Module 3, lesson 1 · 40 minutes

The panic myth: what people actually do

The film version of an emergency, people fleeing blindly over each other, is very rare. What people actually do is slower, more social and more helpful, and that creates different dangers.

You’ll be able to

give Quarantelli’s definition of panic and his later conclusion about how rare it is;
describe affiliative behaviour and the survivor studies of Drury, Cocking and Reicher;
use the World Trade Center evidence to explain why delay, not flight, is the usual problem in an evacuation.

On 11 September 2001, the people in tower 1 of the World Trade Center (WTC 1) had every reason to run. Here is what they did. On floors up to 76, the median time before people started to evacuate was 3 minutes; near the impact zone, floors 77 to 91, it was 5 minutes. In a survey of 440 survivors, 70% said they talked to others before leaving, and 46% gathered personal items.

That is not panic. It is something close to the opposite: people taking time to understand what was happening, checking with each other, picking up their things. Module 2 showed that crowds die of physics, not fear. This module asks the next question: if people do not panic, what do they do, and how should that change the way we plan for them?

What “panic” would mean

To test the panic story you need a definition. The classic one is from the sociologist Quarantelli, in 1954. Panic, in Quarantelli’s classic definition, is “an acute fear reaction marked by loss of self-control which is followed by nonsocial and nonrational flight”. He argued it arises from a sense of “possible entrapment… collective powerlessness… individual isolation”.

Every part of the definition is a claim you can check against what people do: is there a loss of self-control? Is the flight nonsocial, every person for themselves? Is it nonrational? Nearly fifty years later, in 2001, Quarantelli’s conclusion was short: “panic flight is very rare”.

What people do first: they delay

The World Trade Center evacuation on 9/11 was studied in depth both by the US National Institute of Standards and Technology (NIST) and by the UK’s HEED project, which interviewed 271 survivors (the project’s own introduction says “almost 300”).

What World Trade Center survivors did before leaving, and how long they took to start. NIST survey (top) and HEED interviews (bottom).

The HEED data show the same pattern in more detail. Of 119 people whose response times could be established, 13% started in under a minute, 78% took between 1 and 8 minutes, and 9% took longer than 8. On average people completed 4 tasks before setting off: 1.5 seeking information and 2.5 actions. “Collect Items” was reported 143 times, 49% of all action tasks.

None of this is irrational. People do not leave a building on the strength of one loud noise. They look for information, ask the person next to them, call someone, save their work, find their bag. But in a fire, every minute spent on those tasks is a minute not spent moving. In evacuation engineering this period is called pre-movement time. Purser notes that it “often represents a greater component of the total escape time”, and Module 6 builds on that.

The panic story says the danger is people moving too fast. The evidence says the more common danger in an evacuation is people moving too late. Planning that assumes panic tends to focus on calming people down; planning that follows the evidence focuses on getting clear, credible information to them quickly so they start sooner.

What people do next: they stay together

When people do move, they rarely move as isolated individuals. Sime’s 1983 paper on escape to building exits is known for the idea of affiliative behaviour: people move towards the familiar, to the people they came with and to exits they already know, rather than the nearest exit.

Affiliative behaviour is the tendency in an emergency to seek out and stay with familiar people, and to head for familiar places and exits. If that is right, you would expect families to regroup before leaving, friends to wait for each other, and people to head for the door they came in by even when a closer one is marked. To a model that treats each person as a particle heading for the nearest exit, this looks like a mistake. To the people involved it is entirely sensible: the first thing most of us want in an emergency is to be with the people we love.

A venue has a main entrance at the front and four emergency exits along the sides, all clearly signed. In an evacuation drill most people leave through the main entrance and the side exits are under-used. Using this lesson, explain why, and suggest one thing the venue could change.

Affiliative behaviour: people head for the familiar route, the one they came in by, and stay with their group. Signage alone does not overcome that. The venue could make the side exits familiar, for example by using them for routine exit after events, or have staff actively direct people to them, so that a person, not just a sign, offers the alternative.

And they help each other

The panic story predicts selfishness: every person for themselves. Drury, Cocking and Reicher tested that directly.

In the first study they interviewed 21 survivors of 11 different emergencies. They tested two ideas: that a shared identity among the people caught up in an emergency “enhances expressions of solidarity and reduces ‘panic’ behaviour”, and that such an identity can “arise from the shared experience of the emergency itself”. Their analysis “broadly supports these two claims”.

In the second, they collected accounts from over 90 survivors and 56 witnesses of the 2005 London bombings. The findings: “‘Selfish’ behaviors were found to be rare; mutual helping was more common.” The authors’ explanation is that shared danger changes who people think they are: “common fate entails a redefinition of self (from ‘me’ to ‘us’)”.

“even under extremely critical conditions, people helped each other and behaved quite rationally.”

Helbing and Mukerji on the Love Parade video, 2012

The next two lessons build on that “me to us” idea: what a shared identity is, how it forms, and how it can be helped or harmed by the people managing a crowd.

Why the myth matters

If people rarely panic, why does the idea survive? Partly because it makes a vivid story, and partly because it is convenient. Keith Still’s one-line answer reverses the cause and effect: “People don’t die because they panic. They panic because they are dying”. People in a crush do scream, push and struggle. That is a reasonable response to being unable to breathe, not the cause of the crush. Fruin: “Frequently mislabelled a panic, closer investigation usually shows that flight was a reasonable group reaction”.

The practical stakes are high. A manager who believes crowds panic is tempted to hold back information (“don’t start a stampede”), to treat the crowd as a hazard to be contained, and to look for blame among the victims afterwards. A manager who follows the evidence tells people what is happening and what to do, uses the crowd’s own tendency to help, and looks for the cause in density, design and management.

A theatre’s procedure says: “In case of fire, staff should not mention fire, to avoid panic; announce a ‘technical problem’ and ask people to leave.” Using the evidence in this lesson, argue for or against.

Against. Panic flight is very rare, and the common danger is delay: people seek information and finish tasks before they move. A vague announcement like “a technical problem” invites exactly that, as people wait, ask each other and collect belongings. Telling people clearly what is happening and what to do gives them a reason to move now. The evidence also suggests people will help each other if they understand the situation. (Any wording should still be designed and rehearsed; the point is clarity, not alarm.)