Core Study 2:
Piliavin et al. (subway
Samaritans)
Aim
To investigate the effects of 4 situational variables on helping
behaviours of bystanders:
- The type of victim
- The race of victim
- The behaviour of a ‘model’
- The size of the group of bystanders.
Background
Murder of a lady, named Kitty Genovese, in NYC- many people
were either eye or ear witnesses but failed to prevent her murder.
Diffusion of responsibility hypothesis:
- More bystanders= less helping behaviour- responsibility
diffused (someone else will help)
- More helpers= increased helping behaviour- responsibility
diffused (we should also help)
Method
Research method and design
Field experiment: Natural environment- NYC subway.
Independent group design: different days so different people.
4 IVs operationalized as:
- The type of victim: ‘drunk’ or ‘ill’
- The race of victim: black or white
- The behaviour of a model: close or distant + early or late
- The size of the group of bystanders: naturally occurring no.
of passengers in the train.
DV: level of bystander helping
Quantitative data
- Time taken by the 1st passenger to help
- Total no. of helpers
- Race, gender and location in carriage of each helper.
Qualitative data
- Verbal remarks by each passenger
Sample
Opportunity sample: Passengers travelling on an underground
service on weekdays, between Harlem and Bronx, from 11am to
3pm.
Total estimated no. of participants= 4450
Blacks: 45% , Whites: 55%
Mean no. of passengers per carriage= 43
Mean no. of passengers in the critical area =8.5
Procedure
4 teams of student researchers
Standardized procedure
Each trial: 2 male + 2 female students- boarded from diff doors
Female confederates:
- Sat adjacent to ‘critical’ area (where the incident happened)
- Role: observers (recorded data)
Male confederates:
- Role: victim and ‘model’ (helper)
- Victim: stood at the pole, in the center of critical area
- Model: remained standing the whole time
Same route in each trial (7.5 mins journey)
At 70s: victim staggered forward and collapsed- remained lying,
looking up- if no help: model would.
Victim:
- Different males but similar looking
- Ages: 26-35 years
- 3 white, 1 black
- Identical + casual clothing (jacket, old trousers, no tie)
- 38/103 trials: smelled alcohol + carried alcohol bottle
wrapped in a brown bag
- 65/103 trials: sober + black cane
- Identical behaviour
Model:
- White males
- Ages: 24-29 years
- Dressed informally
- Made the victim sit + remained with him until the next
station
Conditions involved in trials:
- Critical/early: model stood in critical area + helped after 70s
- Critical/late: model stood in critical area + helped after 150s
- Adjacent/early: adjacent area + helped after 70s
- Adjacent/late: adjacent area + helped after 150s
- No model condition: model didn’t help the victim during the
trial but at the next stop.
Results
Majority of helpers= men
Spontaneous help (before model or in no model condition)= 80%
victims
60% trials= more than one helper
Victims helped:
- The type of victim: cane (62/65 trials, before model), drunk
(19/38 trials)
- The race of victim: black cane= white cane
black drunk= lesser than white drunk
(same-race helping observed)
Effect of modelling: difficult to analyze b/c only SLIGHTLY higher
helping rate in early (70s) model intervention than late (150s)
D.O.R hypothesis not supported
According to hypothesis:
- 3 person group= higher helping rate
- 7 person group= lower helping rate (responsibility diffused)
In experiment:
- 3 person group= lower helping rate
- 7 person group= higher helping rate
Data from observers:
- Majority of helpers were male
- In 20% trials, people moved away from critical area
- Trials with no help= higher no. of comments
- Drunk victim= more comments than others
Cost-benefit model:
- Emergency= increased arousal (either disgust or sympathy)
- Therefore, individual acts in order to reduce the arousal.
- Weighs up costs and benefits: if cost (risk) is higher= doesn’t
help, if benefits are higher= helps.
Some participants helped directly/indirectly (asked other to help)
Others left the room/thought that victim shouldn’t be helped
(drunk victim)
Conclusion
Many people would help strangers spontaneously.
Factors which affect people’s decision to help:
- The type of victim (cane one helped more than drunk)
- The gender of helper (men more than women)
- People are more likely to help same-race victims (especially
if drunk)
- Longer emergency: lesser chances of help (people find other
ways to cope up with the arousal).
Strengths and weaknesses
Field experiment
Good ecological validity: participants were ordinary passengers +
unaware- natural behaviour.
Independent measures design
Less control over extraneous variables: weather/train delays
could affect participants’ mood + helping behaviour- so lower
validity and reliability.
Methodological issues: same route, some participants might have
become a part twice- demand characteristics.
Lower generalizability: all subway passengers from NYC-
unrepresentative sample, not possible to predict bystander
behaviour in other countries.
Around 4500 participants: mix ethnicities and genders- large
sample is representative, improves validity.
Quantitative data:
- No. of helpers
- Time taken to help
- Objective
- More reliable b/c two observers
Qualitative data:
- Remarks + movements of passengers
- Thought and behaviours related to helping studied in depth
Ethics:
- Consent not taken
- Not debriefed
- Deceived: they believed the victim to be in need of help
- Participants might be distressed: guilty of not
helping/concerned about victim’s well-being afterwards.
