key: cord-0860840-gslz8xfa
authors: Ahmed, Saifuddin; Hsueh-Hua, Vivian Chen; Chib, Arul Indrasen
title: Xenophobia in the Time of a Pandemic: Social Media Use, Stereotypes, and Prejudice against Immigrants during the COVID-19 Crisis
date: 2021-04-30
journal: Int J Public Opin Res
DOI: 10.1093/ijpor/edab014
sha: 6d5776909b29acde7d562af688ec8b882a91e7ac
doc_id: 860840
cord_uid: gslz8xfa

This study examines the relationship between social media use, disease risk perception, social and political trust, and out-group stereotyping and prejudice during a social upheaval. Analyses of primary data collected during the COVID-19 outbreak in Singapore found that disease risk perception is positively related to stereotyping and prejudice against Chinese immigrants. Individuals who used social media for news were more likely to stereotype and express prejudice. However, those who engaged in frequent heterogenous discussions, and had more extensive social networks, were less likely to stereotype and express prejudice. Higher social and political trust was also associated with lower stereotyping and prejudice. Finally, moderation effects of network characteristics on the relationship between risk perception, social trust, and prejudice were observed.

higher tendency to encounter counter-attitudinal information compared to traditional media venues (Anspach, 2017) . Therefore, individuals face varying degrees of expoure to cross-cutting content that goes beyond their personal choices, and is further influenced by the news sources they subscribe to, the composition of their network, and algorithms for ranking content (Bakshy et al., 2015) . Combined with the fact that social context cues are weakened in social media communication, some users would be likely to challenge majoritarian views. Consequently, exposure to the news on social media, often with an endorsement from friends and family, has been found to reduce prejudice against selected minority groups but not for all out-groups (Davidson & Farquhar, 2020) .

On the other hand, social media has been found to be a breeding ground for hate-speech (Ben-David & Fernández, 2016 ) amongst a broader anti-immigrant sentiment (Benkler et al., 2017) . Soral and colleagues (2020) found that those who frequently use social media reported higher levels of islamophobia and acceptance of anti-Muslim hate speech as compared to those who primarily used traditional media. Due to a lack of directionality in expected findings, the following research question is proposed:

How is social media news use related to stereotypes and prejudice against Chinese immigrants?

There is an emerging consensus on the role of factors leading to the counter-attitudinal effects of social media, such as users' network size and network heterogeneity. A more extensive social network increases the likelihood of information opportunities and gives individuals a greater sense of social cohesion (Chan, 2016; Burke et al., 2011) , a previously demonstrated factor in reducing prejudice against out-groups (Herreros & Criado, 2009) . Similarly, engagement within a diverse network increases the opportunity to interact with groups dissimilar http://mc.manuscriptcentral.com/ijpor SOCIAL MEDIA USE AND PREJUDICE 5 to self. Consistent with contact theory, interaction with diverse groups predicts less prejudice towards groups different from oneself (Allport, 1954) . Exchanges within a heterogenous network also force individuals to reconsider their views (Smith et al., 1996; Keele & Wolak, 2008) and increase the likelihood of persuasion (Huckfeldt et al., 2004) . Levitan (2011) also found that when individuals are exposed to differing views in heterogeneous social networks, it destabilizes their prejudice toward out-groups. To summarize, the main findings of the literature suggest that having more extensive networks and engagement within heterogeneous networks is advantageous to limiting biases and promoting acceptance of inter-group differences. However, due to a lack of studies examining online network characteristics' relationship with antiimmigrant attitudes, the following research questions are proposed (instead of hypotheses):

How are network size (RQ2) and discussion network heterogeneity (RQ3) associated with stereotypes and prejudice against Chinese immigrants?

Group conflict theories argue that people become hostile toward an out-group because they perceive a threat from that out-group (Stephan & Stephan,1996; Verkuyten, 2009 ). This threat perception can happen at either the individual level or at the group level. Simply put, individuals can perceive a threat from an out-group to either negatively affect themselves or their in-group at large (McLaren, 2003) . Empirical evidence supports the view that individuals are not necessarily rational and are often worried about their own community (Funk, 2000) . Attitudes against out-groups is then rationalized in terms of a concern for their community, and is often observed after specific crises events (Ahmed & Matthes, 2017; Prati & Pietrantoni, 2016) .

