Blaming minorities during public health crises: post-COVID-19 substantive and methodological reflections from the UK

ABSTRACT

Using an original survey fielded during the COVID-19 pandemic, this paper contributes to understanding the phenomenon of blaming minorities during health crises and public perceptions of minorities more generally. We pose direct and indirect (split-sample) survey questions that gauge explicit blame of minorities, and potential implicit blame of particular groups and intergroup bias. Findings reveal that significant numbers tend to explicitly blame minorities for the spread of COVID-19; when asked about behaviors of the UK’s two largest religious minority groups – Muslims and Hindus – clear majorities blame these groups, with smaller percentages appearing to blame the country’s dominant ingroup. We test hypotheses drawn from theories of perceived threat, locus of control and authoritarianism: blaming minorities is expected to be associated with COVID-19-related (disease) threat, generally low sense of personal control, concern about the country’s lack of control over COVID-19, and general need for social conformity.

KEYWORDS:

• COVID-19
• blame
• prejudice
• minorities
• opinion
• threat

Introduction

COVID-19 prompted a centuries-old phenomenon of blaming minorities for the spread of deadly disease (Cohn 2018).¹ It also revealed how little was understood about how widespread this phenomenon was among mass publics. Understanding the extent of this phenomenon is important because blaming minorities during public crises such as the COVID-19 pandemic can reveal underlying animosities that may not necessarily be detected outside of times of crisis, as well as the potential for intergroup conflict and a general cultural environment that may sustain discrimination and hate crime against minorities both during and outside of a health crisis (Dancygier and Green 2010; Gray and Hansen 2021; Ng 2020).

Pandemics and other crises do not strike in a vacuum but instead arrive amid pre-existing social and political conflicts; public reactions to these crises are likely to reflect these conflicts. The COVID-19 pandemic in particular arrived at a time of high levels of public division over ethnic and religious diversity in European (and other) democracies (Hainmueller and Hopkins 2014). Thus, a few prominent instances of blaming minorities over social (and other) media (Awan 2020; Clarke 2020) may not be overly surprising or unexpected. The extent of blame of minorities among wider publics is still unclear, however: Do sentiments expressed in a small number of prominent tweets or in statements by a handful of politicians reflect the general public’s views, for instance? In addition to the limited understanding of the extent of this phenomenon, there is also limited systematic analysis of the likely causes of blaming minorities during health crises.

Using the case of the UK, where pre-existing conflict over diversity was very high by the start of the pandemic (Sobolewska and Ford 2020), we investigate: (1) the extent to which there was large-scale blaming of minorities for the spread of COVID-19; (2) the impact of asking about specific groups in a split-sample format, which may suggest potentially higher levels of blame of particular groups and intergroup bias; and (3) the likely causes (predictors) of blame. Our analysis of the potential predictors of blame of minorities draws on theories of outgroup hostility, particularly perceived threat, locus of control, and authoritarianism. Application of perceived threat theories to the COVID-19 context suggests that extreme fear of contracting COVID-19 and fears over the economic consequences of COVID-19 would increase the likelihood of targeting minorities as sources of blame for the spread of this deadly disease. Locus of control theories lead us to the expectation that those who generally feel less in control of their lives and those who felt the country did not have much control over the COVID-19 situation would be likely to blame minorities for the spread of COVID-19. Finally, authoritarians are already known to be hostile to minorities and groups they perceive as failing to conform; we expect authoritarians to be likely to perceive that minorities in particular were failing to conform by failing to comply with British norms and values generally (e.g. by living in large households) and failing to conform to lockdown rules (e.g. by continuing to participate in religious festivals and services), and thus be of particular blame for the spread of COVID-19. Our models include demographic control variables and control for general perceptions of ethnic and religious minorities in the UK.

In the remainder of the paper, we discuss the COVID-19 context in the UK and existing research on public responses to health crises and to the COVID-19 pandemic, before turning to the main theoretical perspectives that inform our hypotheses. We then discuss the public opinion survey conducted for our analyses including the measurement of key variables and how hypotheses will be tested, before then discussing the levels of blame of minorities suggested by our public opinion survey results and the extent of support for our hypotheses. The findings and their implications are discussed in the final section of the paper.

In brief, our findings suggest that a relatively large minority of the British public explicitly blamed minorities in general or expressed uncertainty regarding whether minorities were to blame, with only approximately 40 percent rejecting this idea. Higher levels of blame are revealed with more indirect questions about behaviors of people from the two largest religious minorities in the UK (Muslims and Hindus) and there is evidence of intergroup bias when we compare responses to questions about blaming people from these groups with blaming the behavior of the dominant ingroup in the UK. Findings from our theory-driven hypothesis tests suggest that authoritarianism and generally holding negative views of minorities (a control variable) are both likely to be relevant to understanding overt expressions of blame of minorities. Responses to our indirect (split-sample) questions suggest potentially wider blame and hostility that is not necessarily captured by our direct measure of blaming minorities and is less explicable using established theories. Separate analyses of minority and non-minority respondents suggest that among non-minority respondents, health and economic threats were both significantly associated with blaming Muslims, while those with a strong sense of individual locus of control were slightly less likely to blame Hindus, lending some support for these theories among non-minority respondents.

