Wednesday, 12 September 2012

Hillsborough and ‘crowd control’

The Independent Report into the 1989 Hillsborough football crowd tragedy has found that police made ‘strenuous efforts’ to shift blame for the 96 deaths to fans themselves. In today’s news coverage, the disaster itself is sometimes described as a ‘failure of crowd control’. These words echo the conclusions of the 1990 Taylor enquiry into the tragedy, which recommended all-seater stadia for English football grounds.

The words ‘crowd control’ are worth considering for a moment. Whenever my own research and specialism is described as ‘crowd control’ (which is quite often), I try to explain that this is wrong in important ways. I teach a module on 'crowd safety management' to crowd safety and events professionals, and the classic statement of the issue we refer to when we discuss this topic is that by Fruin, who is perhaps worth quoting at length:

‘Although the terms crowd management and crowd control are often used interchangeably, there are important differences. Crowd management is defined as the systematic planning for, and supervision of, the orderly movement and assembly of people. Crowd control is the restriction or limitation of group behavior.’ (Fruin, 2002, p. 6).

Fruin goes on to say Inappropriate or poorly managed control procedures have precipitated crowd incidents rather than preventing them.

Or, the way I put it in my lecture: Approaching the crowd with a view to crowd control risks undermining crowd safety.'

Why? Part of the reason, I would argue, hinges on the dynamic relationship between crowd behaviour and the representations of crowds held by those responsible for crowd management (or crowd control). For crowds only need to be ‘controlled’ when someone thinks they represent a problem.

Broadly, there are two dominant types of representation of 'crowd problems' needing ‘control’. These can be found in both popular discourse and in academic accounts. The first suggests that, in crowds, there is a process of submergence, whereby individuals lose their sense of self and hence their intelligence and self control, becoming ‘swept up’ by any sentiment or behaviour spreading through the mass. The second representation, convergence, suggests that problematic crowds consist of problematic individuals – people who are already lacking in intelligence and self control – who in the crowd act as they do alone ‘only more so’, as Floyd Allport put it. These two accounts are often employed in loose combination in order to explain what is seen as irrational or mindless behaviour in crowds.

In my lecture on ‘crowd management versus crowd control’ on the crowd safety management module, one of our key examples is the Hillsborough tragedy. As the report published today reminds us, certain newspapers, most notably The Sun, claimed that the crushing of fans in the football stadium was due to ‘football hooliganism’. The journalists and the police said the crowd that day was inherently a problem because it represented the convergence of drunken individuals seeking gratuitous violence and ‘disorder’.

The cruel irony is that these fans that were vilified were the very people trying to save others, and in some cases risking their own lives to do so. As part of our ESRC-sponsored study of mass emergency behaviour, we interviewed some survivors of Hillsborough tragedy, who described a deeply harrowing scene yet also one of great humanity, as illustrated in the following extracts (taken from Drury, Cocking, & Reicher, 2009):

‘The behaviour of many people in that crowd and simply trying to help their fellow supporters was heroic in some cases. So I don’t think in my view there was any question that there was an organic sense of … unity of crowd behaviour. It was clearly the case, you know.. it was clearly the case that people were trying to get people who were seriously injured out of that crowd, it was seriously a case of trying to get people to hospital, get them to safety.. I just wish I’d been able to.. to prevail on a few more people not to.. put themselves in danger.’

‘It should be source of great pride to those people I think because you know, they were clearly in control of their own emotions and their own physical insecurity I mean a lot of people were very.. as I was you know.. you’re being pushed, you’re being crushed when you’re hot and bothered you’re beginning to fear for your own personal safety and yet they were I think controlling or tempering their emotions to help.. try and remedy the situation and help others who were clearly struggling’

 Picture courtesy of

However, it wasn’t simply after the event that the police explained the disaster as due to drunken ‘hooligans’. Crucially, this view of the crowd was at the forefront of all their planning and the responses they made during the tragedy. This has been shown both in published research and is clear from numerous statements in the Independent Report published today:

‘It is evident from the disclosed documents that SYP [South Yorkshire Police] were preoccupied with crowd management, segregation and regulation to prevent potential disorder ... The Fire Service, however, raised concerns about provision for emergency evacuation of the terraces’
(p. 11)

And again on page 12:

a policing and stewarding mindset predominantly concerned with crowd disorder ... the delay in realising that the crisis in the central pens was a consequence of overcrowding rather than crowd disorder.’

