, and window of policy opportunity. These considerations highlight the critical need to understand how various audiences make sense of messages about policies that have potential to influence racial equity. ⁴⁶ Although studies abound, we lack guidance on the role(s) that strategic messaging might play in building support and mobilizing action for policies that advance racial equity or rallying opposition to policies that undermine it.

Study Objectives

This paper seeks to reduce gaps in knowledge about messaging for racial equity policy through a scoping review of peer‐reviewed messaging studies published between January 1990 and April 2021 that examine support or opposition to a variety of policies with plausible or documented racial equity implications. We explore studies of both strategic messaging (in which messages clearly identify a strategic goal related to policy support or mobilization) and nonstrategic news media portrayals (in which messages still often convey information about a problem, its causes, and/or potential solutions). We seek to identify broad trends in how researchers across a variety of academic disciplines approach the study of such messaging, document broad patterns of results, interrogate conclusions drawn from the authors of these studies, illuminate gaps in the evidence base, and identify priorities for future research.

Methods

Scoping reviews provide an overview of topics by systematically mapping large bodies of literatures. Unlike systematic reviews that synthesize literature on a more narrowly defined (and often discipline‐bound) question, scoping reviews synthesize research from multiple fields that work on topic areas to provide coherent insights into the conceptual state of research. ⁴⁷ , ⁴⁸ Scoping reviews map literatures in terms of the volume, nature, and characteristics of relevant research. They allow scholars to identify gaps in the literature in ways that do not impose strict constraints that characterize meta‐analyses or other synthesis techniques. ⁴⁸ , ⁴⁹ , ⁵⁰

Identifying Racial Equity Public Policy Domains

The first step in our review involved an effort to identify US public policy domains that have strong evidence or potential to influence racial equity. We do not claim to offer a definitive assessment of the evidence related to policy impacts on racial equity; rather, we developed a working classification scheme to guide our subsequent article retrieval and analysis.

We identified racial equity–related policy domains from a pool of public issues identified by CQ Almanac (https://library.cqpress.com/cqalmanac/index.php) and the US Government Accountability Office (GAO) (https://www.gao.gov/key_issues/overview#t=1). Although CQ Almanac covers all significant legislation and controversial public issues considered by lawmakers in the US Congress, the GAO provides independent, objective, and nonpartisan information facilitating the work of the US government and Congress. We screened the public issue topics and policy domains identified by CQ Almanac and the GAO to retrieve those that the author team deemed to be potentially relevant to equity of outcomes by race and/or ethnicity. To make these assessments of relevance to racial equity, we explored publicly available arguments and evidence about health and social policies in the United States by reviewing racial equity advocacy organization websites and reports, as well as available peer‐reviewed and gray literature on the impacts of various public policies on health and social outcomes across racial groups.

This process revealed several different types of public policies with potential implications for racial equity. Some policies are framed explicitly in terms of their potential impact on racial equity (such as affirmative action and reparations), whereas others may have favorable impacts on racial equity despite not being discussed exclusively in those terms (such as the Affordable Care Act (ACA) or Social Security). Other domains of public policy have been associated with their effects on widening racial inequity (such as redlining and/or stop‐and‐frisk policing policies), whereas other public policy effects on racial equity, for better or worse, are highly dependent on the differential ways through which they are implemented (such as welfare or unemployment benefits). Thus, we defined racial equity policy domains as public policies that have impact on racial equity, whether or not they are designed explicitly for equity, into the following four categories: (a) policy domains typically framed explicitly to advance racial equity, (b) policy domains not exclusively framed for racial equity but with evidence of equity‐promoting effects, (c) policy domains with evidence of inequity‐widening effects, and (d) policy domains with implications for racial equity that are highly contingent on the details of their implementation. The final list of policies was iteratively reviewed by the author team to identify any potential omissions or inappropriate categorizations. Team discussions were used to resolve any disagreement and to reach consensus on our operational classifications. Table 1 features the list of policies included in this review.

Table 1.

Racial Equity Policy Domains

Policy Domains Typically Framed Explicitly to Advance Racial Equity	Policy Domains Not Exclusively Framed for Racial Equity but With Evidence of Equity‐Promoting Effects	Policy Domains With Evidence of Inequity‐Widening Effects	Policy Domains with Implications for Racial Equity That Are Highly Contingent on Policy Implementation
Abolishing residential segregation Affirmative action Baby Bonds Criminal justice reform Environmental justice Policing reform Reparations	Abortion/reproductive/contraceptive rights Affordable Care Act Child Care & Development Block Grant Deferred Action for Childhood Arrivals Earned Income Tax Credit Equal pay initiatives Federal job guarantee Felon voting rights reform Head Start Health care reform Medicaid expansion Raising the minimum wage Social security SNAP SNAP for Women, Infants, and Children	Death penalty Family cap Gentrification Gerrymandering Hospital/enterprise zoning Mandatory minimum sentencing Predatory lending Redlining Restrictive immigration policies Single‐family zoning Stop‐and‐frisk policy Three strikes law Voter ID laws	Affordable housing COVID‐19 policies Disease prevention policies Family leave School vouchers Social safety net Social welfare Unemployment benefits Vaccination mandate

Abbreviation: SNAP, Supplemental Nutrition Assistance Program.

Search Strategy and Inclusion Criteria

The next step of our process involved a search for studies examining the effects of various message strategies on public support for racial equity–related policies and/or their cognitive and/or emotional antecedents. We identified outcomes that both directly speak to levels of public support for racial equity policy domains (supporting or opposing specific policies, intentions to advocate for or against specific policies, intentions to engage in collective action surrounding policies) or indirectly via their impact on cognitive and/or emotional antecedents of policy support/mobilization. We developed a working list of outcomes, mediators, moderators, and message strategies iteratively, starting with a set of variables identified a priori and then adding concepts to a working model of message strategies and outcomes based on our ongoing review of studies identified as relevant to racial equity policy support. Figure 1 shows the final model that is inclusive of all concepts that emerged as relevant during the review.

Figure 1.

Message Strategies, Psychological Processes, and Outcomes

We classified mediators of policy support and/or mobilization into four broad categories: attributions (who is responsible for causing or addressing a health or social issue related to racial equity), racial attitudes and beliefs (including awareness of racial disparities, stereotypical beliefs, and/or racial group centrism), affect and emotions (including prejudicial feelings, empathy and sympathy, fear, guilt, and anger), and policy beliefs (whether respondents believe a policy is likely to be effective and/or beliefs about whether policymakers could come to agreement on such a policy in the first place). We also compiled a list of potential moderators of message effects on these outcomes, including identity‐related factors (racial group identity, education and income, party identification) and ideology‐related factors (racial and political ideology, values, and beliefs about other domains of policy). Finally, we characterized message strategies into two categories based on theories of message effects: message framing (various ways that a message discusses a problem, its causes, and potential solutions to it) ²⁶ and media priming (the use of specific words or cues, often very brief, that make some preexisting beliefs or values accessible and shape thoughts or opinions on an issue). ²⁸ , ²⁹

We limited our search to research studies published between January 1990 and April 2021 and used electronic databases (e.g., Google Scholar, JSTOR, PubMed, and Communication & Media Complete) to search for articles that used terms such as “message,” “communication,” “media,” or “message effects,” in conjunction with “race,” “racial,” “ethnicity,” or “ethnic,” and “policy,” or “[the name of the policy]” (e.g., “affirmative action”) anywhere in the article. This search strategy identified a broad set of articles, which we then screened for their relevance to the topic of our scoping review. We then reviewed the reference list of retrieved articles to identify other studies that could have been overlooked using the keyword search. This process identified 204 potentially relevant articles.

We went through each potentially relevant article to determine its inclusion in the review based on the following standards: (1) The study used one or more randomized experiments to establish causal relationships between message exposure and outcomes, (2) the study context was based in US policy and used US samples of members of the general public (e.g., not policymakers or health care professionals), (3) the study was peer reviewed, and (4) the outcomes measured were related to racial equity policies (such as policy attitudes/support or advocacy intentions). We excluded studies that targeted individual behavioral changes (e.g., interventions to reduce racial bias among doctors in health care settings). For the first three policy domains, in which policies are either explicitly designed to promote racial equity or have clear‐cut implications on racial equity/inequity, each study that met the four inclusion standards above were included in the review. For the last policy domain, in which the impact of the policies on racial equity is ambiguous, only those studies that featured message stimuli related to race/ethnicity (e.g., using racial cues in a message) were included because such message manipulations focused on specific race‐related aspects of the policy that had clear racial equity implications (e.g., when the message featured people of color as the beneficiaries of the policy through racial cues). The final list consisted of 55 papers or articles reporting on 80 different studies.

Message Strategies and Study Characteristics Examined in the Review

We categorized the message strategies featured in each included study. We examined several different message strategies that appeared in the corpus of relevant work, defined here.

Racial Cues

Messages with racial cues identify the race of the people involved in an issue or event very briefly, either via images that provide racial information or via text (“John is a [Black or White] man who has experienced [the issue at hand]”; “Poverty is a major problem among [Black or White] adults”).

Racial Stereotypes

Racial stereotype messages not only specify the race of the people involved in an issue or event but also include additional race‐related stereotypical information in message manipulations.

Nonracial Cues

Messages with nonracial cues include features that could invite racial connotations but do not include explicit mention of racial identity. These cues could include policy implementation details (e.g., how welfare payments have been distributed), political primes (asking people to think about different kinds of values), or language that is not explicitly race‐related but has potential for negative connotations (e.g., referring to immigrants as “aliens”).

Social Comparison Frames

Messages with social comparison frames explicitly compare social conditions and/or outcomes among racial groups. These studies differ from simple racial cues (depictions of a Black or White person in the study) in that they embed race‐specific comparison information in the description of these issues and/or in the discussion of the impacts of relevant policies. Most social comparison framing studies compare responses to short statements emphasizing racial differences that are embedded within longer messages.

Causal Frames

Messages with causal frames include explicit statements about the root cause (e.g., personal, social, or structural) of racial inequity in a social issue or health condition.

Responsibility Frames

Messages with responsibility frames often describe who is responsible for addressing the social issue at hand (in this case, racial inequity). These messages usually involve longer‐form messages (usually several paragraphs in length, often modeled after print news stories) that either feature isolated individual cases of an issue (often called an episodic frame) or emphasize its broader societal context (often called a thematic frame). ⁵¹

Diversity Frames

Messages with diversity frames describe the value of diversity for organizations and society, often comparing different interpretations regarding the social and societal implications of diversity.