Blaming minorities in the context of health crises

Though health crises are not always associated with blaming minorities and other vulnerable groups, often they are (Cohn 2018). It has been argued that when the disease or virus is deadly and spreading rapidly, blaming outgroups can be an unintended mechanism to cope with “fear of the unknown, loss of control, and related social, political, and economic consequences” (Dionne and Turkmen 2020: E215). Outgroup behaviors and cultural norms, often perceived to be inferior to those of the dominant ingroup, can become the object of blame for the origins and/or spread of disease and viruses (Briggs and Mantini-Briggs 2003; Nelkin and Gilman 1988). Recent examples include Africa and African culture being linked to the outbreak of Ebola, “unhygienic” Chinese cultural habits being linked to severe acute respiratory syndrome, or SARs (Gilles et al. 2013; Kapiriri and Ross 2020), and, of course, more recently, Chinese cultural habits being linked to the start of the COVID-19 pandemic. In the UK, the Chinese-origin population is relatively small compared to other ethnic minorities (<1%; see GOV.UK 2020) and attention there turned to some of the country’s larger minority groups (Awan 2020; Clarke 2020).

The UK experienced among the highest number of COVID-19 deaths in the world at various points during the pandemic (Cuthbertson 2021). In the years leading up to this health crisis, research on immigration to the UK and other European countries had already documented the discomfort among domestic populations with immigration and the ethnic and religious diversity it has produced (Hainmueller and Hopkins 2014; Sides and Citrin 2007; Sniderman, Hagendoorn, and Prior 2004). In the UK in particular, the public response to immigrant-origin diversity has been mixed (McLaren and Johnson 2007). While evidence suggests gradual movement towards greater tolerance in the UK (McLaren, Neundorf, and Paterson 2021), it also suggests continued discomfort with some groups (Sobolewska and Ford 2020), particularly Muslims (Storm, Sobolewska, and Ford 2017). It is, therefore, perhaps unsurprising that there were prominent instances of blaming ethnic and religious minorities for the spread of COVID-19 in the UK over social media, with accusations that individuals from certain minority groups were ignoring social distancing rules, continuing to socialize with extended family, and still being allowed to attend religious services and festivals when everyone else was in strict lockdown (Awan 2020; Clarke 2020). We currently have little evidence of the extent to which these prominent instances of blaming minorities over social media reflected wider public perceptions of minorities’ behavior during the pandemic, however.

There is also limited comparative research on blaming minorities for the spread diseases which might provide some clues as to the extent of this phenomenon among mass publics. Research on “othering” in previous health crises such as Ebola and Zika outbreaks suggests that there might not have been extensive blaming or othering of minorities in these contexts (Adida, Dionne, and Platas 2020; Kam 2019), but it seems unlikely that the sense of threat during these outbreaks would have been as severe as that presented by COVID-19. Indeed, Kam’s (2019) survey (fielded in 2016) shows that the majority of Americans were not very concerned about Ebola or Zika. In contrast, YouGov polls indicated that 60 percent or more in both the U.S. and UK were scared that they would contract COVID-19 at some point, with the majority of people being fearful of this for much of the April 2020–March 2021 period (YouGov 2022). That is, the very real, large-scale threat of COVID-19 was more likely to have triggered underlying anti-minority hostility and association of this threat with minority behaviors than previous pandemics, prompting wider-spread public blaming of minorities in countries like the UK.

Systematic comparative research on blaming minorities specifically in the COVID-19 context is still somewhat limited, though growing. Some research on who or what is to blame for COVID-19 in New Zealand suggests that nearly 40 percent of respondents identified Chinese/Asians as being to blame in an open-ended question (Nguyen et al. 2021), while in Germany just under 20 percent explicitly mentioned people from various ethno-racial minority backgrounds in an open-ended question (Bogatzki, Glaese, and Stier 2023). In the U.S., Reny and Barreto (2022) suggest that anti-Asian attitudes were activated during the COVID-19 crisis, resulting in increased xenophobic behaviors such as avoiding Chinese restaurants and xenophobic policy preferences including supporting a reduction in the number of visas for high-skilled immigrants from Asia and requiring a mandatory quarantine of all travelers from Asia. Research also suggests that evaluations of migrants from China in the U.S. and UK became relatively negative during the pandemic (Croucher, Nguyen, and Rahmani 2020; Dhanani and Franz 2021; Fernández-Reino 2021; Griffin, Sides, and Tesler 2020; Lambe et al. 2021; Perry, Whitehead, and Grubbs 2021; Reny and Barreto 2022). Our analysis contributes to this area of research on blaming minorities during health crises – and specifically the COVID-19 pandemic – by providing some understanding of the extent of and reasons for public blaming of minorities in this context and the impact of survey questions themselves on answers to these questions.

