Rallying around the unwaved flag: national identity and Sweden’s controversial Covid strategy

Abstract During the Covid-19 pandemic, Sweden soon found itself alone and heavily criticised for its internationally deviant strategy. Why did this approach nevertheless receive widespread public support in Sweden? This article develops and tests a novel rally around the flag (RRTF) theory, which places national identity centre stage. In contrast to previous work, the article differentiates between national pride, national chauvinism, uncritical patriotism, national attachment, and ethnic nationhood conception – expecting only the first three of these to be related to supporting the Swedish Covid approach. Survey data from a politically representative sample of Swedes support the expectations. It is concluded that the popular support for the ‘Swedish experiment’ is to be understood as an RRTF – although in this case the flag was ‘unwaved’. This suggests that even national identities that draw their pride from institutional rather than cultural superiority lend themselves to RRTF processes.

internationally deviant handling of the pandemic (Kantar Sifo 2022). In this article we seek to understand why.
We argue that this support for the authorities' handling of the pandemic, and for the widespread narrative that lockdowns are dangerous and undemocratic, is the result of a rally around the flag process (RRTF) with three dimensions of national identity as the main mechanisms: national pride, national chauvinism and uncritical patriotism. Our study has major implications for the national identity and RRTF literatures. The latter has already sought to explain heightened trust in authorities and support for the incumbent in Sweden (Esaiasson et al. 2021) and elsewhere during the pandemic (cf. Bol et al. 2021;Schraff 2021) -but not previously demonstrated that even national identities that draw their pride from institutional rather than cultural superiority lend themselves to RRTF processes.
Note that when it comes to the pandemic, the Swedes have not just trusted their authorities, they have also largely supported their substantial policy approach -and this remained the case even two years into the crisis. For example, in the spring of 2021 a study showed that Sweden had a smaller share of people (33%) who were critical of how their country was handling the pandemic than any of the other European countries included. This placed Sweden close to South Korea, despite a five times higher national death toll per capita (Pew Research Center 2021).
These attitudes are puzzling for two main reasons. First, the Swedish strategy is far from an obvious policy success. In 2021, a third of the Swedes report that they have been impacted by the illness, either by themselves having fallen seriously ill or by suffering bereavement (to be compared with 19% in neighbouring Denmark) (Krastev and Leonard 2021). Both governmental and non-governmental agencies have presented highly critical reports of the Swedish handling of the pandemic. Moreover, the Swedish death toll per capita has superseded that of the neighbouring countries, which Sweden often places itself close to on other ratings. At the time of writing, in February 2022, Sweden's Covid deaths per capita are more than twice the size of Denmark's, and more than five times as many as Norway's (Our World in Data 2022).
Secondly, one of the foremost characteristics of this particular pandemic is that many ordinary citizens have been engaged in comparing death figures, policies and national leaders against those in other countries. Thus, in the UK, the other country that first also took a similar path to Sweden, by mid-March there was strong pressure in the public opinion to impose the lockdown that the government had sought to avoid (Baldwin 2021).
In Sweden, by contrast, the clearer it became that the country was left virtually alone and the more fiercely this path was questioned internationally, the more support it received in the Swedish polls. By the summer of 2020 a majority believed the country was handling the pandemic not only well but better than any other country, and especially among citizens with higher education, left-leaning political opinions, and high trust in experts (Jönsson 2021) -e.g. the very same group of voters who tended to support strict lockdowns in other European countries (Oana et al. 2021).
What, then, could explain this puzzle? Previous research on Sweden in the pandemic has sought to understand the Swedish governance model (Pierre 2020), media debate Nygren and Olofsson 2021), trust in and support for the government (Esaiasson et al. 2021;Johansson et al. 2021a;2021b;Nielsen and Lindvall 2021), and the emotions elicited by the Swedish strategy (Renström and Bäck 2021).
To our knowledge there are only two attempts, however, to explain the attitudinal roots of the popular support in Sweden for the country's handling of the pandemic. Both are very brief and in Swedish, pointing to either high trust  or a strong commitment to individual freedom (Bjereld and Demker 2020) as the main explanations. While we study how these attitudes relate to support for the Swedish Covid approach as well, we expect and indeed find that several dimensions of national identity are more important.
Our study contributes to several scholarly debates. First, it adds to previous research on Sweden's puzzling response to , and, second, more broadly, to the emerging discussion of the darker sides of Nordic national identities (Simonsen and Bonikowski 2020). Third, as we expand on in the next section, previous studies of RRTF in the pandemic have largely overlooked national identity as well as the more problematic sides of institutional trust. Ours places both of these centre stage, thus heeding the recent calls to study the role of national hubris for nations' Covid responses of (Lincoln 2020), to 'examine the impact of different types of identity appeals during a pandemic' (van Bavel et al. 2022), including 'distinct measures for nationalism and patriotism' , in the sense of typically exclusionary versus inclusionary views on the nation (Lim and Prakash 2021). Fourth, we also develop the existing theory of how national identity comes into play in RRTF processes even further (Feinstein 2018(Feinstein , 2020. Indeed, we undertake the first theoretical and empirical study of a rally around what, following Michael Billig (1995), we call the unwaved flag. We thus draw attention to the more sinister side of civic and anti-militaristic nationhood, which tends to fly under the radar of nationalism scholars who warn mainly against aggressive ethno-nationalism.
