Good Samaritanism: an underground phenomenon?

Piliavin, Rodin and Piliavin (1969).

In New York, in the early sixties, Kitty Genovese was murdered. This murder was unusual as many people who, alerted by her screams, had looked out of their apartment windows to watch witnessed it. Even though the attack had taken place over a period of about half hour, nobody called the police until it was too late to save Kitty from the final fatal stab wound.

In an effort to explain why people had not helped, Darley and Latane (1968) performed a laboratory experiment, in which a subject heard another 'subject' (really a tape recording of an actor) having an epileptic fit. This 'subject' was believed to be in a nearby room. When the real subject thought he was the only one listening, the chances of his going to help were much higher than when he thought he was one of many subjects listening in (in separate rooms). This tendency for people not to help when part of a group was labelled Diffusion of Responsibility. Another experiment, this time by Latane and Rodin (1969) found that few people in a large group would go to help an experimenter, who had pretended to fall off a chair in an adjoining room.

The above experiments are not that ecologically valid (or lack Mundane realism), for several reasons. The experiments were carried out in a laboratory, the tasks were not that true to life, the victim was not in sight and the subjects were college students. In order to conduct an experiment that was more true to life, Piliavin et al used a New York subway train. Certainly, the results would tell us something about the way New Yorkers respond to emergencies on their subway system; but you must decide whether the findings generalise to other people (e.g. Clactonites) and other public places (e.g. the high street or on a bus).

The Experiment.

A victim collapses on the subway during a non-stop 7½-minute journey, some time between the hours of 11 a.m. and 3 p.m. on a weekday. Sometimes he is helped either after a short while or after a number of minutes by a man (known as the model). The victim is either a black man or a white man acting as if he is drunk in one condition, and as if he is sober, but unsteady on his feet, in another (he carries a black cane). Two female observers record what happens.

The independent variables are:

Drunk or cane
Black or White
Early, Late or No model
Model initially sitting in the critical area or adjacent area.
The number of people on the train.

The Dependent variables are:

The time taken to help.
The race of the helper.
The percentage of trials in which passengers (subjects) left the critical area.
The number of comments made.

The results found were:

The cane victim, regardless of his race, was helped more often and sooner than the drunk victim (62 out of 65 trials for the cane victim compared to 19 out of 38 trials for the drunk victim).
There was a tendency for same race help in the drunk condition.
Significantly more men helped than women. Typical comments made by women passengers were “it’s for men to help him”; “I wish I could help him – I’m not strong enough”; “I never saw this kind of thing before – I don’t know where to look”.
The more people on the train the greater chance of somebody helping. This goes against the diffusion of responsibility theory. Also note that the response level for a seven-person group is higher than would be predicted by the response rates of smaller groups. This would suggest an effect of affiliation. Comparing this study with earlier laboratory studies suggest why the “diffusion of responsibility” hypothesis was not supported.
The initial position of the model had no effect on helping behaviour.
As time went by more comments were made and more people left the critical area.
The early model was more likely to encourage others to help also, than the later model. For 60% of 81 trials where spontaneous help was given the victim received it from two or more helpers. Once one person helped the fact that the victim was drunk did not affect whether others helped.

Piliavin explains the results by proposing a model of emotional arousal (empathy, being close to the emergency and the length of time the emergency continues) and its reduction (by helping, going to get help, leaving the scene and believing the victim does not deserve help), as well as a cognitive appraisal of the situation in terms of the costs and rewards of helping or not helping. For example, a cost of helping might be embarrassment or physical harm; the cost of not helping might be self-blame or censure from others; the rewards of helping might be praise and the rewards of not helping would be getting on with one's own business.

As a useful exercise you would do well to consider whether or not the results are explained by the theory. Several ethical guidelines are broken. You should consider whether the experiment could have been more ethical, without changing the validity of the experiment.

A major limitation of this study, considering it was the Kitty Genovese murder that initiated this kind of research, is that it looks at the inactivity of bystanders rather than why it is that some men attack innocent women. Howitt (1991) points out that these original studies perpetuate the myth that all rapes are carried out by deviant men against questionable women. It’s as if psychology at that time was itself a passive bystander to crime.

Full text