Research on public perceptions of immigrants and “immigrant-origin minorities” suggests that the public express more hostility and discrimination towards some migrants and minorities compared to others, e.g. those perceived to be more ethnically or racially distinct (Hainmueller and Hangartner 2013; Sobolewska, Galandini, and Lessard-Phillips 2017), and so survey questions that ask about specific groups may reveal different levels of hostility than those that ask about minorities in general. Moreover, blame of minorities is likely to be revealed in the form of intergroup bias, or a tendency to evaluate one’s own group (in-group) more favorably than those of other (out-)groups (Hewstone, Rubin, and Willis 2002). The existence of this type of bias has been illustrated by Minimal Group Paradigm (experimental) research showing that even when people are arbitrarily assigned to different groups, individuals tend to favor their own group and show bias against members of other groups (e.g. Tajfel 1970; Turner 1982). If such bias occurs even outside of specifically threatening contexts (i.e. in lab experiments), we would expect it to occur in threatening contexts like the COVID-19 pandemic and especially against existing outgroups like religious minorities:

H1: Levels of blame of minorities for the spread of COVID-19 will be higher than levels of blame of the dominant ingroup.

In terms of theoretical explanations for blaming minorities, we focus on theories of perceived threat, locus of control and authoritarianism as likely explanations for variation in the extent of blaming minorities. Theories of intergroup conflict suggest that hostility to outgroups is likely to be a result of perceived threat posed by these groups (Blumer 1958; Quillian 1995): feelings of threat occur “when members of one group perceive that another group is in a position to cause them harm” (Stephan, Ybarra, and Morrison 2017, 43). In the case of the COVID-19 pandemic, the potential harm to health was considerable especially prior to the rollout of vaccinations. Perceived disease threat in particular is connected to derogation of potential outgroups like immigrants and minorities, who are likely to be blamed for spreading the disease (Green et al. 2010). In the UK allegations were made by politicians and on social media about minorities not cooperating with lockdown rules (Awan 2020; Clarke 2020), against an environment that was already hostile to some groups (particularly Muslims; see Storm, Sobolewska, and Ford 2017); this (alleged) failure to cooperate could be perceived as contributing to the spread of COVID, thus increasing the threat of catching it and becoming seriously ill from it (or worse). We thus expect that:

H2: Individuals who have had COVID-19 or express higher levels of concern about catching it are more likely to blame minorities for the spread of COVID-19.

Other threats that may be associated with pandemics – particularly economic threats – may also be important in explaining blaming of outgroups. Economic threat in the context of intergroup relations is normally argued to be caused by perceived competition for resources with an outgroup. As noted by Quillian (1995) in the context of anti-immigration sentiment, economic circumstances and prejudice may also be linked due to scapegoating, i.e. blaming subordinate groups for economic hardship (see also Campbell, Wong, and Citrin 2006). Perceived failure of minorities to cooperate with lockdown regulations was likely to be perceived as contributing to an increase in COVID-19 cases (Awan 2020; Clarke 2020), resulting in continued lockdowns, with potentially severe negative economic consequences. We thus expect that economic threats caused by the pandemic may prompt the blaming of outgroups like minorities, particularly for those in precarious personal economic circumstances:

H3: Individuals expressing higher levels of concern about their personal economic circumstances are more likely to blame minorities for the spread of COVID-19.

Economic threats posed by COVID-19 may also take the form of more collective, sociotropic concerns (Sides and Citrin 2007) – concerns about the country’s economy more generally. These concerns may also prompt scapegoating of minorities for this situation:

H4: Individuals expressing higher levels of concern about the impact of COVID-19 on the economy are more likely to blame minorities for the spread of COVID-19.

Feelings of control, or locus of control, have been shown to be important in explaining attitudes to immigration (Harrell, Soroka, and Iyengar 2017) and are also likely to be relevant in explaining blaming of minorities during the COVID-19 pandemic. Locus of control in general refers to perceptions of causal attributions as either internal or as a result of external factors (Lefcourt 1976), i.e. perceiving “one’s personal situation as either under one’s own control (internal) or beyond one’s control (external)” (Harrell, Soroka, and Iyengar 2017, 247). People who believe themselves to be in control are argued to feel less threatened and generally more open to those who are different from themselves (e.g. immigrants). This, in turn, may mean that such individuals are less likely to be threatened by other outgroups such as minorities during a pandemic and thus less likely to blame minorities:

H5: Individuals expressing a relatively high level of internal locus of control are less likely to blame minorities for the spread of COVID-19.