In the context of global health crises, an increased risk-perception of diseases has reportedly influenced negative attitudes toward members of out-groups and foreign immigrants http://mc.manuscriptcentral.com/ijpor SOCIAL MEDIA USE AND PREJUDICE 6 (Faulkner et al., 2004; Navarrete & Fessler, 2006) . More recently, evidence from Italy suggested that the risk perception of Ebola was positively associated with prejudice against African immigrants (Prati & Pietrantoni, 2016) . The heightened discrimination against out-groups can happen due to an emotional response to the threat of the disease or a lack of knowledge of the disease (Piţigoi et al., 2018) . Thus, within the context of COVID-19 in Singapore, there is a possibility that those who had a high risk-perception of the disease would stereotype and be prejudiced against Chinese immigrants. Therefore, the following hypothesis is proposed:

Risk perception of COVID-19 will be positively associated with stereotypes and prejudice against Chinese immigrants (H1).

Beyond the direct effect of risk perception, we also explore if network characteristics can varyingly affect the influence of risk perception on anti-immigrant attitudes. When users engage in information consumption or conversations within extensive and more diverse discussion networks, there is a possibility that it can shape their risk perception. Heterogenous networks provide essential context for information interpretation (Bloom & Levitan, 2011) . Some scholars argue that individuals within diverse networks do not necessarily follow a normative attitude, but the information environment provides them with counter-attitudinal evidence that increases their uncertainty (Visser & Mirabile, 2004) . The perceived severity of COVID-19 has been found to increase cyberchondria on social media, a condition where anxiety regarding an individual's health is formed or aggravated (Laato et al., 2020) . As such, there is a possibility that within the explored setting, network characteristics could influence the relationship between risk perception and negative attitudes by raising the uncertainty about COVID-19 and altering the degree of perceived threat. Based on the above discussion, the following research question is proposed: How do network characteristics (size and heterogeneity) moderate the relationship between COVID-19 risk perception and stereotypes and prejudice against Chinese immigrants? (RQ4)

The Singapore government called for the citizens to build social trust during the outbreak to better combat the disease situation (U-Wen, 2020). Social trust, defined as 'trust in strangers and trust in people with whom we are not previously acquainted' (Herreros & Criado, 2009) , is often considered a critical factor for more integrated societies (Li et al., 2005) . Those with high levels of social trust extends it not just to in-groups but also to people from diverse backgrounds, including immigrants. Social trusters are better suited to ignore the systematically biased heuristics on racist and cultural stereotypes and trust strangers (Herreros & Criado, 2009 ). As such, numerous studies have found an association between higher social trust and positive attitudes toward out-group and immigrants (Berning & Ziller, 2017) and an acceptance of multiculturalism (Jung et al., 2017) . To summarize, higher levels of social trust are expected to be associated with more positive views and reduced prejudice against immigrants. Thus, we propose the following hypothesis: Social trust will be negatively associated with stereotypes and prejudice against Chinese immigrants (H2).

Political trust is generally defined as citizens' belief or confidence that the government will work to produce outcomes consistent with their expectations (Citrin, 1974; Hetherington, 1998) . High levels of political trust relate to higher compliance and cooperative attitudes of citizens and support of government policies (Abrajano & Alvarez, 2010; Marien & Werner, 2019) . Low political trust, on the other hand, can undermine the legitimacy of government decisions and the willingness to obey legislation (Lindström, 2008) . If people do not trust http://mc.manuscriptcentral.com/ijpor SOCIAL MEDIA USE AND PREJUDICE 8 institutions to protect their rights and interests, they may be less tolerant of immigrants (Halapuu et al., 2013) . People who have higher trust in political institutions are more tolerant of immigrants and do not overly worry about possible threats that immigrants might represent (Paas & Halapuu, 2012) .

Conversely, individuals who feel political alienation are more likely to blame immigrants for problems happening in society (Espenshade, 1996) . Thus, low political trust is also related to xenophobia and refusal to interact with immigrants (Canetti-Nisim et al., 2004; Hjerm, 2005) . In short, we expect political trust to influence anti-immigrant attitudes against Chinese immigrants.

The following hypothesis is proposed:

Political trust will be negatively associated with stereotypes and prejudice against Chinese immigrants (H3).

