key: cord-0974744-bucrxsma authors: Mellon, Jonathan; Bailey, Jack; Prosser, Christopher title: How Do Coronavirus Attitudes Fit into Britain’s Ideological Landscape? date: 2021-05-29 journal: Parliam Aff DOI: 10.1093/pa/gsab030 sha: f06b587c9763f14bed56ece6440903d7a2f5e1f3 doc_id: 974744 cord_uid: bucrxsma Coronavirus upended British politics in 2020 but where does it fit into the ideological map of party competition? Recent British elections have seen a shift from economic left–right competition between the major parties to competition on the cultural (liberal–authoritarian) dimension, most notably in terms of the issues of immigration and membership of the European Union. Using British Election Study data from June 2020, we find that coronavirus attitudes fall primarily onto the traditional economic left–right dimension, with left-wing voters more willing to make economic sacrifices of various types to reduce infections. However, more draconian coronavirus measures (such as fining or imprisoning those who violate the coronavirus rules) are most supported by voters who score high on authoritarianism. We show that the structure of coronavirus attitudes puts the Conservative government in a difficult position where many steps it takes to reduce infections risk alienating its core economic right-wing vote. British Prime Minister Boris Johnson entered 2020 on a high. After over 20 years of opposition, coalition and hobbled government, the Conservative Party had at last won a sizeable majority. What's more, they had done so by hoovering up seats in the opposition Labour Party's historic heartlands. Despite months of parliamentary wrangling, the UK was finally set to leave the European Union early in the new year. In a post on social media, a buoyant Johnson said that 2020 would be 'a fantastic year for Britain' (Johnson, 2020) . In reality, it was to be anything but. In the months that followed, the global coronavirus pandemic would bring the UK's greatest public health crisis in a century, the Prime Minister himself would be hospitalised, and more than 150,000 Britons would lose their lives to coronavirus over the course of the pandemic (UK Government, 2021) . Despite the government's initial reluctance to impose lockdown measures, the UK had almost ground to a halt by mid-March. This saw all but essential workers told to stay at home and schools and businesses shut. As one would expect, the economic consequences were dire. UK GDP shrunk by 20% (BBC News, 2020b) and one-quarter of the country's workforce was soon sat idle, with 80% of their wages paid for by the state (BBC News, 2020a) . Though all major political parties backed these measures, they were not without criticism. Some argued that they were economically disastrous (Reuters, 2020) , draconian (Picheta, 2020) and illiberal (Snowdon, 2020) . Others simply refused to abide by them at all (Pearson, 2020) . Political research views policy measures like these through one of two lenses. The first is the lens of valence politics, which holds that voters care most about competence. When it comes to the pandemic, everyone agrees what this means: minimising deaths. What matters, then, is how the government performs. If it does well, and few die, then voters should reward it with their votes. If it does badly, and many die, then voters should punish it by voting for someone else instead. The second is the lens of spatial politics, which holds that voters' policy preferences matter most. That is, voters care about more than just competence, they care about which policies are implemented. As such, they make their own trade-off between, say, lives and livelihoods, then vote for the party whose platform is most similar to their own opinions. In this article, we take such a spatial view and examine how voters' attitudes towards coronavirus policy measures map onto existing political divides. In light of their impact, these measures could have serious political consequences. Yet given the UK's complicated politics, it is not at all clear what these consequences should be. After the 2016 Brexit referendum, British politics underwent considerable realignment . Rather than the traditional economic left-right divide, EU attitudes and liberal-authoritarian social values now also structure British voting behaviour . The consequences of coronavirus containment measures, thus, depends on how they correlate with voters' political values. Were voters' liberal-authoritarian values to correlate with their coronavirus policy attitudes, they would reinforce this realignment. But if they were to cut across them, the pandemic might put existing coalitions under strain. Likewise, if left-right economic values are the primary correlate of coronavirus attitudes, the crisis might undermine recent realignments entirely, shifting British politics back towards the traditional economic divide between the major parties. There are good reasons to expect left-right and liberal-authoritarian values to structure voters' attitudes towards policies that seek to change individual behaviour in response to the coronavirus crisis too. Although few voters can be properly described as 'ideological' (Converse, 1964) , core values play an important role in structuring political attitudes and evaluations (Feldman, 1988 (Feldman, , 2003 Tetlock, 2012) . Indeed, we already