In the next section we survey the field, and develop our theory by combining insights from the literatures on RRTF and national identity. In section three we describe the Swedish case, suggesting that the media debate as well as elite rhetoric during the pandemic created a nation disrupting event in which Swedishness was at stake. In the fourth section, we present our data: a large scale Yougov survey of a politically representative Swedish sample, collected in February 2021 (N = 2000). In section five, we describe our main results: that institutional national pride, uncritical patriotism and national chauvinism predict support for the Swedish strategy, to a greater extent than either individualism or trust. In section six, finally, we discuss our findings and contributions.

Rallying around three dimensions of national identity
We define a 'rally around the flag' (RRTF) as the support that national leaders and other national symbols receive in times of major international crises, 'irrespective of the wisdom of their policies' (Kritzinger et al. 2021). Such rallies tend to appear earlier than information allowing for an assessment of the outcome and cost of the policies pursued (Feinstein 2018). The typical RRTF case is thus a military or terrorist attack leading to a surge of patriotic approval for the current president or prime minister (Dinesen and Jaeger 2013) and his or her policies (Baker and Oneal 2001). Since 'limited interest has been devoted to investigating the duration of the effect' (Johansson et al. 2021a), this particular issue remains unclear. Some studies find effects lasting for weeks or months only (Hetherington and Nelson 2003) while others speak of 'rally periods' (Feinstein 2016a) lasting over a year (Gaines 2002).
The Covid-19 pandemic has been connected to RRTF effects such as support for the incumbent, the country's pandemic policies and health authorities (Nielsen and Lindvall 2021), and trust in public institutions (Baekgaard et al. 2020), including in Sweden (Esaiasson et al. 2021). Some suggest these processes -including an increased satisfaction with democracy -resulted from lockdowns, and thus not an RRTF (Bol et al. 2021). Others object that they stem from rising national death tolls triggering collective Angst (Schraff 2021). It may also be that such rally effects and emotions vary by country (Kritzinger et al. 2021).
Threats of different kinds, moreover, elicit different policy reactions. One study found that while perceived realistic threat led to higher support for social distancing, perceived symbolic threat (to the national identity and American democracy) undermined this policy support (Kachanoff et al. 2021). However, research on these individual level mechanisms remains surprisingly scant (Schraff 2021). A recent overview calls for more 'systematic research on individual-level effect heterogeneity in the rally effect' (Hegewald and Schraff 2020).
Our study seeks to further our knowledge of these individual level mechanisms of the RRTF process. It also complements the aforementioned literature in two other respects: it places national identity centre stage, and reminds us that rally induced trust may be problematic. In the current pandemic, the F -as in Flag -has rarely been measured in studies of RRTF effects, even when recurrent references are made to 'feelings of national affiliation' , 'patriotism' , and 'in-group symbols' , not least in studies of Sweden (Esaiasson et al. 2021;Johansson et al. 2021aJohansson et al. , 2021b. Moreover, the predominant view of rally effects in the pandemic is positive, despite the fact that too much trust may counteract the citizen vigilance needed for democratic accountability Lenard 2008). As Devine et al. (2021) pointed out already at the onset of the pandemic, 'too much trust may lead citizens to (naively) believe that government is effectively managing the pandemic when it is not' . It may also undermine compliance with health regulations when trusted authorities signal non-compliance (Goldstein and Wiedemann 2022). In Sweden high trustors have been more reluctant to wear face masks (Johansson et al. 2021b) and those satisfied with the authorities' handling of the pandemic have also been less compliant with health-promoting recommendations (Renström and Bäck 2021).
In the following, we build on what we also know from numerous RRTF studies outside of the pandemic: a crisis itself is not enough, the leaders and media must also narrate it as a crisis for people's social identity (Kam and Ramos 2008). In particular we build on and expand the theory of Yuval Feinstein (2016aFeinstein ( , 2016bFeinstein ( , 2018Feinstein ( , 2020, who has brought together social identity theory with qualitative historical analysis of political rhetoric and media narratives, and survey data from various crises in the US and Israel. Feinstein's conclusion is that in order for a crisis to evoke a rally, it must be perceived as a 'nation-disrupting event' , which elicits a 'deep sense of distress among the public' , including fear (at the nation's perceived power loss) and anger (at the nation's perceived status loss). It is thus not the crisis itself that underpins rally effects, but a certain interpretation of it -via elite rhetoric, symbolic actions of the political leaders and the media -as presenting a symbolic threat to the national status. This happens when the crisis at hand is seen as 'pertaining to the "true" essence or virtue of the nation, which should be celebrated or requires protection' (Feinstein 2020). In sum, it is the experience of symbolic threat to the nation that buttresses an uncritical spirit of defensiveness vis-à-vis the symbols of the nation.