The sense of lack of control over the COVID-19 pandemic due to external sources is also likely to impact blaming of minorities for the spread of COVID-19. In their study of anti-immigration sentiment, for instance, Harrell, Soroka, and Iyengar (2017) illustrate the importance of sense of societal control in the context of immigration, showing that a sense of loss of societal control over immigration is relevant to understanding anti-immigration sentiment. Similarly, a lack of sense of societal control over the COVID situation may prompt blaming minorities for the spread of COVID-19:

H6: Individuals expressing higher levels of concern that the UK lacks control over the spread of COVID-19 are more likely to blame minorities for the spread of COVID-19.

Finally, people who hold authoritarian values are known to be more likely to denigrate groups they perceive as failing to conform, including ethnic and religious minority groups (Altemeyer 1981; Cohrs and Stelzl 2010; Feldman and Stenner 1997). Authoritarians are, therefore, already likely to have been alert to (perceived) failures by minorities to comply with lockdown restrictions, given their (perceived) more general failure to comply with British norms and values (McLaren and Johnson 2007). This suggests that:

H7: Holding stronger authoritarian values is associated with higher levels of blaming of minorities for the spread of COVID-19.

Data and methods

We investigate our hypotheses using a public opinion poll commissioned by the authors, fielded between 8 and 10 February 2022 by Deltapoll among a sample of 758 respondents aged 18 or over based in the UK. Like other Internet survey companies (e.g. YouGov), Deltapoll maintains a large panel of subscribers from which samples are drawn for projects like ours, with respondents paid small incentives to complete the survey. Deltapoll uses quota (nonprobability) sampling and reweighting to produce samples from their pool of respondents that are nationally representative in terms of age, gender, educational attainment, social grade, region, and vote in the most recent (2019) general election (for other scholarly research that uses Deltapoll, surveys, see Johns and Kölln 2020; Karyotis et al. 2021; Neundorf and Pardos-Prado 2022; and Simonsson, Fisher, and Martin 2021).

Measures of all variables are provided in Appendix A. We present findings in the next section primarily using univariate bar charts and graphs showing multivariate OLS coefficients and confidence intervals. Our dependent variable – blaming of minorities for the spread of COVID-19 – is measured with two survey items. These items were embedded with other items asking more generally about whether various groups were particularly to blame for the spread of COVID-19 (e.g. young people, older people, those not wearing masks, not getting vaccinated, etc.; see Appendix A). The items measuring our dependent variables include a question which asks about blame of ethnic and religious minority groups generally. Research on prejudice has long pointed to the difficulty of capturing the full extent of prejudice in the modern day in public opinion surveys due to social desirability issues (Quillian 2006) and a strong feeling of need to control prejudice in the European context (Blinder, Ford, and Ivarsflaten 2013). Therefore, in addition to the direct (and general) question about minorities, we included indirect questions designed to ask about blaming of religious minorities, using a split-sample format that focused on the main religious holidays of the two largest religious minority groups and the main holiday of the dominant group in the UK – Christmas – all of which occurred at various points during the height of the pandemic in 2020. Approximately one-third of respondents were asked about the extent to which people visiting family for Ramadan/Eid were particularly to blame for the spread of COVID-19, another one-third were asked the same question but for Diwali, and the final one-third were asked about people visiting family for Christmas.² Ramadan/Eid and Diwali were chosen as the minority religions for the split sample item due to Muslim and Hindu faiths being the largest minority religions in the UK (ONS 2019). Christmas was chosen as the main religious holiday of Britain’s dominant group – even for those who are not religious – with 86 percent of the British public claiming to celebrate Christmas in a YouGov poll conducted in December 2020 (YouGov 2020). H1 (on intergroup bias) suggests that there will be more expressions of blame of Muslims and Hindus than blame of those visiting family for Christmas. In addition, if the theories discussed above (perceived threat, locus of control, and authoritarianism) provide key explanations of blaming of minorities in particular during the pandemic, their effects should be stronger for those who received the Ramadan/Eid or Diwali prompt than for those who were asked about people visiting family during Christmas.

In order to isolate levels of blame of minorities from a general tendency to blame, for our multivariate models we center responses to the former items around respondents’ general tendency to blame. The latter is measured using responses to the other potential sources of blame that appear in this question sequence (see Appendix A). In addition to testing the theoretical hypotheses specified in H2–H7, our multivariate models control for: age, gender (self-identified; binary), level of education, whether oneself and both parents were born in UK or not (binary), non-white (self-identified; binary variable), Christian (binary), having friends who identify with a different ethnicity or religion than the respondent (binary), and whether the respondent reported residing in the East Midlands (see footnote 2). In the models of the direct question about blame of minorities, we also control for whether the respondent received the Ramadan/Eid or Diwali prompt in the earlier question about whether visiting family was a main cause of the spread of COVID-19 (with the Christmas prompt as the comparison category).³ In addition, the above-mentioned pre-existing conflict over diversity in the UK leads us to expect that general perceived threat posed by minorities is likely to be relevant to expressions of blame of minorities for the spread of COVID-19; we therefore present separate models that control for general perceived threat posed by minorities (see Appendix A for measurement). It is important to note, however, that measures of general perceived threat posed by minorities were part of the same cross-sectional survey as our outcome variables and themselves may have been impacted by negative perceptions of minority behavior during the pandemic, i.e. they may not be completely independent of our measures of blame of minorities for the spread of COVID-19 and so we interpret the models that include the control for general perceived threat posed by minorities with some caution. As shown below, this variable displays different effects on blaming of minorities during the pandemic when using direct measures of the latter compared to our split sample (Ramadan/Eid, Diwali, and Christmas) items and thus it is important in helping to illustrate the possibility that different measures of anti-minority hostility may capture differing aspects of this phenomenon and suggest different potential causes of it.