Strategy Frames

Messages with strategy frames tend to feature lengthier message manipulations that depict policymaking as a strategic move to win votes.

We also characterized studies by the policy domain (using the four categories described in Table 1), policy‐related outcomes measured, psychological mechanisms (potential mediators), and moderators linked to policy support based on the aforementioned logic model (Figure 1). We conducted an in‐depth analysis of each study's policy contexts, message strategies, message length, sample characteristics, and study design (cross‐sectional or longitudinal). We paid close attention to the consistency of results across message strategies, study samples, research teams, and policy domains to inform our assessment of the evidence. We present our findings with studies organized in three different ways: overall, by policy domain, and by message strategy.

Results

Overall Summary of Study Characteristics

We present overall summary statistics on the message strategies employed, message length, sample composition, and study designs for all 80 studies included in this review in Table 2. The most frequently studied message strategy was racial cues (45% of studies), followed by social comparison framing (24%) and nonracial cues (14%). Other framing strategies comprised 11% or fewer of the available studies. Most studies (91%) featured relatively short message manipulations (a few sentences); only a handful (18%) tested longer‐form messages. Most studies also employed general population (e.g., non–race‐specific but typically majority White samples) (49%) or White‐only samples (45%), with few oversampling historically minoritized racial/ethnic groups (5%). All studies were cross‐sectional (100%); none offered assessments of message impact over time.

Table 2.

Overall Summary of 80 Studies Included in Review^a

Study Features	Number of Studies (%)
Message strategy
Racial cues	36 (45%)
Social comparison framing	19 (24%)
Nonracial cues	11 (14%)
Responsibility framing	9 (11%)
Causal framing	6 (8%)
Racial stereotypes	5 (6%)
Diversity framing	4 (5%)
Strategy framing	3 (4%)
Message manipulation length
Short	73 (91%)
Long	14 (18%)
Sample composition
General population sample (non–race‐specific but typically majority White sample)	39 (49%)
White‐only sample	36 (45%)
Oversampling of Black respondents	4 (5%)
A “non‐Black sample”	1 (1%)
Study designs
Cross‐sectional survey experiment	80 (100%)
Longitudinal survey experiment	0 (0%)

^a The proportions may not add up to 100% because some studies tested multiple message strategies. Short message manipulations are typically a few sentences or less (a phrase or even a single word).

Summary of Studies Organized by Policy Domain

We next describe message strategies employed in studies within each of the four policy domains outlined in Table 1. A study may fall into multiple policy domains if the outcome measures involve more than one type of policy. Table 3 lists specific studies and characteristics.

Table 3.

Summary Table of Studies Included in the Review by Policy Domain

Policy Domains (Policies)	Author (Year)	Message Strategy and Manipulation Length	Sample and Design
Policy domains typically framed explicitly to advance racial equity Affirmative action: 18 studies Race‐related/targeted policies (aggregate): 3 studies Government aid to Black people: 2 studies Government responsibility to address racial inequality: 1 study	Craig and Richeson, 2014 (studies 2 and 3a) 52 Eibach and Purdie‐Vaughns, 2011 (studies 1 and 2) 53 Fleischmann and Burgmer, 2020 (study 2) 54 Huber and Lapinski, 2006 (studies 1 and 2) 55 Iyengar, 1996 (racial inequality study) 51 Iyer and colleagues, 2003 (study 2) 56 Lowery and colleagues, 2012 (studies 2, 3, and 4) 57 Lowery and colleagues, 2006 (studies 2 and 3) 58 Nelson and Kinder, 1996 (studies 3 and 4) 59 Ramasubramanian, 2011 60 Reyna and colleagues, 2006 (studies 1 and 2) 61 Richardson, 2005 62 Trawalter and colleagues, 2016 (studies 4 and 5) 63	22 studies in total (some studies adopted multiple message strategies) Social comparison framing: 9 studies (short: 7; long: 2) Racial cues: 6 studies (short) Diversity framing: 4 studies (short) Racial stereotypes: 2 studies (short) Nonracial cues: 1 study (short) Responsibility framing: 1 study (long)	White sample: 17 studies General population sample (non–race‐specific but typically majority White sample): 4 studies Non‐Black sample: 1 study All the studies employ cross‐sectional designs
Policy domains not exclusively framed for racial equity but with evidence of equity‐promoting effects Health care reform/health insurance expansion: 10 studies Abortion access: 2 studies Childcare subsidies: 1 study Government intervention to address health disparities: 1 study Justice reinvestment: 1 study Social security: 1 study	Cappella and Jamieson, 1996 64 Cassese and Barnes, 2019 65 Detenber and colleagues, 2007 66 Gollust and Lynch, 2011 (study 3) 67 James and van Ryzin, 2017 68 Knowles and colleagues, 2010 69 Liebertz and Bunch, 2021 70 Martin and colleagues, 2017 71 Rigby and colleagues, 2009 72 Springer and Harwood, 2015 73 Tesler, 2012 74 Valentino and colleagues, 2018 (studies 1, 2, 3, and 4) 75 Wozniak, 2019 76	16 studies in total (some studies adopted multiple message strategies) Racial cues: 8 studies (short) Social comparison framing: 2 studies (short) Nonracial cues: 2 studies (short) Strategy framing: 2 studies (long) Causal framing: 1 study (short) Responsibility framing: 1 study (long)	General population sample (non–race‐specific but typically majority White sample): 10 studies White sample: 5 studies Oversampling Black people: 1 study All the studies employ cross‐sectional designs
Policy domains with evidence of inequity‐widening effects Death penalty: 4 studies Mandatory minimum sentencing: 2 studies Punitive crime policies: 2 studies Punitive/restrictive immigration policies: 2 studies Voter ID laws: 2 studies Abortion ban: 1 study Deportation: 1 study Family cap: 1 study Police brutality: 1 study Police search: 1study Stop and frisk: 1 study Three strikes law: 1 study	Butler and colleagues, 2018 (studies 1 and 2) 77 Fridkin and colleagues, 2017 78 Gilliam and Iyengar, 2000 79 Gross, 2008 (studies 1 and 2) 80 Hetey and Eberhardt, 2014 (studies 1 and 2) 81 Hurwitz and Peffley, 1997 (Carjack experiment) 82 Lee and colleagues, 2008 (study 2) 83 Nelson and Oxley, 1999 (study 2) 84 Ommundsen and colleagues, 2014 85 Peffley and Hurwitz, 2007 86 Peffley and colleagues, 2017 87 Peffley and colleagues, 1997 (police search experiment) 88 Simon and Jerit, 2007 89 Sirin and colleagues, 2016 90 Valentino and Neuner, 2017 91 Wilson and Brewer, 2013 92	19 studies in total (some studies adopted multiple message strategies) Racial cues: 7 studies (short) Social comparison framing: 7 studies (short) Nonracial cues: 3 studies (short) Responsibility framing: 2 studies (long) Racial stereotypes: 1 study (short) Causal framing: 1 study (short) Strategy framing: 1 study (long)	General population sample (non–race‐specific but typically majority White sample): 11 studies White sample: 5 studies Oversampling racial/ethnic minorities: 3 studies All the studies employ cross‐sectional designs
Policy domains with implications for racial equity that are highly contingent on policy implementation Welfare policies: 7 studies Immigration policies: 6 studies Policy solutions to poverty: 3 studies Crime policies: 2 studies Government spending: 2 studies COVID‐19 policies: 1 study Early release: 1 study Government responsibility to address crime: 1 study Government responsibility to address unemployment: 1 study Government investment in health research: 1 study Government assistance to the homeless: 1 study Policy solutions to religious radicalism: 1 study Prison furlough: 1 study Societal assistance with health care costs: 1 study	Brader and colleagues, 2008 (studies 1 and 2) 93 Ellis and Faricy, 2020 (studies 2 and 3) 94 Feezell and colleagues, 2021 95 Gollust and colleagues, 2010 96 Gollust and Lynch, 2011 (study 1) 67 Gross and Wronski, 2021 (studies 1 and 2) 97 Hannah and Cafferty, 2006 98 Harell and Lieberman, 2021 99 Huber and Lapinski, 2006 (studies 1 and 2) 55 Hurwitz and Peffley, 1997 (prison furlough experiment; rehabilitation experiment; preventive policies experiment) 82 Iyengar, 1990 100 Iyengar, 1996 (crime study; unemployment study) 51 Knoll and colleagues, 2011 101 Merolla and colleagues, 2013 102 Peffley and colleagues, 1997 (welfare experiment) 88 Valentino, 1999 103 Valentino and colleagues, 2013 104 Valentino and colleagues, 2018 (study 4) 75 Wallace and Wallace, 2020 105	26 studies in total (some studies adopted multiple message strategies) Racial cues: 18 studies (short) Nonracial cues: 5 studies (short) Responsibility framing: 5 studies (long) Causal framing: 4 studies (short) Racial stereotypes: 2 studies (short) Social comparison framing: 1 study (short)	General population sample (non–race‐specific but typically majority White sample): 14 studies White sample: 12 studies All the studies employ cross‐sectional designs

Policy Domains Typically Framed Explicitly to Advance Racial Equity

Twenty‐two studies examined the effects of message strategies on policy domains typically framed explicitly to advance racial equity. Among these, the vast majority (18 studies) focused on the issue of affirmative action. Other policy topics included race‐targeted policies in aggregate, government aid to Black people, and responsibility to address racial inequality. The modal study in this category tested the effects of social comparison framing (nine studies), although others in this category tested the effects of racial cues (six studies), diversity framing (four studies), racial stereotypes (two studies), nonracial cues (one study), or responsibility framing (one study).

Policy Domains Not Exclusively Framed for Racial Equity but With Evidence of Equity‐Promoting Effects

Sixteen studies examined the effects of message strategies on policy domains not exclusively framed for racial equity but with evidence of equity‐promoting effects. Among these, most of the studies focused on health care reform (10 studies). Other topics included abortion access, childcare subsidies, justice reinvestment, and social security. The modal study in this category focused on the effects of racial cues (eight studies), with a handful of others exploring social comparison frames (two studies), nonracial cues (two studies), strategy framing (two studies), causal framing (one study), and responsibility framing (one study).

Policy Domains With Evidence of Inequity‐Widening Effects. Nineteen studies examined the effects of message strategies on policy domains with evidence of inequality widening effects. These studies focused on punitive crime policies, restrictive immigration policies, death penalty, mandatory minimum sentencing, three strikes law, stop and frisk, intensive deportation policies, voter ID laws, family cap, abortion ban, etc. The most common message strategies studied in this category of policies were racial cues (seven studies) and social comparison framing (seven studies), with a few others on nonracial cues (three studies), responsibility framing (two studies), racial stereotypes (one study), causal framing (one study), and strategy framing (one study).

