Pandemic threat and intergroup relations: how negative emotions associated with the threat of Covid-19 shape attitudes towards immigrants

ABSTRACT

Behavioural immune system theory predicts that attitudes towards immigrants become more hostile during times of increased threat from pathogens, as pandemic threat triggers aversive emotional responses, which in turn foster outgroup hostility. We test this notion in the context of the current Covid-19 crisis. Combining both original individual-level survey data of around 6,000 European respondents during the second Corona wave in winter 2020/2021 and regional data of pandemic threat in a multilevel design, we show that Covid-19 pandemic threat exposure in 105 European regions is indeed associated with more negative attitudes towards immigrants. Moreover, hierarchical path models indicate that Covid-19-induced anger fosters anti-immigrant attitudes, while we find no evidence that disgust as a pivotal avoidance-oriented emotion is crucial. In contrast to conventional wisdom, individuals reacting with fear to the pandemic hold more immigrant-friendly orientations. Taken together, our results indicate that the behavioural immune system (BIS) appears as a compelling obstacle to inclusive orientations. However, our findings challenge the notion that every emotion triggered by the BIS translates the pandemic threat into negative attitudes towards immigrants.

KEYWORDS:

• Behavioural immune system hypothesis
• affective intelligence theory
• attitudes towards immigrants
• Covid-19
• pandemic

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 This article was written as part of a research project on ‘The Politics of Public Health Threat’ that is financially supported by the Swiss National Science Foundation (SNSF, Grant No. 100017_204507) and the Berne University Research Foundation (25/2020). In this context, reference should also be made to the contributions by Filsinger and Freitag (2022) and Wamsler et al. (2022), who are elaborating theoretically and empirically similar designs in a coherent research program.

2 It must be noted that a few studies find no empirical support for this hypothesis (Adam-Troian and Bagci 2021; Cashdan and Steele 2013; Tybur et al. 2016; van Leeuwen and Petersen 2018).

3 Johns Hopkins University & Medicine, Coronavirus Research Center, URL: https://coronavirus.jhu.edu/ [20.1.2022].

4 The term ‘second Corona wave’ does refer to a very general European trend and is not intended to deny that European countries were affected by the pandemic to different degrees at different times. This overall trend is illustrated, for example, by the New York Times, which uses average deaths and hospitalizations in Europe to speak of a spring wave (peak: April 9, 2020) and an autumn wave (peak: November 28, 2020) (The New York Times, 04.12.2020, URL: https://www.nytimes.com/interactive/2020/12/04/world/europe/europe-covid-deaths.html [31.05.2021]).

5 Even though we should be careful about considering any pandemic as a direct trigger for the BIS (see Ackerman, Tybur, and Blackwell 2021), recent research suggests that the Covid-19 pandemic has had noticeable effects on BIS-related factors (Szymkow, Frankowska, and Gałasińska 2021).

6 Here, a reference can also be made to more classical explanations of outgroup hostility and prejudice such as intergroup threat theory or integrated threat theory. According to these theoretical approaches, (perceived) threats to the existence and well-being of the ingroup or its members as well as to the sociocultural identity of the ingroup are the main drivers of intergroup hostility, basically in line with the BIS hypothesis (Brewer 2010; Kachanoff et al. 2021).

7 Anger is located within an emotional cluster called aversion (Brader and Marcus 2013). The terms ‘aversion’ and ‘anger’ are sometimes used interchangeably in the relevant literature, i.e., naming the same appraisal dimension. Whether one understands the two as synonymous or sees anger as part of the broader concept of aversion, the underlying neural process is the same (Marcus et al. 2019, 121).

8 While the relevant literature generally agrees that fear leads people to search for more information about the threat, it has to be noted that the consequences of this increased information consumption on beliefs and attitudes are less clear and must not always lead to more unbiased and reasoned judgments. For example, Gadarian and Albertson (2014) show that feeling anxious about immigration leads people to seek out more threatening information about immigration and to consider this information as more persuasive.

9 NUTS (‘nomenclature of territorial units for statistics’) is the European standard for referencing the subdivisions of countries and is generally based on existing national administrative subdivisions. We refer to different stages of this hierarchical system for the different countries to ensure that our regional entities represent the highest (politically) meaningful subnational levels in each country (NUTS1 in Germany, France and the United Kingdom, NUTS2 in Italy and Spain, NUTS3 in Switzerland).

