The Politicization of COVID-19 and Anti-Asian Racism in the United States: An Experimental Approach

Abstract

The deadly outbreak of the COVID-19 pandemic has accompanied a worldwide surge in anti-Asian hate crimes and racial violence. In this paper, I experimentally assess the downstream effects of the health crisis on the racial attitudes of the American public. Survey respondents were randomly assigned to different messages about COVID-19 and its association with China and answered a battery of racial attitude questions, including a new measure of anti-Asian racial resentment. Across all outcome measures, I find null effects for both treatment messages, which suggests that racialized views toward Asians may be stable individual-level dispositions that have shaped American responses to the pandemic. Findings from this study have important implications for research on the far-reaching societal and political consequences of the pandemic in the United States and beyond.

Introduction

The deadly outbreak of COVID-19, known as the coronavius, has exerted tremendous social and economic costs in all affected countries around the world. With its one of the first reported cases in China, the virus has been closely associated with the Asian country, leading to a worldwide surge in racialized violence and discrimination against overseas Chinese and broader Asian communities (Jeung and Lee, 2021; Russell, 2020). A small but growing literature in American politics has empirically examined the rise of anti-Asian sentiment and public discourses during the pandemic (e.g., Kim and Kesari, 2021; Nam, Sawyer and Style, 2022; Reny and Barreto, 2022). Based on evidence from cross-sectional and panel surveys, these studies have uncovered noticeable associations between the degree of individual-level exposure to COVID-19 and the level of anti-Asian attitudes among the American public. As Reny and Barreto (2022) note, however, extant research has not fully addressed the underlying issue of causality, especially with regard to whether the coronavirus and its connection to China have significantly increased levels of anti-Asian feelings or entrenched racial attitudes formed prior to the pandemic have instead shaped individual responses to the crisis and policy preferences.

To help answer this question, this paper presents findings from an original survey experiment that directly assesses the presumed downstream effects of the health crisis on the racial attitudes of the American public, focusing on attitudes toward Chinese and Asian Americans and public policies concerned with the minority population. More specifically, this pre-registered survey experiment was designed to test (1) whether information about COVID-19 and its origin in China significantly leads to increased levels of anti-Asian sentiment and (2) whether such effects are more pronounced among Republicans and individuals with higher prior levels of general racial conservatism. To provide a robust assessment of anti-Asian racial animus, the experiment employs diverse measures of racial attitudes including the conventional favorability and stereotype scales and a novel measure of anti-Asian racial resentment. The data suggest that neither the information about the virus itself nor the message emphasizing its Chinese origin has noticeable effects on the public’s views toward Chinese and Asian Americans. This finding is in line with the view that racial attitudes toward Chinese and Asian Americans are more or less entrenched dispositional traits that work to shape, rather than be substantively shaped by, individual responses to the health crisis (Reny and Barreto, 2022). In the discussion section, I will also address potential alternative sources of null results including saturation effects and social desirability bias as well as limitations of the experimental design choice. Findings from this study will have important implications for research on the far-reaching socio-political consequences of the COVID-19 crisis in the United States and beyond.

COVID-19 and anti-Asian racism in the United States

Researchers have long documented the prejudicial linking of infectious diseases with specific identity groups such as the association of the severe acute respiratory syndrome (SARS) with ethnic Chinese and the acquired immunodeficiency syndrome (AIDS) with homosexual individuals, leaving irreparable marks of stigma on the target group (Ali, 2008; Keil and Ali, 2006; Fullilove and Fullilove III, 1999). Social psychologists demonstrate that such a process of stigmatization directly affects the victimized through discrimination, automatic stereotype activation as well as overt violence (Major and O’Brien, 2005; Quinn and Chaudoir, 2009). The association of COVID-19 with China has similarly accompanied the stigmatization of ethnic Chinese and broader Asian populations, as the number of reported anti-Asian hate crimes has soared around the world, especially in America and other western societies (Russell, 2020).

In the United States, critics have attributed rising anti-Chinese and anti-Asian American racial animus at least partly to the politicization of the pandemic including President Trump’s reference to COVID-19 as “Chinese virus” or “China virus” (Aratani, 2020; Tavernise and Oppel, 2020). Mass media and other political elites such as Secretary of State Mike Pompeo also insisted on using such labels as “Wuhan virus,” further strengthening its conflation with China and, by extension, ethnic Chinese people (Su et al., 2020). Past research demonstrates that racially divisive and inflammatory political messages have strong “emboldening effects” on the public’s prejudiced attitudes against marginalized groups in society (Lajevardi and Abrajano, 2019; Lajevardi and Oskooii, 2018; Newman et al., 2019; Siegel et al., 2019; Valentino, Wayne, and Oceno, 2018). American elites’ explicit – and strategic – use of such racially charged terms, in other words, appears to have stoked mass xenophobic and discriminatory reactions to ethnic Chinese people within the country.