know that these values are associated with different attitudes, behaviours, and beliefs. For instance, right-wing and socially-authoritarian voters in Britain tend to be more likely to vote for the Conservative Party (Surridge, 2018; Hobolt, 2018) and tend also to have voted to Leave the EU (Norris and Inglehart 2019) . The nature of the items used to construct these questions also gives us some potential clues as to how different voters might respond to coronavirus policy measures (Heath et al., 1994) . Left-right values tend to concern issues of economic inequality, economic rights and economic justice. Liberal-authoritarian values, instead, tend to concern social transgressions, social rules and social conventions and traditions. We might, thus, expect left-wing voters to be more supportive of coronavirus policy measures that address the negative consequences of the virus even in spite of their potentially negative economic effects than similar right-wing voters. Equally, we might expect socially conservative voters to be more keen to punish those who break the rules than similar socially liberal voters. Initial work from Hicks (2020) suggests that both EU referendum and economic attitudes may structure overall support for lockdowns. Nevertheless, coronavirus policy preferences are an understudied aspect of the crisis compared with predictors of compliance. So far, most social science work on the pandemic has focused on individual behaviours. For instance, research in the USA has found partisan, religious and ideological differences in behaviours in response to the pandemic (Graham et al., 2020; Gollwitzer et al., 2020; Perry et al., 2020; Pickup et al., 2020; Samore et al., 2020) . Other predictors of coronavirus compliance (or not) include fatalism (Akesson, 2020) , trust in government (Devine et al., 2020; Dryhurst et al., 2020) , moral foundations (Harper et al., 2020) , lower risk perceptions (De Neys et al., 2020) , pro-social messaging (Jordan et al., 2020; Van Der Linden and Savoie, 2020) , pro-sociality itself (Campos-Mercade et al., 2020), civic duty (French Bourgeois et al., 2020) , gender (Cassino and Yasemin, 2020) , personality (Brouard et al., 2020) , impulse control (Kuiper et al., 2020) and many more. 1 Using new data from the British Election Study Internet Panel , we show that coronavirus policy attitudes in Great Britain vary according to party and policy. As suggested, we find that voters' coronavirus policy attitudes fit onto the economic left-right spectrum, with left-wing voters in favour of policies that restrict economic activity to tackle the coronavirus. 1 It should be noted that most of these papers have not yet been peer-reviewed given the fast-moving situation. Yet more authoritarian people are also more likely to support more punitive measures. The result of this is that the average Labour Party voter is much more likely to support strong responses to the coronavirus, while the average Conservative voter is more sceptical. This poses a particular challenge to the Conservative party, which has had to impose many policies that their voters are sceptical of. Our results suggest that there may be space for another party to attract Conservative voters by taking a harder line against coronavirus measures. To look at attitudes towards coronavirus policies, we use data from wave 20 of the British Election Study Internet Panel (BESIP). Data collection was conducted between 3 and 22 June 2020. This was past the peak of deaths in April and May, but still at a point in time where the UK was reporting between 52 and 155 coronavirus-related deaths per day (GOV.UK 2020). The UK was still under a substantial lockdown, with non-essential workers advised to stay at home, and citizens told to avoid contact with people outside of their households. However, there was a vigorous public debate about the correct time for lifting restrictions and which restrictions would be lifted (Reuters, 2020) . We use two primary measures of willingness to support policies to reduce the coronavirus. First is a general question about the trade-off between the economy and coronavirus infections: 'Some people think that the government should do everything it can to reduce the number of coronavirus infections, even if it greatly damages the economy. Others think that the government should try to save the economy, even if it increases the number of coronavirus infections. Where would you place yourself and the parties on this scale?' We also use a battery of items on support for specific coronavirus policies: 2 'Which of these measures, if any, do you think Britain should currently be taking to tackle coronavirus? Please tick all that apply:' • Encouraging people to wash their hands more often • Recommending people keep more than 2m away from other people 2 For both of these batteries we report results dropping "don't know" responses. 