Policy Domains With Implications for Racial Equity That Are Highly Contingent on Policy Implementation

Twenty‐six studies examined the effects of message strategies on policy domains with implications for racial equity that are highly contingent on policy implementation. A wide range of policy topics were featured (e.g., welfare, immigration, crime, and health), but many of these were described at some level of abstraction (e.g., government responsibility to address unemployment or other social issues). Thus, the effects of these policies on racial equity may be dependent on the specific ways they are implemented. The most common message strategy tested in this policy category was the effects of racial cues (18 studies), with several others testing the effects of nonracial cues (5 studies), responsibility framing (5 studies), causal framing (4 studies), racial stereotypes (2 studies), or social comparison framing (1 study).

Summary of Evidence for Each Message Strategy

Table 4 presents a detailed summary of study characteristics and results organized by the eight message strategies that were tested in this body of work. We offer high‐level summaries of the findings here but invite readers to review the table for examples of each study category.

Table 4.

Studies Included in the Review Organized by Primary Message Feature Tested ^a

Author (Year)	Message Strategy and Length of Manipulation	Sample and Design	Policy in Focus	Effects on Policy Support	Mediator	Moderator
Racial cues
Huber and Lapinski, 2006 (study 1) 55	Racial cues (an ad using racial language with explicit reference to Black people vs. neutral language without racial reference) (short) Control condition (a public service announcement)	2634 non‐Hispanic White US adults (cross‐sectional design)	Government aid to Black people (policy domain 1) Affirmative action (policy domain 1) Welfare work requirements (policy domain 4) Government spending (policy domain 4)	Anti‐Black predispositions predicted support for policies regardless of racial cue types.	–	–
Huber and Lapinski, 2006 (study 2) 55	Racial cues (an ad using racial language with explicit reference to Black people vs. neutral language without racial reference) (short) Image (an African American woman vs. a Caucasian woman) (short)	3,733 non‐Hispanic White US adults (cross‐sectional design)	Government aid to Black people (policy domain 1) Affirmative action (policy domain 1) Welfare work requirements (policy domain 4) Government spending (policy domain 4)	Anti‐Black predispositions predicted support for policies regardless of racial cue types.	–	–
Iyer and colleagues, 2003 (study 2) 56	Racial cues (measurements about European Americans as the perpetrators of racial discrimination vs. African Americans as the victims of racial discrimination) (short)	250 European American/White undergraduate students (cross‐sectional design)	Affirmative action (policy domain 1)	Measurements about European Americans as the perpetrators of racial discrimination increased guilt, which in turn increased support for compensatory policy. Measurements about African Americans as the victims of racial discrimination increased sympathy, which in turn increased support for equal opportunity policy.	Belief in discrimination Guilt Sympathy	–
Reyna and colleagues, 2006 (study 1) 61	Racial cues (specifying that “Blacks” or “women” benefit from affirmative action policies) (short)	893 White US adults (cross‐sectional design)	Affirmative action (policy domain 1)	Respondents were more likely to oppose affirmative action for Black people than for women. People who were more educated showed less difference in their support for affirmative action for Black people versus women.	–	Education
Reyna and colleagues, 2006 (study 2) 61	Racial cues (asking respondents about their support for affirmative action for African Americans/Black people and women, respectively) (short)	184 White US adults from the greater Chicago area (cross‐sectional design)	Affirmative action (policy domain 1)	Respondents were more likely to oppose affirmative action for Black people than for women.	–	–
Cassese and Barnes, 2019 65	Racial cues (“To what extent do you favor providing government subsidies for child care to assist ________ working mothers?” Respondents were randomized to one of the beneficiary groups below: poor, middle class, White, Black, middle‐class White, middle‐class Black, poor White, or poor Black.) (short)	663 White adults (cross‐sectional design)	Childcare subsidies (policy domain 2)	Mentioning Black working mothers as the recipients of child care subsidies reduced support for the policy compared with mentioning White working mothers. Mentioning poor Black mothers reduced support for the policy compared with mentioning poor White mothers or poor mothers. People with higher racial resentment were less likely to support childcare subsidies when Black mothers or poor Black mothers were mentioned as beneficiaries, but not when White mothers were mentioned.	–	Racial resentment
Knowles and colleagues, 2010 69	Racial cues (specifying the health care reform as President Obama's plan vs. President Clinton's 1993 plan) (short)	130 US adults (cross‐sectional design if only focused on this specific assessment)	Health care reform (policy domain 2)	Specifying that the health care reform was President Obama's plan reduced support for the reform among people with racial prejudice.	–	Racial prejudice
Tesler, 2012 74	Racial cues (mentioning that some people proposed health care reform policies [control] vs. specifying that these policies were from President Clinton's 1993 reform plans vs. these policies were proposed by President Obama) (short)	Nationally representative US adult sample (cross‐sectional design)	Health care reform (universal coverage and a government‐administered health insurance plan) (policy domain 2)	Racial resentment was more negatively associated with support for health care reform when the proposal was mentioned as from Obama than Clinton. Racial stereotyping was more negatively associated with support for health care reform when the proposal was mentioned as from Obama than Clinton or the control condition.	–	–
Valentino and colleagues, 2018 (study 1) 75	Racial cues (news stories about a controversial campaign advertisement: describing pro‐ and anti‐ACA activists as inner‐city and suburban residents vs. describing pro‐ and anti‐ACA activists as Black people and White people) (short) Control (irrelevant topic)	2,394 voting‐age White Americans (cross‐sectional design)	Health care reform (policy domain 2)	Regardless of racial cue types, symbolic racism reduced support for health care reform.	–	–
Valentino and colleagues, 2018 (study 2) 75	Racial cues (reports depicting the clash over passage of the Health Care Bill as between inner‐city and suburban residents vs. between Black people and White people) (short) Control (irrelevant topic)	234 White US adults (cross‐sectional design)	Health care reform (policy domain 2)	Regardless of racial cue types, symbolic racism reduced support for health care reform.	–	–
Valentino and colleagues, 2018 (study 3) 75	Racial cues (news stories about a controversial campaign advertisement: describing pro‐ and anti‐ACA activists as inner‐city and suburban residents vs. describing pro‐ and anti‐ACA activists as Black people and White people) (short) Control (irrelevant topic)	321 White US adults (cross‐sectional design)	Health care reform (policy domain 2)	Regardless of racial cue types, symbolic racism decreased support for health care reform.	–	–
Valentino and colleagues, 2018 (study 4) 75	Racial cues (news stories about a controversial advertisement focusing on the government programs to assist the poor vs. Black people) (short) Control (irrelevant topic)	3,114 White US adults (cross‐sectional design)	Health care reform (policy domain 2) Welfare policies (policy domain 4)	Regardless of racial cue types, symbolic racism decreased support for social welfare, including health care reform.	–	–
Wozniak, 2019 76	Racial cues (mentioning of communities that would receive justice reinvestment funds: high crime communities vs. high poverty communities vs. communities with many residents on welfare vs. rural communities vs. inner city communities vs. African American communities vs. communities) (short)	1,812 non‐Hispanic White and Black adults (about half of the respondents are Black adults) (cross‐sectional design)	Justice reinvestment (policy domain 2)	Mentioning African American communities as receiving justice reinvestment funds made respondents less likely to allocate the funds to health care and economic development compared with the control condition.	–	Race Racial resentment Perceived fairness of the justice system
Fridkin and colleagues, 2017 78	Racial cues (an introductory paragraph describing an altercation between a White police officer and an African American professor using a law‐and‐order frame vs. a police brutality frame vs. a race frame vs. no frame) (short)	225 undergraduate students (cross‐sectional design)	Police brutality (policy domain 3)	Police brutality frame reduced support for the police officer's actions compared with the law‐and‐order frame. No significant difference in effects among other frames.	Perceived racism in law enforcement	–
Gilliam and Iyengar, 2000 79	Racial cues (a videotaped news story about crime without showing the identity of the perpetrator vs. with a White male as the perpetrator vs. with an African American male as the perpetrator) (short) Control condition (no crime story)	2,331 Los Angeles metropolitan area residents (cross‐sectional design)	Punitive crime policies (policy domain 3)	White respondents became more supportive of punitive crime policies when the crime story showed an African American perpetrator compared with the no crime story condition. Black respondents became less supportive of punitive crime measures when watching a crime story that did not mention the identity of the perpetrator.	Dispositional attributions Racism	Race
Hurwitz and Peffley, 1997 (carjack experiment) 82	Racial cues (judgments of punishment of a White suspected carjackers vs. a Black suspected carjacker) (short)	501 White US adults (cross‐sectional design)	Punitive crime policy (policy domain 3)	Stereotypical beliefs about Black people increased support for more severe punishment to a Black suspect.	/	Stereotypical beliefs
Sirin and colleagues, 2016 90	Racial cues (photo of an undocumented immigrant detainee, White vs. Black vs. Latino vs. Arab, accompanying a news story about the undocumented immigrant detainee requiring medical treatment outside the detention facility) (short)	671 US adults (about one‐third of the respondents are Anglo, one‐third are African Americans, and one‐third are Latino/a) (cross‐sectional design)	Intensive deportation policies (policy domain 3)	Latino/a and African American respondents had higher levels of empathy than Anglo respondents when the photo depicted a racial minority detainee, which in turn was associated with more opposition to deportation policies.	Empathy	Race
Brader and colleagues, 2008 (study 1) 93	Ethnic cues (photo and name of an immigrant (Latino vs. European) depicted in a news story about a governors’ conference on the issue of immigration) (short) Tone of the news report (positive vs. negative consequences of immigration for the country) (short) Control condition	354 White US adults (cross‐sectional design)	Immigration policies (policy domain 4)	News stories about the negative consequences of immigration increased anxiety among White respondents, reduced support for immigration, and mobilized them to send a message against immigration to Congress when the story featured a Latino immigrant compared with a European immigrant.	Anxiety	–
Brader and colleagues, 2008 (study 2) 93	Ethnic cues (photo and name of an immigrant (Latino vs. European) depicted in a news story about a governors’ conference on the issue of immigration highlighting its negative consequences) (short) Featuring the immigrant worker as high skilled vs. low skilled (short) Control condition	220 White US adults (cross‐sectional design)	Immigration policies (policy domain 4)	News stories about the negative consequences of immigration featuring a low‐skilled immigrant worker increased anxiety among White respondents and mobilized them to send a message against immigration to Congress when the story featured a Latino immigrant compared with a European immigrant.	Anxiety	–