Findings: levels of blame

Our measures of blame of various groups used a 0–10 (not at all to blame – very much to blame) scale. We first summarize the results for our dependent variables by reducing this scale, combining responses of 0–4 to form a “Not to blame” category, 6–10 combined to form a “To blame” category, with 5 – which was labelled “Neutral” in the questionnaire – as the Neutral point. As shown in Figure 1, when respondents were asked directly about the extent to which ethnic or religious minorities were to blame for the spread of COVID-19 in the UK, less than 30 percent leaned towards saying they were, with approximately 40 percent leaning towards believing minorities were not to blame. Thirty percent leaning towards blaming minorities suggests that the blaming of minorities was not likely to be limited to only a handful of individuals who were posting negative sentiments about minorities on social media during the pandemic. Moreover, given that the question asked which groups are particularly to blame, the 30 percent who gave a neutral response also suggests lack of disagreement with the notion that minorities were particularly to blame. This is consistent with the work of Caller and Gorodzeisky (2022) indicating relatively high levels of explicitly racist views in the UK. On the other hand, 40 percent leaning towards believing minorities were not to blame suggests the possibility that large segments of the British public reject the notion that minorities were specifically to blame – or were unwilling to state that this was their perception.

Figure 1. Blame of minorities for spread of COVID-19.

Note: See Appendix A for question wording.

Turning now to our split sample question responses (Figure 1), these suggest that slightly less than 50 percent of respondents leaned towards blaming those who visited family for Christmas for spreading COVID-19, whereas closer to 60 percent of respondents blamed people who visited family for Ramadan/Eid and Diwali. The mean differences in responses between those receiving the Christmas prompt and those receiving either the Ramadan/Eid or Diwali prompt were statistically significant at p ≤ 0.001 in independent samples t-tests. Thus, as predicted in H1, there is a significantly – and we would argue substantively – higher level of blame of outgroups than ingroups and the results suggest generally high levels of blame of religious outgroups, potentially more so than when asking generally about blaming minorities.

How does the blaming of minority groups compare with the levels of blame placed on other groups and behaviors? Figure 2 shows the mean responses to all of the blame items included in this question sequence (see Appendix A), which further illustrates that when we ask directly about ethnic and religious minorities, the average level of blame is relatively low – in fact it is the second lowest response in this figure, with older people receiving the least blame. However, in the split-sample question, respondents who received the Ramadan/Eid and Diwali prompts generally leaned more toward blaming these groups. In fact, those who received the Ramadan/Eid and Diwali prompts leaned more toward blaming these groups than respondents who saw the Christmas prompt, and even blamed those visiting family for Ramadan/Eid and Diwali more than they blamed immigrants, who in the context of a pandemic might be seen logically as being a source of spreading of the virus (e.g. migrants who arrived during the pandemic). Other groups for which the responses leaned towards blame included: young people, those refusing to wear masks or get vaccinated, UK holiday-makers, overseas holiday-makers, and tourists. In fact, comparing these other items to the split-sample prompt responses, we see that blame of those visiting family for Ramadan/Eid and Diwali – which has not been established as being connected with the spread of COVID-19 – is only slightly lower than blame being placed on factors that are thought to be connected to the spread of COVID-19 (e.g. not being vaccinated and not wearing a face covering).⁴

Figure 2. Perceived blame for the spread of COVID-19 in the UK.

Note: Items shown in same sequence in which they appear in the questionnaire; Total N = 758; Christmas N = 247; Ramadan/Eid N = 264; Diwali N = 247.

Turning to the centered versions of our blame of minorities items (centered around general tendency to blame), the means of these variables (see Figure 3) again illustrate that blaming minorities in particular was far lower than blaming other groups listed in this question sequence. On the other hand, the centered means for blaming people for visiting family during Ramadan/Eid and Diwali are higher than blaming minorities in general, suggesting a higher propensity to blame these groups – but this is also the case for blaming people for visiting family for Christmas. However, the differences between these means still suggests intergroup bias as outlined in H1: after centering the variables, the mean for blaming people for visiting family for Ramadan/Eid is 0.52 higher than the mean for blaming people who visited family for Christmas on the 0–10 scale and this difference is statistically significant (p = 0.0003). There is no evidence of similar prejudice towards Hindus (based on the response to the Diwali question; for this difference of means, p = 0.1260). Thus, there is still some support for H1 when using the centered variables, in the case of Muslims in particular. This is consistent with other research from the UK suggesting continued discomfort with Muslims (Storm, Sobolewska, and Ford 2017).