We further explore if the network characteristics of social media can separately affect the influence of trust levels on stereotypes and prejudice against Chinese immigrants. Individual's social and political trust can vary depending on the information they get from their social networks and the conversations they engage within their network. Previous research indicates that high levels of heterogeneity can lower generalized trust and increase the perceived threat from different out-groups and thereby increase prejudice (see Stolle et al., 2018) . Moreover, network size has also been found to be negatively associated with social trust, as more extensive networks do not foster close associations that enhance trust within its members (Shen & Gong, 2018) . While both heterogeneity and size are more commonly found to be negatively associated 

Negative stereotype was adapted from an existing validated measure (Vedder et al., 2016; Velasco González et al., 2008) . Respondents were asked to what extent (1 = strongly disagree to 5 = strongly agree) do they agree that Chinese immigrants in Singapore are a) violent, b) dishonest, c) intelligent, d) friendly, e) arrogant, f) kind. Positive items were reverse-coded. A scale of negative stereotype was created by averaging the six responses, M = 3.02: SD = .66, α =.80.

Feelings for Chinese immigrants were adapted from a frequently used existing measure (Abramowitz & Webster, 2018; Valentino et al., 2013; Verkuyten & Yildiz, 2009 ). Respondents were asked to rate their feelings towards Chinese immigrants in Singapore (0º very cold to 100º very warm). The item was reversed so that a higher value represents more unfavorable feelings, M = 52.23; SD = 24.27.

http://mc.manuscriptcentral.com/ijpor SOCIAL MEDIA USE AND PREJUDICE 10 COVID risk perception was measured by asking the respondents if they thought (1 = strongly disagree to 5 = strongly agree) that the residents in Singapore are at a major risk of the coronavirus infection, M = 3.61; SD = 1.04.

Social (M = 3.76; SD = 1.19) and political trust (M = 4.49; SD = 1.30) were single-item measures asking respondents how much they trust (1 = never to 7 = always) people in society and the government (Ceron, 2015; Gershtenson et al., 2006) . Social media news use was measured by asking the respondents how frequently (1 = never to 7 = several times a day) do they get news and information from social media sites, M = 5.52; SD = 1.62 (Ahmed, 2020; Lee et al., 2017) .

Network size was measured by asking respondents to estimate the approximate number of friends (1 = 0-50 people to 6 = more than 2000 people) they have on the social network that they use the most, M = 2.00, SD = 1.21.

Discussion network heterogeneity was assessed using a previously validated four-item scale (Diehl et al., 2016; Choi et al., 2017) . Respondents were asked how frequently (1 = never to 7 = several times a day) they talk to people on social media who are from a different a) race, b) social class, and c) whose views are different than theirs, and d) who disagree with them. A Social media news use was found to be positively associated with both negative stereotypes (β = .105, p < .001) and unfavorable feelings (β = .146, p < .001). Simply put, those respondents who frequently used social media as a news source reported higher levels of stereotypes and prejudice against Chinese immigrants (RQ1).

Recall RQ2 and RQ3 examined the relationship between network characteristics and antiimmigrant attitudes. The results suggested that both, having a larger online network (RQ2: stereotype: β = -.072, p < .001; feelings: β = -.146, p < .001) and a frequent engagement within a heterogeneous discussion network (RQ3: stereotype: β = -.087, p < .001; feelings: β = -.150, p < .001) are negatively associated with stereotypes and prejudice against Chinese immigrants.

The threat perception results suggest that COVID-19 risk perception was positively associated with both negative stereotype (β = .161, p < .001) and unfavorable feelings (β = .201, p < .001). Therefore, those who felt that Singaporean residents were at a greater risk of coronavirus infection reported higher levels of stereotypes and prejudice against Chinese immigrants than those who did not (H1 supported). Both social (stereotype: β = -.218, p < .001; feelings: β = -.187, p < .001) and political trust (stereotype: β = -.234, p < .001; feelings: β = -.191, p < .001) were found to be negatively associated with negative stereotype and unfavorable feelings. Consistent with previous research, it can be inferred that those who had higher levels of social and political trust were less likely to stereotype and be prejudiced against the Chinese immigrants (H2 and H3 supported).