Such an interpretation of there being a threat to the nation's status will resonate more with some citizens than others, depending on their investment in the national identity at stake, both in terms of strength and type. Not all dimensions of national identity invite the sensitivity to status threat and hurt in-group pride that we expect to be crucial for rallies. Feinstein has investigated this in the Israeli case. He follows the tradition of distinguishing between what is sometimes called 'good' patriotism and 'bad' nationalism (Kosterman and Feshbach 1989), and finds that national pride does not lead to RRTF, while feelings of national superiority and ethno-nationalism, on the contrary do (Feinstein 2018).
While this is an important addition to the RRTF literature, we wish to develop it even further, based on the observation that the national identity literature increasingly speaks of several sub-dimensions (Gustavsson and Stendahl 2020;Huddy and Khatib 2007), the effects of which may be benign or malign depending on the situational context (Huddy and del Ponte 2020). In the following, we will thus distinguish between the following five dimensions of national identity, only some of which we believe affect our outcome variables.
1. National attachment (sometimes also called 'national belonging') centres on the cognitive aspect of seeing oneself as close to and part of the national in-group. 2. National pride picks out a more affective aspect of national identity, as it involves a positive emotion towards this identity and the achievements connected to it. 3. National chauvinism (sometimes also called 'hubris' or simply 'nationalism') is the stance that one's country and co-nationals are superior to others. In contrast to national attachment and pride, it thus involves a comparative element, and is sometimes also seen as containing a wish to dominate others. 4. Uncritical patriotism (sometimes also called blind patriotism), is sometimes collapsed with national chauvinism (Huddy and Khatib 2007), and highly correlated with it (Huddy and del Ponte 2020). Yet it is still distinct in theory, since in contrast to the former it does not exclude an acknowledgement that the country is in fact doing poorly, while it does take the stance that even if that is the case one should not openly acknowledge this. This dimension of national identity consists of uncritical allegiance to the country, an unwillingness to admit its errors, and an attitude of 'my country right or wrong' (Schatz et al. 1999). 5. Ethnic/civic nationhood conception concerns the criteria by which people specify the normative content of their national identity: ethnic markers (e.g. descent), versus more civic ones (e.g. law abidance).
As for RRTF effects, we follow Feinstein (2018) in not expecting any from dimension (1). If rallies are indeed bolstered by caring about the national status, and thus taking perceived threats to it deeply to heart, then merely feeling attached to one's nationality need not suffice, for this does not mean that one must also be invested in the national honour, and thus motivated to defend it in times of crisis by rallying around the national symbols. Indeed, Federico et al. (2021) found a negative relationship between this particular dimension and caring about the reputation of one's nationality, and it is clear from several studies that feeling attached to one's nationality is distinct from seeing it as superior to others, which is what status is ultimately about (Huddy and del Ponte 2020). Roccas et al. (2006) further discovered that this dimension is positively related to acknowledging group-based guilt -as opposed to suppressing it, for example by acting defensively whenever national symbols become an object of international critique.
In contrast to Feinstein (2018), however, we do expect an RRTF effect from national pride (2). National pride is particularly strongly related to institutional trust (Gustavsson and Stendahl 2020) and compliance with state authorities (Gangl et al. 2016), which lies close to the outcome we seek to understand in this article. We also know from a recent overview by Huddy and del Ponte (2020) that once a national identity threat arises, national pride -which is otherwise largely benign -invites uncritical support for national leaders. Finally, a Swedish study has shown that satisfaction with the Swedish authorities' way of handling the pandemic is indeed related to pride. The authors further suggested this was most likely pride of a hubristic rather than an authentic type, since this stance also had the anti-social effect of undermining compliance with health-promoting behaviour (Renström and Bäck 2021).
Note that taking pride in national achievements is sometimes shown to be different from being generally proud of one's nationality, and more prone to social desirability bias; in some countries saying one is proud to be X (nationality) is simply less politically correct than in others (Meitinger 2018;Miller and Ali 2014). In the case of Sweden, this may be especially relevant, since while Swedes score close to the EU average on other dimensions of national identity (including general national pride), they score high above (2.033) the EU average (1.493) on what we would call 'institutional national pride' , e.g. pride in the scientific and technological achievements of the country, the way the country's democracy works, and the country's economic development (see Online appendix B, Table 6). As we will see in the next section, in the public debate in Sweden on the country's approach to Covid-19 this approach was often presented as symbolising precisely these three achievements, while its critics and alternative approaches were depicted as threats to them. In this case, we thus expect this type of more specific national pride to also boost a rally effect.