And again on page 15:

‘the ‘prevention of hooliganism’ and ‘public disorder’ was the main priority. The custom and practice that had evolved within SYP for packing the pens was concerned primarily with controlling the crowd.’

‘Crowd control’ meant treating fans like animals, neglecting their safety. As Fruin says, crowd control is the ‘restriction or limitation’ of crowd behaviour – and this is done when someone regards that crowd as a ‘problem’. Crowd control is achieved with means such as barriers or coercion, which risk injury and even fatalities in the crowd.

These days, the leading crowd event safety experts agree that many problems in crowd events – including some of the most well-known crowd disasters – are due to problems in crowd management. Examples would example the failure to plan for sufficient space for the size and flow speed of the crowd, and the failure to communicate adequately with the crowd.

This kind of analysis, which moves attribution for crowd disasters away from the supposedly inherent psychological problems of the crowd (whether of ‘convergence’ or ‘submergence’) to deficiencies in management and planning, is a positive development. It suggests that crowd disasters are not simply something that ‘just happens’ from time to time due to the inherently primitive psychology of the crowd; rather, crowd disasters are preventable through improvements in knowledge about, and hence to the practice of, crowd safety management. It is far too late for the Hillsborough tragedy, of course, but the increased scientific interest in the psychology of crowd safety management, in both research and academic contexts, is a vital contribution to the critique of ‘crowd control’ and hence to safer crowd events in the future.


Drury, J., Cocking, C., & Reicher, S. (2009). Everyone for themselves? A comparative study of crowd solidarity among emergency survivors. British Journal of Social Psychology, 48, 487-506.

Fruin, J. J. (2002). The causes and prevention of crowd disasters. Originally presented at the First International Conference on Engineering for Crowd Safety, London, England, March 1993 (Revised exclusively for, January 2002.)

Wednesday, 1 August 2012

The ideology of the madding crowd: Learning lessons from our history

A brief commentary to coincide with the anniversary of the August 2011 riots, by Clifford Stott & John  Drury, published in The Occupied Times of London 16 (August, 2012)

Friday, 20 July 2012

At the Olympics, transport chiefs must trust the wisdom of crowds

Crowd control [sic] is often driven by a fear of panic and selfish behaviour, but this view is out of date – communication is key

Wednesday, 28 March 2012

The logic of ‘panic buying’

This week, the government (and the mass media) have been criticized for promoting ‘panic-buying’ of petrol in the context of a possible tanker drivers strikeHere at Sussex University, we have been looking at the use of the word ‘panic’, and at the kinds of behaviours that are sometimes labelled in this way. We have also worked alongside emergency planners and crowd safety management professionals on this question of communication in possible emergency situations. Based on this work, there are three simple points to make about the current situation of supposed ‘panic-buying’.

Run on the Seamen's Savings' Bank during the so-called ‘panic’ of 1857 (courtesy of Wikipedia commons)


1, Don’t say ‘don’t panic’

The first point is that the advice ‘don’t panic’ is worse than useless. When those in authority say ‘don’t panic’, or even ‘stay calm’, the rest of us know by this that there is definitely something we should be anxious about – for why else would they be warning us in this way? And why else should they be so concerned about our emotional responses if not because they think there is something that we might ‘panic’ about? In the present case, the advice that there was no crisis, but that people should fill up their cars just in case, was as good as advice to act exactly as though a crisis was imminent and therefore to fill up as much as we can.