Hurwitz and Peffley, 1997 (prison furlough experiment) 82	Racial cues (prison furlough programs to help violent White prisoners vs. model White prisoners vs. violent Black prisoners vs. model Black prisoners) (short)	501 White US adults (cross‐sectional design)	Prison furlough programs (policy domain 4)	Stereotypical beliefs about Black people reduced policy support when mentioning violent Black prisoners but not in other conditions.	–	Stereotypical beliefs
Hurwitz and Peffley, 1997 (rehabilitation experiment) 82	Racial cues (a 25‐year‐old Black vs. White prisoner hoping to be released early) (short)	501 White US adults (cross‐sectional design)	Early release (policy domain 4)	Stereotypical beliefs about Black people reduced perceived likelihood of rehabilitation for a Black prisoner.	–	Stereotypical beliefs
Hurwitz and Peffley, 1997 (preventive policies experiment) 82	Racial cues (asking respondents about their approval of government spending to help people get jobs and to have more drug rehabilitation programs in White neighborhoods vs. Black neighborhoods) (short)	501 White US adults (cross‐sectional design)	Crime prevention policies (policy domain 4)	Stereotypical beliefs about Black people did not predict policy support regardless of racial cues.	–	–
Valentino and colleagues, 2013 104	Racial cues (asking how worried respondents are about “the effect that immigrants from Latin America vs. East Asia vs. Africa vs. Eastern Europe are having on communities across the United States”) (short) Control condition (same question but did not specify any region)	421 White US adults (cross‐sectional design)	Immigration policies (policy domain 4)	Specifying immigrants as from East Asia, Africa, or East Europe reduced concerns about immigration compared with immigrants from Latin America or the control condition in which no region is mentioned.	–	–
Racial stereotypes
Nelson and Kinder, 1996 (study 4) 59	Racial stereotypes (negative stereotypic images of Black people vs. counterstereotypic images of Black people vs. no images of Black people) (short)	84 non‐Black undergraduate students (cross‐sectional design)	Affirmative action (policy domain 1)	Negative stereotypic images of Black people strengthened the association between attitudes toward Black people and support for affirmative action compared with both the counterstereotypic images of Black people condition and the control condition combined.	Group centrism	Racial attitudes
Ramasubramanian, 2011 60	Racial stereotypes (pictures of media exemplars: positive in‐group and out‐group exemplars vs. negative in‐group and out‐group exemplars vs. positive in‐group and negative out‐group exemplars vs. negative in‐group and positive out‐group exemplars) (short)	363 White undergraduate students (cross‐sectional design)	Affirmative action (policy domain 1)	Nonsignificant direct effects of media exemplars on policy support. However, stereotypical out‐group exemplars increased stereotypical beliefs about out‐group, internal attributions of out‐group failures, and prejudicial feelings, which in turn decreased support for affirmative action policies.	Stereotypical beliefs Attributions of failure Prejudicial feelings	–
Peffley and colleagues, 1997 (police search experiment) 88	Racial stereotypes (“police see two young [Black vs. White] men about 20 years old. They are [using foul language vs. well‐dressed and well‐behaved] and walking very near a house where the police know drugs are being sold. The police search them and find that they are carrying drugs. Do you think this is a reasonable search?”) (short)	1,650 White adults (cross‐sectional design)	Police search (policy domain 3)	Hostility stereotype was associated with approval of police search when the suspects were Black men and using foul language.	Stereotypical beliefs	–
Peffley and colleagues, 1997 (welfare experiment) 88	Racial stereotypes (asking respondents their support for a welfare program designed to help [“blacks” vs. “new immigrants from Europe”] and people who will benefit from the program [“want to work their way out of their problems” vs. “have had trouble hanging on to their jobs”].) (short)	737 White adults (cross‐sectional design)	Welfare program (policy domain 4)	Work ethic stereotype was associated with support for welfare when the beneficiaries of the welfare program were Black people who had trouble holding their jobs.	Stereotypical beliefs	–
Valentino, 1999 103	Racial stereotypes (5‐second mug shots of two White suspects vs. Black suspects vs. Asian suspects vs. Hispanic suspects vs. no suspects in a gang‐related crime story) (short) Control condition (no crime news story)	289 US adults in the Los Angeles area (cross‐sectional design)	Crime policies (policy domain 4) Welfare policies (policy domain 4)	Featuring a crime story (regardless of the race of the suspects depicted) made respondents more likely to use Clinton's performance on crime to evaluate him. Featuring a crime story with Black suspects made respondents more likely to use Clinton's performance on welfare to evaluate him.	–	–
Nonracial cues
Fleischmann and Burgmer, 2020 (study 2) 54	Nonracial cues (abstract vs. concrete thinking about the affirmative action policy: asking respondents to think about “why” the policy is implemented vs. “how” the policy is implemented.) (short)	201 US adults (cross‐sectional design)	Affirmative action (policy domain 1)	Nonsignificant effects between priming people to think about the policy abstractly (about “why”) vs. concretely (about “how”). However, for people who perceived a level of discrimination against African Americans, thinking abstractly about affirmative action increased positive attitudes toward affirmative action compared with thinking concretely about affirmative action.	–	Perceived discrimination
Detenber and colleagues, 2007 66	Nonracial cues: intensity of the protest featured in the television news coverage (high intensity vs. low intensity) (short) Issue position of the protest (prolife vs. prochoice) (short)	256 undergraduate students, the University of Delaware (cross‐sectional design)	Abortion rights (policy domain 2)	Featuring the protest as high intensity increased criticisms of the protestors and reduced perceived support for the protest group. Such effects were independent from the issue position of the protest.	–	–
James and van Ryzin, 2017 68	Nonracial cues (political prime (asking respondents questions about the role of government) vs. health care needs prime (asking respondents about affordability of and access to health care)) (short)	661 US adults (cross‐sectional design)	ACA (policy domain 2)	Priming did not significantly affect beliefs about ACA. However, political priming (compared with health care needs priming) polarized respondents’ evaluations of pro‐ACA evidence along party lines: Democrats rated pro‐ACA evidence as stronger, whereas republicans rated the same evidence as weaker. Political priming also made respondents more likely to choose the indicators of ACA's performance based on their party ID.	–	Political party ID
Ommundsen and colleagues, 2014 85	Nonracial cues (labeling people who entered the United States without authorization “illegal immigrants” vs. “illegal aliens” vs. “undocumented immigrants”) (short)	274 undergraduate students (cross‐sectional design)	Punitive immigration policies (policy domain 3)	The label “undocumented immigrants” reduced support for punitive immigration policies compared with “illegal immigrants.” The label “illegal aliens” reduced support for punitive immigration policies compared with both “illegal immigrants” and “undocumented immigrants.”	–	–
Simon and Jerit, 2007 89	Nonracial cues (using the word “baby” vs. “fetus” vs. both words when discussing partial‐birth abortion in a news article) (short) Control condition: no message	185 respondents (about equal number of undergraduate students and adults) (cross‐sectional design)	Abortion ban (policy domain 3)	Support for partial‐birth abortion ban was lower among people who read the news article using the word “fetus” compared with the word “baby,” both words, and the no‐message control.	–	–
Wilson and Brewer, 2013 92	Nonracial cues (asking for opinions about voter ID laws vs. stating that these laws “are necessary to keep people who aren't eligible to vote from voting” vs. these laws “can keep people from voting multiple times” vs. these laws “can actually prevent people who are eligible to vote from voting” vs. these laws “are unnecessary because voter fraud is very rare”) (short)	906 US adults (nationally representative sample) (cross‐sectional design)	Voter ID laws (policy domain 3)	Stating that voter ID laws might “actually prevent people who are eligible to vote from voting” reduced support for the laws. Such effects only appear among Democrats. Other conditions had similar levels of policy support.	–	Political party ID
Ellis and Faricy, 2020 (study 2) 94	Nonracial cues (welfare program delivery mechanisms: direct spending in which citizens receive checks purchasing groceries vs. tax expenditure in which citizens receive tax credits) (short)	US nationally representative sample (cross‐sectional design)	Welfare programs (policy domain 4)	Symbolic racism had stronger effects on support for programs delivered through direct spending than tax expenditure.	–	–
Ellis and Faricy, 2020 (study 3) 94	Nonracial cues (direct spending vs. tax expenditure similar to study 2 but in three different contexts: paying for necessities, health care, and wage subsidies) (short)	US nationally representative sample (cross‐sectional design)	Welfare programs (policy domain 4)	Symbolic racism had stronger effects on perceived deservingness when welfare programs were delivered through direct spending than tax expenditure in the context of health care and paying for necessities, but not wage subsidies.	–	–
Knoll and colleagues, 2011 101	Nonracial cues (labeling people who entered the United States without authorization “illegal immigrants” vs. “undocumented immigrants”) (short)	496 likely Iowa Caucus‐goers (cross‐sectional design)	Immigration policies (policy domain 4)	The label “undocumented immigrants” increased support for the temporary work policy compared with “illegal immigrants.”	–	–
Merolla and colleagues, 2013 102	Nonracial cues (labeling people who entered the United States without authorization “illegal immigrants” vs. “undocumented immigrants” vs. “unauthorized immigrants”) (short) Nonracial cues (using the term “amnesty” vs. “opportunity to eventually become citizens”; mentioning that the DREAM Act will benefit illegal immigrants “who came to the US as young children” or not; using the term “law” vs. “Constitution” when describing birthright citizenship) (short)	2,188 US adults (cross‐sectional design)	Immigration policies (policy domain 4)	Nonsignificant effects of these cues on support for legalization, DREAM Act, or birthright citizenship. The term “opportunity to eventually become citizens” increased support for legalization compared with the word “amnesty.” Mentioning that the DREAM Act will benefit illegal immigrants “who came to the US as young children” increased support, especially among Republicans. Using the term “Constitution” compared with “law” when describing birthright citizenship increased support for the policy among Democrats and Independents but not Republicans.	–	Political party ID
Wallace and Wallace, 2020 105	Nonracial cues (specifying qualifying criteria for US citizenship: “enroll in college” vs. “join the military” vs. “join the military or enroll in college” vs. the control condition “grew up in the United States”) (short)	1,000 US adults (cross‐sectional design)	Immigration policies (policy domain 4)	Specifying the criteria for US citizenship as “join the military” or “join the military or enroll in college” increased support for such people to become citizens compared with the control condition (especially among Republicans/Conservatives), whereas featuring the criteria for US citizenship as “enroll in college” had little impact.	–	Political party ID Political ideology
Social comparison framing
