Impact: From riots to crowd safety

In the first of an occasional series, John Drury describes his pathway to impact.

Meet the author

‘My Impact Case Study – submitted to the British Psychological Society’s portal – seeks to recognise and understand collective resilience in crowds of survivors. The research covered in the case study was carried out with my colleagues Steve Reicher (University of St Andrews), Chris Cocking (Brighton University), Damian Schofield (State University of New York), Andy Burton (Nottingham Trent University), Paul Langston (University of Nottingham), David Novelli (University of Hertfordshire), and Clifford Stott (University of Leeds), and so this journey includes them too.

The Impact Case Study is concerned with the collective psychology of emergencies and disasters. My principal focus has been how these events affect (or sometimes create) crowds. In fact, it is crowds that fascinate me, and the work on emergencies and disasters was an extension of research on a different type of crowd event. Here I explain how that fascination began.’

One day, in March 1990, when I was in my first year as an undergraduate student, I heard on the radio breaking news of a huge riot taking place in London. This seemed very important for two reasons. The first reason was that politically many people seemed to think that riots were irrelevant, ineffective and even counter-productive for any cause. Yet the significance of the poll tax riot for both the anti-poll tax movement and the legitimacy of the Conservative government’s ‘flagship’ policy soon became evident. The movement maintained its momentum and an announcement was made only a year or so later that the hated tax was to be abolished. The event therefore had a profound effect on the political landscape and certainly galvanised many of those who wanted change.

The second reason why the riot was important was that I had just been introduced to the topic of crowd behaviour on my degree course. I was in the middle of reading the relevant literature when the riot took place. In particular, I had just read Steve Reicher’s paper on the St Pauls riot which was published in the European Journal of Social Psychology (Reicher, 1984). I loved the way that in this paper Steve not only outlined a new model in which crowd behaviour was understood as meaningful and identity-based, but also put the boot hard into those established theories that pathologised the crowd – from Le Bon’s ‘submergence’ to Zimbardo’s de-individuation theory. I remember wishing that I had written that paper. But I also felt excited by the fact that the topic of riots could be the focus of academic study.

The poll tax riot was the subject of my final-year empirical project. I wanted to emulate the St Pauls study. But after starting my PhD under Steve Reicher’s supervision, the research emphasis shifted away from the limits of crowd behaviour to issues of change and empowerment, both within and beyond the crowd. On the one hand, crowds express people’s social identities: the contours of crowd behaviour reflected people’s identity-based understandings of self and world. This was what the St Pauls study had shown so vividly. But, on the other hand, crowd events could also alter the understandings and identities of the participants. People became empowered (or disempowered). The boundaries of the group sometimes changed. And sometime the very issue that the protest was about changed within the event – for example, from student fees to the right to protest itself. These psychological changes within an event could help explain why some crowd events changed over time – from peaceful to conflictual, for example. They also helped explain why people sometimes were transformed as individuals through their experiences in crowd events and became ‘different’ people.

We felt that we had some success in displacing irrationalism in the study of crowd conflict, and re-configuring the field as now part of collective action. We wanted to extend some of the ideas to another crowd domain. Paralleling the ‘mob mentality’ of irrationalist account of riots, notions of ‘mass panic’ pathologised collectivity in emergencies. And like the notion of the ‘mad mob’, these ideas were pervasive. We were often asked to comment on cases of ‘crowd panic’, but didn’t yet have the systematic research evidence to substantiate the arguments we were making. Our aim therefore was to apply some of the social identity principles to the domain of mass emergency behaviour.

An emergency plan that changed course
Our original research plan was to compare examples of two types of emergency evacuations – one in which the crowd ‘panicked’ and one in which the evacuation was orderly – and then see if social identity processes explained the difference. The hypothesis was that in the orderly crowd there was a shared identity, whereas in the ‘panicking’ crowd, there wasn’t. However, several things occurred that changed this plan somewhat and led us into slightly unexpected directions.

First, once we started the research we struggled to find many instances of ‘panic’ to study. We advertised in order to recruit interviewees who had been involved in emergency evacuations. But most of them described at least some instances of cooperation and order emerging in the evacuating crowd. We couldn’t find clear instances of crowd ‘panic’.

