Piliavin, I M, Rodin, J and Piliavin, J A (1969), Good Samaritanism: An Underground Phenomenon? Journal of Personality and Social Psychology, 13(4): 289–99

Psychology Being Investigated

Bystander Effect and Diffusion of Responsibility – The study explores how the presence of other bystanders affects an individual’s likelihood of helping a person in need. This concept, known as the bystander effect, suggests that individuals are less likely to offer help when other bystanders are present, due to a diffusion of responsibility. The theory was initially proposed by social psychologists John Darley and Bibb Latané following the infamous Kitty Genovese murder in 1964.

Cost-Reward Model of Helping Behaviour – This model, as applied in the study, looks at the decision-making process behind helping behavior. It posits that individuals weigh the potential costs (e.g., embarrassment, danger) against the rewards (e.g., self-satisfaction, public praise) when deciding whether to help someone. This model is consistent with the principles of social exchange theory, which suggests that human interactions are driven by the desire to maximise rewards and minimise costs.

Influence of Victim Characteristics – The study examines how characteristics of the victim, such as their apparent condition (drunk vs. ill) and race, influence the likelihood of receiving help. This aspect of the study ties into theories related to empathy, prejudice, and in-group vs. out-group dynamics.

Modeling and Social Influence – The research also investigates the impact of modeling on helping behavior. According to social learning theory, people are influenced by observing the actions of others. In the context of this study, the presence of a model (a person who offers help) was hypothesized to increase the likelihood of bystanders offering assistance.


Kitty Genovese Incident and Bystander Effect – The study was influenced by the social and psychological impact of the Kitty Genovese murder in 1964, which led to a surge in research on bystander behavior. This incident highlighted the bystander effect, where the presence of other people inhibits an individual’s likelihood of helping in an emergency due to a diffusion of responsibility.

Diffusion of Responsibility – The concept of diffusion of responsibility, particularly explored by John Darley and Bibb Latané, forms a crucial theoretical background for this study. Their research indicated that people are less likely to help in an emergency when they believe others are also present, thereby diffusing the responsibility of action. The study builds upon the experiments conducted by Darley and Latané, such as their study involving bystanders who heard an epileptic seizure over earphones. These studies form a significant part of the theoretical and experimental background for this study.


The primary aim of the study by Piliavin et al. was to provide more information from a field setting, as opposed to a laboratory setting, to understand bystander behavior in real-life situations.

The study aimed to explore the dynamics of helping behavior in a more realistic and natural environment, moving beyond the controlled conditions of the laboratory which offer greater control but less reality. This shift towards a field setting was seen as crucial for gaining a deeper and more applicable understanding of how bystanders react in actual emergencies​​.


Research Method – The study used a field experiment approach.

Sample Size and Demographics – The subjects were approximately 4,450 men and women who traveled on the 8th Avenue Independent (IND) branch of the New York subways.

Sampling Technique – The study used a convenience sample of subway passengers during the experiment.


Team Composition and Entry – Each trial involved a team of four Columbia General Studies students, comprising two males and two females. These teams boarded the train using different doors. There were four different teams, and each team’s members always worked together for the data collection of 103 trials. The location of the experimental car was varied from trial to trial by each team.

Roles and Placement in the Train

  • The female confederates took seats outside the critical area of the train. They were responsible for recording data as unobtrusively as possible throughout the duration of the ride.
  • The male members of the team, one playing the role of the model and the other as the victim, remained standing. The victim always stood next to a pole in the center of the critical area.


  • Model’s Intervention (if necessary)
    • If the victim did not receive any assistance by the time the train slowed to a stop, the model would step in to help the victim to his feet​​.
  • Approximately 70 seconds after the train departed and passed the first station, the victim staggered forward and collapsed.
  • The victim would then remain supine on the floor, looking at the ceiling, until help was received.


  • Victim Condition – The victims were either in a ‘drunk condition’ (smelling of liquor, carrying a liquor bottle in a brown bag, 38 trials) or a ‘cane condition’ (appearing sober and carrying a black cane, 65 trials). Victims were dressed identically in both conditions.

Location and Timing

The experiment took place during the express run between the 59th Street and 125th Street stations, providing a consistent 7-½ minute duration for each emergency scenario.

Controls and Variables

  • Measured Variables – The primary variables measured were the time taken for the first passenger to help, the race, sex, and location of every helper, and the total number of individuals who came to the victim’s assistance.
  • Manipulated Variables – The conditions of the victim (drunk or cane) and the race of the victim were the primary manipulated variables.
  • Data Collection Techniques – Observational data were collected by the female confederates, including the details of the incident, responses of the bystanders, and other relevant observations.