9% of respondents gave a don't know response to the first battery and 8% to the second battery for a total of 12% who said don't know to either battery. We do not treat don't knows as not agreeing with the policy because it is only possible for the respondent to say don't know to the battery as a whole, so it is not possible to distinguish which items respondents felt unable to answer. Nonetheless the level of don't know responses to these questions are well within the normal range for survey responses. • Only allowing essential shops (e.g. supermarkets, pharmacies, post offices, etc) to stay open • Banning people from leaving their homes except for essential work, grocery shopping and accessing medical care • Banning gatherings of more than two people • None of these • Don't know 'And which of these measures, if any, do you think Britain should currently be taking to tackle coronavirus? Please tick all that apply' • Imprisoning people who do not follow the coronavirus rules • Fining people who do not follow the coronavirus rules • Closing schools • Requiring people to wear masks in public • Asking people to download an app to track who they have been in contact with • None of these • Don't know To measure the two dimensions of politics we use the long-standing (Evans et al., 1996) batteries of five left-right and liberal-authoritarianism items (see the Appendix for the items). These scales are well-validated and known to be broadly stable over time. Further, the correlation between the two value scales is generally low to zero, though it does appear to have increased a small amount over time (Surridge, 2012) . We can think of the specific coronavirus policy items that we use here in one of two ways. The first is as a set of repeated measurements, nested within individual respondents and individual items. Most often, researchers use multilevel models to account for this nested structure. This allows one to pool information across the items, but also allows each to respond to covariates in their own way. The second is a series of items that measure each respondent's latent attitude towards coronavirus policy. Here, the most common approach is to use psychometric models informed by item response theory (IRT). This assumes that each respondent has some underlying, continuous attitude towards coronavirus policy, and that each item taps into one of its unique facets. Fortunately, multilevel and IRT models are often mathematically equivalent (Doran et al., 2007) . In this case, we model our data using multilevel linear probability models. These treat voters' support for various coronavirus policies as a function of either their age, gender, class, having a degree, employment status, sector, vote at the 2019 general election, vote at the 2016 referendum on EU membership, left-right attitudes or liberal-authoritarian attitudes. We then allow each item and each individual to have their own random intercepts, and use random slopes to allow the effect of the particular covariate in the model to vary over each specific item. We first describe the overall patterns of support for different coronavirus policies and how they relate to each other. Figure 1 shows the proportion of respondents who said that they thought a particular policy should be in place at the present time. The popularity of different coronavirus policies varied dramatically, from 76% for encouraging handwashing to just 20% support for imprisoning rule breakers. Interestingly, support for policies does not map perfectly onto the policies that were in place at that time. While some existing policies, such as social distancing (71% support), were strongly supported, other policies that were in place, such as schools being closed (40% support), were much less popular. Similarly, support for mandatory mask-wearing (which was not yet recommended by the government) was relatively popular at 57%. Overall, the pattern of support is relatively intuitive, with more punitive, demanding or intrusive policies receiving lower support. The BES data also include party positions on one broad aspect of coronavirus policy, the trade-off between reducing infections and protecting the economy. Figure 2 (top panel) shows that at the time of the wave 20 fieldwork voters were heavily skewed towards preferring reducing infections over saving the economy. On average, respondents scored themselves at 3.5 on the 0-10 scale. This means that voters rated themselves as more willing to favor economic sacrifices over greater infections than they rated either the ruling Conservative party (6.6) or the Labour opposition (3.9), although the average voter was far closer to Labour than to the Conservatives. While voters judge the smaller parties as being more in line with public opinion, it is difficult to directly compare these ratings because voters are much less likely to be able to place these small parties on the infections/economy scale as we see in Figure A3 in the Appendix. The one exception to this is the Scottish National Party (SNP) who most Scottish voters are able to place on the scale. This is likely due to the Scottish government having substantial devolved power over coronavirus policy making it easier to place them in this policy space. The SNP are rated as being more concerned about coronavirus infections than the economy, with the average position (2.9) being very close to the average voter on the issue. While the Conservative position on the infection-economy scale is well outside that of the average voter, it is more in line with the position of Conservative voters. The bottom panel of Figure 2 shows the average self-placement score on the infections-economy scale of each party's 2019 voters. Conservative voters are the most pro-economy of any party's supporters (with the median