Social psychology research
The bystander effect was first demonstrated in the laboratory by
John M. Darley and
Bibb Latané in 1968 after they became interested in the topic following the
murder of Kitty Genovese in 1964.
 These researchers launched a series of experiments that resulted in one of the strongest and most
replicable effects in
 In a typical experiment, the participant is either alone or among a group of other participants or confederates. An emergency situation is
staged and researchers
measure how long it takes the participants to intervene, if they intervene. These experiments have found that the presence of others inhibits helping, often by a large margin.
 For example, Bibb Latané and Judith Rodin (1969) staged an experiment around a woman in distress. 70 percent of the people alone called out or went to help the woman after they believed she had fallen and was hurt, but when there were other people in the room only 40 percent offered help.
Variables affecting bystanders
Emergency versus non-emergency situations
Latané and Darley performed three experiments to test bystander behavior in non-
 Their results indicated that the way in which the subjects were asked for help mattered. In one condition, subjects asked a bystander for his or her name. More people provided an answer when the students gave their name first. In another condition, the students asked bystanders for a dime. When the student gave an explanation, such as saying that their wallet had been stolen, the percentage of people giving assistance was higher (72%) than when the student just asked for a dime (34%). Additional research by Faul, Mark, et al., using data collected by EMS officials when responding to an emergency, indicated that the response of bystanders was correlated with the health severity of the situation.
According to Latané and Darley, there are five characteristics of emergencies that affect bystanders:
- Emergencies involve threat of harm or actual harm
- Emergencies are unusual and rare
- The type of action required in an emergency differs from situation to situation
- Emergencies cannot be predicted or expected
- Emergencies require immediate action
Due to these five characteristics, bystanders go through cognitive and behavioural processes:
- Notice that something is going on
- Interpret the situation as being an emergency
- Degree of responsibility felt
- Form of assistance
- Implement the action choice
Notice: To test the concept of "noticing," Latane and Darley (1968) staged an emergency using Columbia University students. The students were placed in a room—either alone, with two strangers or with three strangers to complete a questionnaire while they waited for the experimenter to return. While they were completing the questionnaire smoke was pumped into the room through a wall vent to simulate an emergency. When students were working alone they noticed the smoke almost immediately (within 5 seconds). However, students that were working in groups took longer (up to 20 seconds) to notice the smoke. Latané and Darley claimed this phenomenon could be explained by the social norm of what is considered polite etiquette in public. In most western cultures, politeness dictates that it is inappropriate to idly look around. This may indicate that a person is nosy or rude. As a result, passers-by are more likely to be keeping their attention to themselves when around large groups than when alone. People who are alone are more likely to be conscious of their surroundings and therefore more likely to notice a person in need of assistance.
Interpret: Once a situation has been noticed, a bystander may be encouraged to intervene if they interpret the incident as an emergency. According to the principle of
social influence, bystanders monitor the reactions of other people in an emergency situation to see if others think that it is necessary to intervene. If it is determined that others are not reacting to the situation, bystanders will interpret the situation as not an emergency and will not intervene. This is an example of
pluralistic ignorance or
social proof. Referring to the smoke experiment, even though students in the groups had clearly noticed the smoke which had become so thick that it was obscuring their vision, irritating their eyes or causing them to cough, they were still unlikely to report it. Only one participant in the group condition reported the smoke within the first four minutes, and by the end of the experiment, no-one from five of eight groups had reported the smoke at all. In the groups that did not report the smoke, the interpretations of its cause, and the likelihood that it was genuinely threatening was also less serious, with no-one suggesting fire as a possible cause, but some preferring less serious explanations, such as the air-conditioner was leaking.
 Similarly, interpretations of the context played an important role in people's reactions to a man and woman fighting in the street. When the woman yelled, "Get away from me; I don't know you," bystanders intervened 65 percent of the time, but only 19 percent of the time when the woman yelled, "Get away from me; I don't know why I ever married you."