We also expect an RRTF effect from national chauvinism (3) and uncritical patriotism (4), both of which seem likely to make a person sensitive to threats against the status of -or even just to critique ofthe national in-group. National chauvinism revolves around a feeling of national superiority, and the humiliating feeling that this superiority has been challenged is crucial to the rally process, according to Feinstein (2018). This dimension has also shown itself to be associated with militaristic aggression and out-group animosity (de Figueiredo and Elkins 2003). Uncritical patriotism makes critique and questioning of the national symbols feel like a threat, and has also been shown to be linked to perceptions of foreign threat, perceived importance of symbolic behaviours and selective exposure to information that makes one's country come across as attractive and successful (Schatz et al. 1999).
Finally, and again in contrast to Feinstein, we do not expect an RRTF effect from an ethnic as opposed to civic nationhood conception (5). What should matter is how emotionally invested one is in the national identity under threat, whatever its content. We suggest Feinstein's (2018) finding that an ethnic nationhood conception matters in the 2014 war in Gaza to be contingent on the ethno-national threat in that particular context. In other contexts, by contrast, liberals -who tend to disavow ethno-nationalism -have been shown to be as vulnerable to system justification bias induced patriotism as conservatives (Van der Toorn et al. 2014), and no less likely to rally around the flag in the face of collective anxiety (Porat et al. 2019). This suggests that in situations where the threat at hand is national -as opposed to specifically ethno-national -the ethnic dimension of nationhood should not be expected to invoke a rally effect.
In sum, we specify the following hypotheses: H1: National attachment will not be associated with support for the Swedish Covid approach.
H2: National pride will be positively associated with support for the Swedish Covid approach.
H3: National chauvinism will be positively associated with support for the Swedish Covid approach.
H4: Uncritical patriotism will be positively associated with support for the Swedish Covid approach.
H5: Ethnic nationhood will not be associated with support for the Swedish Covid approach.
According to our theory, however, in order for any RRTF to take place there first needs to be a crisis that is narrated as a symbolic threat to the national identity. We will now suggest this is what happened with the Covid-19 pandemic in Sweden.

Sweden and the pandemic media debate
Rally effects are typically studied in contexts of belligerent flag-waving, such as military crises in the US or Israel. However, in line with Michael Billig (1995) we believe it is high time also to study RRTF situations evoked by a more banal nationalism and in largely civic and anti-militaristic contexts, where rallies may fly under the radar because they are directed at largely unwaved flags.
We thus turn to Sweden, a country which has not been at war for two centuries, and where the population expresses less self-perceived cultural superiority than most other Europeans (Pew Research Center 2017). Indeed, the Swedes are frequently ranked as having the most civic nationhood conception in the world (Larsen 2017).
However, according to the social identity approach that we delineated in our theory section, even in a civic and peaceful country a rally effect should be possible -as long as people's national pride, chauvinism and uncritical patriotism come under threat. And on these attitudes Swedes do not in fact score lower than other Europeans. Table 6 in Online appendix B shows that Swedes follow the European average in terms of national attachment, general national pride and uncritical patriotism. Moreover, Swedes also score slightly above the European average on national chauvinism, and they stand out internationally on institutional national pride (e.g. pride in the scientific and technological achievements of the country, the way the country's democracy works, and the country's economic development).
While these data corroborate that they are indeed much less inclined towards an ethnic conception of nationhood than the average European, at the same time they suggest Swedes do not lack national hubris. It is just that in this national context, such hubris may be related to the more post-cultural and institutional aspects of the national identity of which Swedes are considerably proud. Indeed, in our survey data we find a positive and statistically significant correlation in Sweden between national chauvinism and holding a civic nationhood conception (see Online appendix F, Table 9).
Let us now turn to the debate over Covid-19 in Sweden. Media studies of it have shown that there was a 'burst of nationalism in the media' (Andersson and Aylott 2020) and that Swedish 'exceptionalism' was a major theme (Jerneck 2021) throughout 2020 (Baldwin 2021;Bjurwald 2021). 'The most common storyline, through which both defences and critiques were articulated, was the pressure on Sweden from the outside, which required the country to defend itself ' , conclude Nygren and Olofsson (2021).
We want to specify this further, suggesting the crisis was narrated not only as a threat for the Swedes' public health and the economy, but also for their national status and their main sources of institutional national pride. In the terminology of Feinstein (2020), we suggest this created a 'nation disrupting event' , with the opposition suspending its critique of the government in order to promote national unity, and prominent opinion leaders treating national leaders with reverence rather than critique, which was framed as divisive. Such a situation tends to evoke a pervasive sense that the 'core elements of the collective national identity are at stake, which in turn produces a deep sense of distress among the public' .
The key to this process was how in Sweden the seemingly dry and technical issue of how to handle a pandemic quickly became nationalised, thus mobilising the feelings towards the national identity that we have seen in Table 6. It became the Swedish way, both for its critics in the international press, and for its defenders.