Conservative voter placing themselves at the midpoint of the scale). In addition, Conservative voters see the party's position as considerably more moderate than other voters do, with the median Conservative voter placing the party at the midpoint as well (see the middle panel of Figure 2 ). However, it is likely that self-placement and party placement are endogenous to vote choice. Figure 3 shows how the support for the various coronavirus measures differs across 2019 voters of the political parties (compared with 2019 Conservative voters). There is a considerable variance across items in how they split or crosscut parties. Labour and SNP voters are significantly more likely to support closing schools, closing shops, making people stay home and requiring mask-wearing than Conservative voters. These differences are substantial, Labour voters are 30-percentage points more likely to support closing schools than Conservative voters. However, there are other policies which do not map onto the existing party divisions, with 2019 Conservative voters being similarly or more likely to support intrusive coronavirus measures, such as fining and imprisoning those who break coronavirus rules or creating an app to track contacts as Labour voters. Overall, the pattern appears to be that Labour voters are more willing to support burdensome policies (shutting down economic activity in various ways), while Conservative voters are more open to measures that could be described as authoritarian but are relatively unlikely to affect the economy. This likely presents more of a challenge for the Conservatives than the Labour. Labour's voters are supportive of relatively burdensome measures to tackle the coronavirus, which broadly matches the policy position of the party. However, the Conservatives are in a difficult position because their government has instituted many of the measures which their own voters are sceptical about. For instance, only 25% of Conservative voters supported closing schools at the time of the BES wave, a policy that the Conservative government reluctantly introduced. This pattern means that the Conservatives will be in tension with their voters about introducing many measures to tackle the coronavirus. Figure 3 also shows how support for the policies maps onto Brexit divisions. While much of British political discourse has been subsumed into the argument over Europe, coronavirus attitudes seem to be less related to EU referendum voting than to party divisions. This suggests that coronavirus attitudes have the potential to weaken the EU cleavage. Importantly, the patterns of support for items do not map neatly onto the popularity of different policies. Closing schools and implementing a tracking app are similarly popular among the population but school closing shows a strong party gradient in support whereas implementing a tracking app is not heavily split along party lines. One lens for understanding these cleavages is through values. Figure 4 shows where each coronavirus measure falls on the economic left-right and liberal-authoritarian dimensions. The first thing to note is that despite recent realignments in the British electorate , the economic left-right dimension tends to do a better job of predicting the items than does the cultural liberalauthoritarian dimension. Having economic right-wing values predicts much lower support for closing schools, shops, social distancing mask-wearing and banning gatherings. However, right-wing economic values are unrelated to fining or imprisoning rulebreakers, handwashing and actually positively predict support for a tracking app. Although there has been a popular perception that support for policies to tackle coronavirus has been higher among liberals, the effects are small and sometimes in the reverse direction. Coronavirus policies that limit individuals' autonomy (banning gatherings and staying at home) or are highly punitive (fining or imprisoning rulebreakers) are actually more supported by respondents with authoritarian values than liberals. The dimensional correlations give clues to the patterns of party support and Brexit support we saw in the previous section. Vote choice in Britain is structured around both the left-right and liberal-authoritarian dimensions, with the Conservatives appealing to a mixture of authoritarians and economically right-wing voters and Labour appealing to a mixture of liberals and economically left-wing voters. However, Brexit is almost entirely a liberal-authoritarian issue. The few coronavirus measures that 2019 Conservative voters do support tend to be those that are predicted by having authoritarian values. We can also look at how coronavirus attitudes play out among demographic groups. Figure 5 shows the patterns of support for coronavirus measures by age, gender, education, class, working status and job sector. The largest gender differences are for mask-wearing (which women support more than men in line with other studies (Haischer et al., 2020) ) and imprisoning rulebreakers (with men expressing higher support than women). Age follows a pattern that is somewhat different to other patterns we have seen so far. Older people are significantly more likely to support the least demanding policy of encouraging handwashing. However, they have significantly lower support for closing schools and