Frequency of Help – The study found that the victims received help quite frequently, surpassing earlier laboratory results. The victim with a cane received spontaneous help in 62 of 65 trials, while the drunk victim received spontaneous help in 19 of 38 trials. This difference was not due to the number of potential helpers in the cars.

Latency of Response – The mean and median response times did not support the idea of diffusion of responsibility. In fact, response times were consistently faster for larger groups (7 or more individuals) compared to smaller groups (1 to 3 individuals). This challenges the expected decrease in response speed as group size increases, a phenomenon known as the diffusion of responsibility effect.

Major Findings

  • An apparently ill person (cane condition) is more likely to receive aid than a person who appears to be drunk.
  • The race of the victim had little effect on the race of the helper, except when the victim was drunk.
  • The longer an emergency continues without help, the more likely it is that someone will leave the area of the emergency.
  • Contrary to the expectations from the diffusion of responsibility theory, the study did not find a decrease in the speed of responding as group size increased.

Results by Victim Condition and Race

The graph above visually represents the key findings from the Piliavin et al. study on the “Good Samaritanism: An Underground Phenomenon.” It illustrates the percentage of times help was offered to different types of victims (either appearing as cane users or drunk) based on their race (either white or black). This visualization helps in understanding how the likelihood of receiving help varied depending on the perceived condition and race of the victim. For instance, victims with a cane (both white and black) were more likely to receive help compared to those who appeared drunk. ​

  • For white victims, the percentage of trials in which help was given was 100% in the cane condition and 73% in the drunk condition.
  • For black victims, the percentage of trials in which help was given was 100% in the cane condition and 67% in the drunk condition.


Differential Response Based on Victim’s Condition – An individual who appears to be ill is more likely to receive aid than one who appears to be drunk. This finding highlights the role of victim characteristics in influencing bystander intervention.

Diffusion of Responsibility – The study suggests that the concept of diffusion of responsibility, where bystander intervention decreases as the number of bystanders increases, may not be as straightforward as previously thought. In larger groups, the increase in the probability that someone will help might offset the deterrence to action resulting from a larger number of observers.

Comparison with Laboratory Studies – The study emphasises the differences between field studies and laboratory studies in understanding bystander behavior. It cautions against making direct generalisations from field studies to laboratory research due to the differing conditions and dynamics.

Need for Further Research – The study concludes that more work is needed in both natural and laboratory settings to fully understand the conditions under which diffusion of responsibility occurs or does not occur.


High Ecological Validity – The study boasts high ecological validity due to its real-life setting. By conducting the experiment in a natural environment (a New York City subway train), the study captured authentic behaviors and reactions of bystanders to emergencies. This high level of ecological validity ensures that the findings are more likely to be applicable to real-world situations, providing valuable insights into human behavior outside controlled laboratory settings.

Control of Variables – Despite being a field study, it effectively controlled for key variables such as the type and race of the victim, the presence or absence of a model, and group size. This control allowed the researchers to isolate the effects of these variables on bystander behaviour.

Detailed and Systematic Observation: The study employed detailed and systematic observation techniques. The observers (female confederates) were trained to record specific and relevant data points unobtrusively, ensuring that the data collected was both comprehensive and accurate. This approach not only provided depth to the study’s findings but also added to its overall reliability, as the data was collected systematically and methodically.


Ethical Concerns – The study involved deception, as the participants (subway passengers) were not aware they were part of an experiment. This raises ethical concerns, particularly regarding the informed consent of the participants. Additionally, the study potentially caused distress to the passengers who witnessed what they believed to be real emergencies.

Lack of Diversity in Sample: The study was conducted on the New York City subway and involved a convenience sample of subway passengers. This sample primarily consisted of individuals who happened to be present in the subway cars during the trials. While the study did involve a large number of trials (103 in total) and thus a substantial number of observations (approximately 4,450 passengers), the sample might not adequately represent the broader population. Subway passengers in New York City could have specific characteristics or behaviours not generalisable to other groups or settings. This limitation in sample diversity affects the study’s external validity, as its findings about helping behaviour might not be applicable to different populations or cultural contexts.

Possible Observer Bias: The study employed observers (female confederates) who were aware of the study’s aims and were part of the experimental team. These observers were responsible for recording the reactions of the bystanders during each trial. Since they were informed about the purpose of the study, their observations and interpretations could have been influenced by their knowledge and expectations. This awareness raises the potential for observer bias, where the observers’ perceptions and recordings of the events could be unintentionally skewed to align with the study’s hypotheses. Observer bias can affect the reliability and objectivity of the data collected, potentially impacting the study’s conclusions.