State epidemiologist Anders Tegnell noted that 'while people in other countries need obligations in order to do things' , there is a 'cultural difference' in that voluntary recommendations are 'a form of Swedish culture' . The political editor of the left-of-centre newspaper Aftonbladet, Sweden's largest publication, went even further, defending the authorities' approach as 'what Swedishness is' , since it was based on the 'uniquely high trust' of the Swedes (Andersson and Aylott 2020). Those who opposed this strategy were called to stop and 'let Sweden be Sweden' , to quote the popular writer Jonas Gardell. Rallying around the PHA and their way of handling the pandemic became the Swedish thing to do, since it was understood as showing trust in government agencies, and supporting a strategy that built on such trust (Gustavsson 2021).
At the same time, 'banal' reminders of nationhood were commonplace. Members of government started wearing a small Swedish flag on their lapels. In 2020, the PM gave two televised 'speeches to the nation' , the third and fourth of this kind ever to have been given in Sweden. Both ended by appeals to people to behave responsibly 'for Sweden' . 'Our society is strong' , stressed the PM, immediately followed by 'Our government agencies toil day and night' . One of these agencies, The Swedish Institute, which promotes trust and interest in Sweden abroad, started producing weekly reports on how the Swedish strategy was depicted in the foreign media, and the potential effect on the image of Sweden abroad. Their results received large coverage in the national media. Citizens were reminded of these concerns when minister of foreign affairs, Ann Linde, received critical questions about the Swedish death toll in a German interview that went viral in June 2020. The minister replied with visible indignation: 'It sounds as if you want to throw us out of the EU!' (Gustavsson 2021).
State epidemiologist Tegnell came to personify the Swedish handling of the pandemic. Since early March he led televised daily press briefings which came to resemble a new national ritual, and were referred to as 'the entire Swedish people's campfire' . In April, the record number of viewers was 1.5 million, i.e. 15% of the population. On these occasions, Swedish reporters asked much fewer critical questions than their foreign counterparts . Several support groups for Tegnell were created on Facebook, including what was then Sweden's largest, with many prominent public figures in its ranks. Newspapers were filled with praise: 'a better general in this war could not be had' , Tegnell was 'our guide to the galaxy' , 'a superhero' , a 'national father' figure and the 'national soul incarnated' (quoted in Gustavsson 2021).
All this fits well with Feinstein's (2020) observation that 'when the media coverage of a war or security crisis takes on this celebratory character, the coverage charges the event with symbolic meaning, portraying it as a test of what most members of the nation perceive to be the nation's core elements such as profound solidarity among members, the common qualities of members (e.g. courage and determination), and shared values' . More specifically, the shared values that were now invoked each related to one of the main sources of national institutional pride in Sweden: the country's science, democracy and economy.
First, Tegnell and the Swedish approach to Covid-19 that he symbolised were described in terms of their scientificness. Tegnell himself declared that when other countries started locking down it was as if 'the world had gone mad' . In the Swedish press, other responses such as lockdowns were described as 'hysterical' and 'irrational' . The choice was perceived to stand between 'reasonable experts and panicking politicians' , and critics were dismissed as 'self-proclaimed experts' or 'hobby epidemiologists' . Opponents of Tegnell were portrayed as conspiracy theorists and tin foil hats driven by an overwhelming fear of death, even when they were scholars (Gustavsson 2021).
Secondly, the Swedish approach and the dominant role it gave to the experts at the PHA was perceived as characterised by its democratic openness. Not locking down was to safeguard 'the open society' , and doing so as pandering to authoritarian and populist urges. 'In many other countries the police have to run out on the street and beat people and force them indoors' , noted the Director of the PHA. Critics of the Swedish strategy were also often narrated as threatening Swedish democracy, by being aggressive and unstable, and potentially connected to Russian 'troll armies' . Criticising the strategy or the PHA was to 'undermine public authorities in a crisis' (Andersson and Aylott 2020). The minister of defense warned that any talk of Swedish authorities' presumed inability to handle the crisis might play into Putin's hands (quoted in Gustavsson 2021).
Thirdly, the Swedish Covid-19 strategy was perceived to safe-guard the Swedish economy in a situation where it was under severe threat. When the UK locked down, leaving Sweden alone, an interview with the CEO of the Third Swedish National Pension Fund received much attention. In it, she warned that even relatively mild restrictions would lead to an economic depression similar to Germany's in the 1920s, with mass unemployment and civil unrest. During the same time, the polls also show that Swedes became more worried about the economy than the risk of serious illness (Gustavsson 2021).
In sum, the Swedish Covid approach was narrated as an embodiment of -and its critics as well as any alternatives, such as lockdowns, as threatening -the three main sources of national institutional pride among the Swedes. These are Swedish science, Swedish democracy and the Swedish economy. The mentor of state epidemiologist Anders Tegnell, professor Johan Giesecke, exemplified all three concerns in his frequent media appearances during 2020, in which he said that the Swedish authorities followed the science, and further restrictions would threaten both the Swedish democracy and economy. He declared that all other countries were 'wrong' in their approach (Andersson and Aylott 2020).
