When and how people perceive threats

‘Panic’ is a popular view of how people respond in emergencies.^{Reference Clarke7} There are various definitions of panic, but one thing that distinguishes the concept from related notions such as fear and flight is the idea of overreaction to a perceived threat.

A basic empirical problem with the concept of panic in the context of mass emergencies is that of measurement:^{Reference Sime and Canter8} in an emergency, how do we know if fear is ‘excessive’? At the time, what counts as an overreaction is typically a subjective judgement. Post hoc, it is easier to judge, but then it is no longer a description of someone's mental state and is not explanatory.

For example, at the beginning of the COVID-19 crisis, was the extra shopping that some people carried out – so-called ‘panic buying’ – really excessive? On what criteria? If people believe (a) that they may have to stay home for prolonged periods of time, or (b) that other people will soon empty the shelves, or both (a) and (b), it is logical to buy additional goods. It may be excessive from the point of view of the observer, but psychologically it might be an understandable reaction to a perceived threat.

The panic concept is sometimes used to claim that public overreaction to signals of danger is a major cause of death in emergencies.^{Reference Forsyth9} But most sources suggest that people are more likely to die from ignoring or reacting too slowly to indications of danger than they are from overreacting.^{Reference Canter6} Indeed, public underreaction to signals of threat in emergencies and disasters is common.^{Reference Atwood and Major10} This pattern of response has sometimes been characterised as evidence of an ‘optimistic bias’.^{Reference Kinsey, Gwynne, Kuligowski and Kinateder11} But there are a number of group-level psychological factors that determine the extent to which people respond to (versus discount) signals of impending emergency.

The first factor is the historical context. For example, if there has not been an earthquake locally in 100 years, people may not see the point in engaging in preparedness. But when they have had experience of quakes and other disasters, preparedness activities increase.^{Reference Becker, Paton, Johnston, Ronan and McClure12} In fact, signal detection theory^{Reference Wormwood, Lynn, Barrett and Quigley13} would suggest that, given the frequency and magnitude of genuine threats, the belief that ‘it won't happen to us’ can actually reverse, and people become extremely vigilant about potential signals of danger. For example, in London in 2017, following a series of terrorist attacks earlier in the year, hundreds of people in Oxford Street fled in fear from a noise that turned out to be harmless.

The second factor is self-relevance. Self or identity exists at many levels, from personal identity (e.g. ‘I’ as an individual with unique personal characteristics) to the many social identities that are important to us based on our group memberships (e.g., national identity [‘us’ as British people’] or religious identity [‘us’ as Muslims]). The social categories we belong to can affect whether we perceive threats as relevant,^{Reference Stapel, Reicher and Spears14} and can facilitate either adherence to protective behaviours or the adoption of risky behaviours.^{Reference Cruwys, Stevens and Greenaway15} If the threat of COVID-19 is understood principally in relation to the individual, then those individuals least at risk (those young and healthy) may well feel it is unnecessary to change their behaviours. Indeed, they might also feel they have a right to take risks with their own health.^{Reference Reicher and Drury16} Thereby they continue to act in ways that put others at risk of infection. This perceived lack of self-relevance may be one reason why younger people have been among those with lower levels of adherence with public health regulations such as staying at home during ‘lockdown’.^{Reference Foster17} By contrast, where people perceive ‘us’ as being at risk, they are less likely to discount that risk.^{Reference Stapel, Reicher and Spears14}

Most information about impending threat in a potential or actual emergency is indirect, or socially mediated: people do not see flames, rather they hear a fire alarm or they see others escaping. The third factor in perceiving threat is therefore communication about that threat, which needs to come from a trusted source. A source that provides too many false positives can lead to false negatives in behavioural response, as in the poor public response rate to fire alarm bells.^{Reference Proulx18} Conversely a trusted source can convince there is a threat, even when it is novel and invisible. In the early weeks of the pandemic in the UK, trust in the government was high,^{Reference Fancourt, Bu, Mak and Steptoe19} as were public concerns about the threat of the virus²⁰ (and those who denied the threat, for example, by drawing upon conspiracy theories or through mistrust of science,^{Reference Biddlestone, Green and Douglas21} were in the minority).