General bystander effect research was mainly conducted in the context of non-dangerous, non-violent emergencies. A study (2006) tested bystander effect in emergency situations to see if they would get the same results from other studies testing non-emergencies. In situations with low potential danger, significantly more help was given when the person was alone than when they were around another person. However, in situations with high potential danger, participants confronted with an emergency alone or in the presence of another person were similarly likely to help the victim.
 This suggests that in situations of greater seriousness it is more likely that people will interpret the situation as one in which help is needed and will be more likely to intervene.
Degree of Responsibility: Darley and Latané determined that the degree of responsibility a bystander feels is dependent on three things:
- Whether or not they feel the person is deserving of help
- The competence of the bystander
- The relationship between the bystander and the victim
Forms of Assistance: There are two categories of assistance as defined by Latané and Darley:
- Direct intervention: directly assisting the victim
- Detour intervention. Detour intervention refers to reporting an emergency to the authorities (i.e. the police, fire department)
Implementation: After going through steps 1-4, the bystander must implement the action of choice.
In one study done by Abraham S. Ross, the effects of increased responsibility on bystander intervention were studied by increasing the presence of children. This study was based on the reaction of 36 male undergraduates presented with emergency situations. The prediction was that the intervention would be at its peak due to presence of children around those 36 male undergraduate participants. This was experimented and showed that the prediction was not supported and was concluded as "the type of study did not result in significant differences in intervention."
A meta-analysis (2011) of the bystander effect
 reported that "The bystander effect was attenuated when situations were perceived as dangerous (compared with non-dangerous), perpetrators were present (compared with non-present), and the costs of intervention were physical (compared with non-physical). This pattern of findings is consistent with the arousal-cost-reward model, which proposes that dangerous emergencies are recognized faster and more clearly as real emergencies, thereby inducing higher levels of arousal and hence more helping." They also "identified situations where bystanders provide welcome physical support for the potentially intervening individual and thus reduce the bystander effect, such as when the bystanders were exclusively male, when they were naive rather than passive confederates or only virtually present persons, and when the bystanders were not strangers."
An alternative explanation has been proposed by
Stanley Milgram, who hypothesized that the bystanders′ callous behavior was caused by the strategies they had adopted in daily life to cope with
information overload. This idea has been supported to varying degrees by empirical research.
Timothy Hart and Ternace Miethe used data from the National Crime Victimization Survey (NCVS) and found that a bystander was present in 65 percent of the violent
victimizations in the data. Their presence was most common in cases of physical assaults (68%), which accounted for the majority of these violent victimizations and less likely in
robberies (49%) and
sexual assaults (28%). The actions of bystanders were most frequently judged by victims as "neither helping nor hurting" (48%), followed by "helping" (37%), "hurting" (10%), and "both helping and hurting" (3%). Half of the attacks in which a bystander was present occurred in the evening, where the victim and bystander were strangers.
Ambiguity and consequences
Ambiguity is one factor that affects whether or not a person assists another in need. In some cases of high ambiguity, it can take a person or group up to 5 times as long before taking action than in cases of low ambiguity. In these cases, bystanders determine their own safety before proceeding. Bystanders are more likely to intervene in low ambiguity, insignificant consequence situations than in high ambiguity, significant consequence situations.
Understanding of environment
Whether or not a bystander intervenes may have to do with their familiarity of the environment where the emergency occurs. If the bystander is familiar with the environment, they are more likely to know where to get help, where the exits are, etc.
 Bystanders who are in an environment in which they are not familiar with the surroundings are less likely to give help in an emergency situation.
Priming the bystander effect
Research done by Garcia et al. (2002) indicate that priming a social context may inhibit helping behavior.
 Imagining being around one other person or being around a group of people can affect a person's willingness to help.
Cohesiveness and group membership
Group cohesiveness is another variable that can affect the helping behaviour of a bystander. As defined by Rutkowski et al., cohesiveness refers to an established relationship (friends, acquaintances) between two or more people.
