Bandura, Ross and Ross found that the children who had witnessed the model behaving violently towards the doll, displayed more violent behaviours than those who had not. If generalising the results, exposure to models of violent behaviour leads to the participant displaying violent behaviours and they concluded from this research that exposure to media violence may cause imitation of the behaviours in real life [3].
The research is weakened as the experiment lacks in ecological validity as it was carried out in unnatural conditions that were unlike everyday life and therefore the results cannot be applied to society. There is also the problem of the sample of participants used, which is often small, due to the cost in time and money of laboratory experiments and unrepresentative of society, even if picked at random so the results cannot be generalised. .
Field experiments have also been used in this area of research and are much more ecologically valid than laboratory experiments, and involve behaviours that take place in a more natural environment. The researcher manipulates the independent variables but has very little control over any other variables, which may lead to deficit reliability. In this type of research, violent behaviour is usually monitored in the participants before, during and after exposure to media violence. A study by Friedrich and Stein (1973) took this form as nursery school children were monitored in their classroom for a four-week period. They found that children who were originally aggressive in their behaviour, were effected by violent cartoons and displayed more anti-social behaviour but those who were initially classed as neutral showed no change in behaviour [4].
This research adds some support to the Bo-bo doll study, but can be criticised for not having a representative sample as all the participants used were in the same class at the same school, adding difficulties to generalising the results.