2, Belief that there is ‘panic’ makes it logical to act individualistically

The advice ‘don’t panic’ is based on the mistrust by the government of the (potentially irrational) ‘masses’. In turn it sows mistrust. It does this in two ways. First, the advice communicates that the government is holding something (scary) back from the public, so creates mistrust between public and government. Second, reference to ‘panic’ risks creating mistrust amongst members of the public themselves. When the government and the mass media tell us that our neighbours are ‘panic-buying’, we imagine those around us acting individualistically, rushing to hoard goods for themselves. And if we imagine that everyone else is acting this way, it becomes foolishly self-sacrificial to do otherwise oneself. Rather, it becomes completely logical to look after number one. The result is that we trust our neighbours less to act patiently and in the collective interest.

3, ‘Panic buying’ is not panic

The third point is that the large queues and the rush to stock up does not really fit the criteria for panic. Journalists reporting from the scene of petrol queues note with surprise the calm, boredom and stoicism of the people waiting in line. ‘Panic’ is a clinical condition marked by an excessive fear and anxiety reaction, the physiological signs of which include raised heart rate, shortened breath, sweating and so on. This excessive anxiety prompts reckless, uncontrolled behaviour. People queuing for petrol in so-called ‘panic buying’ may be acting in their personal interest rather than the collective interest, but this is cognitively driven rather than an instinctive ‘flight or fight’ response. The queuing and bulk-buying  is logical - given people's reasonable beliefs about others’ behaviour and their reasonable mistrust of the authorities.  And there is no spread of irrational beliefs or emotions through these crowds. As I have argued elsewhere, there is no evidence to support the argument that crowds are susceptible to the uncritical spread of simple emotions through so-called ‘contagion:‘mass panic’ is a myth

It is often the mass media, more than the politicians, who mobilize the ‘panic’ cliché to describe consumer queuing and stock piling. It is perhaps encouraging to report, therefore, that not all journalists are so lazy. After an interview we gave on Radio Leicester to discuss our ESRC research project on mass emergency behaviour, the reporters were so convinced by the argument that they made it a policy not to use the term in future.

However, we don’t think that solutions to the problems produced by mutual mistrust and individualism lie only or essentially in changing the messages given out by those in authority. It was notable that, in his nervous back-tracking, David Cameron referred to ‘resilience’ as just a quality of the state and its functionaries. We have been exploring the informal resilience of crowds, which expresses itself in solidarity and is a function of shared social identity among crowd membersWe are currently running some experiments to show that, when people think of themselves in terms of their social identities (e.g., ‘me as a member of my community’) they will be more cooperative, less ready to push in to queues, and more willing to share dwindling supplies with strangers than when they think of themselves in terms of their personal identities (e.g., ‘me as distinct from others in my community’). If the outcomes of the behaviours labelled as 'panic-buying' are dysfunctional for the collective (e.g., more acute shortages for most as the minority hoard goods) it is precisely because people in the crowd are acting reasonably as individuals,  rather than as members of a psychologically united crowd; cf. Mintz, 1951.)


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. DOI:10.1348/014466608X357893

Drury, J., Cocking, C., & Reicher, S. (2009). The nature of collective resilience: Survivor reactions to the 2005 London bombingsInternational Journal of Mass Emergencies and Disasters, 27, 66-95.

Mintz, A. (1951). Non-adaptive group behavior. Journal of Abnormal and Social Psychology, 46, 150-159.

Sunday, 1 January 2012

2011 - The year of contagion?

In 2011, four of the major news events of the year concerned mass social influence or the spread of behaviours. The so-called Arab Spring is said to have begun with a self-immolation in Tunisia. The protests there escalated until President Ben Ali was forced to step down in mid-January. Egypt’s uprising also began with a self-immolation and eventually forced President Mubarek out of office as more people joined in. Uprisings of varying forms also took place in Yemen, Bahrain and eventually Libya, with unrest reported across the Arab region as a whole.

In July, the plunging financial markets across the world were said to be subject to ‘contagion’, suggesting that ‘fear’ and ‘panic’ were rapidly spreading among investors.