According to our theory, if these signals of a national identity threat were picked up by the public, we should expect to find a relationship between national pride, chauvinism and uncritical patriotism, on the one hand, and support for the Swedish approach and a negative lockdown narrative, on the other, whenever a threat to the national status arose. We shall now argue that this threat was the clearest a year into the pandemic, when Sweden's death toll in Covid was at an all-time high, having been hit hard by a second wave that many had expected the country to avoid due to herd immunity.

Methods and variables
Note that the crisis we are expecting to set into motion the RRTF process for some citizens (those high on national pride, etc.) is not the pandemic itself, nor that of Sweden's covid approach becoming a national symbol, but the nation disrupting event of this symbol coming under serious threat and in need of defending. We thus conducted a survey on a politically representative sample of the Swedish population (n = 2019) during February 3-12, 2021, between the second and third wave of the virus, when it had become painfully clear that the positive effect of avoiding a second wave due to herd immunity (which the Swedish authorities had suggested earlier) would not materialise.
At this point, the vaccines had not yet been rolled out, the 'Delta' variant was on the rise, and the country had just reached an over-all peak in the death toll (see Figure 1). Younger patients were filling Covid wards, and many other countries had re-imposed lockdowns. Swedes were thus again reminded that their country was handling the pandemic differently, and this led to a fierce public debate, based also on a first, rather critical, report from December 2020 by the Corona Commission, a national inquest into the Swedish handling of the pandemic (Gustavsson 2021). All this, we suggest, was likely to lead to the feelings of anxiety and national humiliation that Feinstein deems crucial for the rally process. Indeed, our survey data from this time reveals that 20% of the respondents agreed that the foreign critique of the Swedish strategy felt 'like bullying and smearing' , and 35% that it made them 'worried about Sweden's reputation' .
The respondents were recruited via Yougov's online-based consumer panel where members receive emails about available surveys. Quota sampling was used to get a representative sample of the Swedish population in terms of gender, age, location and political party preference. As can be seen Table 7 in Online appendix D, the quota sampling resulted in a fairly representative sample, although highly educated respondents are overrepresented, and immigrants underrepresented. To be on the safe side, controls for age, gender, education and foreign background are included in all the models, and a control for political party preference is included in selected models since political data is missing for many observations. The inclusion of these controls does not affect the results to a meaningful degree. For the effects of these background characteristics on the dependent variables see Table 8 in Online appendix E.
Two main dependent variables are used to capture support for the Swedish Covid approach: overall confidence in the authorities' handling of the spread of the coronavirus (0-4, 0 = Very low, 4 = Very high), and the degree to which the respondents agree (0-4, 0 = strongly disagrees, 4 = strongly agrees) with five negative statements regarding lockdowns that were common in the public debate, such as lockdowns being undemocratic or dangerous to human mental health. The results remain largely the same if we recode the dependent variables into binary variables (1, 0) and run logistic regression instead of OLS (comparing Tables 1-3  below to Tables 17-19 in Online appendix K).
As our main independent variables, we use six variables capturing different sub-dimensions of national identity. All the variables are based on standard items from previous research and following this research they are treated as continuous rather than categorical (Gustavsson and Stendahl 2020;Huddy and del Ponte 2020;Huddy and Khatib 2007). Most of them are based on several statements to make the results less dependent on individual statements and increase generalisability. 'Uncritical patriotism' captures the degree to which the respondents on average disagree with four common statements such as 'there are certain phenomena in today's Sweden that I am ashamed of ' (cf. Huddy and Khatib 2007;Miller and Ali 2014). 'National chauvinism' captures the degree to which the respondents on average disagree with three common statements often used, such as 'Sweden is a better country than most other countries' (cf. Huddy and del Ponte 2019; Gustavsson and Stendahl 2020). 'Institutional pride' captures the degree to which the respondents on average felt proud over the way democracy works, the economic achievements, and the scientific and technological achievements of Sweden, while 'national pride' captures the degree to which the respondents overall felt proud of being Swedish. We use two different variables capturing national pride since taking pride in specific achievements is somewhat different from being proud generally of one's country, as already discussed in the theory section (see also Meitinger 2018;Miller and Ali 2014). 'National attachment' captures the degree to which the respondents felt close to Sweden (cf. Gustavsson and Stendahl 2020;Miller and Ali 2014). Finally, 'ethnic nationhood' captures the degree to which the respondents think that in order to be truly Swedish it is important to have been born in Sweden and have Swedish ancestors (cf. Gustavsson and Stendahl 2020). Since religion is less important in Sweden than most other countries, we chose to not include religion when measuring this more ethnic conception of national identity. A detailed description of all the variables and the survey questions they are based on is provided in Table 4 in Online appendix A.
Previous research has shown that our different dimensions of national identity are both theoretically and empirically separate from each other (cf. Gustavsson and Stendahl 2020;Huddy and Khatib 2007), and we also find their correlation coefficients to be generally below 0.5. The only exceptions are the moderate correlations between the two types of pride, which is to be expected, as well as between national pride and national attachment (0.5 < r, see Table 9 in Online appendix F for the correlation matrix). The Cronbach's alphas also show an acceptable level of internal consistency (α > 0.6) in regard to the most important indices (see Online appendix G, Table 10).
