John M. Darley, Bibb Latané · 1968

The bystander effect experiments

When several people witness an emergency, does each of them become less likely to help?

  • New York University and Columbia University, New York
  • Replicated, with caveats

The question

In March 1964, Kitty Genovese, a 28-year-old woman, was attacked and killed near her home in Kew Gardens, Queens, in New York. Two weeks later The New York Times ran a front-page story by Martin Gansberg claiming that 38 neighbours had watched the attack and that none of them had called the police while it was happening. The Times’s metropolitan editor, A. M. Rosenthal, expanded it into a short book, Thirty-Eight Witnesses.

Two young psychologists in New York, John Darley and Bibb Latané, doubted the usual explanation, which was that people had simply become apathetic. They suspected something more ordinary: perhaps it was precisely because so many people knew what was happening that no one acted. Their question was simple. Does the number of people who witness an emergency change whether any one of them helps?

How it was done

The seizure study (Darley & Latané, 1968). Students at New York University were told they would discuss the personal problems of college life with other students over an intercom. To keep things anonymous, each sat alone in a small room, and the experimenter said he would not be listening. Only one microphone worked at a time, for about two minutes per person.

In reality, every other voice was a recording. Early on, one “student” mentioned that he was prone to seizures. Later, when his turn came round again, he seemed to have one: about a minute in, his speech broke down, and roughly two minutes into his turn he went silent. The real participant could not see or talk to the others, and could not know whether anyone else was already doing something.

The key change was the size of the group the participant believed was listening: just himself or herself and the victim, one other person as well, or four others. Helping meant leaving the room to tell the experimenter’s assistant down the hall. If six minutes passed without that, the session ended.

The smoke study (Latané & Darley, 1968). Male students at Columbia University sat in a waiting room filling out a questionnaire. Then smoke began to pour in through a wall vent until it was hard to see. Some were alone. Some sat with two other people who were working with the researchers and calmly ignored the smoke. Others were in groups of three genuine participants.

What came out

In the seizure study, the effect of the imagined crowd was clear:

These figures come from the 52 women in the main comparison. Of the 72 students who took part, two were excluded because they did not believe the emergency was real; the other 18 took part in variations of the one-other-listener condition.

Whether the other listener was a man, a woman or a medical student made no clear difference, and of 16 personality and background measures, only one (the size of the community a student grew up in) was linked to how quickly they reported it.

Crucially, those who did not go were not indifferent. Many showed visible signs of nerves, such as shaking hands and sweaty palms, and seemed, if anything, more upset than those who went for help. The authors did not think these students had chosen to do nothing. They were stuck between doing something and fear of overreacting or interfering.

The smoke study gave a similar picture. Alone, 75% of students reported the smoke. With two calm strangers ignoring it, only 10% did; the rest kept filling in their forms in a room full of smoke. In groups of three real participants, someone reported it in only 38% of groups. Afterwards, non-reporters offered many reassuring explanations, from steam or smog to a “truth gas” meant to make them answer honestly.

Darley and Latané proposed two mechanisms:

They later described helping as a series of decisions, each of which can fail: notice the event, interpret it as an emergency, decide that it is your responsibility, choose how to help, and act. They themselves warned that real people do not go through these steps so neatly.

What really happened (replications and critiques)

The Genovese story was not what the newspaper said. In 2007, psychologists Rachel Manning, Mark Levine and Alan Collins went back to the trial records and other documents, building on research by local historian Joseph De May Jr. They found no evidence for 38 eyewitnesses. There were two attacks, not three, and the second took place inside a building, out of sight of most neighbours. Of the five neighbours who testified at the trial, only three had seen Genovese and the attacker together, and none reported seeing a stabbing. After the first attack, a neighbour shouted from a window and the attacker ran off. Later accounts say residents called the police, and that a neighbour, Sophie (or Sophia) Farrar, went to her side. Genovese was still alive when the police arrived.

In 2016, in its obituary of the man who killed her, The New York Times itself called its 1964 front-page story flawed, acknowledging that it had greatly overstated how many people witnessed the attack and how much they saw. The documentary The Witness (2015) follows her younger brother, Bill Genovese, as he spent years looking for what really happened. He found a neighbour who had called the police and a friend who stayed with his sister until help came; most others had only heard cries.

Manning and colleagues are careful to separate the myth from the science. They describe the bystander effect as among the best-replicated results in the field, and its validity does not depend on the newspaper story being true. Their concern is that the “parable of the 38 witnesses” taught generations of students to see groups mainly as a reason people do not help. All ten social psychology textbooks they examined told the story, and all said no one intervened until after Genovese died.

The effect is real, but not the whole picture. In 2011, Peter Fischer and colleagues combined 105 independent effect sizes from studies with more than 7,700 participants. Overall, the presence of others did reduce helping. But the effect was weaker when the emergency was clearly dangerous, when a perpetrator was present, and when helping carried physical costs. One explanation they discuss: dangerous situations are recognised as emergencies more quickly, and the alarm they cause pushes people to act.