Next, the analyses investigated the moderation effects of network characteristics on the relationship between COVID-19 risk perception, stereotypes and prejudice (RQ4). It was found that the interaction between disease risk perception and network heterogeneity (β = -.350, p < http://mc.manuscriptcentral.com/ijpor .01) was statistically significant in predicting unfavorable feelings. The relationship is plotted in Figure 1a . None of the other interactions involving COVID-19 risk perception were significant.

Further, the analyses investigating the moderation effects of network characteristics on the relationship between trust factors (social and political), stereotypes and prejudice found that the interactions between social trust and network size (β = -.268, p < .05) and social trust and network heterogeneity (β = .579, p < .001) were statistically significant in predicting unfavorable feelings. These interactions are plotted in Figures 1b and 1c . Similarly, the interaction between social trust and network heterogeneity (β = .366, p < .001) was also found to be statistically significant in predicting negative stereotypes. This interaction is plotted in Figure 1d . Based on the results presented in Table 1 and Figure 1 , it can be inferred that the moderation effects were as expected from the literature, except for the interaction between social trust and network heterogeneity on unfavorable feelings (see Figure 1c ).

[ Table 1 and Figure 1 here]

This is one of the first studies to provide empirical support for the conjecture that the COVID-19 outbreak has had negative social consequences. Overall, the findings largely correspond with the existing literature, suggesting that higher risk perception of the disease and using social media for news purposes are related to higher levels of stereotypes and prejudice against Chinese immigrants. On the other hand, we find that having a more extensive and diverse social network is associated with lower levels of stereotypes and prejudice. Social and political trust is also found to be negatively associated with the outcome variables.

The relationship between social media news consumption, stereotypes, and prejudice may reflect selective exposure, wherein individuals are consuming news sources that are biased http://mc.manuscriptcentral.com/ijpor SOCIAL MEDIA USE AND PREJUDICE 13 against immigrants. Recent research does confirm that social media news use is associated with more significant prejudice than traditional news media (Soral et al., 2020) . While these findings may dampen the optimism, the role of the social network characteristics suggests that existing stereotypes and prejudices can be lessened if users are exposed to information from multiple sources and engage in conversation with different groups. This underscores the importance of extensive information and diverse discussion environments in breaking away from echo chambers that can reinforce existing biases promoting extremism.

We also found the risk perception of COVID-19 to be positively associated with stereotypes and prejudice. The outbreak has possibly provoked xenophobic attitudes because some in-group members have blamed the Chinese for the disease. The in-group favoritism and out-group prejudice are in line with existing sociological theories explaining prejudice. 3 Further, this negative rhetoric is not limited to civic discourse, because global political leaders too have labeled the disease to be a 'Chinese virus,' 'Wuhan pneumonia,' 'China Wuhan pneumonia' (Everington, 2020; Orbey, 2020) . More efforts are required to control panic based on misinformation and raise public knowledge regarding the disease.

While these findings indicate that it is difficult to control how individuals engage in information seeking, we observe social and political trust to be negatively associated with stereotypes and prejudice. This is perhaps a reflection of policy interventions in the country. The Singapore government addressed the rise in anti-China sentiment by praising China's efforts in tackling the pandemic. They simultaneously addressed the infodemic by issuing online clarifications on misinformation regarding Covid-19 (Republic of Singapore, 2020), urging the public not to share fake messages and unfounded rumors.

Frequent heterogenous discussions are also associated with lower levels of stereotypes and prejudice across varying risk-perception groups. Among individuals with low social trust, having a more extensive social network and frequently engaging in conversations within heterogeneous networks are advantageous in reducing stereotypes and prejudices. We note a limitation, unexplained by current theorizing, in the significant interaction suggesting that among individuals with high social trust, engaging in conversations within heterogeneous networks is associated with greater unfavorable feelings against immigrants.

Also, the significant moderation effects are primarily observed for the less reactive measure of prejudice in unfavorable feelings and not for the explicit measure of negative stereotypes. It is plausible that online network characteristics are more likely to moderate the effects of trust and risk perception on general feelings toward immigrants than negative stereotypes, since feelings are cognitively momentary (Clore, 1992) but stereotypes are more deeply rooted (Bell, 1992) . Further research is required to monitor if the patterns are replicated in other contexts.

Next, it is essential to acknowledge the study limitations. First, the study utilizes crosssectional data, which limits causal inferences. Second, some of the independent variables were measured using single-item measures. Future studies should use multi-item scales for greater robustness of the relationships. Third, we have investigated a specific multicultural context. Future studies should explore this framework in other contexts to examine how the relationships play out.