Figure 3. Blame of minorities for spread of COVID-19 (centered).

Note: Items centered around general tendency to blame (see paper text); Total N = 758; Christmas N = 247; Ramadan/Eid N = 264; Diwali N = 247.

Unfortunately, we do not have enough observations to fully examine perceptions of minorities towards one another or towards non-minorities due to small Ns: the total Muslim N = 22 and for Hindus and Sikhs combined N = 10. There are 105 (self-identified) ethnic minorities in our sample, and Appendix C provides results for all of our analyses, comparing minorities and non-minorities. These suggest that minority respondents are slightly less likely to blame minorities (in general) than the non-minority respondents (see Appendix Figures C1 through C4), as might be expected. For minorities, the difference between the Christmas and Ramadan means is 0.5511, suggesting that – like non-minorities – minorities tend to believe Muslims in particular were more to blame for the spread of COVID-19 than those celebrating Christmas. However, this difference is not statistically significant, presumably due to small Ns (the split sample format meant that the 105 minority respondents were split into three groups). The difference between blaming Christmas and Diwali is even smaller and not significant (see Appendix Figures C3 and C5).

Findings: sources (covariates) of blame of minorities

We now examine the predictors of blame of minorities using the centered versions of our dependent variables (Figure 4). Recall that our dependent variables are measured on 0–10 scales (which have been standardized for the purpose of illustrating the effects in coefficient plots below) and so we use OLS to estimate the effects of independent variables. We begin with responses to the general question about blaming minorities. The top left panel of Figure 4 suggests that those who have had COVID-19 are the least likely to blame minorities in general for the spread of COVID-19. However, this variable is not significant in the models of blaming Ramadan/Eid or Diwali (or Christmas).

Figure 4. Blame minorities generally, Ramada/Eid, Diwali and Christmas (all centered).

Note: Figure shows multivariate OLS coefficients and 95% confidence intervals using standardized variables, for two models: one that excludes perceived general threat from minorities (light dot/line); and one that includes perceived general threat from minorities (dark dot/line). Full Models Ns: Minorities question = 744; Ramadan/Eid = 233; Diwali N = 240; Christmas N = 240.

Among the other hypothesized predictors, authoritarianism appears to be a significant (but somewhat weak) predictor of blaming minorities in general but this effect disappears when general perceived threat posed by minorities is added to the model. The latter is strongly related to authoritarianism in our dataset (Pearsons r = 0.46), suggesting that authoritarians are more threatened by minorities (consistent with existing literature), and thus that authoritarianism is likely to impact blaming of minorities during the pandemic via a more general sense of threat presented by minorities. (As noted above, it is also possible that general perceptions of minorities as measured in our survey may have been impacted by feelings of blame of minorities for the spread of COVID-19). Nevertheless, whether direct or indirect, authoritarianism appears to be connected to blaming minorities during the pandemic. Authoritarianism has no discernible effect in the models of blaming Ramadan/Eid or Diwali (or Christmas), though.⁵

Being worried about catching COVID-19, potential economic threat, and locus of control measures are all statistically insignificant in all models in Figure 4. Among the control variables, in the Ramadan/Eid models, age and being Christian achieve statistical significance in the models that do not control for general perceived threat posed by minorities: older people and those who identified their religion as Christian were more likely to blame people visiting family for Ramadan/Eid (but the effects are relatively weak). In the Diwali model, those with university level education are less likely to blame Diwali for the spread of COVID-19 than those without higher education; those with higher education are, on the other hand, more likely to blame Christmas for the spread of COVID-19 than those without higher education. Education has no discernible impact on blaming minorities generally or blaming people who visited family for Ramadan/Eid. People in the East Midlands are less likely to blame Christmas for the spread of COVID-19. Finally, general perceived threat posed by minorities is significant in blaming minorities generally and blaming Ramadan/Eid but not Diwali or Christmas.