Second, I discovered a rich literature on mass emergencies and alternative approaches to understanding them. Most of this literature wasn’t in social psychology, but some of the ideas could be extended to develop a social psychological analysis that applied specifically to crowds. Some of the best conceptual and empirical critiques of the concept of ‘panic’ I read at this time were in papers by Norris Johnson, Charles Fritz and Jonathan Sime. And then, following 9/11, there emerged a powerful critical analysis of the meaning and implications of psychological vulnerability versus resilience in emergencies; the arguments of Russell Dynes, Simon Wessely and Frank Furedi were particularly inspiring. One of the exciting things about these arguments was the way they showed that notions of collective vulnerability – as in the discourse of ‘mass panic’ – were not simply an (incorrect) description of behaviour, but operated as the justification for policies and practices of control and coercion that could undermine ‘natural’ tendencies to resilience. As has been argued by discourse analysts (another early inspiration), social constructions do things.

There was one experience in particular that brought many of these issues together, and that was the 7 July London bombings in 2005. Chris Cocking, Steve Reicher, Andy Burton, Damian Schofield and I were presenting the research on mass emergency behaviour at the Royal Society Summer exhibition in London. On the day of the bombing, Chris and I were at Waterloo East tube station on our way to the exhibition venue. (Somewhat presciently, one of our exhibits was a simulation of an evacuation of London tube station during a fire.) Just before 9am, we were on the busy platform waiting for a train when we were evacuated. At the time, we didn’t know that there had been terrorist attacks. We followed the grumbling crowd making their way out of the station, annoyed by the inconvenience. The rest of our journey across London was by foot.

Some hours later, we heard about soot-covered people emerging from tube stations and then about the deaths and injuries. As well as the shock and sadness for those who had been killed, I realised that it was important to try to look at how survivors behaved during the event. Since the crowd on the tube is one of the best examples of a crowd lacking a shared identity, then we might have expected disordered selfish behaviour as people affected by the bombing tried to evacuate. And yet our preliminary findings based on interviews with survivors of other emergencies led to a hypothesis that was actually closer to some of the ideas we developed in the collective action research: identities can change in crowds.

Soon there were many detailed reports of cooperation, orderly queuing and even self-sacrifice among survivors of the bombings that made any notion of ‘mass panic’ during 7 July impossible to sustain. People shared water with each other, they tied tourniquets, and they tried to support each other emotionally as well as practically. In most cases, this support took place amongst commuters who were complete strangers to each other.

When we interviewed survivors and analysed their accounts plus the huge amount of secondary data produced in the wake of the event (including quotes from witnesses in media reports, personal accounts and detailed statements in the London Assembly report), it gave support to the idea that survivors had shifted from seeing themselves simply as ‘me’ (in relation to other commuters) to ‘us’ (the group of people affected by the explosion). And this strong sense of ‘we-ness’ was associated with cooperation and concern for others.

The theoretical focus in the research on mass emergency behaviour then became this question of shift or change in social identities. Some crowds affected by emergencies already have one (or more) shared identities – the crowds of supporters at the Hillsborough disaster is an example. But in these cases the meaning of the ‘we’ seems to change, along with the boundaries. Liverpool fans we spoke to said the crowd identity was no longer a matter of team allegiance but a common ‘humanity’. Other crowds, like those at transport hubs, shopping centres and so on, are what Steve Reicher has dubbed physical crowds (Reicher & Drury, 2011): they are simply people in the same place at the same time. But, under certain conditions, they can become psychological crowds (‘us’, ‘we’), something that has a number of important implications for how one perceives and behaves towards others in the crowd. Shared understandings, trust and the motivation to help all seem to increase when this shift occurs. Of course some crowd evacuations are disorderly and characterised by individualistic behaviour, but many more exhibit this emergent sociality that serves to protect people within the crowd.

So instead of ‘panic’ another concept and terminology was needed. ‘Resilience’ seemed to be a good way a way of talking about adaptive crowd behaviour in emergencies, but what did that mean in this context? In many definitions, ‘resilience’ means the preservation of an existing entity in the face of adversity; but the psychological crowd was created not preserved in the cases we looked at. In many definitions, ‘resilience’ refers to a disposition or attribute; but how could that be the case for a fleeting phenomenon like a crowd forming in an emergency?

Moreover, just as notions of psychosocial vulnerability (such as public ‘panic’) have been used to justify policies we might see as coercive and illegitimate – as Dynes (2003) had argued, for example – discourses of ‘resilience’ too have been used in various ways not all of them beneficial. Notions of resilience are evident in policies like the ‘Big Society’ and serve to justify cuts to services; and in the Second World War a similar notion was implied in discredited propaganda such as ‘Britain can take it’ [see also the forthcoming Edgar Jones ‘Looking back’ article on citizens at war].