shops. There are relatively weak relationships between education levels and support for different coronavirus measures in Figure 5 . This provides further support for the idea that coronavirus attitudes fall around economic interests rather than cultural and social divides. There are clear class divisions on coronavirus policy in Figure 5 . Working-class voters are more supportive of measures that tackle the coronavirus by limiting economic activity (closing schools, staying home and closing shops (the latter insignificantly)) or through the punitive measure of imprisoning rulebreakers. However, middle-class voters are relatively more supportive of measures that require individual action but probably do not limit economic activity (tracking app, handwashing and mask-wearing). The divides by working status are limited to only a subset of the measures, with respondents who were not in the labour force or were furloughed expressing greater support for mask-wearing, social distancing and handwashing but otherwise expressing similar levels of support for coronavirus policies compared with those who were working. This pattern goes against the idea that respondents who are most likely to be exposed to coronavirus would express greater support for policies to reduce its spread. This may reflect the fact that people in the workforce bear the largest compliance burden with coronavirus measures. Furloughed respondents were also more likely to support closing schools compared with those who were working, which may reflect the difficulty people with children have had in managing work and childcare. Finally, furloughed respondents were more likely to support fining rulebreakers, although the reasons for this difference are not immediately obvious. The sectoral differences are also limited to a few of the items. People working in the charity sector were generally more supportive of coronavirus measures that involved personal behavior including handwashing, mask-wearing, using a tracking app and social distancing. Public sector employees showed similar patterns but with smaller differences, with significantly more support for handwashing, using a tracking app, social distancing and closing shops. 3 Our results show that even by our fieldwork date in June 2020, the British public already had structured views towards the coronavirus. While strong measures to tackle the coronavirus were popular in general, 2019 Conservative voters were much less likely than the Labour voters to support many of the policies that impose economic costs on Britain. This scepticism reflects the fact that attitudes towards many coronavirus measures are polarized along the economic left-right dimension, likely reflecting the heavy toll on business and the economy of many measures to tackle the coronavirus. Taken together, these results suggest that support for policies to tackle coronavirus are most strongly an economic left phenomenon and that liberals are, on average, actually slightly less likely to support policies tackling coronavirus. However, the results also show that different aspects of coronavirus policy tap into different parts of the existing political landscape. 3 The sector analysis uses a Bayesian model (fitted in Stan using brms), because the original lme4 version struggled to reach convergence. However, the Conservatives will risk losing their newly won working-class supporters if they take a more relaxed stance towards coronavirus policies, as these policies are substantially more popular among working-class voters. They will also be out of line with the views of the public as a whole. Our results suggest that at the time of the fieldwork, the public were generally much more willing to support policies to tackle coronavirus than they perceived the government to be. Of course, these data reflect only a snapshot of British political competition, and competition around coronavirus could shift as parties and policies change. There is a some evidence of movement in issue linkages from the one other study of ideological linkages to British coronavirus attitudes (Hicks, 2020) . However, polling related to the January 2021 lockdown reveals a similar pattern of political support to what we find here (Ibbetson, 2021) ; high overall support for the lockdown, but with Conservative and Leave voters being the most likely to oppose the measures. One question that should be tackled in future research is how ideological positions interact with objective risks in determining coronavirus attitudes. Previous research has shown that many policy preferences are best explained by understanding a voter's social position combined with whether or not they have particularist or universalist policy preferences (which is closely related to their liberal-authoritarian position) (Mellon and Prosser, 2017) . Overall then, the coronavirus crisis appears to be politicised around left-right economic competition in strong contrast to the increasing tendency of British politics to revolve around liberal-authoritarian issues, such as immigration and the EU. If the coronavirus crisis has a long-lasting impact, it could push political competition back towards its traditional economic roots. However, it may also open up opportunities for challengers to the Conservatives such as the Brexit party to compete for its more lockdown-sceptical