As shown in Appendix Figure C7, the results for non-minority respondents largely mirror those reported above. For the minorities in the sample, authoritarianism had no discernible impact on perceptions that minorities were generally to blame for the spread of COVID-19. Holding generally negative perceptions of minorities has a slightly stronger effect for minority respondents than non-minorities but also has a larger confidence interval due to the smaller N. For the Ramadan/Eid prompt (Figure C8), being worried about catching COVID-19 (disease threat) and about not being able to pay bills (economic threat) were both significantly associated with blaming Ramadan/Eid among non-minorities (in the full model) but not for minorities, suggesting that the (non-)findings reported in Figure 4 masked support for H2 and H3 among non-minorities. Age was also significant for non-minorities but not for minorities in Ramadan/Eid models, suggesting that it is particularly older white respondents who blamed Muslims for the spread of COVID-19. In the case of the Diwali prompt (Figure C9), non-minority respondents with high individual locus of control were slightly less likely to blame Diwali, but this does not quite achieve statistical significance in the full model (dark lines in the figures). Concern about COVID-19 cases coming into the country is significant among minorities who received the Diwali prompt, perhaps reflecting concern about the alleged delay by the UK government in imposing travel restrictions on travellers from India after a new COVID variant emerged there and cases there rose significantly (Webber 2021) – of which minorities may have been more aware and concerned.

Discussion and conclusion

This is the first study to examine the extent and causes of blaming of minorities for a public health crisis in the UK using a representative survey of the British public. Our hypotheses drew from theories about perceived threat from potential outgroups, the impact of locus of control on perceptions of these groups, and the relationship between authoritarian values and perceptions of outgroups like ethnic and religious minorities. Among our full sample, we found limited support for the notion that those who had or were worried about catching COVID-19, were worried about the personal and country-level economic impact of COVID-19, had a low sense of personal locus of control, or high level of concern about the lack of control over COVID-19 by the UK government were more likely to blame minorities. Our additional analyses of minority and non-minority respondents suggest that among non-minority respondents, those who were expected to feel most threatened and to blame minorities as a result – i.e. those who felt COVID-related disease threat and personal economic threat most acutely – did indeed tend to blame Ramadan/Eid participants for the spread of COVID-19. That is, disease and economic threats appeared to translate into blame of Muslims in particular. Authoritarianism was shown to be related to blaming minorities in general – particularly among non-minorities (see Appendix Figure C7) – before our control for general perceptions of minorities was added to the models. Authoritarianism was not at all significantly related to blaming Muslims or Hindus, however, suggesting the possibility that when faced with a serious threat to safety or security, even those who are ostensibly libertarian may inadvertently express similar levels of blame of outgroups as those who appear to be more authoritarian.

Our findings also suggest that those with higher levels of education (one of our control variables) tended to be significantly less likely to blame those visiting family for Diwali and more likely to blame those visiting family for Christmas. Education was not significant in the models of blaming those visiting family for Ramadan/Eid, however. That is, those with university-level education were equally likely to blame those visiting family for Ramadan/Eid as those without higher education, suggesting widespread anti-Muslim hostility that is unaffected by education, which itself would normally be strongly associated with lower levels of overt prejudice; this finding lends support to research that questions the role of higher education in fostering tolerance (see Weber 2022). Christians were somewhat more likely to blame those visiting family for Ramadan/Eid but not those visiting family for Diwali or Christmas, thus revealing additional evidence of intergroup bias in the case of blaming Muslim cultural practice versus blaming the dominant ingroup. Non-white minorities (self-identified) were less likely than white respondents to blame Hindus (Diwali) for the spread of COVID-19; alternatively, white respondents were more likely to blame people visiting family for Diwali than non-white respondents (see Figure 4), also revealing intergroup bias.

Finally, though generally negative perceptions of minorities (our final control variable) were associated with increased blaming of those visiting family for Ramadan/Eid, the effect was relatively weak and it was not even significant in the Diwali models (nor in the Christmas models, as expected). The weak impact of general negative perceptions of minorities on blaming those visiting family for Ramadan/Eid combined with the lack of significance of general negative perceptions of minorities on blaming those visiting family for Diwali further suggests that our split sample items may have captured some extent of unconscious or implicit prejudice. That is, people who did not report being uncomfortable about the impact of minorities on British values, jobs and other resources in our control variable questions (see Appendix A) were equally likely to blame those visiting family for Diwali and only somewhat less likely to blame those visiting family for Ramadan/Eid than those who did report being generally uncomfortable about the impact of minorities. We take this as evidence that our indirect survey items may have captured a different aspect of hostility to and blame of minorities than seen when measuring attitudes to minorities more directly (as with our control variable). This suggests the need for survey researchers to experimentally ask about specific groups in survey questions that attempt to measure anti-minority sentiment.

More generally, our findings suggest that large segments of the British public may have agreed with some of the sentiment expressed over social media (Awan 2020) and may have held some minorities to blame for spreading COVID-19. As noted above, this is consistent with research indicating relatively high levels of explicit racism in the UK (Caller and Gorodzeisky 2022). While COVID-19 is no longer a major threat in the UK, it seems unlikely that the UK and other countries will avoid similar health crises in the future. Our study suggests that the centuries-old phenomenon of blaming minorities for the spread of disease may not be limited to a small handful of particularly vocal individuals. Instead, negative attitudes to minorities held by large segments of the public may sustain an environment in which the vocal individuals are willing to make statements that are potentially damaging to intergroup relations and to the groups being targeted. Blaming of specific minority groups also potentially undermines the “we-feeling” required for large-scale public cooperation with government-recommended health measures, while creating an environment in which hate crime can flourish (Abrajano and Hajnal 2015; Dancygier and Green 2010; Gray and Hansen 2021; Ng 2020).