We also control for the two main alternative explanations that have been raised in previous research on Swedish attitudes to the country's pandemic approach: social trust ), here measured with the standard measure of generalised trust, and a commitment to individual liberty (Bjereld and Demker 2020), here measured with an individualism index inspired by Oyserman et al. (2002). Since the Swedish approach to Covid was often described as based on social trust and as prioritising individual freedom, one might expect high trusting as well as individualistically oriented persons in particular to approve of it, and expect it to be successful.

Analysis and results
We start by analysing the individual level relationships between the different dimensions of national identity and confidence in the authorities' handling of the pandemic (Table 1). The results show that uncritical patriotism, national chauvinism, institutional national pride, national pride and interpersonal trust are all associated with higher such confidence, while the reverse is true for individualism and ethnic nationhood. All eight coefficients are statistically significant at the 0.1 percent level. In line with our theoretical expectations, the largest coefficients (comparing standardised ones, see Tables 14-16 in Online appendix J for unstandardised coefficients) are those of uncritical patriotism, national chauvinism, institutional national pride and national pride. These variables also account for a major part (14-31%) of the observed variation in the dependent variable.
Uncritical patriotism and institutional pride account for more of the variation in the dependent variable than any other explanatory variable (e.g. trust, or individualism) or background variable (e.g. education level or voting for left-leaning parties, cf. Table 8 in Online appendix E). The size of these coefficients is considerable. For example, moving between the two extremes on the uncritical patriotism scale (0-4) is associated with moving from very low to very high in confidence in the authorities' handling of the pandemic (0-4) (for unstandardised coefficients, see Tables 14-16 in Appendix J).
The same general pattern can be found when looking at the results for the negative lockdown narrative index in Table 2. All correlation Table 1. effects of different dimensions of national identity on overall confidence in the swedish corona response (ols, standardised variables 0-1). coefficients (except for individualism) are again statistically significant at the 0.1 percent level. The largest coefficients are again found for uncritical patriotism, national chauvinism, institutional national pride and national pride. The latter three variables also account for the largest part of the variance in the dependent variable (between 7 and 9.5 percent of the observed variation), followed by trust (5 percent). At the same time, we note that both the coefficients and the adjusted r-squared are much smaller than in the previous tables, suggesting that none of the variables predict attitudes to lockdowns as well as they predicted confidence in the Swedish authorities' overall response. Finally, the results in Table 3 show that even when controlling for all the different dimensions of national identity, trust and individualism in the same model, the relationships between uncritical patriotism, national chauvinism, institutional pride and national pride, on the one hand, and confidence in the Swedish authorities' response, on the other, remain statistically significant below the 1 percent level (Model 1). When controlling for political party preference, all of these coefficients except the one for national chauvinism remain statistically significant (Model 2). The relationships between the independent variables and the lockdown narrative index are largely the same, however here the coefficient for uncritical patriotism is no longer statistically significant (Models 3-4). The results remain largely the same when also adding controls for risk aversion, concerns about effects of the pandemic on notes: all models include background controls for age, gender, education and foreign background. party controls: voted for left leaning parties. + p < .10, *p < .05, **p < .01, ***p < .001.
health and personal finances, as well as health status (see Table 23 in Appendix M).
In summary, hypotheses 2 and 3 are all given clear support, since national chauvinism, institutional national pride and national pride all have a strong relationship with the dependent variables that is independent from the other variables. Hypothesis 4 about uncritical patriotism also receives some support, although this is less strong. In addition, we also see strong independent relationships between uncritical patriotism, national chauvinism and institutional national pride and other related outcomes such as confidence in the government and in the PHA (see Table 12 in Appendix H). Interestingly, uncritical patriotism and national chauvinism also clearly suppress the negative relationship between being concerned about the effects of the pandemic on one's own or the family's health, on the one hand, and all our dependent variables, including confidence in the corona response, on the other (see interaction effects in Appendix L, Tables 20-22).
Looking at the other variables, the coefficients for national attachment, ethnic nationhood and individualism are not statistically significant in any models. Hence, in line with Hypotheses 1 and 5, these cannot independently explain confidence in the Covid response, or a critical stance towards lockdowns. This is not the case for trust, however, where the coefficients are statistically significant in all models. Hence while trust is arguably less important than many of the dimensions of national identity (based on the coefficients and the r2 value), it is still clearly an important predictor of the dependent variables.

Concluding discussion
We set out to understand the puzzle of why the Swedish public has largely supported the country's internationally deviant approach to Covid-19. Our results point to the conclusion that this resulted from a rally around what we call the 'unwaved' flag. By this we mean a process that takes place in major crises, and whereby national leaders and policies receive support as symbols of a national identity that is perceived to be threatened, irrespective of their rationality. In this case the flag was 'unwaved' , moreover, in that the authorities and policies that came to embody the national identity were not explicitly nationalist or belligerent -after all, they were a Public Health Agency, a state epidemiologist and a policy approach to a crisis that was not a military one -and that the national identity at hand revolved around pride in national institutions, rather than in cultural traditions or ethnicity. Nevertheless, trusting in and supporting the Swedish authorities and their handling of Covid-19 seems to have become the 'Swedish' thing to do in this crisis, and a way of defending supposedly threatened Swedish values, as well as a national reputation under attack. Hence the results demonstrate that even national identities that draw their pride from institutional rather than cultural superiority lend themselves to RRTF processes.