More importantly, this politicization of the pandemic has adversely affected broader Asian American communities, giving rise to a historic surge in reported hate crimes against the minority population regardless of their diverse ethnic and national origin identities (Jeung and Lee, 2021).Scholars have explained such a rapid racialization of COVID-19 as rooted in deep-seated American stereotypes of Asians who have been lumped together as a monolithic racial group (Kimura, 2021; Li and Nicholson Jr, 2021). Furthermore, dating back to the late 19th-century Chinese Exclusion Act and the internment of Japanese Americans during WWII, Asian Americans have long been viewed as “perpetual foreigners” in American society, unassimilable to the mainstream and sometimes even more loyal to their countries of origin (Kim, 1999; Wu, 2002). Against the backdrop of such longstanding anti-Asian racial stereotypes, the pandemic has ushered in the revival of what critics have termed the ‘yellow peril’ syndrome in America today (Zhang, 2021).

A question remains as to whether the pandemic and its association with China have led to a net increase in anti-Chinese and anti-Asian sentiments among the American public or, alternatively, stable individual-level racial attitudes formed prior to the pandemic have shaped behavioral responses to the health crisis. To help adjudicate these two possibilities, I present a survey experiment designed to directly assess the presumed short-term effects of the message linking COVID-19 with China on the racial attitudes of the American public. In the survey, I include separate measures of attitudes toward Chinese and Asian Americans to examine if the observed effects are restricted to views toward ethnic Chinese population. The experiment thus tests the following first hypothesis:

Hypothesis 1 Reading about the message emphasizing the Chinese origin of COVID-19, the respondents will express higher levels of anti-Chinese American and anti-Asian American sentiments compared to the control group (H1).

By including an additional treatment arm on the identical COVID-19 message without any reference to China, I also examine the possibility that the widespread linking of the coronavirus with China would instantly lead to heightened racial animus upon hearing about the health crisis. In addition, by including a battery of racial attitude measures that concern other racial and ethnic minorities, ¹ I will estimate the distinct adverse consequences on Chinese and Asian Americans while probing any spillover effect to broader racial conservatism. ²

Next, I examine theoretically meaningful heterogeneous treatment effects based on the public’s partisan identity and prior beliefs about race. First, given the fact that President Trump publicly emphasized the association between COVID-19 and China, I expect that Republicans would be more receptive to the message and express higher levels of racial conservatism upon receiving it, compared to Democrats and Independents. Second, given the above discussion on how the American public tends to lump together diverse ethnic groups under the single racial category of “Asian” (Junn and Masuoka, 2008; Kim, 1999; Le Espiritu, 1992), I expect to find that respondents’ pre-existing essentialist view of race would significantly moderate the treatment effect on attitudes toward Asian Americans. Specifically, building on the literature on racial essentialism in social psychology (Haslam, Rothschild, and Ernst, 2002; Jayaratne et al., 2006), I hypothesize that those who believe in the fixed nature of racial categories will be more likely to express higher levels of anti-Asian American sentiment upon receiving the treatment messages. With the experimental data, I thus test the following second hypothesis:

Hypothesis 2 The treatment effect will be larger among Republicans and those with higher levels of beliefs on racial essentialism (H2).

Experimental design

To test my hypotheses, I recruited a national sample of 2,025 Americans through Lucid Theorem (Coppock and McClellan, 2019) during the early phase of the pandemic in May 2020. In line with the preregistered survey design, ³ the sample size was determined based on the power analysis taking into consideration that the anticipated effect size between the two treatment conditions would be relatively small to detect. According to the power analysis, as can be found in the appendix, about 600 subjects per experimental condition were judged to be sufficient to yield a large enough power for the experiment. The obtained sample was well balanced on key demographic covariates including age, gender, and race ⁴ and included 1,852 respondents who self-identified as white, black, or Latino/Hispanic Americans. The analyses below relied on data from this final set of respondents.