 Experiments have been done to test the performance of bystanders when they are in groups with people they have been acquainted with. According to Rutkowski et al., the social responsibility norm affects helping behavior. The norm of social responsibility states that "people should help others who are in need of help and who are dependent on them for it." As suggested by the research, the more cohesive a group, the more likely the group will act in accordance to the social responsibility norm. To test this hypothesis, researchers used undergraduate students and divided them into four groups: a low cohesive group with two people, a low cohesive group with four people, a high cohesive group with two people, and a high cohesive group with four people. Students in the high cohesive group were then acquainted with each other by introducing themselves and discussing what they liked/disliked about school and other similar topics. The point of the experiment was to determine whether or not high cohesive groups were more willing to help a hurt "victim" than the low cohesive groups. The four member high cohesive groups were the quickest and most likely groups to respond to the victim who they believed to be hurt. The four member low cohesive groups were the slowest and least likely to respond to the victim.
Altruism research suggests that helping behaviour is more likely when there are similarities between the helper and the person being helped. Recent research has considered the role of similarity, and more specifically, shared group membership, in encouraging bystander intervention. In one experiment (2005), researchers found that bystanders were more likely to help an injured person if that person was wearing a football jersey of a team the bystander liked as opposed to a team the bystander did not like. However, when their shared identity as football fans was made salient, supporters of both teams were likely to be helped, significantly more so than a person wearing a plain shirt.
The findings of Mark Levine and Simon Crowther (2008) illustrated that increasing group size inhibited intervention in a street violence scenario when bystanders were strangers, but encouraged intervention when bystanders were friends. They also found that when gender identity is salient, group size encouraged intervention when bystanders and victims shared social category membership. In addition, group size interacted with context-specific norms that both inhibit and encourage helping. The bystander effect is not a generic consequence of increasing group size. When bystanders share group-level psychological relationships, group size can encourage as well as inhibit helping.
These findings can be explained in terms of self-categorization and
empathy. From the perspective of
self-categorization theory, a person’s own social identity, well-being is tied to their group membership so that when a group based identity is salient, the suffering of one group member can be considered to directly affect the group. Because of this shared identity, referred to as self-other merging, bystanders are able to empathize, which has been found to predict helping behaviour. For example, in a study relating to helping after eviction both social identification and empathy were found to predict helping. However, when social identification was controlled for, empathy no longer predicted helping behaviour.
In discussing the
case of Wang Yue and a later incident in
China, in which
CCTV footage from a
subway showed passengers fleeing from a foreigner who fainted,
UCLA anthropologist Yunxiang Yan asserted that the reactions can be explained by deeply seated historical cultural differences in Chinese
agrarian society, in which there was a stark contrast between how individuals associated with
ingroup and outgroup members, saying, "How to treat strangers nicely is one of the biggest challenges in contemporary Chinese society...The prevailing ethical system in traditional China is based on close-knit community ties, kinship ties." He continued, "A person might treat other people in the person's social group very, very nicely...But turn around, when facing to a stranger, and (a person might) tend to be very suspicious. And whenever possible, might take advantage of that stranger."
Diffusion of responsibility
Darley and Latané (1968) conducted research on diffusion of responsibility.
 The findings suggest that in the case of an emergency, when people believe that there are other people around, they are less likely or slower to help a victim because they believe someone else will take responsibility. People may also fail to take responsibility for a situation depending on the context. They may assume that other bystanders are more qualified to help, such as
police officers, and that their intervention would be unneeded. They may also be afraid of being superseded by a superior helper, offering unwanted assistance, or facing the legal consequences of offering inferior and possibly dangerous assistance. For this reason, some legislations, such as "
Good Samaritan Laws" limit
liability for those attempting to provide medical services and non-medical services in an emergency.
Organizational ombuds practitioners' research
A 2009 study published by
International Ombudsman Association in the Journal of the International Ombudsman Association suggests that—in reality—there are dozens of reasons why people do not act on the spot or come forward in the workplace when they see behavior they consider unacceptable.
The most important reasons cited for not acting were: the fear of loss of important relationships in and out of the workplace, and a fear of "bad consequences." There also were many reasons given by people who did act on the spot or come forward to authorities.