In August, the term ‘contagion’ was mobilized again – although more often the term used was ‘copy-cat’ – to ‘explain’ the spread of riots from Tottenham to Hackney, Ealing, Clapham, Birmingham, Liverpool, Salford and other places.

Then there was the ‘Occupy’ movement. This began in Wall Street, New York, but soon spread to other cites in the USA as well as to the UK, in particular to St Paul’s, London.

The terminology used in each of these news stories tell us a lot about how the events were evaluated. Specifically, the term ‘contagion’ is not simply descriptive, but serves to construct events as negative and pathological.

In the UK media, the ‘Arab Spring’ ‘uprisings’ were largely described as a good thing, while the UK ‘riots’ were condemned. In fact, there was a very well-worn repertoire which was used to characterize the events in London, Birmingham, Manchester and so on: converge, submerge, and copy-cat. The riots were said to be about the ‘convergence’ of people who were ‘simply criminals’, their ‘submergence’ in the crowd (which supposedly leads to being ‘swept up’, ‘carried away’ and hence loss of self-restraint), and then the supposedly mindless ‘copy-cat’ riots, as ‘other people’ did the same thing elsewhere. The effects of investor behaviour on financial markets was also seen as bad; however, while the Occupy movement undoubtedly seen by some commentators as bad, this was not usually on a par with an illness. Perhaps this was because of their ‘moral’ message; bad behaviour in the markets was already the target. But maybe it was also because these events, unlike the riots, were not ‘violent’.

What’s wrong with ‘contagion’?

‘Contagion’ and associated metaphors are therefore not neutral terms used simply to describe. They are also weapons. In nineteenth century France, the notion that crowd behaviour is pathological became systematized and rationalized through the introduction to social science of a medical discourse (not only ‘contagion’, but also ‘feverishness’ and ‘delirium’) by the psycho-historian Taine.1

Of all the ways to describe social influence and the rapid spread of ideas, beliefs, feelings and behaviours, why choose a disease metaphor? Why choose a term that likens the process to a sickness or illness? The concept buys you the implication that the spread is bad, or toxic, debilitating or harmful in some way. But it also implies that the spread is uncritical, simple, non-cognitive, primitive – that anyone is susceptible to just anything. Contagion is implied to be a non-rational process, maybe a deeply irrational one.

Whether the spread of a behaviour, idea or sentiment is good or bad may be a value judgement. But whether or not just any behaviour spreads to just anyone is an empirical question.

As Milgram and Toch (1969)2 pointed out in their critique of Le Bon, if behaviours and sentiments spread  through contagion all those simply ‘submerged’ in a crowd, why aren’t riot police affected by the words of crowd demagogues? Why do the protestors in the crowd behave one way and the riot police another way? What are the limits of spread?

The riot in the St Pauls district of Bristol, in 1980, afforded a systematic test of the idea that just any behaviours do not spread indiscriminately among just anyone. The riot was the first of the big 1980s urban riots. The event which was suggested to have set it off was a police raid on a local café. It was certainly a violent event – there were attacks on the police and on property; over 20 police vehicles were damaged. However, analysis of the event showed that there were clear limits to the spread of behaviours that simply do not make sense in terms of ‘contagion’. In particular, there were limits in terms of action. Essentially, the only people that were attacked were the police. The property that was damaged comprised banks, the benefits office, the rent office and the post office, and expensive shops owned by ‘outsiders’. But homes and small locally owned shops were actively protected. And there was disapproval when someone threw a missile at a bus. In other words, behaviours that were not consistent with the shared social identity of those in the crowd – St Pauls residents – did not spread.

Recent research has added to this by showing how the spread of the 1981 riots – Toxteth, Brixton, Handsworth and others – was not random. Participants in the different cities did not mindlessly ‘copy-cat’ just any other person doing any action simply because they heard of it and had seen it on their televisions. Through a combination of mathematical modeling and oral history, the research was able to show that the riots spread partly through shared subcultural networks and partly when people recognize themselves and their own experiences in the others they saw rioting.3 This point will be developed below.

Can contagion be rehabilitated?

