Piliavin et al. · subway Samaritans

Approach and research focus

Social approach. The field study investigates how ordinary passengers respond when a stranger collapses, and whether the victim’s apparent condition, race, other bystanders and a helping model relate to assistance.

Background: Earlier laboratory studies suggested that helping can fall as the number of bystanders increases. Piliavin and colleagues observed a staged emergency in a crowded, face-to-face public setting to examine helping under natural travel conditions and the possible role of diffusion of responsibility.

Aim and procedure

About 4,450 passengers were present across 103 trials on New York’s Eighth Avenue subway between 59th and 125th Streets, on weekdays from late morning to mid-afternoon in 1968. Four student confederates worked as a male victim, a male model ready to help, and two female observers. The victim collapsed about 70 seconds into the express journey. In 65 trials he appeared ill and carried a cane; in 38 he smelled of alcohol and carried a bottle in a bag. The model’s presence and position/timing varied. Observers recorded whether and when passengers helped, passenger numbers, helper characteristics, movement and spontaneous comments. Riders were unaware that a trial was taking place.

Results and conclusion

Before the model intervened, the cane victim received spontaneous help in 62 of 65 trials; the drunk-appearing victim received it in 19 of 38. Median spontaneous-help latency was 5 seconds for the cane condition and 109 seconds for the drunk condition when non-respondents were assigned 400 seconds. Across 81 trials with a spontaneous first helper, 60% involved more than one helper and 90% of first helpers were men. The study did not find the simple prediction that larger groups consistently delayed help; the authors noted limits in interpreting crowd-size patterns.

Passengers often helped, especially when the victim appeared ill rather than drunk, so the study challenged a blanket claim that bystanders usually remain passive. The results did not establish that crowd size never matters: the subway was a constrained setting, help was often immediate and the study’s measures and conditions limit the diffusion test.

Evaluate the evidence

A field experiment recorded behaviour in a real public setting with many unsolicited passengers, giving the response ecological relevance and reducing awareness of observation.

The 103 trials were concentrated on one New York route, weekday daytime hours and a single era; the passengers are not 4,450 independent trials, and the setting limits generalisation.

Victim condition was not perfectly balanced across teams and victim race; the authors reported one black victim and uneven condition counts. Confederates and observers may also differ in how consistently they staged and coded events.

A visible collapse, helper presence and crowd size are difficult to isolate in a natural train. The model eventually intervened, so outcomes after the preset point are not spontaneous passenger helping.

Ethics and issues/debates

Passengers were unsuspecting and could not give prior informed consent; the staged collapse could cause concern or distress, and observers recorded behaviour in a public place without individual permission.

The model provided assistance if passengers did not help by the planned point, limiting the duration of the apparent emergency. The article describes the ethical tension between observing natural behaviour and protecting uninformed bystanders.

The field result applies to this particular crowded, face-to-face and time-limited situation; it does not refute every form of diffusion of responsibility.

The apparent illness of the victim affected both the frequency and speed of help; situational interpretation and perceived responsibility may influence action.

The findings can inform everyday discussion of helping, but should not be used to promise what an individual will do in a different emergency.