This practitioners' study suggests that the "bystander effect" can be studied and analyzed in a much broader fashion. The broader view includes not just a) what bystanders do in singular emergencies, b) helping strangers in need, when c) there are (or are not) other people around. The reactions of bystanders can also be analyzed a) when the bystanders perceive any of a wide variety of unacceptable behavior over time, b) they are within an organizational context, and c) with people whom they know. The practitioners' study reported many reasons why some bystanders within organizations do not act or report unacceptable behavior. The study also suggests that bystander behavior is, in fact, often helpful, in terms of acting on the spot to help,and reporting unacceptable behavior (and emergencies and people in need.) The ombuds practitioners' study suggests that what bystanders will do in real situations is actually very complex, reflecting views of the context and their managers (and relevant organizational structures if any) and also many personal reasons.
In support of the idea that some bystanders do indeed act responsibly,
Gerald Koocher and Patricia Keith Spiegel wrote a 2010 article related to an NIH-funded study which showed that informal intervention by peers and bystanders can interrupt or remedy unacceptable scientific behavior.
What Would You Do?
John Quiñones' primetime show,
What Would You Do? on ABC, tests the bystander effect. Actors are used to act out (typically non-emergency) situations while the cameras capture the reactions and actions of innocent bystanders. Topics include cheating on a millionaire test, an elderly person
Non-computer versus computers: computer mediated intervention
Research suggests that the bystander effect may be present in
computer-mediated communication situations.
 Evidence demonstrates that people can be bystanders even when they cannot see the person in distress. In the experiment, 400 online chat groups were observed. One of two confederates were used as victims in each chat room: either a male victim whose screen name was Jake Harmen or a female victim whose screen name was Suzy Harmen. The purpose of the experiment was to determine whether or not the gender of the victim mattered, if the size of each chat group had any effect and if asking for a person's help by directly using their screen name would have any effect. Results indicated that the gender of the victim had no effect on whether or not a bystander assisted the victim. Consistent with findings of Latané and Darley, the number of people present in the chat room did have an effect. The response time for smaller chat groups was quicker than in the larger chat groups. However, this effect was nonexistent when the victim (Suzy or Jake) asked for help from a specific person in the chat group. The mean response time for groups in which a specific person was called out was 36.38 seconds. The mean response time for groups in which no screen name was pointed out was 51.53 seconds. A significant finding of the research is that intervention depends on whether or not a victim asked for help by specifying a screen name. The group size effect was inhibited when the victim specifically asked a specific person for help. The group size effect was not inhibited if the victim did not ask a specific person for help.
Children as bystanders
Although most research has been conducted on adults, children can be bystanders too. A study conducted by Robert Thornberg in 2007 came up with seven reasons why children do not help when another classmate is in distress. These include:
dissociation, embarrassment association, busy working priority, compliance with a competitive norm, audience modelling, and responsibility transfer.
 In a further study, Thornberg concluded that there are seven stages of moral deliberation as a bystander in bystander situations among the Swedish schoolchildren he observed and interviewed: (a) noticing that something is wrong, i.e., children pay selective attention to their environment, and sometimes they don't tune in on a distressed peer if they're in a hurry or their view is obstructed, (b) interpreting a need for help—sometimes children think others are just playing rather than actually in distress or they display pluralistic ignorance, (c) feeling empathy, i.e., having tuned in on a situation and concluded that help is needed, children might feel sorry for an injured peer, or angry about unwarranted aggression (empathic anger), (d) processing the school's moral frames—Thornberg identified five contextual ingredients influencing children's behavior in bystander situations (the definition of a good student, tribe caring, gender stereotypes, and social-hierarchy-dependent morality), (e) scanning for social status and relations, i.e., students were less likely to intervene if they didn't define themselves as friends of the victim or belonging to the same significant social category as the victim, or if there were high-status students present or involved as aggressors—conversely, lower-status children were more likely to intervene if only a few other low-status children were around, (f) condensing motives for action, such as considering a number of factors such as possible benefits and costs, and (g) acting, i.e., all of the above coalesced into a decision to intervene or not. It is striking how this was less an individual decision than the product of a set of interpersonal and institutional processes.