But what interested me about notions of resilience when applied to ‘communities’ in UK policy since 9/11, including in the new civil contingencies framework, was that the approach is something of a double-edged sword for the state. On the one hand, the need for ‘community resilience’ is now clearly recognised, in the sense that the government acknowledges that the professional responders will not reach the emergency in time or in sufficient numbers and therefore the public must deal with the emergency themselves – at least to some degree. On the other hand, the more independent, empowered and active these ‘communities’ are, the more there is the possibility that they might develop in all sorts of unpredicted directions and do things the government hadn’t anticipated or desired. It’s also interesting to note in this context that the government’s guidance on ‘community resilience’ (2011) lists crowds (‘communities of circumstance’) alongside geographical communities and others as types of ‘community’. In a sense, therefore, the fashion for ‘resilience’ at the time was an opportunity to legitimise autonomous action in emergencies as a good thing, taking advantage of that double-edged sword.

Moreover, in psychology there were other ways of talking and thinking about ‘resilience’ (such as ‘bouncing forward’ instead of simply ‘bouncing back’), so I chose the term, prefixed by ‘collective’, as a way of referring to a particular set of behaviours and perceptions in order to create a new way of talking about crowd psychology in emergencies. Thus we defined collective resilience as ‘the way a shared identification allows groups [and crowds] of survivors to express solidarity and cohesion, and thereby to coordinate and draw upon collective sources of support and other practical resources, to deal with adversity’ (Drury et al., 2009, p.502).

Talking to practitioners
At this point, there was another factor that went to shape the work and led indirectly to the impacts described in the case study. One of the things I liked about academia was the fact that we can study things that don’t necessarily have a use. I had certainly not sought to translate the research on rebellious crowds into something ‘useful’. It was important knowledge for its own sake, to my view. Yet with the research on mass emergencies there was suddenly an applied interest and a demand for the ‘implications’ of the work. People asked me what our findings meant for practice. Chris Cocking, who was the research fellow on the evacuations research, had a different perspective from mine: he was impatient with academia and was more interested in producing knowledge that could be used. So together we began to draw out the implications of the findings of our research for emergency preparedness and response, such as the importance of communication and trust in relations between responders and crowd.

This was at a time before the Impact agenda of both the Research Excellence Framework and the ESRC were quite as prominent as they are now, but we thought of various strategies for dissemination, the main one being an accessible free report on the research, which we distributed to over 35 organisations. Following this, I got a number of invitations to speak to practitioner events, and I found I enjoyed these. One of these invitations turned out to be significant in terms of the further development and application of the research. It was an event organised by the Joint Royal Colleges Ambulance Liaison Committee, which I hadn’t heard of before, and I’m not sure how they found my name. I spoke about the London bombings research and our ideas about crowds and resilience. These interested Richard Williams who was at the event and who was working for the Department of Health Emergency Preparedness Division on the new NATO guidance on psychosocial care for people affected by emergencies and disasters. The evidence and the concepts we were developing fitted well with the aims of the stepped model of care he was putting into the guidance, which suggested that while a minority need psychiatric care, the majority have the capacities to care for themselves and recover without expert intervention. Our work showed that group processes in crowds – including crowds of strangers – was key to that. The meeting was extremely fortuitous, for Richard, unusually, was a psychiatrist sympathetic to ideas in social psychology and whose specialism was in mental health strategy in emergency response. He had the interest and the vision to translate some of the social identity ideas into some of the Department of Health guidance.

Through Richard Williams and his colleague in the Department of Health, Verity Kemp, I made contact with Richard Amlôt in what was then the Health Protection Agency (now Public Health England). There was one unusual type of emergency scenario I wanted to study, and it was the same type of emergency for which PHE colleagues were seeking an academic partner for research: mass decontamination of chemical, biological, radiological and nuclear (CBRN) incidents. The perceived threat of ‘dirty bombs’ from Al Qaeda meant that CBRN mass decon was a government priority at that time. Emergency procedures in such events were unlike those in any other emergency, where the public is expected simply to leave the scene, for in these events those affected needed to shower in specially designed tents to prevent the toxic agents spreading. This was obviously a very delicate operation for the fire service who would be leading the procedure, but all the research and drills had neglected the psychological side – the stress, the question of dignity, communication and the relationship between responders and public.

Richard Williams, Richard Amlôt and James Rubin (King’s College London) and I designed a programme of research combining crowd behaviour hypotheses on the role of social identity with investigations into public–responder interactions. Holly Carter (PHE) was the PhD student who carried out the research. She demonstrated that communications from the responders could improve the efficiency of the decon process through legitimising the responders and enhancing identification between them and public. Through her interventions, the public became active participants in the decon process rather than passive recipients of care (Carter et al., 2015). Holly’s work is now starting to transform practices not only in the Fire and Rescue Service in the UK, who are changing their training to include the role of perceived legitimacy and shared social identity – and hence emphasising ‘soft skills’, but also organisations in the US who manage CBRN emergency decon incidents.