For those in the mass media casting about for words to describe the rapid spread of a protest or direct action movement, whether in Algeria or Florida‘contagion’ is as an easy, off-the-shelf cliché. Looking at the origins of the term historically, the problems are clear. For the early ‘crowd scientists’, the crowd was an alien other – a source of irrational behaviour and hence of society’s ills. They sought to combat the crowd, and pathologizing it was a means of achieving that – for the attribution of irrationality served to rationalize increased coercion.

In some areas of contemporary academia, though, there has been an attempt to rehabilitate the concept. An example is network analysis, which have attempted to redefine contagion as a form of rational social decision-making.

But why try to reclaim this notion with all its baggage? Wouldn’t a more neutral concept like social influence or spread be less problematic? It might be possible to redefine ‘contagion’ as rational by stipulation; but the trouble is that it already has 100 years of history in which it refers to an irrational and pathological process.

In psychology, ‘contagion’ persists in the concept of ‘emotional contagion’. This is said to be an often non-conscious transmission of mood between people. The implication is, again, that this is primitive and hence non-cognitive; and the origins and indeed pathologizing implications of the concept appear to be ignored by many who use the term.

From a social identity perspective, the problem with the concept of ‘emotional contagion’, and indeed the types of paradigms that have been used to show evidence for it, is that there is a neglect of the meaning-defining role of group memberships. Implicit in studies where people ‘catch’ moods and emotions off others is shared group membership. Of course others’ laughter and tears is ‘infectious’ – in a shared ‘human’ context. But imagine a context where the people expressing emotion are so deeply ‘other’, where their world is so different from or irrelevant to yours, that the source of their emotions and feelings are alien too. Will you catch them then? We will be running some experiments to show that so-called ‘emotional contagion’ effects will be diminished in ‘alien’ contexts, but increased in those contexts where shared group membership is salient.

Social identification and empowerment: An alternative to contagion

An alternative to ‘contagion’ as an explanation for the spread of behaviours in the ‘Arab Spring’, the UK riots and the ‘Occupy’ movement is in terms of social identification and empowerment. In this account, whether in Tunisia, Tottenham and New York, only those who identified with the participants in these events – those who felt ‘this is us’ – would have felt encouraged or uplifted by them and hence have felt able to join in.4 

In these cases, identifying with the crowd (as ‘us’) in each event also means recognizing the same enemy. For those involved in the riots, the police in Tottenham were not distinguished from the police in Hackney, for example: they were just the police, and equally illegitimate and vulnerable in each case.

Therefore, where ‘we’ act and ‘they’ are seen to retreat in an event elsewhere, ‘we’ can infer that that ‘the same’ action may be possible here too. Only those who identify are empowered. There is no need to posit a different kind of theoretical account for what has happened in the ‘Arab Spring’ and the Occupy movement than in the case of the UK riots. The actions spread – they became a movement – because the relevant people took strength from the fact that they, as a physically dispersed but psychologically unified group, were seen to be able to objectify their understanding of how the world should be. We have been testing these ideas directly with some simple experiments. However, all the coverage and participant accounts produced following each of the movements fits this analysis.


1 McClelland, J. S. (1989). The crowd and the mob: From Plato to Canetti. London: Unwin Hyman.
2 Milgram, S. & Toch, H. (1969). Collective behavior: Crowds and social movements. In G. Lindzey & E. Aronson (eds.), The Handbook of Social Psychology. Volume 4. (Second edition). Reading, Mass: Addison Wesley.
3 Ball, R. (2011). Violent urban disturbance in England in 1981. Unpublished PhD Thesis. University of West of England.
4 The case of the markets is different, for a number of reasons. First, there is a question over whether the market should be understood as constituting a psychological crowd at all. Second, unlike the three social movements, the markets are not an otherwise subordinate social group whose action changes their social relations. Therefore, third, if there is any shared social identification – if people in the financial markets do use others’ conduct as a guide for their own conduct – this social influence may well be as bounded (to ‘people like us’) as it is in any other ‘crowd’, but it can hardly be understood as a process of empowerment.