Figure 1 visualizes the design of the survey experiment. ⁵ First, after providing their consent to participate in the study, respondents were asked a battery of standard demographic questions. ⁶ A shortened version of the racial essentialism scale (No et al., 2008) was placed at the beginning of the demographic questionnaire so as to minimize any priming effects of answering the racial attitude measure. Next, after reading a common introductory prompt, participants were randomly assigned to one of the three experimental conditions. First, the control condition delivered a placebo message on pet product sales. Second, the ‘Coronavirus condition’ reminded the participants of the rapid ongoing spread of COVID-19 in the United States with information on the total number of confirmed cases as of the survey fielding date. The last ‘Chinese origin condition’ conveyed the same message as the previous condition with one additional note emphasizing that the virus originated from China (see Appendix 1.1 for full details).

Figure 1. Experimental Design.

One alternative experimental design would be to present the last treatment message that refers to COVID-19 as the “Chinese virus” or “China virus” so as to emphasize its association with China. For this study, I chose to employ a more general treatment message without any reference to such politicizing – and unavoidably partisan – rhetoric. First, focusing exclusively on the US-specific labels of the virus would make the study difficult to replicate and compare with results from other studies with samples outside the country. Follow-up studies will be able to employ a similar study design to probe the downstream effects of alleging the link between COVID-19 and China on the rise of anti-Asian sentiment beyond the American context. Alternatively, it is possible to use both the “Chinese virus” label and the Chinese origin message in the treatment. This, however, would compromise the tractability of treatment effects: it would be hard to assess to what extent the treatment effects derive from respondents’ partisan attitudes and views of President Trump. In short, the current design leverages the advantage of abstracting away from the specific political debates and contexts in America involving President Trump’s rhetoric on COVID-19 and utilizes a more general frame that implicates China as the origin of the virus.

To answer the outcome measures, respondents were then asked to express their level of (dis)approval of statements from the following measures of negative sentiment toward ethno-racial minorities in American society: Asian American Resentment (Kim, 2022), ⁷ Latino American Resentment (Sergio and Ocampo, 2020), and Symbolic Racism (Kinder and Sears, 1981; Sears, 1988). ⁸ To provide a more comprehensive list of racial attitude measures, I also administered the standard racial stereotype and favorability scales as well as racial policy questionnaire regularly employed in the literature and the American National Election Studies. As explained above, target racial groups included both Chinese Americans and Asian Americans for the stereotype and favorability scales. After answering the outcome measures, all respondents completed a factual manipulation check question. ⁹

Results

All key covariates were well balanced across experimental conditions (see Table 2 in the appendix). First, as shown in Fig. 2, the data reveals that the differences in mean responses to racial attitudes toward Chinese and Asian Americans between experimental conditions are minimal. None of the treatment effects as compared to the control group on all outcome measures was found to be statistically significant, both before and after controlling for potential moderators including race, partisanship, and political ideology (see Table 7 in the appendix for full regression results). In the appendix, I replicate this finding with one-tailed t-tests between all experimental conditions. Furthermore, these null results remain robust to subsetting the data to white, African, or Hispanic/Latino American respondents as well as to those who passed the manipulation check. ¹⁰

Figure 2. Treatment effects on attitudes toward Asian and Chinese Americans.

Note: This figure summarizes ordinary squares regression coefficients for treatment dummies with 95% confidence intervals based on results in Table 7 in the appendix. All outcome measures are rescaled to range from 0 to 1. AA: Asian American. CA: Chinese American.

While these multiple comparisons and subgroup analyses increase the chance of a Type I error, I find null effects for both treatment conditions on all outcome measures – also including racial attitudes and policy views toward other racial minorities (see Table 22 in the appendix). Note also that the experiment has enough statistical power (.9) to detect a small effect size (d = .2), making the chance of a Type II error sufficiently low. Additionally, despite the null treatment effects, I find that some of the moderators including party identification and political knowledge significantly explain the variation in key outcome measures (see e.g., Table 7 in the appendix). To summarize, I cannot reject the null hypothesis that exposure to information on COVID-19 itself or its association with China did not shift American racial attitudes, including views toward Chinese and Asian Americans as captured by various measures – anti-Asian racial resentment, racial stereotypes and favorability, and public policy preferences.

Next, Fig. 3 summarizes the results from tests on heterogeneous treatment effects based on respondents’ party affiliation and baseline levels of racial essentialism (H2). Here again, across all outcome measures, being Republican or high racial essentialist did not significantly moderate the respondents’ reaction to the experimental stimuli. In other words, I cannot reject the null hypothesis that those who identify with the Republican party and believe in the fixed nature of race and racial categories were not more likely to express anti-Chinese and anti-Asian feelings upon exposure to the treatment messages.

Figure 3. Interaction effects based on party identification and racial essentialism.