10 Since both variables are strongly skewed towards lower values, they were logarithmized for the subsequent analyses.

11 Since we cannot be entirely sure whether the objective indicators are in fact behind these emotions and attitudes or whether they are a result of certain omitted variables, we control for subjectively perceived health threat in additional analyses (see tables A5 and A6 in the online appendix). Our central findings remain unaffected.

12 One of the items referred to people from other countries but of the same race or ethnic group to most people in the host country, while the other asked about people from other countries who are of a different race or ethnic group to most people in the host country.

13 The overall index investigates anti-immigrant attitudes through two dimensions: one relating to policy preferences regarding the openness of the host country for immigrants and another linked to perceived threats related to immigrants. Confirmatory factor analysis revealed that the two-dimensional model has satisfactory fit with the empirical data (model vs. saturated χ2[4] = 105.17, p < 0.01; CFI = 0.994; TLI = 0.986; SRMR = 0.012; RMSEA = 0.064; CD = 0.988).

14 The TDDS is an 21-item instrument with three seven-item subscales to assess pathogen, sexual and moral disgust (Tybur, Lieberman, and Griskevicius 2009). As we are not interested in disgust related to sexual and moral concerns, we only use the pathogen disgust subscale.

15 Marcus et al. (2006) regard worry as a central dimension of fear.

16 Comparing the fit statistics, our results indicate that the two factor solution is superior (model fit one factor solution: model vs. saturated χ2[2]= 1041.18; CFI = 0.856; TLI = 0.568; SRMR = 0.075; RMSEA = 0.289; CD = 0.784 vs. model fit two factor solution: model vs. saturated χ2[1]= 48.391; CFI = 0.993; TLI = 0.961; SRMR = 0.012; RMSEA = 0.087; CD = 0.928). Using Cronbach's Alpha, we also tested the internal consistency of our two emotion scales (although Cronbach's Alpha is usually only recommended for a minimum of three items). Despite the small number of items, we get convincing degrees of internal consistency (anger = 0.73; fear = 0.75).

17 Data on the regions’ properties were obtained from the websites of the respective countries’ statistical offices (Germany: www.destatis.de; France: www.ined.fr and www.insee.fr; Italy: www.istat.it; Switzerland: www.bfs.admin.ch; Spain: www.ine.es; United Kingdom: www.ons.gov.uk).

18 (Multilevel) path models are understood as the simplest form of (multilevel) structural equation models.

19 The direct effect represents the coefficient for the relationship between the independent and dependent variable adjusted for the mediators in the model. The total effect is the sum of this direct effect and the indirect effects of the mediators in the model, i.e. the effects of the independent variable on the dependent variable running through the mediators (O’Rourke and MacKinnon 2015).

20 It should be noted that the inclusion of country dummies in our models leads to a reduction in the variance of our main variables, resulting in less substantive results. The shrinking variance may be an indication that national policies were striving for equality of conditions at the regional level even in federal states, but this was accomplished with varying degrees of success across countries. Moreover, including these fixed effects, the number of observations or units at level two (regions) is severely reduced (N= between 12 and 26) to perform a meaningful multilevel analysis which not only focuses on individual-level effects but also on level-2 predictors.

21 As indicated above, the arousal of anger is likely to be triggered not only by the infectious threat posed by Covid-19, but also by the pandemic's social and economic concomitants and consequences. Against this background, we conducted additional analyses on the drivers of anger, considering for example the subjectively perceived health, financial and social threat posed by Covid-19, the assessment of pandemic containment measures and unemployment status and regional unemployment rate to capture ego-tropic and socio-tropic risks linked to the pandemic (cf. Borbáth et al. 2021). Results from these additional analyses show that while financial and social concerns as well as a negative assessment of the enacted containment measures significantly relate to the experience of anger during the pandemic, our objective threat indicators and the subjectively experienced health threat further contribute to the explanation of this emotional reaction (see table A4 in the online appendix). We further re-estimated our models including these same factors as additional controls so as to also consider them in the formation of anti-immigrant attitudes. Controlling for these alternative explanations does not substantially change our main findings (see tables A5 and A6 in the online appendix).

22 We refrained from including disgust as an aversive emotional experience since disgust is part of the affective appraisal dimension of anger or aversion (Brader and Marcus 2013). Additional analyses not documented here indicate that disgust (as an aversive emotional state) is triggered by regional-level pandemic threat and, like anger, fosters anti-immigrant attitudes. The corresponding results are available upon request.