Why People Don't Always Step In: The Psychology of the Bystander Effect

When everyone is watching, who takes action?

A shopping bag splits open in a busy pedestrian square. Apples roll towards the kerb, and its owner crouches beside scattered belongings. One passer-by slows down. Another glances across, then looks at the people nearby. A third wonders whether assistance would be welcome. For a few seconds, several people notice the same difficulty while each waits for somebody else to respond.

This fictional moment is ordinary rather than dramatic, but it exposes an important social puzzle. Other people can make us feel that helping is somebody else's responsibility. They can also provide reassurance, practical support and the beginnings of a coordinated response. The bystander effect describes circumstances in which the presence of others inhibits an individual's intervention. It does not mean that crowds inevitably abandon people who need assistance. Understanding helping requires examining how people interpret situations, relate to one another and decide what they can safely do.

The discovery of the bystander effect

John Darley and Bibb Latané challenged explanations of non-intervention based simply on selfishness. In their 1968 intercom experiment, students believed they were discussing personal difficulties with other participants. A recording then simulated someone experiencing a seizure. Participants thought they were the only witness, or that one or four other listeners were present. Believing that more people could respond reduced and delayed reporting to the researcher (Darley & Latané, 1968).

Another experiment introduced smoke while male students completed questionnaires, either alone, alongside two deliberately passive accomplices, or with other genuine participants. The social setting influenced whether the smoke was reported: apparent calm could make an ambiguous situation seem less alarming (Latané & Darley, 1968).

These designs separated influences that everyday observations entangle. However, a simulated emergency among students cannot establish the frequency of helping across public life. An early review documented both substantial evidence of inhibition and important boundary conditions (Latané & Nida, 1981). Even early field experiments involving staged collapses on New York subway trains did not reproduce the expected diffusion pattern (Piliavin et al., 1969). Context mattered from the beginning.

Diffusion of responsibility

Imagine noticing an unattended suitcase blocking a station entrance. When you are alone, the connection between noticing and responding is direct. In a crowd, you might assume that someone closer has alerted station staff. Each person can make that assumption without anyone checking whether assistance has actually been requested.

Diffusion of responsibility concerns this weakening of personal obligation when other potential helpers are present. Responsibility has not disappeared objectively; it feels less specifically assigned. The intercom experiment was especially informative because participants could not see the other supposed listeners' reactions. Their mere presumed presence affected behaviour.

Delay is therefore not a reliable measure of empathy. Darley and Latané described visible anxiety among some people who failed to report the simulated emergency. Concern and indecision can coexist. Recognising this helps explain inaction without excusing its consequences or pretending that motives can be read from a brief observation.

Pluralistic ignorance — When everyone misreads the situation

Now imagine hearing a loud crash behind a closed shop door. You look around for clues. Everyone else appears calm, so you infer that nothing serious has happened. They may be making exactly the same inference from your expression.

Pluralistic ignorance arises when people misread others' public behaviour as evidence of their private beliefs or feelings. Several uncertain people can collectively produce an appearance of certainty. The smoke experiment illustrates how this process may influence the interpretation of an unfamiliar event. Trying to avoid embarrassment can make someone conceal the very concern that would help others recognise a problem.

This differs from diffusion of responsibility. Pluralistic ignorance concerns whether help is needed; diffusion concerns whether I should provide it. Someone may recognise an emergency clearly yet assume another person will act.

The wider phenomenon extends beyond emergencies. Research on students' drinking attitudes found that participants overestimated peers' comfort with campus drinking practices (Prentice & Miller, 1993). It demonstrates mistaken norm perceptions, rather than directly testing emergency intervention.

Making concern visible can change the information available to a group. In the shop-door example, saying “I heard a crash; shall we check?” gives others a chance to acknowledge their uncertainty. This is a plausible application of the mechanism, not evidence that one particular phrase has been proven to overcome hesitation in every setting.

The psychology of deciding to help

Latané and Darley's (1970) decision model organises helping into five stages:

1. Notice the event. Attention may be absorbed by a conversation, a phone or an urgent destination.
2. Interpret it as requiring intervention. Distinguish ordinary activity from a situation in which assistance is needed.
3. Accept personal responsibility. Recognise a possible role instead of assuming somebody else has acted.
4. Determine how to help. Identify an appropriate response within your abilities and the situation's risks.
5. Act on the decision. Translate intention into a concrete step, which might include seeking assistance.

This is a useful framework, not a claim that every real decision follows five conscious, strictly ordered calculations. People can act quickly, revise interpretations or coordinate while assessing what is happening.

