Observing the Bystander Effect

We've all heard about the bystander effect: Supposedly, you can lay in the street dying of grievous injuries, and people will walk right past you, step over you, wheel their carts around you. Not a single person runs to your assistance. Why not? Perhaps they all assume someone else will do it. Perhaps they all assume that since nobody else is helping, you don't need help, so why bother. In fact, the way we've all heard it, the more people there are around, the less likely it is any of them will help you! It's called the bystander effect, and it's long been considered one of those fundamentals of the social sciences. But then you may have heard that newer research has recently disproven it. Today we're going to have a close look at the origins of the bystander effect, whatever this newer research is that's said to overturn it, and finally come to a conclusion that's best supported by the facts.

The event that kicked off research into the bystander effect (or bystander apathy, as it was called at the time) was the very horrible rape and murder of 28-year-old Kitty Genovese in New York in 1964. The New York Times reported that some 38 people saw or heard the attack, but that none intervened to help. This shocked the nation, and got everyone talking about bystander apathy.

It made no difference at the time, but more than 40 years later, the New York Times published substantial investigative journalism that found the original article's facts were greatly exaggerated. Only a few people had seen or heard anything, few thought it was an attack, and three people did render assistance to Genovese, tragically too late. But the basic fact that some small number of earlier witnesses did know what was going on and yet did nothing still supports the idea of bystander apathy.

Genovese's death did launch the field of research into this concept. Perhaps the most famous example is the Smoke Filled Room study. In 1968, a pair of social psychology researchers, Bibb Latane and John Darley, set up a series of experiments searching for an explanation for cases like the Kitty Genovese murder. Their hypothesis was that when other people are around, we're all less likely to intervene in some emergency. They did a number of different experiments. In this one, they took a room and put a row of chairs in it, and then recruited subjects to fill out some questionnaire. They divided the subjects into two groups. The first group consisted of subjects who went into the room and sat down to fill out the questionnaire — all by themselves, with nobody else in the room.

But trouble was brewing. After a few quiet minutes filling out the form, ominous thick smoke began leaking into the room through a vent. Recognizing the evident danger, the vast majority of subjects immediately left the room to advise the researchers. When by themselves, they were Johnny on the Spot.

Subjects in the second group, however, had a somewhat different experience. The chairs were all filled with other people also working on the same forms — and, unbeknownst to the subject, those others were all confederates who were part of the experiment. When the smoke began, the confederates' job was to ignore it. Each subject was the only person in the room who noticed or cared that they were all about to burn to death. However, perhaps thinking that those others knew better and knew the smoke wasn't a problem, or maybe that others were closer to the vent or to the door and would solve it first, the subjects almost never took any action.

In the end, 75% of solo subjects intervened in the smoke, and just 10% of subjects surrounded by confederates did. The difference was astonishing, and appeared to confirm Latane and Darley's hypothesis. They attributed it to pluralistic ignorance, in which you adopt the lack of concern shown by the others around you, assuming that their lack of action means there's no need for action; and also to diffused responsibility. This is when there are so many more people around than are needed for a certain task that there's no point in you being the one to step up and do it. Everyone around is responsible, so in a sense, no one is.

They did other experiments as well, including one where a subject walked down the street when a woman in distress appeared. People who were walking by themselves voluntarily offered assistance 70% of the time; but when experimenters gave the subject a stranger to walk along with them, the subjects offered assistance only 40% of the time. Having only a single other person around whose reaction the subjects were unsure of was enough to discourage intervention by the majority of subjects.

Latane and Darley's paper concluded, in no uncertain terms:

A victim may be more likely to get help, or an emergency be reported, the fewer people who are available to take action.

And thus the bystander effect entered the public body of knowledge in social psychology. A single person, or a small group, is more likely to take action than is a large group.