Crowd safety and the future
I was encouraged by my then Head of School to run a continuing professional development (CPD) course on crowd behaviour commercially for relevant professionals. This seemed like an interesting thing to do as way of engaging with interested end-users, but there was a problem. My university (Sussex) did not award credits for the CPD I ran, which would limit its attractiveness. The resolution of this problem led indirectly to further research and further impact. I was aware that Buckinghamshire New University was running crowd safety management courses, so I approached Chris Kemp who was Dean there to see if Bucks could offer accreditation for my course. But he had a better suggestion: that I run a version of the CPD course as a module on their crowd safety management degree programme.

Running the module at Bucks on top of my Sussex duties means extra work, but has brought me into contact with many interesting people involved in crowd safety management in the live events industry. Many of these run stewarding and security companies and pass on some of the ideas from my course to the people they train themselves.

Running the Bucks course made me think more carefully about how to present the research in a way that these professionals could understand and use – something I am still working on. The first part of this ‘translation’ work has been to start to show how beliefs about crowd ‘panic’ rationalise crowd safety management practices that have been shown to produce the very anxiety and disorganisation in the crowd that the managers seek to avoid (Drury et al., 2013). The second part, more provocative, has been to show that crowd behaviour based on social identity processes can produce the crowd self-regulation, mutual social support and spontaneous coordination we gloss as ‘safety’ and ‘resilience’ (Drury et al., 2015). Put differently, the research had sought to show that crowds can not only respond adaptively in disasters but can actually contribute to preventing disasters when the professionals themselves are overwhelmed.

This dialogue with crowd safety professionals in the live events industry is my current focus and it is where I see the impact of this work in the future. Crowd safety managers and others are becoming more professionalised and seek accredited training. That training, and the official guidance used in the industry, is increasingly informed by scientific research. This impact work is about more than emergencies – it’s also about those crowd events that pass off normally. When they manage these events, crowd safety professionals operate with beliefs about crowd psychology, even if they are not fully conscious of them. In my ongoing dialogue with these groups, my role has been to help them to reflect on their beliefs and to give them the latest research findings on how those beliefs impact on crowd behaviour.

I still think it’s crucial to hold on to the idea of pure research, in these Impact-agenda driven times. I continue to do research on collective action, and currently have three PhD students working in this area: Sara Vestergren, Patricio Saavedra Morales and Atalanti Evripidou. For me, this work essentially advances theory and contributes to the accumulation of knowledge. It is part of the wider body of work that challenges pathologising accounts of collectives, but it will not be part of an Impact Case Study. (I don’t get many requests from activists asking me how we can enhance crowd empowerment, but I’d be happy to try to help them if I did get such an approach!) It’s important that we don’t expect everyone in our Departments to produce an Impact Case Study and that we continue to value pure research. It just so happens that I’ve been lucky in choosing a research topic – crowds in emergencies – that has practical implications that I actually enjoy drawing out. Talking to practitioners – like the stewards at Roskilde Festival that I met last week – and seeing them get excited by the research is one of the most rewarding parts of my job.


- John Drury is a Reader in Social Psychology at the University of Sussex
[email protected]


Carter, H., Drury, J., Rubin, G.J. et al. (2015). Applying crowd psychology to develop recommendations for the management of mass decontamination. Health Security. 13(1), 45–53.
Drury, J., Cocking, C. & Reicher, S. (2009). Everyone for themselves? A comparative study of crowd solidarity amongst emergency survivors. British Journal of Social Psychology,  48, 487–506.
Drury, J., Novelli, D. & Stott, C. (2013). Psychological disaster myths in the perception and management of mass emergencies. Journal of Applied Social Psychology, 43, 2259–2270.
Drury, J., Novelli, D. & Stott, C. (2015). Managing to avert disaster: Explaining collective resilience at an outdoor music event. European Journal of Social Psychology, 4, 533–547.
Dynes, R.R. (2003). Finding order in disorder: Continuities in the 9-11 response. International Journal of Mass Emergencies and Disasters, 21, 9–23.
Reicher, S.D. (1984). The St Pauls riot: An explanation of the limits of crowd action in terms of a social identity model. European Journal of Social Psychology, 14, 1–21.
Reicher, S. & Drury, J. (2011). Collective identity, political participation and the making of the social self. In A. Azzi, X. Chryssochoou, B. Klandermans, & B. Simon (Eds.) Identity and participation in culturally diverse societies: A multidisciplinary perspective (pp.158–176). Oxford: Wiley-Blackwell.

BPS Members can discuss this article

Already a member? Or Create an account

Not a member? Find out about becoming a member or subscriber