Situational pressures also matter before responsibility becomes the central issue. In a field experiment involving theology students, those placed under greater time pressure were less likely to stop for someone apparently needing assistance (Darley & Batson, 1973). The experiment tested helping under particular conditions; it did not establish that hurried people have less compassionate personalities.

The Kitty Genovese story — Separating history from psychological myth

The murder of Kitty Genovese in New York in 1964 profoundly influenced public discussion of helping. A widely repeated account described 38 witnesses watching without responding. That story became a familiar illustration of collective indifference.

It should not be presented as established fact. Examining archival material, Manning, Levine and Collins (2007) found that the evidence did not support the iconic account of 38 fully informed, inactive witnesses. What different people could see, hear and understand was not equivalent to watching one clearly recognisable event together. The blanket claim that nobody responded is also unsupported.

Genovese was a person whose life was taken, not an experimental demonstration. Correcting the historical narrative respects that distinction. It also leaves a separate question intact: what do controlled studies reveal about social influences on helping? A flawed origin story does not invalidate independently obtained findings. As our article on reconstructive memory explores from another perspective, a compelling account still needs corroboration.

What modern research actually reveals

Danger changes the pattern

Fischer and colleagues' (2011) meta-analysis combined 105 independent effect sizes from more than 7,700 participants. The overall standardised effect was g = −0.35, indicating less individual helping when others were present. This is a standardised difference, not a 35% reduction in the probability of assistance.

Inhibition was weaker in situations perceived as dangerous, when perpetrators were present and when intervention involved physical costs. Other bystanders could sometimes offer potential support. An earlier experiment had similarly found inhibition under lower perceived danger but comparable helping with and without another bystander under higher danger (Fischer et al., 2006).

Danger can make a need clearer while also making intervention harder. Neither result means that danger always increases helping or that joining a confrontation is advisable. A meta-analysis summarises varied studies; comparisons between study characteristics do not isolate one universal causal mechanism.

One person's probability and a group's probability

Philpot and colleagues (2020) examined CCTV recordings of 219 public conflicts in the United Kingdom, the Netherlands and South Africa. At least one person intervened in 90.9% of incidents: approximately nine out of ten. Larger numbers of bystanders were associated with a higher probability that somebody intervened.

The distinction is crucial. The probability that a particular individual helps is different from the probability that at least one person helps. A larger pool of potential helpers can increase the latter even when each person's likelihood is lower. These findings therefore do not disprove the entire bystander effect.

Indeed, a separate CCTV study observed 764 bystanders across 81 violent incidents. Individual intervention became less likely with increasing bystander numbers, while social relationships strongly predicted helping (Liebst et al., 2019).

Consider a purely hypothetical comparison. One person alone might have a higher chance of helping than any particular member of a larger group. Yet the group contains several opportunities for somebody to respond. We cannot simply multiply those chances as if decisions were independent: people watch, influence and assist one another. Researchers therefore need to specify whether their unit of analysis is the person or the whole incident before interpreting a crowd-size result.

Video captures behaviour in genuine events without researchers staging an emergency. It cannot reveal every motive, establish that crowd size caused an outcome or represent every emergency. Recorded intervention also need not be successful intervention. Experiments offer stronger control; observational footage offers realism. Together they support a conditional account of helping.

Why groups can encourage helping

A crowd is not simply a collection of interchangeable witnesses. Friends can communicate, anticipate one another's responses and make assistance feel collectively supported. Shared identity can also connect people who have never met.

Across four experiments, Levine and Crowther (2008) showed that relationships, salient group membership and context-specific norms could alter whether increasing group size encouraged or inhibited intervention. The presence of friends differs psychologically from the presence of unrelated strangers. Group norms matter too: belonging can support assistance or pressure people to remain uninvolved.

In field experiments involving football supporters, an apparently injured stranger was more likely to receive help when wearing an in-group team's shirt. Making a broader football-supporter identity salient extended helping towards a rival supporter (Levine et al., 2005). The boundaries of “us” were flexible, although broader identification did not make helping universal.

Observational research likewise found that conciliatory third-party behaviour could involve collective action (Levine et al., 2011). Interviews with emergency survivors linked shared identity with solidarity, while relying on retrospective accounts rather than experimental manipulation (Drury et al., 2009).

Meaningful connection and physical proximity are different. Our discussion of social connection and belonging considers that distinction more broadly. Strangers can develop solidarity, but merely placing people together does not guarantee a shared commitment to help.

The bystander effect in the digital age

A harmful message in a group chat presents different uncertainties from an incident in a square. Who has actually read it? Is apparent joking welcome? Would a public reply expose the target to further attention? Online intervention might involve private support, reporting or refusing to amplify harassment.