Communication and hence influence regarding threat occurs not only through deliberate attempts (such as alarms and warnings), but also through observed examples of others’ conduct. What our peer groups, family, community and others appear to feel safe doing can serve to convey what it is safe for us to do. In all these cases, trustworthiness is linked to shared identity. When the source is seen as one of ‘us’, we are more likely to see their messaging and behaviour as relevant, more likely to listen, and more likely to follow their instructions or their example. This has been demonstrated in both ‘normal’ life^{Reference Turner22} and in relation to the COVID-19 pandemic.^{Reference Haslam, Jetten, Reicher, Haslam and Cruwys23}

Adherence to the required public health behaviours

Most of the required public health preventive behaviours during the COVID-19 pandemic involved some sacrifice of personal freedoms. Staying home and other aspects of lockdown were costly and onerous for most people, and much more for some groups than for others.^{Reference Templeton, Guven, Hoerst, Vestergren, Davidson and Ballentyne24} Before lockdown in the UK, some doubted whether the public had the mental strength to endure these privations over time. This was most clearly expressed in early March by the Chief Medical Officer, who stated ‘There is a risk if we go too early people will understandably get fatigued and it will be difficult to sustain this over time’.^{Reference Conn, Lawrence, Lewis, Carrell, Pegg and Davies25} This notion of ‘fatigue’ as a form of psychological frailty^{Reference Reicher, Drury and Stott4} was quite separate from the expected stresses of quarantine^{Reference Rubin and Wessely26} and the mental health toll for many people (contrary to some claims, the notion of ‘behavioural fatigue’ was not suggested by behavioural scientists advising the UK Government; rather, those participating in SPI-B, the SAGE subgroup on behaviour, criticised the concept^{Reference Conn, Lawrence, Lewis, Carrell, Pegg and Davies25,27} ). This contemporary concern with public mental fragility parallels government fears in the Second World War of public panic, shelter mentality and widespread breakdown in response to the Blitz.^{Reference Jones28} In both cases, there was a debate around the extent to which coercion would be needed to enforce public adherence.

Research on behaviour in other public health emergencies has shown that people would be willing to undergo prevention and mitigation measures at high personal cost, where these are perceived as legitimate public health interventions and where the relevant authority is respected and is seen as in-group (rather than as ‘other’). A good example is mass casualty decontamination, which is a procedure undertaken to remove contaminants from the skin of a potential casualty in the event of a chemical, biological, radiological or nuclear incident. Because the procedure can involve removing clothes in a public place, those affected may perceive decontamination as more threatening than the incident itself. Reviews of the literature find that any failure by responders to consider casualties’ psychosocial needs can lead to failures of the procedure, including people leaving the scene still contaminated and therefore endangering their families and communities.^{Reference Carter, Drury, Rubin, Williams and Amlôt29} The reviews also showed that the use of coercive measures (threats, force, shouting) had a backfire effect rather than enhancing effective public engagement. However, when responders managing the procedure explained the importance of decontamination and provided regular updates about their actions, this increased perceptions of the legitimacy of the procedure among casualties. In turn, this enhanced shared identification between emergency responders and members of the public. Shared identification predicted: reduced public anxiety;^{Reference Carter, Drury, Amlôt, Rubin and Williams30} greater public adherence with the procedure, cooperation,^{Reference Carter, Drury, Amlôt, Rubin and Williams30,Reference Carter, Drury, Rubin, Williams and Amlôt31} and greater speed and efficiency of the decontamination process.^{Reference Carter, Drury, Rubin, Williams and Amlôt32} Subsequent research has demonstrated that effective engagement with the decontamination procedure is also increased when casualties perceive the effectiveness of the measures.^{Reference Symons, Rubin, Carter and Amlôt33}

Although staying at home and distancing more generally during lockdown were experienced as onerous, adherence rates were very high in the UK. (Adherence rates for self-isolation were much lower.^{Reference Smith, Amlôt, Lambert, Oliver, Robin and Yardley34} As well as the commitment, this is likely to be because of material constraints, such as not having the financial support or support for shopping needs, and not understanding what self-isolation actually requires.^{Reference Webster, Brooks, Smith, Woodland, Wessely and Rubin35,36} ) This was the case on both behavioural measures^{Reference Aguilar-Garcia37} and self-report measures across multiple sample.^{Reference Fancourt, Bu, Mak and Steptoe19,38} These high rates of adherence suggest that the public were indeed psychologically able to make the required sacrifices.^{Reference Drury, Reicher and Stott39}

