social psychology : pilavin Flashcards
aim
To study the factors affecting whether people would help a
collapsed man on the New York underground.
method
Experimenters faked collapse on New York underground
trains, and the number of people who helped and the time taken
to help were recorded. The race, apparent responsibility of the
victim (ill or drunk), the presence of a model helper, and the
number of passengers present were varied.
results
: 79% of victims received help. Help was more likely if the
victim appeared to be ill. There was some increased tendency for people to help those of their own race. The number of bystanders
made little difference, and most people were helped before the
model could initiate helping.
conclusion
Provided people are in a closed environment where
they cannot simply leave, they are likely to help someone in need.
Helping is most likely when the victim is seen as not responsible
for the situation and is the same race as helpers. The number of
bystanders is not important in this situation.
CONTEXT
This study is concerned with bystander behaviour. Bystanders are people who
witness events and have to choose whether to intervene or not. Recently there
has been a lot of debate over ‘have-a-go-heroes’ who put themselves at risk to
intervene and attempt to stop crimes taking place. Most of the time bystanders
can help without putting themselves at risk. However, surprisingly often we
choose not to act to help people in need.
CONTEXT : The Kitty Genovese murder
Psychological research into bystander behaviour was triggered by a murder
that took place in New York in 1964. Excerpts from the New York Times article
describing the incident are shown opposite.
Some of the details of the story as it was reported at the time have since been
challenged. Given the layout of the block, it would not have been possible for
anyone to have seen the whole incident, so each person would have seen just
fragments of the event. Also, the area was not actually as quiet as the article
implies – one neighbour said that rows between couples leaving a local bar
were common late at night. Given these facts, we cannot be sure that 38 people
really saw, correctly interpreted, and chose to ignore the murder. However, the
Genovese murder captured the public imagination and stimulated psychological
research into bystander behaviour.
CONTEXT : diffusion of responsibility
Latané and Darley (1968) proposed that the key issue in deciding whether we
help or not is whether we see it as our personal responsibility to do so. One
reason why groups of people do not help individuals in need is that responsibility
is shared equally among the group so that each person has only a small portion
of responsibility. They called this idea diffusion of responsibility. In a series
of lab experiments they demonstrated that the more people who are present in
an emergency, the less likely people are to help.
AIM
Piliavin et al. wanted to extend early studies of bystander behaviour in
several key ways. First, they wanted to study bystander behaviour outside the
laboratory, in a realistic setting where participants would have a clear view of the
victim. Second, they wanted to see whether helping behaviour was affected by
four variables:
1 The victim’s responsibility for being in a situation where they needed help
2 The race of the victim
3 The effect of modelling helping behaviour
4 The size of the group.
By observing variations in the size of the group, they were able to test whether
diffusion of responsibility occurred.
METHOD : Participants
An estimated total of around 4,550 passengers travelled in the trains targeted
by the researchers. These were all regarded by the researchers as ‘unsolicited
participants’. An average of 43 were present in each carriage in which the
procedure was conducted, and a model average of eight were in the immediate or
‘critical’ area. The racial mix of passengers was estimated as 45% black and
55% white.
METHOD : Design and procedure
The study was a field experiment carried out on trains on the New York subway.
The procedure involved a male experimenter faking collapse on a train between
stops, in order to see whether he was helped by other passengers. One particular
stretch of track was targeted where there was a 7.5 minute gap between two
stations.
Experimenters worked in teams of four, two females to record the results, and
two males who would play the roles of victim and model helper. There were four
teams, one containing a black male. Each male taking the role of victim took part
in both drunk and ill conditions. Seventy seconds after the train left a station the
victim would stagger and fall. He then lay still on his back with eyes open, not
moving until helped. Between six and eight trials were run on a given day, between
11am and 3pm. Four independent variables were manipulated in the procedure:
1 Victim’s responsibility: operationalised as carrying a cane (ill – low
responsibility) or smelling of alcohol and carrying a bottle wrapped in a
paper bag (drunk – high responsibility).
2 Victim’s race: operationalised as black or white.
3 Presence of a model: operationalised as whether a male confederate; either
close to or distant from the victim; helped after 70 or 150 seconds.
4 Number of bystanders: operationalised as however many people were
present in the vicinity.
Four males, aged 24–29, and identically dressed in casual clothes, took the role
of models of helping behaviour. Four model conditions were applied to both
apparently drunk and ill victims:
*Model stood in the critical area and helped after 70 seconds
*Model stood in the critical area and helped after 150 seconds
*Model stood in the adjacent area and helped after 70 seconds
*Model stood in the adjacent area and helped after 150 seconds.
The dependent variable, helping, was measured in the following ways:
*Time taken for first passenger to help
*Total number of passengers who helped.
In addition the gender, race and position of each helper was noted. Qualitative
data was also gathered in the form of comments from passengers.