Two experiments found that a larger apparent audience reduced participants' felt responsibility and willingness to intervene in cyberbullying scenarios (Obermaier et al., 2016). Another scenario experiment involving 453 Flemish students found that incident severity and other bystanders' identity influenced helping intentions (Bastiaensens et al., 2014). Intentions in a study are informative but cannot be equated with actions during sustained harassment.

A systematic review identified relationships, social context, empathy and confidence in one's ability to help as relevant factors (Domínguez-Hernández et al., 2018). A September 2026 review reached a similarly contextual conclusion. Its adolescent-focused synthesis drew on a 42-report core corpus within a wider map of 49 reports. Most evidence was cross-sectional and self-reported; findings concerning platform design and sociocultural norms were comparatively limited or mixed (Wang et al., 2026).

Perceived anonymity, visible audience numbers and reporting tools may influence responsibility or the costs of responding. However, anonymity can mean different things: whether the aggressor, target or potential helper is identifiable. Persistent posts, uncertain readership and invisible private support further complicate comparisons with physical crowds. It would be premature to claim that a particular button or audience counter reliably causes intervention across platforms.

In an earlier chat-room study, addressing a help request to a named person shortened response times and largely removed the audience-size association (Markey, 2000). That concerned a simple information request, not threatening harassment. The result supports examining direct communication while keeping the limits of transfer clear.

Can we become more effective bystanders?

Understanding these mechanisms offers practical possibilities, but awareness alone should not be treated as a proven remedy. Training studies provide stronger evidence than an appealing slogan.

A meta-analysis of school bullying-prevention programmes found a modest improvement in bystander intervention (Polanin et al., 2012). A later synthesis of 49 intervention studies across bullying and cyberbullying reported a small average effect, d = 0.25 (Chen et al., 2025). Programmes and outcome measures varied. These findings do not establish that every component works independently or that a brief lesson reliably changes adult behaviour in public emergencies.

Several responses are nevertheless consistent with the evidence and sensible safety principles:

• Clarify uncertainty. If it is safe, ask whether assistance is wanted rather than treating surrounding silence as proof that everything is fine.
• Take a defined role. Decide on one useful step, such as alerting station staff, rather than assuming that someone has done so.
• Communicate directly. Request a specific person's help with a specific task, and check whether it has been completed.
• Choose a proportionate response. Keep a safe distance from danger and seek appropriate assistance. Helping need not involve physical confrontation.
• Support without amplifying harm. Online, consider private contact or reporting options and the target's preferences before redistributing abusive material.

Evaluation should also distinguish recognition, confidence, intentions and observed behaviour. Feeling better prepared after a workshop is valuable, but does not by itself show that assistance increased months later. Nor should success be defined only as confronting an aggressor: quiet support and effective requests for assistance may matter, while a conspicuous response can sometimes increase exposure to harm.

These are evidence-informed applications, not a separately validated universal checklist. The safest useful action depends on what is happening and what someone can reasonably do. Practising concrete options may improve preparedness, but claims about lasting behavioural change require direct evaluation.

The psychology of helping is more hopeful than we think

Return to the scattered shopping in the square. One person asks whether help would be useful. Another retrieves an apple before it reaches the road. A third holds the replacement bag open. The same group that initially appeared hesitant has become a source of coordinated assistance.

This fictional ending is not a promise about every crowd. People can fail to respond, and the consequences can be serious. Yet laboratory inhibition, everyday uncertainty and widespread real-world intervention belong within the same explanation. Helping depends on attention, interpretation, responsibility, relationships, norms, skills and perceived risk.

Psychology gives us reasons to examine hesitation before calling it indifference, and reasons to recognise solidarity alongside failure. Communities make assistance more likely when people can communicate concern, see practical ways to contribute and understand responsibility as something they can share through action. The presence of others can become a resource when somebody helps turn uncertainty into a response.

Written by Arkadiusz Dawidowicz · Social Psychology. Sources checked on 28 October 2026.