Before we end, we would like to discuss some policy implications. The first implies that social media companies should enforce more significant content moderation on their platforms.

The COVID-19 health crises witnessed a flood of conspiracy theories against Chinese http://mc.manuscriptcentral.com/ijpor SOCIAL MEDIA USE AND PREJUDICE 15 communities being propagated on social media platforms (Sardarizadeh & Robinson, 2020) .

Since fake news is more likely to be seen as novel than real news (Vosoughi et al., 2018) it is necessary to apply firmer guidelines to restrict false claims against minority communities. The second concerns policy implications at the national level. The findings suggest that increased social and political trust are associated with lower stereotyping and prejudice against Chinese immigrants. Thus, it is in the interest of governments worldwide to implement societal interventions targeted to increase social and political trust within the citizenry to achieve greater harmony within communities.

To conclude, this study indicates that the effect of social media must be understood with a broader focus on citizen's news consumption behavior along with network characteristics. The results also emphasize that social and political trust and threat-perception are essential predictors of stereotypes and prejudice.

http://mc.manuscriptcentral.com/ijpor Footnotes 1 Previous studies have used a variety of ways to analyze anti-immigrant attitudes. Some have used feeling thermometers (Abramowitz & Webster, 2018; Valentino, Brader, & Jardina, 2013; Verkuyten & Yildiz, 2009) , while others utilized explicit prejudice measures focusing on negative evaluations about out-groups (Dhont, Roets, & Van Hiel, 2013; McLaren, 2003) . Some also focus on opinions regarding policies concerning immigrants (Berg, 2009; Scheepers, Gijsberts, & Coenders, 2002) . Hainmueller and Hopkins (2015) suggest that such variations in operationalization make it challenging to compare the findings across studies. Therefore, we adopt an approach to measure anti-immigrant attitudes via multiple measures -negative stereotypes and feeling thermometer. 

A one-way ANCOVA was conducted to determine a statistical difference between different ethnic groups on the negative stereotype about Chinese immigrants controlling for age, gender, education, income, social media use, social and political trust. There is a significant effect of ethnicity on negative stereotypes after controlling for the above-mentioned variables Table C1 , and the estimated marginal means are plotted in Figure C1 . Table D1 , and the estimated marginal means are plotted in Figure D1 .

Pairwise-Comparisons 

Assessing the causes and effects of political trust among US Latinos

Negative partisanship: Why Americans dislike parties but behave like rabid partisans

Concurrent media news use and gender-based political participation inequality in a low-income democracy

Media representation of Muslims and Islam from 2000 to 2015: A meta-analysis

The Nature of Prejudice

Contact hypothesis in ethnic relations

The new personal influence: How our Facebook friends influence the news we read

Exposure to ideologically diverse news and opinion on Facebook

Myths, stereotypes, and realities of Black women: A personal reflection

Study: Breitbart-led right-wing media ecosystem altered broader media agenda

Hate speech and covert discrimination on social media: Monitoring the Facebook pages of extreme-right political parties in Spain

We're closer than I thought: Social network heterogeneity, morality, and political persuasion

Core networks and whites' attitudes toward immigrants and immigration policy

Social trust and radical right-wing populist party preferences

How news content influences anti-immigration attitudes: Germany

Social capital on Facebook: Differentiating uses and users

Political involvement in "mobilized" society: The interactive relationships among mobile communication, network characteristics, and political participation

Social capital and social exclusion: Immigrants from the former Soviet Union in Israel

Internet, news, and political trust: The difference between social media and online media outlets

Social network sites and political engagement: Exploring the impact of Facebook connections and uses on political protest and participation

Business Insider Singapore. 4 in 10 people in Singapore are migrants -and almost 1 million are from Malaysia: UN

Influencing myself: Selfreinforcement through online political expression

Investigating effects of social media news sharing on the relationship between network heterogeneity and political participation

Rumour about maid's death due to coronavirus is fake news, says MOH. The Straits Times

The covid-19 social media infodemic

Cognitive phenomenology: Feelings and the construction of judgment

Trump's Discordant Display of Nativism in a Pandemic

Echo chamber or public sphere? Predicting political orientation and measuring political homophily in Twitter using big data