Real life looks different again. In 2020, Richard Philpot, Lasse Liebst, Mark Levine and colleagues analysed 219 CCTV recordings of aggressive public conflicts, from arguments to fights, in Lancaster (UK), Amsterdam (Netherlands) and Cape Town (South Africa). In 90.9% of them, at least one bystander intervened, and on average almost four people did. Intervention rates did not differ between the three countries. And more bystanders meant a higher chance that someone helped.

This does not contradict the lab results. The CCTV study measured whether anyone in a group helped, not how likely each individual was to help, and each person’s share of responsibility may still shrink in a crowd. But it shifts the question from “Why does no one help?” to “What helps people help well?” The authors also note limits: the footage had no sound, came from nightlife areas, and cannot show cause and effect.

Ethics

The seizure study depended on deception: participants believed they were hearing a real medical emergency happen to a fellow student. That was distressing for many of them, and the experimenters saw signs of strong anxiety in those who did not go for help.

Once the session ended, the assistant explained the deception and talked through how participants felt. According to the authors, every participant described the study as interesting, said they understood it, and felt the deception had been necessary and justified. Whether a review board would approve the same design today is an open question: deliberately staging a medical crisis would be examined closely, and any approval would require strong justification and careful aftercare.

What it means today

The original studies show how easily uncertainty and shared responsibility can freeze ordinary, caring people. The newer research shows that in real conflicts, people usually do step in. Both point to some practical habits. These are applications of the findings rather than things the 1968 studies tested directly:

Would you have passed?

Coming soon. You’re filling out a form when something starts to go wrong in the room, and the other people around you don’t react. Decide when, or whether, to act, then compare yourself with the 75% who reported the smoke when alone and the 10% who did beside two calm strangers. One honest caveat: knowing this is a test about helping changes everything, which is exactly why the original studies relied on deception.

Questions people ask

What is the bystander effect?

It is the finding that a person is less likely to help in an emergency, or slower to help, when they believe other people are also present. In Darley and Latané's 1968 seizure study, 85% of students who thought they were alone with the victim reported the seizure before it ended, against 31% of those who thought four others could hear it.

Did 38 people really watch Kitty Genovese being killed and do nothing?

No. That account comes from a 1964 New York Times article that later research and the Times itself described as greatly exaggerated. Only a few people saw part of what happened, the second attack took place out of most neighbours' sight, and at least one neighbour called the police and another went to help her.

Is the bystander effect real?

In the lab, yes: it has been found many times. A 2011 meta-analysis of more than 7,700 participants confirmed it, but found it weaker in dangerous emergencies. In real-life public conflicts filmed by CCTV, at least one bystander stepped in about nine times out of ten, and more bystanders made help more likely, not less.

How can I make sure I get help in an emergency?

The research suggests reducing doubt and shared responsibility: make it clear that this is an emergency and ask a specific person to do a specific thing, such as calling an ambulance. This is an application of the findings rather than something the 1968 studies tested directly.

Sources

  1. Darley, J. M., & Latané, B. (1968). Bystander intervention in emergencies: Diffusion of responsibility. Journal of Personality and Social Psychology, 8(4), 377–383. doi:10.1037/h0025589
  2. Latané, B., & Darley, J. M. (1968). Group inhibition of bystander intervention in emergencies. Journal of Personality and Social Psychology, 10(3), 215–221. doi:10.1037/h0026570
  3. Latané, B., & Darley, J. M. (1969). Bystander 'apathy'. American Scientist, 57(2), 244–268. www.truthaboutnursing.org/research/orig/latane_and_darley/bystander_apathy.pdf
  4. Latané, B., & Darley, J. M. (1970). The unresponsive bystander: Why doesn't he help? Appleton-Century-Crofts.
  5. Manning, R., Levine, M., & Collins, A. (2007). The Kitty Genovese murder and the social psychology of helping: The parable of the 38 witnesses. American Psychologist, 62(6), 555–562. doi:10.1037/0003-066X.62.6.555
  6. Gansberg, M. (1964, March 27). 37 who saw murder didn't call the police. The New York Times, p. 1.
  7. McFadden, R. D. (2016, April 4). Winston Moseley, 81, killer of Kitty Genovese, dies in prison. The New York Times. bendbulletin.com/2016/04/07/his-killing-of-genovese-shocked-the-nation
  8. Solomon, J. (Director), & Genovese, W. (Executive Producer) (2015). The Witness [Documentary film]. www.wfae.org/npr-arts-life/2016-06-02/the-witness-looks-back-at-those-accused-of-ignoring-a-murder
  9. Fischer, P., Krueger, J. I., Greitemeyer, T., Vogrincic, C., Kastenmüller, A., Frey, D., Heene, M., Wicher, M., & Kainbacher, M. (2011). The bystander-effect: A meta-analytic review on bystander intervention in dangerous and non-dangerous emergencies. Psychological Bulletin, 137(4), 517–537. doi:10.1037/a0023304
  10. Philpot, R., Liebst, L. S., Levine, M., Bernasco, W., & Lindegaard, M. R. (2020). Would I be helped? Cross-national CCTV footage shows that intervention is the norm in public conflicts. American Psychologist, 75(1), 66–75. doi:10.1037/amp0000469