voters. with the tradeoff between freedom and coronavirus measures providing the most information. However, two of the binary items (making people stay home and closing shops) provide particularly high levels of information around the middle of the scale where most respondents are. Figure A2 . Test information function for full set of Coronavirus items (note that higher values of theta indicate more support for Coronavirus policies). The horizontal red line reflects the equivalent of 0.7 reliability. Figure A3 . Percentage of voters reporting that they don't know where a party is placed on the 0-10 infections versus economy scale. Combining the information for the items together in Figure A2 shows that the combined set of coronavirus items have impressively high information across much of the possible latent dimension, with only very pro-coronavirus measures respondents measured with less precision. Fatalism, Beliefs and Behaviors during COVID_19 UK Furlough Scheme Extended by Four Months Bank of England 'ready to act' as economy shrinks record 20% Sociodemographic and Psychological Correlates of Compliance with the Covid-19 Public Health Measures in France Prosociality Predicts Health Behaviors during the COVID-19 Pandemic Of Masks and Men? Gender, Sex and Protective Measures during COVID-19 The Nature and Origin of Belief Systems in Mass Publics Moral Outrage and Social Distancing: Bad or Badly Informed Citizens? Trust and the Coronavirus Pandemic: What Are the Consequences of and for Trust? An Early Review of the Literature Estimating the Multilevel Rasch Model: With the Ime4 Package Risk Perceptions of COVID-19 around the World The Measurement of Left-Right and Libertarian-Authoritarian Values: A Comparison of Balanced and Unbalanced Scales Measuring Left-Right and Libertarian-Authoritarian Values in the British Electorate Structure and Consistency in Public Opinion: The Role of Core Beliefs and Values Values, Ideology, and the Structure of Political Attitudes British Election Study Internet Panel Electoral Shocks: The Volatile Voter in a Turbulent World To Follow or Not to Follow: Social Norms and Civic Duty during a Pandemic Partisan Differences in Physical Distancing Predict Infections and Mortality during the Coronavirus Pandemic Coronavirus (COVID-19) in the UK: Deaths Faith in Trump, Moral Foundations, and Social Distancing Defiance during the Coronavirus Pandemic Who is Wearing a Mask? Gender-, Age-, and Location-Related Differences during the COVID-19 Pandemic Functional Fear Predicts Public Health Compliance in the COVID-19 Pandemic The Measurement of Core Beliefs and Values: The Development of Balanced Socialist/Laissez Is Lockdown Approval Rooted in Ideology and Vote Choice? Brexit and the 2017 UK General Election Brits Support New National Lockdown @BorisJohnson: This is Going to Be a Fantastic Year for Britain Don't Get It or Don't Spread It?', Comparing Self-Interested versus Prosocial Motivations for COVID-19, Prevention Behaviors The Intelligent Lockdown: Compliance with COVID-19 Mitigation Measures in The Netherlands Authoritarianism, Social Structure and Economic Policy Preferences Cultural Backlash: Trump, Brexit, and Authoritarian Populism I Refuse to Abide by These Bonkers Rules Any Longer Culture Wars and COVID-19 Conduct: Christian Nationalism, Religiosity, and Americans' Behavior during the Coronavirus Pandemic UK Coronavirus Response Criticized as People Are Filmed by Drones and Stopped While Shopping Novel Coronavirus, Old Partisanship: COVID-19 Attitudes and Behaviors in the United States and Canada Wetherspoon Founder Criticises UK Government for Coronavirus' lockdown Of Pathogens and Party Lines: Social Conservatism Positively Associates with COVID-19 Precautions among U.S. Democrats but not Republicans Lift Lockdown Now or Normalise a Chilling Authoritarianism Unprecedented in Peacetime A Reactive Core? The Configuration of Values in the British Electorate Brexit, British Politics, and the Left-Right Divide Coping with Trade-Offs: Psychological Constraints and Political Implications Coronavirus (COVID-19) in the UK Does Collective Interest or Self-Interest Motivate Mask Usage as a Preventive Measure against COVID-19? No conflicts of interest. Left-right (Evans and Heath, 1995; Evans et al., 1996) :• Government should redistribute incomes • Big business takes advantage of ordinary people • Ordinary working people do not get their fair share of the nation's wealth • There is one law for the rich and one for the poor • Management will always try to get the better of employees Authoritarianism-libertarianism:• Young people today don't have enough respect for traditional British values • For some crimes, the death penalty is the most appropriate sentence • Schools should teach children to obey authority • Censorship of films and magazines is necessary to uphold moral standards • People who break the law should be given stiffer sentences The first question to answer is how closely related attitudes to the coronavirus are to each other. Figure A1 shows the information scores for each coronavirus item across the scale. The three continuous measures provide high information across the scale, Figure A1 . Information scores for each coronavirus item across values of Theta (note that higher values of theta indicate more support for Coronavirus policies).