Craig and Richeson, 2014 (study 2) 52	Social comparison framing (expecting a racial shift in which racial minorities will become a majority of the US population) (long) Control condition (irrelevant topic)	415 White US adults (cross‐sectional design)	Race‐related policies (policy domain 1)	The racial‐shift condition made White respondents have more conservative policy positions than the control condition.	Perceived group status threat	–
Craig and Richeson, 2014 (study 3a) 52	Social comparison framing (expecting a racial shift in which racial minorities will become a majority of the US population vs. expecting a racial shift but with status relations among groups unchanged) (long) Control condition (irrelevant topic)	170 White US adults (cross‐sectional design)	Race‐related policies (policy domain 1)	The racial‐shift condition made White respondents have more conservative policy positions than both the control condition and the condition mentioning that status relations among groups will remain unchanged despite such shift.	–	Baseline political ideology
Eibach and Purdie‐Vaughns, 2011 (study 1) 53	Social comparison framing (progress frame: “How much progress have Americans made towards achieving racial equality?” vs. commitment frame: “How strongly are Americans committed to achieving racial equality?”) (short)	140 White undergraduate students (cross‐sectional design)	Race‐targeted policies in aggregate (policy domain 1)	Commitment frame significantly increased support for race‐targeted policies compared with the progress frame.	–	–
Eibach and Purdie‐Vaughns, 2011 (study 2) 53	Social comparison framing (progress framing vs. commitment framing of civil rights accomplishments varying the instructions, title, and the concluding statement of a message) (short) Control condition (without information about progress or commitment)	75 White undergraduate students (cross‐sectional design)	Affirmative action (policy domain 1)	Commitment frame condition significantly increased support for affirmative action compared with the progress frame and the control condition.	Frame‐relevant thoughts	–
Lowery and colleagues, 2012 (study 2) 57	Social comparison framing (stating that recruiting policies of a fictitious company unfairly “disadvantaged Blacks” vs. “advantaged Whites” vs. in the absence of inequity information) (short) Outcome framing (affirmative action policies increased the percentage of Black employees vs. affirmative action policies decreased the percentage of White employees) (short)	136 White adults (cross‐sectional design)	Affirmative action (policy domain 1)	Mentioning that recruiting policies unfairly “advantaged Whites” increased support for affirmative action (compared with mentioning that recruiting policies “disadvantaged Blacks” or in the absence of inequity information) when the outcome of affirmative action was framed as reducing the percentage of White employees.	–	–
Lowery and colleagues, 2012 (study 3) 57	Social comparison framing (stating that research shows White advantage (higher chances of being hired and higher salaries) vs. Black disadvantage (lower chances of being hired and lower salaries)) (short) Outcome framing (affirmative action policies increased economic opportunities for minorities vs. affirmative action policies decreased economic opportunities for White people) (short)	88 White adults (cross‐sectional design)	Affirmative action (policy domain 1)	Highlighting White advantage increased support for affirmative action (compared with highlighting Black disadvantage) when the outcome of affirmative action was framed as reducing White privilege.	Group esteem	–
Lowery and colleagues, 2012 (study 4) 57	Social comparison framing (manipulation was identical to study 3) (short) Self‐affirmation (ranking a list of 11 values and writing a paragraph to justify their top‐ranked value vs. no such tasks) (short)	155 White adults (cross‐sectional design)	Affirmative action (policy domain 1)	Replicated findings in study 3 among people without affirmation. Social comparison framing did not have significant effects among respondents in the affirmed condition.	–	Self‐affirmation
Lowery and colleagues, 2006 (study 2) 58	Social comparison framing (affirmative action policies increased the percentage of Black employees vs. affirmative action policies decreased the percentage of White employees) (short)	57 White undergraduate students (cross‐sectional design)	Affirmative action (policy domain 1)	Respondents who scored high in White racial identity were less supportive of affirmative action when the message depicted the policy as decreasing the percentage of White employees. White racial identity did not predict support for affirmative action when the policy was framed in terms of increase in Black employees.	–	White racial identity
Lowery and colleagues, 2006 (study 3) 58	Social comparison framing (similar to study 2 plus a condition stating that the percentage of White employees has not changed after the adoption of affirmative action policies) (short)	204 US adults (cross‐sectional design)	Affirmative action (policy domain 1)	Respondents who scored high in White racial identity were less supportive of affirmative action when the message depicted the policy as decreasing the percentage of White employees. White racial identity did not predict support for affirmative action when the policy was framed in terms of increase in Black employees or no change in the percentage of White employees.	–	White racial identity
Gollust and Lynch, 2011 (study 3) 67	Social comparison framing (mentioning a gap in life expectancy among racial groups vs. among gender groups vs. among income groups vs. among education groups) (short)	1,334 US adults (nationally representative sample) (cross‐sectional design)	Government‐financed health insurance (policy domain 2)	Mentioning a gap in life expectancy between African Americans and White Americans made people more likely to attribute disparities in life expectancy to biological reasons. Biological attribution was negatively associated with support for government‐financed health insurance.	Causal attribution	–
Rigby and colleagues, 2009 72	Social comparison framing (featuring health disparities between African Americans and White Americans vs. between the poor and middle class vs. between people with and without college degree) (short)	1,264 adults in Wisconsin (random digit dialing sample) (cross‐sectional design)	Government intervention to close health disparity gaps (policy domain 2)	Featuring racial disparities reduced support for government intervention compared with economic disparities or educational disparities.	Awareness of health disparities Causal attribution	–
Butler and colleagues, 2018 (study 1) 77	Social comparison framing (asking respondents about their support for death penalty “for persons convicted of murder” [control] vs. also mentioning that “most of the people who are executed are African Americans”) (replicating Peffley and Hurwitz, 2007) 86 (short) Racial cues: mugshot and name of a specific defendant facing death penalty because of murder (a White male vs. a Black male vs. no such information/photo) (short)	2,134 US adults (cross‐sectional design)	Death penalty (policy domain 3)	No significant effect for either social comparison framing or racial cues.	–	–
Butler and colleagues, 2018 (study 2) 77	Social comparison framing (asking respondents about their support for death penalty “for persons convicted of murder” [control] vs. also mentioning that “most of the people who are executed are African Americans”) (replicating Peffley & Hurwitz, 2007) 86 (short)	661 White US adults (cross‐sectional design)	Death penalty (policy domain 3)	No significant effect for social comparison framing.	–	–
Hetey and Eberhardt, 2014 (study 1) 81	Social comparison framing (racial disparities regarding the ratio of Black to White male inmates: 45% of the photographs were of Black inmates vs. 25% of the photographs were of Black inmates) (short)	62 White registered California voters (cross‐sectional design)	Three strikes law (policy domain 3)	Portraying more extreme racial disparities in the prison population made White respondents less likely to sign a petition to lessen the punitive three strikes law.	–	–
Hetey and Eberhardt, 2014 (study 2) 81	Social comparison framing (statistics about racial disparities regarding prison population: 60.3% of Black inmates and 11.8% of White inmates vs. 40.3% of Black inmates and 31.8% of White inmates) (short)	164 White New York City residents (cross‐sectional design)	Stop‐and‐frisk policy (policy domain 3)	Portraying more extreme racial disparities in the prison population made White respondents less likely to sign a petition to end the stop‐and‐frisk policy.	Fear of crime	–
Peffley and Hurwitz, 2007 86	Social comparison framing (asking respondents about their support for death penalty “for persons convicted of murder” [control] vs. also mentioning that “most of the people who are executed are African Americans” vs. mentioning that “penalty is unfair because too many innocent people are being executed.”) (short)	1,200 US adults (600 White adults and 600 Black adults) (cross‐sectional design)	Death penalty (policy domain 3)	Social comparison framing reduced support for death penalty compared with the control condition among Black respondents but increased support for death penalty among White respondents.	Causal attribution	Race
Peffley and colleagues, 2017 87	Social comparison framing (asking about respondents’ attitudes toward death penalty: “Some people say that the death penalty is unfair because African Americans convicted of the same crimes as Whites are much more likely to be executed. What about you?” vs. “Some people say that the death penalty is unfair because too many innocent people are being executed. What about you?” vs. baseline “Do you favor or oppose the death penalty for convicted murderers?”) (short)	1,204 US adults (majority White; oversampling of Latinos and African Americans) (cross‐sectional design)	Death penalty (policy domain 3)	The racial argument and the innocent argument decreased support for death penalty among Black Americans. The racial argument polarized support for death penalty based on respondents’ negative Black dispositional explanations.	–	Race Dispositional explanations
Valentino and Neuner, 2017 91	Social comparison framing (news articles about voter ID laws during the midterm election: article about the election mentioning voter ID laws vs. stating that voter ID laws might make some voters unable to cast their ballots vs. indicating that Black Americans will be affected by the voter ID laws the most vs. stating that ID laws prevent voter fraud vs. no information about voter ID laws) (short)	750 US adults (cross‐sectional design)	Voter ID laws (policy domain 3)	The Black disenfranchisement condition and the voter fraud condition increased advocacy and political participation intentions. The Black disenfranchisement condition increased anger, which in turn heightened advocacy and political participation intentions among Democrats.	Anger	Political party ID
Harell and Lieberman, 2021 99	Social comparison framing (featuring that “Blacks are 2.5 times more likely to die from COVID‐19 than Whites” or not) (short) Asking respondents about their estimates of racial disparities related to COVID‐19 mortality rates or not (short)	3,961 US adults (cross‐sectional design)	Policies to combat COVID‐19 (policy domain 4)	Featuring accurate racial disparity information about COVID‐19 death did not affect COVID‐19 policy support among Black respondents. It increased policy support among White respondents who were favorable toward Black people but decreased policy support among White respondents with colder feelings toward Black people.	Awareness of racial disparities	Black affect
Causal framing
Liebertz and Bunch, 2021 70	Causal framing (linking banning abortion to protecting unborn children vs. linking allowing abortion to protecting pregnant women vs. showing both vs. no‐frame control) (short)	831 US adults (cross‐sectional design)	Abortion rights (policy domain 2)	Nonsignificant effects of causal framing compared with the control condition. Associating allowing abortion with protecting pregnant women triggered reactance among the more religious people and people with born‐again Christian affiliation. Linking banning abortion to protecting unborn children triggered reactance among women.	–	Gender Religiosity Born‐again Christian affiliation