The targeting of minorities during the COVID-19 pandemic in the UK and elsewhere also serves as a reminder that the conflict over diversity which, for example, produced the historic Brexit vote (Sobolewska and Ford 2020) is still very much ongoing. Targeting of minorities during the pandemic thus underscores the need for further and continued study of perceptions of ethnic and religious minorities in Britain. Methodologically, our research for this paper revealed the existence of only a surprisingly limited number of studies specifically on public perceptions of minorities in the UK that are based on representative surveys (as opposed to small student samples, e.g. Maio 2017), with much of the most relevant research tending to focus instead on attitudes to immigrants and immigration. Accordingly, an important area for future research is the development of measures that could be introduced into regularly fielded public opinion surveys (British Social Attitudes and European Social Surveys) which would allow us to track public perceptions of minorities over time. This is undoubtedly not an easy task, especially given the above-noted social desirability and the anti-prejudice norm. Our questionnaire constitutes an attempt to grapple with this using a split sample question format but it will need to be adapted for non-COVID times.

Given the potential harm caused to minorities by being blamed for a public health crisis – as well as many other problems – an additional area for (continued) further research lies in understanding factors that are likely to impact attitudes to minorities more generally. For instance, extensive literature has established that positive “contact” (i.e. interaction) with minority groups can reduce hostility to them (Paluck, Green, and Green 2019), but the usefulness of this type of intervention is limited by extensive barriers to meaningful contact such as segregated housing and friendship groups (Enos 2017). Mediated contact with members of minority groups (e.g. through television and film) may also reduce hostility (Alrababa’h et al. 2021; Schiappa, Gregg, and Hewes 2005), presenting the possibility of interventions that could influence public perceptions of minorities on a much larger scale. Generally speaking, the difficult post-pandemic economic recovery in countries like the UK along with other currently unknown public crises and their possible economic repercussions means that the possible scapegoating of minorities requires monitoring and methods to ameliorate this problem require further investigation.

Ethics statement

Approval for conducting the public opinion survey was received from the University of Leicester’s Science & Engineering, Arts, Humanities and Law Research Ethics Committee (17 December 2021; Reference 32327). Participants were provided with a participant information form and were required to click that they consent to participate in the survey before answering any of the survey questions. No identifiable information was shared with the authors, and the current dataset is fully anonymized.

Disclosure statement

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

Additional information

Funding

Funding for the public opinion survey used in this project was provided by the University of Leicester College of Social Sciences Research Development Fund. A previous version of this paper was presented at the Elections, Public Opinion and Parties Annual Conference, Newcastle University, 2–3 September 2022. We thank conference participants for their comments, particularly Sir John Curtice, Geoff Evans, Steve Fisher, and Nicole Martin. We also thank Jason Reifler and Neil Chakraborti for their helpful comments on the questionnaire used in this study.

Notes

1 We use “blame” in the colloquial sense: “to say or think that someone or something did something wrong or is responsible for something bad happening” (Cambridge Dictionary; https://dictionary.cambridge.org/dictionary/english/blame) rather than as used in attribution theory (see, Dionne and Turkmen 2020 for similar usage in the COVID-19 context).

2 Appendix B shows the demographic distribution across the split sample item. Many of these demographic variables are included in our multivariate analyses. We also include a dummy variable for the East Midlands region, since there was relatively wide variation in the percentage of respondents per split-sample group who were from the East Midlands.

3 It might be contended that blame of minorities for the spread of COVID-19 is epidemiologically logical/rational, given that research has shown that many minority groups in the UK live in large multigenerational households, carry out work that puts them into closer contact with people, have underlying health conditions, and were more hesitant about vaccinations. Thus, many who express blame of minorities for the spread of COVID-19 may do so because they understand these causal explanations rather than because they hold these groups morally responsible due to personal or group shortcomings or failures. We address this issue of causal versus moral blame in Appendix D using questions that were included in our questionnaire asking respondents for their views on the reasons for higher minority infection and mortality rates.

4 In the UK, face covering was found to be effective in reducing average weekly COVID-19 infection rates during the pandemic (Heald et al. 2021). Vaccination also reduced overall COVID-19 rate (Moghadas et al. 2021).

5 We have also investigated the possibility that authoritarians, libertarians (e.g., Hetherington and Suhay 2011), and those with left-leaning ideologies (Bogatzki, Glaese, and Stier 2023) who faced high levels of disease or economic threat (i.e., interactive effects) were more likely to blame minorities; none of these interactions were statistically significant.