A number of studies have now looked at predictors of this adherence. In addition to the perceived effectiveness of public health measures,^{Reference Clark, Davila, Regis and Kraus40} many studies have identified group processes including social identification, similar to those evidenced in the decontamination research. Thus the Office for National Statistics data showed that at the same time that adherence rates were high during the strict lockdown, there was a strong sense among members of the public that people were becoming more united,^{Reference Parveen and McIntyre41} suggesting a possible association between the two. The association between group processes, such as shared identification, and adherence to (or support for) public health measures, has now been demonstrated in cross-sectional and longitudinal survey studies as well as in experiments. Thus predictors include perceived norms within valued social groups,^{Reference Goldberg, Gustafson, Maibach, van der Linden, Ballew and Bergquist42} identification with one's community,^{Reference Vignoles, Jaser, Taylor and Ntontis43} sense of public duty,^{Reference Everett, Colombatto, Chituc, Brady and Crockett44} empathy with vulnerable groups,^{Reference Pfattheicher, Nockur, Böhm, Sassenrath and Petersen45} the belief that ‘we're all in it together’ (which was found to be more important than threats of coercion),^{Reference Jackson, Posch, Bradford, Hobson, Kyprianides and Yesberg46} social capital,^{Reference Sharkey47} national identification,^{Reference Van Bavel, Cichocka, Capraro, Sjåstad, Nezlek and Alfano48} horizontal collectivism^{Reference Biddlestone, Green and Douglas21} and belief in shared values of security and responsibility.^{Reference Wolf, Haddock, Manstead and Maio49} Overall, these factors evidence the importance of a sense of collectivity, at the group, community or national level, that has driven adherence behaviours.

There has been some shaming and stigmatising of those who are perceived not to be adhering.^{Reference Prosser, Judge, Bolderdijk J, Blackwood and Kurz50} However, structural inequalities have been shown to affect ability to comply.^{Reference Templeton, Guven, Hoerst, Vestergren, Davidson and Ballentyne24} In the same way that disadvantaged demographics are overrepresented in the infection and death figures,^{Reference Conn, Lawrence, Lewis, Carrell, Pegg and Davies25,51} some groups have been less able to distance and were obliged to go into work on sometimes crowded trains, and were limited to busy public spaces when taking exercise.

As lockdown in the UK eased over May, June and July 2020, there was evidence that public adherence to the rules declined.^{Reference Fancourt, Bu, Mak and Steptoe52,Reference Fancourt, Bu, Mak and Steptoe53} Other developments over the same period suggest possible explanations for this decline more in line with the factors described above than behavioural fatigue.

First, the official messaging changed. In May, the official slogan ‘stay at home’ was replaced with ‘stay alert’, shifting from a clear behavioural imperative to a somewhat vaguer instruction on how to feel. There were also a large number of government announcements on changes to the rules in May, June and July. There is some evidence that official communications were becoming less effective both in communicating risk and in providing information on mitigations. Over this period, perceptions of being unsafe reduced⁵⁴ and public knowledge of the rules declined.^{Reference Fancourt, Bu, Mak and Steptoe55} Further, the most significant relaxation of lockdown – which included being allowed to visit a pub, cafe or restaurant, museum, library, theme park, cinema, hotel, hairdresser or barber – that took place on 4 July, appeared to have a strong signalling effect (not least, perhaps, because these were heralded in the mass media as ‘freedom’ and as the ‘end of lockdown’). Thus, surveys report drops in people's reported adherence to physical distancing shortly afterward.^{Reference Fancourt, Bu, Mak and Steptoe53,56}

Second, there were some significant changes in terms of national unity and the public's relationship with the government that interacted with these problems with communication. A number of sources show that the sense of national unity that was evident at the start of the pandemic declined over time.^{Reference Duffy and Allington57} The notion that ‘we are all in it together’ was flatly contradicted by the evidence that structural differences between groups – the greater economic losses in terms of jobs and pay for some and not others, and the mounting evidence of particular risk among ethnic minorities⁵¹ – was being exacerbated by the pandemic.

A significant feature of this decline in a sense of national unity was a loss of trust in the UK Government. In particular, lack of government repentance over the actions of one special advisor, who admitted breaking the rules on lockdown, served to communicate that there were one set of rules for the privileged and another set for everyone else who had made the sacrifices of staying at home and not visiting family; this decline in trust was in turn associated with a drop-off in adherence.^{Reference Fancourt, Steptoe and Wright58} (Other research^{Reference Jackson, Bradford, Yesberg, Hobson, Kyprianides and Pósch59} suggests that there were two types of public response to this incident. First, there was cynicism about the rules, which was associated with lower levels of physical distancing adherence. Second, those who were angry about the actions of the government advisor were less likely than others to believe that it was acceptable to bend the rules, but more likely than others to comply with the guidelines themselves. This latter pattern is in line with Stephen Reicher's suggestion that the advisor would be seen as an anti-role model.^{Reference Turk60})