Nelson and Oxley, 1999 (study 2) 84	Causal framing (claims that family cap policy will increase responsibility of welfare mothers vs. the policy will worsen poverty among children) (short)	121 undergraduate students (cross‐sectional design)	Family cap (policy domain 3)	Causal framing did not directly affect attitude toward family cap. However, it marginally increased individual attribution beliefs, which in turn increased support for family cap.	Attribution	–
Gollust and colleagues, 2010 96	Causal framing (claims about causes of diabetes or no causal claim) (short) Racial cues (a photo of Black woman, White woman, or a glucose‐testing device) (short)	2,490 US adults (cross‐sectional design)	Government investment in health research (policy domain 4)	Causal claims about genetic causes and social determinants of diabetes increased support for government spending on diabetes research compared with no causal claims. Nonsignificant main effects of racial cues or interaction effects between causal framing and racial cues.	Negative stereotyping	–
Gollust and Lynch, 2011 (study 1) 67	Causal framing (likely cause of heart disease) (short) Racial cues (a White man or an African American man) (short)	1,342 US adults (nationally representative sample) (cross‐sectional design)	Societal assistance with health care costs (policy domain 4)	Respondents were less supportive of societal assistance of health care costs when heart disease was linked to the patient's behavior (smoking) compared with family history. Non‐White respondents were more supportive of societal assistance of health care costs when the patient was an African American man.	Attribution of blame	Race
Gross and Wronski, 2021 (study 1) 97	Causal framing (claims about external reasons for homelessness or no causal claim) (short) Racial cues (Black individuals or White individuals) (short)	691 White US adults (cross‐sectional design)	Government assistance to the homeless (policy domain 4)	Significant effects of claims about the external reasons for homelessness when the person depicted in the message was from one's own racial group.	Sympathy	–
Gross and Wronski, 2021 (study 2) 97	Causal framing (claims about external reasons for homelessness or no causal claim) (short) Racial cues (Black individuals or White individuals) (short) Control condition	1,226 White US adults (cross‐sectional design)	Government assistance to the homeless (policy domain 4)	Significant effects of claims about the external reasons for homelessness when the person depicted in the message was from one's own racial group.	Sympathy	–
Responsibility framing
Iyengar, 1996 (racial inequality study) 51	Responsibility framing (episodic vs. thematic framing of racial inequality) (long) Racial cues (featuring Black poverty or White poverty) (short)	40 to 244 residents of Suffolk County, New York (cross‐sectional design)	Government responsibility to address racial inequality (policy domain 1)	Nonsignificant effects of episodic vs. thematic framing on support for affirmative action. Episodic framing featuring Black poverty increased perceived individual treatment responsibility and reduced perceived societal/government treatment responsibility.	Causal attribution	–
Springer and Harwood, 2015 73	Responsibility framing (stereotypical episodic framing vs. counterstereotypical episodic framing vs. thematic framing) (long) Control condition	218 undergraduate students (cross‐sectional design)	Social Security (policy domain 2)	Episodic framing (featuring either stereotypical or counterstereotypical depictions of older adults) reduced support for Social Security than the thematic framing condition or the control condition. Policy support in the thematic condition did not differ from the control condition.	Attribution of responsibility	–
Gross, 2008 (study 1) 80	Responsibility framing (episodic vs. thematic framing of mandatory minimum sentencing) (long) Racial cues (a White defendant or a Black defendant was featured in episodic framing) (short) Control condition (irrelevant topic)	163 undergraduate students (cross‐sectional design)	Mandatory minimum sentencing (policy domain 3)	Thematic framing and episodic framing featuring a White defendant increased opposition to mandatory minimum sentencing compared with the control condition. Episodic framing featuring a Black defendant did not shift attitude toward mandatory minimum sentencing.	Empathy Aversion	–
Gross, 2008 (study 2) 80	Responsibility framing (episodic vs. thematic framing of mandatory minimum sentencing) (long) Racial cues (a White defendant or a Black defendant was featured in episodic framing) (short)	105 undergraduate students (cross‐sectional design)	Mandatory minimum sentencing (policy domain 3)	Nonsignificant differences between conditions on policy support.	Empathy	–
Feezell and colleagues, 2021 95	Responsibility framing (episodic vs. thematic framing) (long) Racial cues (Americans or Muslim Americans) (short) Issue (poverty vs. religious radicalism)	1,655 US adults (cross‐sectional design)	Policy solutions to poverty and religious radicalism (policy domain 4)	Regarding the issue of poverty, thematic framing increased perceived government responsibility. The pattern of effects was similar whether Americans or Muslim Americans were featured. Regarding the issue of religious radicalism, thematic framing did not increase perceived government responsibility whether Americans or Muslim Americans were featured.	–	–
Hannah and Cafferty, 2006 98	Responsibility framing (episodic vs. thematic framing of poverty) (long) Racial cues (featuring White poor people or Black poor people) (short)	200 undergraduate students (cross‐sectional design)	Policy solutions to poverty (policy domain 4)	Nonsignificant main effects of episodic vs. thematic framing. Featuring White poor people (compared with featuring Black poor people) in news footage made both Black and White respondents feel more strongly that too little was spent on programs to address poverty.	–	–
Iyengar, 1990 100	Responsibility framing (thematic vs. episodic framing) (long) Racial cues (featuring White poor people or Black poor people in episodic framing) (short)	244 residents of Suffolk County, New York (cross‐sectional design)	Government responsibility to address poverty (policy domain 4)	Thematic framing decreased perceived personal responsibility for poverty and increased perceived societal/government treatment responsibility to address poverty. Support for societal/government responsibility was higher when featuring White poor people than Black poor people.	Causal attribution	–
Iyengar, 1996 (crime study) 51	Responsibility framing (episodic vs. thematic framing of crime) (long) Racial cues (featuring a White perpetrator or a Black perpetrator in episodic framing) (short)	40 to 244 residents of Suffolk County, New York (cross‐sectional design)	Government treatment responsibility to crime (policy domain 4)	Episodic framing increased perceived individual treatment responsibility more than thematic framing when a White perpetrator was featured. Crime stories featuring a Black perpetrator increased individual treatment responsibility and reduced societal/government treatment responsibility regardless of framing.	Causal attribution	–
Iyengar, 1996 (unemployment study) 51	Responsibility framing (episodic vs. thematic framing of unemployment) (long) Racial cues (featuring a White unemployed worker or a Black unemployed worker in episodic framing) (short)	40 to 244 residents of Suffolk County, New York (cross‐sectional design)	Government responsibility to address unemployment (policy domain 4)	Nonsignificant effects of episodic vs. thematic framing. Nonsignificant effects of racial cues.	Causal attribution	–
Diversity framing
Nelson and Kinder, 1996 (study 3) 59	Diversity framing (stating that favoring Black people in hiring and promotion gives them unfair advantage vs. favoring Black people in hiring and promotion amounts to discrimination against White people) (short)	614 Americans of voting age (from National Election Study) (cross‐sectional design)	Affirmative action (policy domain 1)	Featuring affirmative action as giving unfair advantage to Black people strengthened the relationship between attitude toward Black people and support for affirmative action compared with the reverse discrimination frame, but the effect did not reach statistical significance.	Group centrism	Racial attitudes
Richardson, 2005 62	Diversity framing (news editorial featuring affirmative action as remedial action vs. diversity is good for all students vs. both vs. none of these) (short)	153 White students at the Michigan State University (cross‐sectional design)	Affirmative action (policy domain 1)	No main effect of framing on policy support. However, featuring that diversity is good for all students weakened the link between interracial attitudes (e.g., guilt and belief in White privilege) and support for affirmative action.	The link between guilt and policy support The link between belief in White privilege and policy support	–
Trawalter and colleagues, 2016 (study 4) 63	Diversity framing (“diversity is good” condition (which hire would be the most valuable for a small company given the goal of diversity) vs. “diversity is fair” condition (which hire would be most fair given the goal of diversity)) (short)	39 White business executives from an MBA program (cross‐sectional design)	Affirmative action (policy domain 1)	Respondents in the “diversity is good” condition thought about the scope of diversity more broadly and were less likely to hire the Black applicant compared with respondents who read the “diversity is fair” message.	–	–
Trawalter and colleagues, 2016 (study 5) 63	Diversity framing (design is similar to study 4 but adding the “diversity is important” condition, message without diversity framing condition, and message without mentioning diversity condition) (short)	209 White US adults (cross‐sectional design)	Affirmative action (policy domain 1)	Compared with the message that does not mention diversity at all, respondents in the “diversity is fair” condition and “diversity is important” condition were more likely to hire the Black applicant. The other conditions did not differ from the “no diversity mentioned” condition. People low in resources were less likely to prioritize the Black candidate, dampening the effects of “diversity is fair” framing.	–	Resources
Strategy framing
Cappella and Jamieson, 1996 64	Strategy framing (news articles focusing on the issue of health care reform itself vs. the groups affected by the issue vs. strategic maneuvering vs. the legislation processes vs. a combination of these) (long) Control condition (news articles on topics not relevant to health care reform)	Respondents were recruited in six media markets (cross‐sectional design)	Health care reform (policy domain 2)	All the experimental conditions increased cynicism toward the policy compared with the control condition.	–	–
Martin and colleagues, 2017 71	Strategy framing (news articles featuring support for ACA as political competition vs. portraying the positive effects of ACA on public health) (long) Exemplar valence (two exemplars with positive experience with ACA vs. one positive exemplar and one negative exemplar vs. two exemplars with negative experience with ACA) (short)	1,056 young Americans who were not insured (cross‐sectional design)	ACA (policy domain 2)	Nonsignificant effects of framing on attitudes toward ACA. Featuring two positive exemplars or two negative exemplars increased anger among Republicans, which in turn reduced support for the ACA.	Anger	Political party ID
Lee and colleagues, 2008 (study 2) 83	Strategy framing (news articles on the issue of immigration with either a strategy frame featuring parties’ positions on the issue as strategic maneuvering to win the midterm election vs. a value frame portraying parties’ positions on the issue as reflecting the important values held by different political parties) (long) Control condition (in the absence of either framing)	338 undergraduate students (cross‐sectional design)	Restrictive immigration policies (policy domain 3)	Neither the strategy frame nor the value frame had significant effects on policy support. However, the strategy frame weakened the association between partisanship and immigration policy support and increased the association between feelings toward minorities and policy support.	–	–