RESULTS
Overall, a higher proportion of people helped than was the case in previous
laboratory experiments: 79% of victims received spontaneous help from
passengers, and in 60% of cases where the victim was helped it was by more
than one person. Most helpers were male.
*Ill versus drunk conditions. In the cane condition, the victim received
help 95% of the time without intervention from a model. In the drunk
condition, this was reduced to 50%. People took longer to help the drunk
victim than the ill one: over 70 seconds in 83% of the drunk trials, but in
only 17% of the cane trials. However the proportion of cases in which more
than one person helped was the same.
*Race of victim. In the cane condition, black and white victims were equally
likely to be helped. However, in the drunk condition, black victims were less
likely to receive help. Also, in the drunk condition, there was a slight same-
race effect – people were a little more likely to help a drunk of the same
race as themselves. The proportion of cases in which help came from
more than one person did not vary by race.
*The effect of modelling. The model intervening after 70 seconds was
more likely to lead to help from other passengers (in nine cases) than the
one intervening after 150 seconds (three cases). However, the researchers
noted that because passengers helped spontaneously in the vast majority
of trials, there were too few cases of helping after modelling to analyse
in detail.
*Number of bystanders. There was no evidence for diffusion of
responsibility. There was a mild effect in the opposite direction –
when more passengers were present, people were slightly more likely to
receive help.
*Other observations. In a significant minority of trials (21 of 103), some
passengers moved away from the critical area. More comments were made
in drunk trials, and more when no passenger spontaneously helped. The
researchers interpreted this as meaning the comments were in response to
passengers feeling uncomfortable about the situation.
CONCLUSIONS
Piliavin and his colleagues admitted that the situation they set up was unusual
in that their participants were trapped in a carriage with a collapsed person and
therefore could not simply walk away as they could normally. In this situation:
*An ill person is more likely to receive help than a drunk person.
*Men are more likely to help another man than women are.
*People are slightly more likely to help someone of their own ethnic group,
especially when they appear drunk.
*There is no strong relationship between size of group and likelihood of
helping. The small correlation between group size and helping behaviour is
positive rather than negative. Therefore there is no support for diffusion of
responsibility.
*The longer an incident goes on, the less likely people are to help (even if
help is modelled), the more likely people are to leave the area, and the more
likely they are to discuss the incident.
CONCLUSIONS : Explaining the findings
The researchers explained the findings in terms of arousal and the costs and
rewards of alternative responses. Perceiving an emergency raises arousal levels.
According to the situation, this can be interpreted either as sympathy or as fear
and disgust. The closer one is to the emergency and the longer it continues,
the more arousal increases. Arousal is also greatest when the bystander can
empathise with the victim. The behaviour of bystanders aims to reduce the
arousal level. This can be achieved in four ways:
1 helping directly
2 leaving to find help
3 leaving the area
4 dismissing the victim as unworthy of help.
Which of these options is chosen depends on the costs and benefits of helping or
not helping. These are shown in Table 1.4. If the benefits of helping and the costs
of not helping outweigh the costs of helping and the benefits of not helping, then
help will be offered.
This model explains neatly the behaviour of participants. The drunk is helped
less because there are greater costs in terms of fear and embarrassment. People
help their own ethnic group more because they can feel more empathy with
them. Late modelling has less effect than early modelling because people will
have found another coping strategy by then.
TABLE 1.4 COSTS AND BENEFITS OF
ALTERNATIVE RESPONSES
costs of helping : benifit - praise , cost - fear, embaressment, effort, disgust.
costs of not helping : costs - self blame, nlame from others, benifits : continueing with work.
EVALUATION : The research method
The study was a field experiment, and field experiments are
associated with particular strengths and weaknesses. The
major strength is the natural environment in which it is
conducted and the opportunity to create realistic situations.
The weaknesses centre around the difficulty in controlling
variables. In this case the researchers put into place a number
of controls. The victims were the same age and sex, and
dressed identically. They collapsed in the same way. The
same stretch of track was used on each trial to ensure that
there was always the same time available, and the timings
before a model helped were kept the same. Nevertheless,
not all conditions could be kept constant. The number and
nature of the passengers boarding the particular carriage was
unpredictable. Some might have seen the incident several
times, and, if so, it is hard to predict how they might have
responded to that.
EVALUATION : Quantitative and qualitative data
Like Milgram, Piliavin et al. collected both quantitative and
qualitative data. However, their emphasis was very much on
the quantitative. This was appropriate, given the aims of the
study. They were interested in how many people helped and
how long it took them to do so under each condition. However,
qualitative data in the form of what comments people made
about victims collapsing was also useful as an indicator that
people were responding to the emergency by justifying why they were not helping. This was important, as this became one
of the strategies for reducing arousal levels in Piliavin et al.’s
model of bystander behaviour.