Neighbourhood support and community mutual aid groups

Government policies such as the community resilience programme^61,62 are a recognition that public involvement, in the form of active support for others affected, is a necessary part of emergency response. (Although governments started officially acknowledging this after 9/11,^{Reference Drury, Novelli and Stott63} some disadvantaged groups have long made their own resilience plans in the knowledge that the state will not provide them with the support and resources they need in times of crisis.^{Reference Charles, Bulley, Edkins and El-Nany64}) Research on emergencies and disasters shows that such public involvement is common.^{Reference Drury65} Of course, not every member of the public provides support; and some emergency events evidence more support behaviours from the public than others. But more lives are saved by the ‘average’ citizen, whether ‘bystander’ or fellow survivor, than are saved by professionals.^{Reference Drury, Carter, Cocking, Ntontis, Tekin Guven and Amlôt5}

How does this public support come about? Research on a variety of emergencies and disasters suggests that an underlying mechanism is that of a shared social identity among those affected.^{Reference Drury65} Shared social identity facilitates support in three ways. It motivates people to give support to others (since their interest is seen as ‘our’ interest).^{Reference Drury65} It increases expected support within the in-group (and hence enables a coordinated response).^{Reference Drury, Novelli and Stott66} And it means that offers of help are perceived as genuine rather than reflecting ulterior motives (which means that the help is more likely to be accepted).^{Reference Haslam, Haslam, Cruwys, Haslam and Jetten67} Most group identities reflect long-standing group memberships; hence, existing social capital is one of the major factors in community resilience.^{Reference Aldrich and Meyer68} But an emergency can also create a completely new group identity or sense of ‘we-ness’ among those affected, based on common fate.^{Reference Ntontis, Drury, Amlôt, Rubin and Williams69} Further, in this context, the supportive behaviour of fellow in-group members can serve to define supportive group norms, and identification with the group increases the influence of such norms.^{Reference Drury, Brown, González and Miranda70}

At the start of the COVID-19 pandemic, there was a massive upsurge in mutual caring behaviours among members of the public.^{Reference Monbiot71} Mutual social support among members of the public took different forms, partly reflecting community members’ differing needs. Moreover, the requirement for some to self-isolate at home meant that there were multiple simultaneous needs. Neighbours helped neighbours with shopping, collecting medical prescriptions, dog walking, providing information and emotional support.^{Reference Hogan72} For example, in one week in April 2020, the Office for National Statistics survey found that nearly two-thirds of adults surveyed said other local community members would support them if they needed help during the pandemic, and over one in three adults said they had done shopping or other tasks for neighbours in the previous week.⁷³ Identification with the community has been shown to be a predictor of helping neighbours in these ways.^{Reference Vignoles, Jaser, Taylor and Ntontis43}

In addition to informal interactions between neighbours, a very large number of organised mutual aid and community support groups were set up. One estimate made in May 2020 was that there were 4300 such groups connecting an estimated 3 million people in the UK.^{Reference Butler74} These ranged from Facebook groups (which acted simply to connect requests for help with offers from volunteers) to those groups that cooked food and carried out deliveries themselves. One research study concluded that such groups have been crucial in the society's response to the pandemic.^{Reference Tiratelli and Kaye75}

A number of studies are now investigating the psychology of volunteering and mutual aid during the pandemic. A survey by Abrams et al^{Reference Abrams, Lalot, Broadwood and Platts-Dunn76} found that, compared with other participants, those who had volunteered to help others in the context of the pandemic reported higher trust in others to follow the guidelines, higher trust in the government, higher compassion for people living in their local area and stronger connections with their family, friends, colleagues and neighbours. Although some mutual aid groups have been ‘emergent’, reflecting new relationships among participants,^{Reference Tiratelli and Kaye75} many others are extensions of existing community groups and are based on existing social capital.^{Reference Tiratelli and Kaye75}

Previous research on ‘altruistic communities’ in other kinds of disaster^{Reference Fritz77} suggests that these groups decline over time as people run out of emotional and physical resources or the state steps in.^{Reference Kaniasty, Norris, Gist and Lubin78} There has been only limited research on how such groups sustain themselves over time. But it would seem that some of the factors that can help include a group identity, a place to meet and talk about the group and its aims, commemorative events, support (but not co-option) from local authorities and alliances with other groups.^{Reference Drury, Carter, Cocking, Ntontis, Tekin Guven and Amlôt5,Reference Ntontis, Drury, Amlôt, Rubin and Williams79}