Note: This figure summarizes ordinary squares regression coefficients for main interaction terms with 95% confidence intervals based on results in Tables 8–9 in the appendix. All outcome measures are rescaled to range from 0 to 1. AA: Asian American. CA: Chinese American.

I additionally tested a set of preregistered auxiliary hypotheses, probing the potential moderating role of individual variations in perceived and actual vulnerability to COVID-19 (see Appendix 4.3 for the list of my main and auxiliary hypotheses). To be specific, I hypothesized that the treatment effects would be more pronounced among individuals who live in regions with higher COVID-19 infection rates, have confirmed cases in personal networks, and have lost their jobs during the pandemic. First, I accessed the publicly available data on the total number of county-level infection cases as of May 17, 2020, one day before the survey fielding date, and merged the dataset with my survey data for analysis. As shown in Table 10 in the appendix, I find no empirical support for the hypothesized moderating role of whether respondents reside in areas with higher infection rates. Tables 11 and 12 in the appendix also suggest that personal distance and economic vulnerability to the pandemic did not significantly moderate individual responses to the treatment messages. Finally, I tested the last auxiliary hypothesis on the potential treatment effect on strengthening public support for hardline foreign policies against China. As shown in Table 23 in the appendix, I find no such effect across all different types of foreign policy measures.

Discussion

The present study focused on how a message emphasizing the severity of the pandemic and its Chinese origin could, at least in the short term, increase negative sentiment against Chinese and Asian Americans. I find no evidence that priming the American public with information on COVID-19 and its association with China significantly boosted anti-Chinese/Asian American or broader anti-minority sentiments during the early phase of the pandemic. Republican supporters and individuals with higher levels of racial essentialism also did not react significantly differently to the treatment messages. For the rest of the paper, I discuss the implications of these findings, the potential limitations of the present study, and suggestions for future research.

First, the results appear to support the view that racial attitudes, including attitudes toward Chinese and Asian Americans, are more or less stable individual-level characteristics (Reny and Barreto, 2022; Tesler, 2015) and that the political controversy over the Chinese origin of COVID-19 did not simply instill or increase anti-Asian racial animus in the minds of the American public. As Reny and Barreto (2022) point out, extant research on anti-immigrant attitudes during the Trump presidency suggests that racially charged political rhetoric works to activate previously formed attitudes, emboldening certain individuals to publicly voice their prejudiced opinion (Newman et al., 2019) In our case, it is conceivable that the politicization of COVID-19 helped incite surging anti-Asian racial violence by allowing individuals to act upon their pre-existing resentment against the minority population – who have long been marginalized as outsiders in American society (Kim, 1999; Wu, 2002).

It is, however, still possible that the current study due to its potential limitations simply failed to detect the causal effect of the current health crisis on levels of anti-Asian sentiment across American society. One potential explanation for the null results is that knowledge about the Chinese origin of the virus was already widespread, and even saturated, among the public by the time of the survey experiment (May 2020). The short treatment messages in the experiment, for this reason, might have been too weak to substantially shift the respondents’ racial attitudes, while their initial exposure to the pandemic and the “Chinese virus” label had already shaped views toward the minority group. Nam, Sawyer, and Style (2022) find with observational survey data that negative sentiment toward Asian Americans significantly increased after the onset of the pandemic. In other words, it is difficult to rule out the possibility that anti-Asian attitudes spiked during the initial phase of the pandemic and the experimental stimuli employed in the present study were too weak to further shift the respondents’ racial attitudes.

It is also possible that respondents under-reported their racial attitudes by correctly inferring the intention of the study, undermining the treatment effects with social desirability bias. While I cannot completely rule out this scenario, it is important to note that many of the respondents’ individual-level characteristics, including party affiliation and political ideology, significantly predicted a substantial variation in reported anti-Asian views in the survey.

Despite its limitations, this study contributes to the emerging literature on the political implications of COVID-19 by probing one potential mechanism of surging anti-Asian discrimination and violence amidst the unprecedented turmoil of the global pandemic. By probing the linkage between the pandemic and rising racial tensions in the United States with an experimental approach, this paper also calls for more experimental studies in political science to assess the competing models of the causal relationship between the current – and future – health crisis and mass ethno-racial attitudes (see e.g., Adida, Dionne, and Platas 2020 for an experimental approach to the Eloba crisis and immigration attitudes). In addition, combined with the literature in political psychology on the long-term effects of pathogen prevalence on political attitudes (e.g., Tybur et al., 2016), follow-up studies should also probe the long-term behavioral and political consequences of the health crisis.