Statistical tests for moderator variables: Flaws in analyses recently proposed

Prejudice and Social Media: Attitudes Toward Illegal Immigrants, Refugees, and Transgender People

Republic of Singapore. Population and Population Structure

The intergenerational transmission of need for closure underlies the transmission of authoritarianism and anti-immigrant prejudice

Taiwan premier says COVID-19 should be called 'Wuhan pneumonia

Evolved disease-avoidance mechanisms and contemporary xenophobic attitudes

The dual influence of self-interest and societal interest in public opinion

Parties, institutional control, and trust in government

Social media, political expression, and political participation: Panel analysis of lagged and concurrent relationships

Is institutional trust related to proimmigrant attitudes? A pan-European evidence

Social trust, social capital and perceptions of immigration

The Political Relevance of Trust in Government

What the future may bring: Xenophobia among Swedish adolescents

Disagreement, ambivalence, and engagement: The political consequences of heterogeneous networks

6 in 10 people in Singapore have received fake COVID-19 news, likely on social media: Survey. Channel News Asia

Cognitive, attitudinal, and emotional effects of news frame and group cues, on processing news about immigration

Why Pandemics Activate Xenophobia

Interaction effects in multiple regression

The influence of personal values, social trust, and political trust on multicultural acceptance

Contextual sources of ambivalence

Statistical notes for clinical researchers: assessing normal distribution (2) using skewness and kurtosis

Principles and practice of structural equation modeling

What drives unverified information sharing and cyberchondria during the COVID-19 pandemic

The effects of news consumption via social media and news information overload on perceptions of journalistic norms and practices

Networking Egalitarianism: Social Network Influences on the Stability and Political Impact of Prejudice

Social capital and social trust in Britain

Social capital, political trust and purchase of illegal liquor: A populationbased study in southern Sweden

Government has debunked 40 instances of fake news on COVID-19: Iswaran. Channel News Asia

Fair treatment, fair play? The relationship between fair treatment perceptions, political trust and compliant and cooperative attitudes cross-nationally

The "black sheep effect": Extremity of judgments towards ingroup members as a function of group identification

Modem racism, ambivalence, and the modern racism scale

Prejudice, discrimination, and racism

Anti-immigrant prejudice in Europe: Contact, threat perception, and preferences for the exclusion of migrants

Examining the role of positive and negative intergroup contact and anti-immigrant prejudice in Brexit

Disease avoidance and ethnocentrism: The effects of disease vulnerability and disgust sensitivity on intergroup attitudes

Trump's "Chinese Virus" and what's at stake in the coronavirus's name. The New Yorker

Attitudes towards immigrants and the integration of ethnically diverse societies

Intergroup contact theory

Assessment of knowledge, attitudes and perceptions regarding Ebola

Knowledge, risk perceptions, and xenophobic attitudes: evidence from italy during the ebola outbreak

COVID-19 (Coronavirus Disease

Coronavirus: US and China trade conspiracy theories

Public opposition to civil rights for legal migrants as a response to perceived ethnic threat

The influence of news media on stereotypic attitudes toward immigrants in a political campaign

Wuhan virus: No ban on travellers from China at the moment, as Govt cautions against overreacting

Americans Are Wary of the Role Social Media Sites Play in Delivering the News. Pew Research Center

Minority and majority influence in freely interacting groups: Qualitative versus quantitative differences

Media of Contempt: Social Media Consumption Predicts Normative Acceptance of Anti-Muslim Hate Speech and

Predicting prejudice

When does diversity erode trust? Neighborhood diversity, interpersonal trust and the mediating effect of social interactions

Social capital and trust help government respond to virus crisis: PM Lee. The Business Times

Immigration opposition among US Whites: General ethnocentrism or media priming of attitudes about Latinos?

Explaining negative outgroup attitudes between native Dutch and Muslim youth in The Netherlands using the Integrated Threat

Prejudice towards Muslims in the Netherlands: Testing integrated threat theory

Support for multiculturalism and minority rights: The role of national identification and out-group threat

Muslim immigrants and religious group feelings: Selfidentification and attitudes among Sunni and Alevi Turkish-Dutch

Attitudes in the social context: the impact of social network composition on individual-level attitude strength

The spread of true and false news online

Adjustment for multiple comparisons