But then in 2019, it was reported everywhere that the science underlying the bystander effect had been debunked. Researchers observed actual public conflicts in the real world and discovered that large groups of bystanders do render assistance, almost all the time. Here's a headline from New Scientist:

BYSTANDER EFFECT: FAMOUS PSYCHOLOGY RESULT COULD BE COMPLETELY WRONG

And from Psychology Today:

NEW STUDY SUGGESTS BYSTANDER APATHY IS NOT THE NORM

And from the science news on New Atlas:

DISPROVING THE BYSTANDER EFFECT: CCTV STUDY FINDS PEOPLE INTERVENE IN CONFLICT MORE THAN EXPECTED

Even the Washington Post trumpeted this shocking new development to the masses:

FORGET WHAT YOU MAY HAVE BEEN TOLD. NEW STUDY SAYS STRANGERS STEP IN TO HELP 90 PERCENT OF THE TIME

Every indication given by the 2019 headlines was that the bystander effect had been turned on its head and disproven. Bystanders do, it turns out, spring into action and intervene; they do not apathetically look the other way in the hope that someone else will solve the problem.

What this new group of researchers did was to study "public conflicts" — basically street altercations of some sort — captured by surveillance cameras. From over 1200 conflicts recorded, 300 comparable conflicts ranging from "animated disagreements" to "grave physical violence" were analyzed in urban areas of three cities: Lancaster in the United Kingdom, Amsterdam in the Netherlands, and Cape Town in South Africa, cities with greatly differing levels of violent crime. In all three places, irrespective of their normal crime level, there was equally about a 90% chance that one or more bystanders would intervene in the conflict, and that chance was highest when there were the most bystanders around. The authors wrote:

According to conventional wisdom, there is an epidemic of bystander non-involvement during public emergencies. Challenging this view, the current cross-national study of video data show that intervention is the norm in actual aggressive conflicts, with more populated settings providing a greater likelihood that someone helps. This is reassuring for potential victims of violence.

Certainly this would indeed appear to overturn not only Latane and Darley but virtually every psychology school and textbook on the planet. How could the Smoke Filled Room results have been so wrong? If the other results we see are true, what could explain the observation that only 10% of the time, the smoke emergency was reported? The explanation lies in the experimental design.

Taking all of these cases into consideration — whether it's Kitty Genovese, the people in the Smoke Filled Room, or the public conflicts captured by the surveillance cameras in the three cities, or any of the other countless cases where there's been an opportunity for a bystander to intervene — there is one conclusion that's apparently supported by all the data. In the great majority of such events, bystanders intervene about 80% of the time. That's a rough number, because I'm comparing all sorts of diverse data sets, but it's within 10% of the true number in all the cases studied. And that number seems to hold constant whether there is one bystander or whether there are one hundred. About 80% of victims are helped, and about 80% of other crises receive rendered assistance of whatever kind.

Yet, in the Smoke Filled Room, intervention was observed only 10% of the time. Why? It's simply because all but one of the bystanders were confederates who were told not to act.

Had Latane and Darley designed their experiment better, there would have been no confederates. All the people in the room would have been subjects. And if that had been the case, the data tells us that we'd have seen that 80%-or-so intervention rate. With only a single subject and half a dozen confederates, we saw the subject look around the room and consider whether to act, but she was invariably dissuaded because of all the other people around. Any one of them could have acted at any moment, for all the subject knew, and so she refrained. The way the experiment was designed was an interesting test of peer pressure, but it was not a fair test of the bystander effect. By controlling the reaction of most bystanders, they broke the math.

So what about Kitty Genovese? She was just unfortunate enough to be among the 20% of cases in which bystanders do not intervene. It doesn't matter whether there were 38 bystanders or just two; the 80% rule seems to hold true. About 20% of the time, a victim in plain view will not receive any help. Because of pluralistic ignorance and diffused responsibility that Latane and Darley correctly identified, each of us is individually less likely to intervene as the number of our fellow bystanders grows; but the likelihood for the group as a whole to intervene remains about the same.

It was too late to help Kitty Genovese, but her death did help prompt the creation of the 9-1-1 emergency phone system. The question is whether making the task of rendering assistance as easy as dialing three digits bumps that basic 80% ratio even higher. I certainly hope so. Perhaps that's the next area of study.

By Brian Dunning
Follow @BrianDunning