Abbreviations: ACA, Affordable Care Act; DREAM, Development Relief and Education for Alien Minors.

^a For each message strategy section of the Table, studies are then grouped together based on policy domains. If a study used more than one message strategy, it was only categorized based on the main strategy it used so that each study only appears once in this table to avoid repeated information.

Racial Cues

The 36 studies focused on this message strategy offer robust evidence, across a variety of policy domains, that among White audiences, even brief reference to racial cues (almost always mentions or portrayals of Black versus White people or populations) tends to undermine (average) levels of support for policies likely to advance racial equity. Most of these studies reported on exclusively or largely White samples of respondents, and many featured samples powered to detect moderate to small differences in message effects (19 studies feature samples with at least 500 respondents, with one including 3,733 respondents).

These patterns were most often observed in messages focused on policies related to both crime and poverty. For example, depicting a Black versus White perpetrator reduced perceived government treatment responsibility to crime in a regional study with a non–race‐specific sample. ⁵¹ Another study found that White respondents were more supportive of punitive crime policies when the story depicted a Black perpetrator. ⁷⁹ Featuring poor people who were Black (versus White) reduced perceived government responsibility to address poverty in a regional study with a non–race‐specific sample ¹⁰⁰ and reduced the feeling that too little was spent on poverty‐reducing programs in a sample of largely White students. ⁹⁸

Racist responses to racial cues among White respondents have also been regularly observed in other policy domains including childcare subsidies, ⁶⁵ affirmative action, ⁶¹ justice reinvestment funds for economic development and health care, ⁷⁶ and immigration policies. ⁹⁰ , ⁹³ , ¹⁰⁴ Apart from directly influencing policy support, racial cues also often strengthened the links between racial attitudes or stereotypical beliefs and policy support in some studies. ⁶⁵ , ⁸² Racist responses to messages were also observed among some audiences in the context of subtler racial cues. For example, specifying that the health care reform plan was proposed by President Obama (compared with President Clinton) made racial resentment and stereotyping stronger negative predictors of policy support in a nationally representative (though still majority White) sample ⁷⁴ and reduced support for the policy among people (again largely White) with racial prejudice. ⁶⁹

Although, on balance, studies have tended to find evidence of White racism in responses to racial cues about policy domains, several studies found no differences in policy support in response to racial group cues. ⁵¹ , ⁷⁷ , ⁸⁰ , ⁹⁶ Other studies confirmed racial bias in audience priors but do not find evidence that these predispositions were exacerbated by racial cues in policy messaging. ⁵⁵ , ⁷⁵ A single study offered evidence that brief racial cues featuring Black people or populations can influence emotional outcomes (sympathy) linked to support for public policies that compensate for racial discrimination, ⁵⁶ whereas another study (one of few designed to assess messaging effects of racial cues among Black respondents) found that depicting a Black patient increased support for government‐financed health insurance among non‐White respondents. ⁶⁷

Overall, however, the general pattern of results revealed clear evidence that racial cues in policy messaging that identify Black people or populations as either disproportionately affected by the social issue at hand or potential beneficiaries of policies to address those issues tend to undermine support for public policies across a wide variety of health and social issues.

Racial Stereotypes

Beyond brief racial cues, the five relevant studies included in this review also demonstrate that exposure to racially stereotypical portrayals of people from historically minoritized groups tends to undermine support for policies to address racial equity. For example, depictions of negative stereotypical images of Black people strengthened the association between racial attitudes and support for affirmative action among non‐Black respondents compared with a no–image control condition (in which policy support was not associated with racial attitudes) and a counterstereotypical image condition (in which policy support was less affected by racial attitudes). ⁵⁹ Other studies reported negative effects of racist stereotypes in the context of welfare policy, ⁸⁸ excessive policing, ⁸⁸ affirmative action, ⁶⁰ and crime. ¹⁰³

Nonracial Cues

A review of the 11 studies in this category makes clear that factors beyond depictions of an issue's causes, its potential solutions, its impact on various groups, and/or the use of racial cues or stereotypes, can shape support for policies with racial equity implications in meaningful ways. These studies largely have involved samples of adult respondents, often designed to be nationally representative, and have focused on policies known or likely to reduce racial inequity but featured message manipulations that do not focus on race or racial implications of the policies. Findings vary depending on the message manipulation studied. For example, messages describing the delivery system of welfare programs as “direct spending” rather than “tax expenditures” strengthened the relationship between symbolic racism and support for welfare in health care but not in the context of wage subsidies. ⁹⁴ Compared with the term “illegal immigrants,” the label “undocumented immigrants” increased support for a temporary work policy ¹⁰¹ and decreased support for punitive immigration policies. ⁸⁵ Mentioning the role of government in health care reform (compared with describing the issue in terms of health care needs) did not significantly change the overall rating of the pro‐ACA evidence, but it polarized respondents’ evaluation of the ACA along party lines, with Democrats rating the same pro‐ACA evidence as somewhat stronger and Republicans rating the evidence as weaker. ⁶⁸ The breadth of nonracial cues and policy contexts in this category precludes broader conclusions.

Social Comparison Framing

Most studies in this large category (19 studies) compared responses to short statements emphasizing racial differences that were embedded within longer messages. Overall, these studies reveal the breadth of ways that social comparisons can shape perceptions of racial inequity and policies to address it. On the one hand, simply comparing rates of social outcomes (crime or incarceration) or health outcomes (life expectancy or COVID‐19) by race appeared to undermine support for policies to address them, likely because respondents defaulted to behavioral (for crime) or biological or behavioral (for health) causal attributions for these disparities. ⁷² , ⁸¹ , ⁸⁷ , ⁹⁹ Also, comparison frames that conveyed information that could be perceived as threatening to White respondents (population shifts away from a White majority, threats to employment among White people) were associated with lower support for racial equity–enhancing policies. ⁵² , ⁵⁷ On the other hand, message strategies that described ways that existing policies disenfranchise, penalize, or harm Black people ⁸⁷ , ⁹¹ and emphasized ongoing commitment to addressing racial disparities ⁵³ were associated with increased levels of support for reform, particularly when descriptions avoided suggesting a zero‐sum threat to White populations. ⁵² , ⁵⁸

Causal Framing

Causal framing in these six studies took the form of short assertions about social or structural causes of an issue or health condition. ⁶⁷ , ⁷⁰ , ⁸⁴ , ⁹⁶ , ⁹⁷ Racism was not included within the causal explanations offered in these studies, although three articles (four studies) also included racial cues via visual or verbal depictions of White or Black adults. ⁶⁷ , ⁹⁶ , ⁹⁷ Overall, these studies indicate that attributing health and social outcomes to structural (but not racial) versus individual behavioral causes can increase support for policies that have potential to advance racial equity.

For example, one large study found that a short description of the social and genetic causes of diabetes reduced negative stereotyping and increased support for government spending on diabetes research. ⁹⁶ A message in the same study attributing diabetes to individual behaviors increased negative stereotypes, however. Adding a photo depicting a White or Black woman did not significantly influence policy support or stereotyping or interact with causal frames in the overall sample. Another nationally representative study found that linking heart disease to individual behavior (e.g., smoking) in a short personal story about diabetes increased individual blame and reduced support for government‐financed health insurance, whereas use of a racial cue (describing a person suffering from diabetes as Black or White) did not influence policy support. ⁶⁷ A third national study concluded that 1‐minute videos briefly highlighting external causes of homelessness for specific individuals (e.g., bank foreclosure) increased sympathy and increased support for government assistance to homeless people, but effects on government support were only observed among White respondents when White homeless individuals were shown. ⁹⁷ This study makes clear that White respondents may exhibit racist responses to Black versus White characters even when depictions describe the structural causes of social outcomes.

Responsibility Framing

Nine studies tested responsibility frames that involved more substantial message designs (usually several paragraphs in length). Thematic framing led to greater perceived government responsibility for solutions to poverty than episodic framing among two different samples of adults, ⁵¹ , ⁹⁵ although similar messages did not shift judgments about government spending on programs to help the poor in a student sample. ⁹⁸ Racial cues also produced different patterns of response in several samples: Featuring poor White (versus Black) people increased perceived government treatment responsibility for poverty in a regional study with a non–race‐specific sample ¹⁰⁰ and increased the feeling that too little was spent on programs to address poverty in a sample of largely White students. ⁹⁸ In another study, thematic framing reduced support for mandatory minimum sentencing (a policy likely to increase racial inequity) compared with a control condition. Although episodic framing increased empathy, it only reduced support for mandatory minimum sentencing when the message featured a White defendant.80 Thematic frames did not shape perceived government responsibility to address racial inequality or unemployment in small regional studies with non–race‐specific samples. ⁵¹

Overall, this pattern of findings suggests that many White respondents respond to depictions of several policy issues (including crime and poverty) in ways that reflect a pattern of racism—when effects were observed, depictions of broader societal responsibility for poverty and crime tended to shift largely White samples toward greater support for governmental intervention only when White characters were depicted. Although several of these studies observed no impact of thematic frames relative to control groups, many of these studies were composed of very limited and homogenous study populations (college students or people living in a single county). Only one study explored racial inequality as the social issue of interest, offering limited guidance on messaging for policies explicitly designed to advance racial equity.

Diversity Framing

A limited body of work in this category (four studies) suggests that there may be conditions under which messages emphasizing the benefits of diversity for various collectives could influence policy‐relevant attitudes, but the evidence base is too mixed and limited to draw broader conclusions.

Strategy Framing

The mixed set of findings in this small category of work (three studies) offers limited opportunities for broader generalization in the context of policies to support racial equity, though conceptually it seems unlikely that framing these policies in strategic terms is likely to have positive impacts on broad population support to address racial inequity.

Discussion

What Do We Know About Messaging to Promote Policies That Advance Racial Equity?

This scoping review identified a variety of clear patterns of findings that are relevant to strategic messaging to promote policies that advance racial equity. The first is the sobering reality that brief cues (whether verbal or visual) that signal the Black identity of victim(s) or perpetrator(s) in the context of a social problem, in the absence of broader contextual discussion of the social, structural, environmental, or economic causes of social issues, can undermine support for racial equity–promoting policies among White respondents. Inclusion of cues that signal racial stereotypes in the context of these portrayals are likely to exacerbate these effects.

