Core study 2: PILIAVIN ET AL.

(SUBWAY SAMARITANS)

Background to the study

- This study is concerned with bystander behavior. Bystanders are people who witness events and
have to choose whether to intervene or not.
- The study is inspired by the case of Kitty Genovese who was murdered in New York, in the early
sixties.
- In 1964 at 3.30a.m, 28 year old Kitty Genovase was stabbed to death by a knife wielding attacker
outside her apartment in New York. She was attacked for 30 minutes and her attacker twice
returned to inflict further injuries. Her screams woke up 38 of her neighbors although no one
intervened or called the police for 20 minutes.
- Following this murder and the apparent apathy of 38 witnesses, where no one offered to help,
many psychologists began to conduct studies into the phenomenon of ‘bystander behavior.

The psychology being investigated

- The psychology being investigated here is diffusion of responsibility.
- Diffusion of responsibility refers to the fact that as the number of bystanders increases,
the personal responsibility that an individual bystander feels decreases. As a consequence,
so does his or her tendency to help. Thus, a bystander who is the only witness to an
emergency will tend to conclude that he or she must bear the responsibility to help, and in
such cases people typically do help. But bystanders diffuse responsibility to help when
others are present.
- Previous studies by Darley and Latane ('Woman falls off chair in next room’ study,
‘Epileptic seizure’ study and ‘Smoke room’ study) showed that the amount of helping
behavior decreases as the group size increases- this idea is called diffusion of
responsibility.
Aim
i) To investigate the factors that influence helping behavior
ii) To test whether diffusion of responsibility will occur in a real life setting.

Method
The study was a field experiment carried out on trains on the New York subway. One particular stretch
of track was targeted where there was a 7.5 minute gap between two stations.

Observation method (covert) was also used.

pg. 1
Experimenters
Students from Columbia University in New York: four male victims (three white and one black, aged 26-
35); four male models (all white, aged 24-29 years); and eight female observers. The experimenters
were divided into four teams of one victim, one model and two observers.

Participants and sampling technique

- 4450 men and women traveling between the Harlem and Bronx between 11a.m and 3pm. unaware
that they were involved in an experiment (‘naive’ or ‘unsolicited participants’).
- Their racial composition was 45% black and 55% white.
- The mean number of passengers per carriage was 43, and the mean number of people in the critical
area was 8.5.
- An opportunity sampling technique was used as it consisted of participants who just happened to
be on that train on that day.
Controls:
- The same 7 ½ minute train journey for all 103 trials
- Victims wore the same clothes and fell over at the same time (after 70 seconds) in the same place
and in the same way.
- Each team member started the journey in the same place

Procedure
- The procedure involved a male experimenter faking a collapse on a train between stops, in order to
see whether he was helped by other passengers.
- Experimenters worked in teams of four consisting of two females to record the results, and two
males who would play the roles of victim and model helper.
- There were four teams. Each male taking the role of victim took part in both drunk and ill
conditions. In one of the teams, the victim was black.
- Seventy seconds after the train left the station the victim would stagger and fall. He then lay still on
his back with open eyes not moving until helped.
- Between six and eight trials were run each day, between 11a.m. and 3pm., for just over ten weeks.
Four IVs were manipulated in the procedure:
i. The type of victim: operationalised as carrying a cane (ill) or smelling alcohol and carrying a
bottle wrapped in a paper bag (drunk).
ii. The race of the victim: operationalised as black or white.
iii. The type of model: operationalised as whether a male confederate was either close or distant
from the victim (critical early/ critical late; and adjacent early/ adjacent late).

pg. 2
 iv. The number of bystanders: operationalised as how many passengers were present in the subway
carriage.
- Four males, aged 24-29 years, all white, and identically dressed in casual clothes, took the roles of
models of helping behavior. When helping, the model raised the victim to the sitting position and
stayed with him until the train reached the next stop. F our model conditions were applied to both
drunk and ill victims:
i. Critical- early: the model stood in the critical area and helped after 70 seconds.
ii. Critical- late: the model stood in the critical area and helped after 150 seconds.
iii. Adjacent- early: the model stood in the adjacent area and helped after 70 seconds.
iv. Adjacent- late: the model stood in the adjacent area and helped after 150 seconds.
DV was the level of bystander helping. This was operationalised as the time taken for the first passenger
to help, as well as the total number of passengers who helped. The race, gender and location of each
helper was also recorded.
Qualitative data was also gathered in the form of comments from passengers.