References

Bastiaensens, S., Vandebosch, H., Poels, K., Van Cleemput, K., DeSmet, A., & De Bourdeaudhuij, I. (2014). Cyberbullying on social network sites. An experimental study into bystanders' behavioural intentions to help the victim or reinforce the bully. Computers in Human Behavior, 31, 259–271. https://doi.org/10.1016/j.chb.2013.10.036

Chen, Q., Lin, W., Wu, Q., & Chan, K. L. (2025). The effectiveness of interventions on bullying and cyberbullying bystander: A meta-analysis. Trauma, Violence, & Abuse, 26(5), 938–954. https://doi.org/10.1177/15248380241297362

Darley, J. M., & Batson, C. D. (1973). “From Jerusalem to Jericho”: A study of situational and dispositional variables in helping behavior. Journal of Personality and Social Psychology, 27(1), 100–108. https://doi.org/10.1037/h0034449

Darley, J. M., & Latané, B. (1968). Bystander intervention in emergencies: Diffusion of responsibility. Journal of Personality and Social Psychology, 8(4, Pt. 1), 377–383. https://doi.org/10.1037/h0025589

Domínguez-Hernández, F., Bonell, L., & Martínez-González, A. (2018). A systematic literature review of factors that moderate bystanders' actions in cyberbullying. Cyberpsychology: Journal of Psychosocial Research on Cyberspace, 12(4), Article 1. https://doi.org/10.5817/CP2018-4-1

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(3), 487–506. https://doi.org/10.1348/014466608X357893

Fischer, P., Greitemeyer, T., Pollozek, F., & Frey, D. (2006). The unresponsive bystander: Are bystanders more responsive in dangerous emergencies? European Journal of Social Psychology, 36(2), 267–278. https://doi.org/10.1002/ejsp.297

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. https://doi.org/10.1037/a0023304

Latané, B., & Darley, J. M. (1968). Group inhibition of bystander intervention in emergencies. Journal of Personality and Social Psychology, 10(3), 215–221. https://doi.org/10.1037/h0026570

Latané, B., & Darley, J. M. (1970). The unresponsive bystander: Why doesn't he help? Appleton-Century-Crofts. National Diet Library catalogue record

Latané, B., & Nida, S. (1981). Ten years of research on group size and helping. Psychological Bulletin, 89(2), 308–324. https://doi.org/10.1037/0033-2909.89.2.308

Levine, M., & Crowther, S. (2008). The responsive bystander: How social group membership and group size can encourage as well as inhibit bystander intervention. Journal of Personality and Social Psychology, 95(6), 1429–1439. https://doi.org/10.1037/a0012634

Levine, M., Prosser, A., Evans, D., & Reicher, S. (2005). Identity and emergency intervention: How social group membership and inclusiveness of group boundaries shape helping behavior. Personality and Social Psychology Bulletin, 31(4), 443–453. https://doi.org/10.1177/0146167204271651

Levine, M., Taylor, P. J., & Best, R. (2011). Third parties, violence, and conflict resolution: The role of group size and collective action in the microregulation of violence. Psychological Science, 22(3), 406–412. https://doi.org/10.1177/0956797611398495

Liebst, L. S., Philpot, R., Bernasco, W., Dausel, K. L., Ejbye-Ernst, P., Nicolaisen, M. H., & Lindegaard, M. R. (2019). Social relations and presence of others predict bystander intervention: Evidence from violent incidents captured on CCTV. Aggressive Behavior, 45(6), 598–609. https://doi.org/10.1002/ab.21853

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. https://doi.org/10.1037/0003-066X.62.6.555

Markey, P. M. (2000). Bystander intervention in computer-mediated communication. Computers in Human Behavior, 16(2), 183–188. https://doi.org/10.1016/S0747-5632(99)00056-4

Obermaier, M., Fawzi, N., & Koch, T. (2016). Bystanding or standing by? How the number of bystanders affects the intention to intervene in cyberbullying. New Media & Society, 18(8), 1491–1507. https://doi.org/10.1177/1461444814563519

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. https://doi.org/10.1037/amp0000469

Piliavin, I. M., Rodin, J., & Piliavin, J. A. (1969). Good Samaritanism: An underground phenomenon? Journal of Personality and Social Psychology, 13(4), 289–299. https://doi.org/10.1037/h0028433

Polanin, J. R., Espelage, D. L., & Pigott, T. D. (2012). A meta-analysis of school-based bullying prevention programs' effects on bystander intervention behavior. School Psychology Review, 41(1), 47–65. https://doi.org/10.1080/02796015.2012.12087375

Prentice, D. A., & Miller, D. T. (1993). Pluralistic ignorance and alcohol use on campus: Some consequences of misperceiving the social norm. Journal of Personality and Social Psychology, 64(2), 243–256. https://doi.org/10.1037/0022-3514.64.2.243

Wang, Z., Wei, X., Saad, S., De Costa, F., & Liu, T. (2026). Factors associated with bystander intervention in cyberbullying: A systematic review. Frontiers in Psychology, 17, Article 1912279. https://doi.org/10.3389/fpsyg.2026.1912279