Comparing rates of incarceration or health outcomes between Black and White people, in the absence of broader contextual information, can also undermine support for policies to address them. These racist responses have been documented across a variety of health and social policy domains, though they appeared most frequently in the context of crime, incarceration, poverty, and unemployment. Although not within the timeframe of this scoping review (which was limited to studies published until April 2021), recent evidence shows that these patterns of racist responses to racial comparisons have persisted in the context of devastating racial inequities in deaths from COVID‐19. ¹⁰⁶ , ¹⁰⁷ Several studies suggest that these patterns may be driven by the fact that White respondents default to behavioral (for crime, incarceration, poverty, and unemployment) and/or biological or behavioral (for health) causal attributions for these outcomes and disparities.

That said, there is also evidence that racist responses are not inevitable when discussing race and social policy. Several studies provide evidence that messages that (a) illuminate the structural causes of social problems; (b) describe the specific ways that existing policies disenfranchise, penalize, or harm Black people; or (c) emphasize ongoing commitment to addressing racial disparities can increase levels of support for social reform among both White and non‐White audiences (a recently published study outside the timeline of our scoping review offers a similar conclusion). ¹⁰⁸ These strategies appear most likely to shift support when they avoid reference to cues that may be seen as threatening by some White respondents (e.g., population shifts away from a White majority, threats to employment among White people) and when they emphasize present and future aspirations over historical progress in reducing racial inequity. Some evidence suggests that messages emphasizing racial justice have potential to galvanize the political left, although the same strategy may undermine support among the political right.

It is important to acknowledge some of the limitations of this scoping review. First, by design, this review only included studies published in peer‐reviewed literature. However, messaging studies of advocacy and narrative strategies are often conducted by nonprofit organizations and communication firms; these data, although quite useful, are not included here. For instance, one influential communication approach developed and tested outside academic peer‐reviewed research is the Race‐Class Narrative project. ¹⁰⁹ Furthermore, publication bias against studies showing null results may also have influenced the number of relevant studies identified in this review. Second, by necessity, the research only included studies published before we completed our searches (April 2021). However, research on communication about racism and racial disparities has increased in the past year, likely as a result of renewed attention to racial injustice spurred by the murder of George Floyd and attention to racial inequities in COVID‐19. The more recently published research does not feature drastically different message strategies than were covered by our review, however, and generally replicates patterns observed in the works described here. ¹⁰⁶ , ¹⁰⁸ , ¹¹⁰ Third, although our inclusion of racial equity–related policy domains was based on our team's scoping review of policy documents, others might identify other policy areas not encompassed in our schema portrayed in Table 1.

What Do We Need to Know About Effective Messaging for Racial Equity Policy?

Despite a seemingly large corpus of research identified in this review (80 studies from 55 papers), another central finding of this review is a surprising lack of breadth and diversity of messaging approaches and study designs in this area. Here, we identify several areas of need in this body of research to inform future strategic messaging to promote racial equity policy.

Need for Longitudinal Designs With Repeated Exposures

The vast majority of studies reviewed here featured very short messages (usually ranging from a few words to a few sentences) that were almost always shown to respondents a single time. None of these studies followed respondents over time, relying on immediate responses within minutes of exposure to a message (one very recently published study offers an exception to this pattern). ¹⁰⁸ Although short‐term reactions may predict subsequent cognitive, emotional, and behavioral outcomes, this is not necessarily so; media priming effects can wear off quickly in the absence of repeated exposure, ¹¹¹ and some messages can exert stronger effects over time. ¹¹² , ¹¹³ Furthermore, messaging surrounding public policies is almost always competitive and dynamic, meaning that policy proponents and opponents respond to one another with countermessaging strategies that shift over time. ¹¹⁴ At a minimum, these observations highlight a need for additional studies that gauge the durability of the patterns of effects observed in this body of work, the impact of repeated exposures to such messaging, and the impact of countermessaging strategies that seek to respond to and/or mitigate the impact of anti–racial equity policy messaging.

Need to Consider Both Short‐Term and Long‐Term Strategies

A related observation involves the need to consider both short‐term and long‐term strategies in efforts to promote racial equity policy. The exclusive focus on short‐term outcomes, combined with a persistent pattern of racist responses to racial cues embedded in policy advocacy messages, has led some scholars and advocates to recommend that advocates of policies with evidence or potential to reduce health inequity should avoid reference to race entirely. This strategy runs the long‐term risk, however, of failing to build collective recognition and momentum to address the deeply embedded structures of racism that sustain racial inequity in the United States. We acknowledge that the studies reviewed here suggest there may be circumstances in which framing policies in terms of factors beyond race may be strategically advantageous for social change—that avoiding racial cues may increase the likelihood of passing a policy that has racial equity–promoting effects under some conditions. But this approach is also short‐sighted and likely counterproductive in the long term because it avoids illuminating the ways that structural racism is embedded in all domains of US public policy and how structural changes will be necessary to eliminate it.

Several studies indicate that it is indeed possible to discuss issues of racism and racial inequity in ways that invite more nuanced understanding of these issues. For example, Nelsen ¹¹⁵ conducted a field experiment in high schools in the Chicago metropolitan area in which he assigned some students to read excerpts from a history textbook that took a critical pedagogy and others to read a traditional textbook. Students of color who were assigned to read the more critical text—which discussed marginalization and systemic injustice—were more empowered to engage in civic and political discourse, and White students gained a greater appreciation for the contributions of people of color in American society. ¹¹⁵ , ¹¹⁶ A larger body of research in developmental and social psychology also consistently finds positive effects of teaching about and discussing issues of race and racism with (particularly young) people. ¹¹⁷ , ¹¹⁸ , ¹¹⁹ However, efforts to formally teach students about these topics have consistently been met with backlash throughout history. ¹²⁰ Therefore, there is a critical need to better understand the conditions under which gains from discussing and teaching more nuanced understandings of racism are possible.

Need to Consider Longer‐Form Messages

As noted earlier, most studies included in this review examined responses to relatively short messages, ranging from brief mentions of a racial group to text passages resembling short news stories. There would be substantial value in exploring the effects of longer‐form messages, including stories that center perspectives from historically marginalized people and populations about ongoing sources of structural racism. One messaging strategy with evidence of success in promoting policy support involves storytelling about social policy. ¹²¹ , ¹²² An emerging body of research has found that short stories that describe the social and structural causes of social problems and center policy solutions can increase public support for policies that address social determinants of health such as predatory marketing of unhealthy products, ² soda and alcohol taxes, ¹²³ unhealthy food environments, ¹²⁴ and harm reduction policies for opioid use disorder. ³⁶ Storytelling approaches have also increased empathy, ¹²⁵ reduced stigma, ¹²⁶ , ¹²⁷ , ¹²⁸ and increased positive attitudes toward members of other social groups ¹²⁹ in a variety of social issue domains, although none of these studies have employed narrative strategies that interrogate structural racism in social policy. There remains a critical need to learn how to tell stories that center and explain the forms, prevalence, and effects of structural racism in contemporary policy debates, all while avoiding the potential for backlash or political polarization that has been observed in prior work. The current media environment also features many platforms that center short‐form messages (e.g., Tweets, TikTok videos, and YouTube or Instagram shorts). Creative approaches will be needed to develop long‐form messages that center structural, social, environmental, and economic causes of racial inequity on these platforms.

Need to Understand Responses to Racial/Ethnic Cues Related to Hispanic/Latino/a, Asian, Indigenous, and Muslim People and Populations, and to Study These Issues With an Intersectional Lens

Our scoping review revealed a near‐exclusive focus on comparing racial cues between Black and White people and populations. There are, of course, important reasons for a focus on Black Americans considering the historical legacy of slavery, oppression, and marginalization in the United States. But Black Americans are not the only group that have experienced marginalization; other non‐White groups also experience structural racism in how public policies shape their health and social well‐being. ¹³⁰ By limiting research to Black versus White, the field misses important nuances that could improve our theoretical understanding of how these processes operate and insights that could be practically applied to issues faced by other groups.

Beyond the dichotomization of race that has occurred over time, there is also a need to consider the intersections between race and other dimensions that shape people's experiences in life, and of public policies. ¹³¹ Without an intersectional analytic frame, we can miss important interactions that ought to be considered in theories and practice. For example, immigrants are frequently portrayed in a negative light in the US media. ¹³² People of color who are immigrants might face increased hostility. ⁹³ Also, although social science research on race in the workplace has largely focused on aggregate Black versus White differences, those experiences vary substantially when one also considers the gender and sexuality of the people in those racialized bodies. Black women have to contend with unique issues that Black men are not subjected to. ¹³³ Moreover, in some contexts, patterns of discrimination can even reverse when sexuality is added to the constellation of variables—because gay men are stereotyped to be effeminate and weak, researchers have found that Black gay men can be advantaged in the hiring process because they are perceived as more educated and less threatening and criminal than non‐gay Black men. ¹³⁴ , ¹³⁵ These examples highlight a few ways in which failures to take an intersectional approach can lead the field to miss important nuances. There is a clear need to broaden research to understand how diversity affects people's experiences and how more diverse portrayals shape responses to public policy debates.

Need to Consider Broader Models of Policy Change

Another striking finding of this review is the near‐exclusive focus in messaging studies on policy support (or its cognitive/ emotional antecedents) among White audiences, whether explicitly (through samples that are White only) or implicitly (through convenience samples or so‐called nationally representative population samples that are majority White). These research design choices imply a causal model of social change that seems to rest on an assumption that majority (White) public support is the primary pathway to policy passage. There would be substantial value in exploring other message strategies, outcomes, and audiences relevant to the policymaking process. This includes the study of messages that may catalyze mobilization and collective action among non‐White populations and social change advocacy organizations, as well as messages that shape policy support and legislative, legal, or administrative action among policymakers themselves.

Conclusion

The events of 2020 have catalyzed increased attention among public health researchers to structural racism as a foundational determinant of health. ¹³⁶ Nevertheless, public awareness of racial inequity remains static, ¹³⁷ , ¹³⁸ and public support for some safety net policies that would advance racial equity even declined during the COVID‐19 pandemic. ¹³⁹ The time has arrived to invest the effort and resources to develop, test, and disseminate communication interventions that advance racial equity through policy advocacy, community mobilization, and collective action. In doing so, we should explore strategies that explicitly address the many ways that public policies create, sustain, and reinforce racial inequity. We hope that this review offers a useful background and framework for this vital area of research and practice.

Funding/Support: This research was supported by the Robert Wood Johnson Foundation (grant nos. 77117 and 79754). The content of this article is solely the responsibility of the authors and does not necessarily represent the official views of the Robert Wood Johnson Foundation.

Conflict of Interest Disclosures: All authors completed the ICMJE Form for Disclosure of Potential Conflicts of Interest. No conflicts were reported by the authors.