In contrast to most previous RRTF studies in the pandemic, we have placed national identity centre stage in both our theory and our data. To recapitulate, we built on Yuval Feinstein's theory of rallies as responses to symbolic threats to the nation. Based on recent research in national identity, we argued that a rally should manifest itself by a positive individual level link to support for the authorities from three dimensions of national identity more specifically: national pride, national chauvinism and uncritical patriotism -but not national attachment and ethnic nationhood.
Our empirical results fit well with the expected pattern. While demographic variables as well as social trust also matter, we find that national pride (both general and institutional), national chauvinism and uncritical patriotism are even more crucial, whereas national attachment and ethnic nationhood are not. Interestingly, prioritising the individual over the collective, or liberty over safety, turns out not to be related to more support for the Swedish Covid response; if anything, individualists tend to be less supportive. This may seem odd given the individualistic content of the Swedish Covid approach. However, it is in line with our interpretation of support for the Swedish strategy as the national norm, which would thus appeal less to individualists.
Our findings contribute to several research literatures. First, they provide crucial empirical knowledge on how the Swedes acted in the pandemic, something which has puzzled the world for over two years. While there have been studies of Swedish nationalism in the public debate and in media, ours is the first to suggest that these signals were also picked up in public opinion. Without the high public support for the Swedish strategy, it is far from certain that Sweden's internationally deviant strategy would have been politically viable.
Secondly, the Nordic countries and Sweden in particular are often held up as an example of how a trust-and consensus-based patriotism may help strengthen democratic citizenship (Berman 2006). Our findings temper this optimism, suggesting that even in such a context national pride, national chauvinism and uncritical patriotism lead to blind support for leaders and the branding of dissent as divisive. This is particularly problematic given that the policy approach the Swedes rallied around in this case came at a high price -especially for immigrants and other vulnerable groups, who suffered disproportionately more in terms of death and serious illness than natives (cf. Hooijer and King 2022). The civic chauvinism we have found in Sweden may also help explain the recent finding that in the Nordic countries civic nationhood is related to islamophobia (Simonsen and Bonikowski 2020).
Thirdly, our findings complement the literature on RRTF during the pandemic, which has given scant attention to national identity, and largely welcomes rallies and the trust in authorities. We are more skeptical, given that our findings remind us that blind trust can be counterproductive in a pandemic Jennings et al. 2021;Johansson et al. 2021b), and also stands in stark contrast to democratic citizenship, which requires vigilance and the ability to criticise authorities (Lenard 2008;O'Neill 2018).
Fourthly, our study provides an important addition to the research front on nationalism and national identity, which especially during the pandemic has tended to equate the negative sides of national identity to ethno-nationalism (Elias et al. 2021) or blatantly hubristic tendencies like collective narcissism (Federico et al. 2021). Even Feinstein's sophisticated theory of how different dimensions of national identity come into play in rally situations has overlooked the risks of what we call rallying around the unwaved flag (Billig 1995). In contrast to this, we have argued that the nation disrupting situation that took place in Sweden during the pandemic invoked mainly civic, subtle and peaceful national values, such as trust in Swedish authorities and in science, support for Swedish democracy, and safeguarding the Swedish economy -but that even so, national pride, hubris and blind patriotism led to an RRTF process.
A drawback with our design is that we cannot tell if this RRTF was sparked by the high death toll and infection numbers, etc., right before our survey in early February 2021, or if we would have found similar results also earlier in the pandemic -and, if so, if we are dealing with several rallies or one rather tenacious one. We thus urge future research to look further into the understudied question of the duration of rallies. Another problem is that we only have cross-sectional data, and thus cannot disentangle the direction of the relationship we have found between certain dimensions of national identity, and trust or support for authorities. We cannot exclude that these two types of attitudes mutually re-inforce one another, as suggested when RRTF are said to include 'a surge of patriotism' (Baker and Oneal 2001). Our overarching assumption, however, has been that feelings of national identity largely precede trust in and support for the Swedish handling of the pandemic, as opposed to vice versa. The RRTF and national identity literatures show that the dimensions which we see are related to rallying around the Swedish approach -e.g. national chauvinism and national pride -tend to be rather stable and closely related to personality (Huddy and del Ponte 2020).
We also believe it would be fruitful for future research to study how these dimensions of national identity behave in largely civic and peaceful contexts like the Nordic countries. It may be that the problems they create in that context escape our attention as long as we only look for flags that are belligerently waved.

Data availability statement
Our quantitative data can be provided on request to enable full replication.