Results/ Findings
There was no significant evidence for diffusion of responsibility as the models aided the victim in only a
few instances.
The cane victim received spontaneous help, 95% of the time in 62 of the 65 trials, even before the
model could offer help.
The drunk also got spontaneous help, 50% of the time in 19 out of the 38 trials (there were less drunk
condition trials and they were acted out mostly by the black student who was the only black individual
in all the four groups).
In 60% of the 81 trials where the victim was helped, he received help from more than one person.
Even when there were females in the critical area, they were less likely to help. Males in the critical area
were always the first responders to the incident. Out of the first 81 helpers in the experiment, 90% of
them were males.
Same race helping was apparent in the study -out of the 65 trials where there was a white victim, 68% of
the helpers were white compared to 16 trials where 50% of whites helped the black victim. This was a
slight difference compared to the significant difference where the drunk was helped mainly by members
of his own race.
Comments were collected on trials in which no one offered to help until after 70 seconds. The female
observers who did not help in the drunk condition tried to justify their inaction through comments such
as: "It's for men to help him," or "I wish I could help him—I'm not strong enough," "I never saw this kind

pg. 3
of thing before—I don't know where to look," "You feel so bad that you don't know what to do."

Conclusion
There is no strong relationship between the number of bystanders (size of the group) and the speed of
helping; Diffusion of responsibility hypothesis was not supported. The larger the group, the faster the
response.
An ill person is more likely to receive help than a drunk person.
More men are likely to help than women.
Same race helping is more likely, especially for the drunk victim than the ill victim.

Piliavin et al. proposed an alternative explanation for their findings known as the Cost-Reward model.
Diffusion of responsibility did not occur. The diffusion of responsibility hypothesis states that the more
people that are present, the less helping behavior occurs. In this experiment, the fastest help came
from the largest groups.
▪ Why? Piliavin et al. created a theory to explain this behavior, called the Arousal: cost-reward
theory.
▪ This theory states that the observation of a situation will create arousal (the type depends on the
situation), which will then cause the subject to choose a response based on a "cost-reward
analysis" by the individual. These include: Costs of helping, such as effort, embarrassment and
possible physical harm. Cost of not helping, such as self-blame and perceived censure from others;
Rewards of helping, such as praise from self, onlookers and the victim; Rewards of not helping, such
as getting on with one’s own business and not incurring the possible costs of helping.
▪ Thus, we do not act out of pure altruism, but as a means of reducing feelings of unpleasant arousal.

Evaluation of the whole study

Strengths
i) The study produces both qualitative and quantitative data which provides an in-depth
perspective into helping behavior.
ii) Research Method: The field experiments were not subject to demand characteristics, as
the travelers on the underground train did not know they were part of an experiment.
Several different experimental conditions that added to the validity of the findings.
iii) There is high ecological validity as the study took place in a natural or realistic
environment- the New York subway system- and the participants were actual commuters
on train A and D. The participants were also not aware they were being observed making
their behavior more natural & valid, and less prone to demand characteristics.

pg. 4
 iv) The study has a high amount of control for example the victim is either a drunk or a cane
victim, the early and late model have a predetermined period in which to help, either after
70 seconds or 2 ½ minutes.

Weaknesses

i) Ethical Considerations: There a number of ethical issues associated with Piliavin et al.’s

study. Participants were unaware that they were taking part in an experiment, therefore
they could not consent to take part and it was also not possible to withdraw from the study
or be debriefed. Furthermore, seeing a victim collapse may have been stressful for the
participants, they also may have felt guilty if they didn’t help, therefore leading
to psychological harm.
ii) Reliability: Because of methodological problems, there were more cane trials than drunk
trials and more white victims than black victims. Therefore the results from each group’s
trials cannot always be reliably compared with each other. The victims and helpers were all
male and the results cannot be generalized to females or a mixed group. Further
research could investigate if women are more likely to help if the victim and the helper are
both female.

pg. 5