The question, the predictions and the theory

The research question

Earlier studies had already shown that children copy adults. But in those studies the adult was usually still in the room. A child who copies an adult who is watching might simply be following the adult's lead. That is not strong evidence of learning.

So Bandura, Ross and Ross asked a sharper question:

If a child watches an adult behave aggressively, will the child later behave in the same way in a different room, when the adult is no longer there?

If the answer is yes, the child must have taken something away from watching. The behaviour has been learned, not just prompted.

The four hypotheses

A hypothesis is a specific, testable prediction. The paper makes four predictions in its introduction, written as ordinary prose. The numbers below are this guide's, and they match the numbering in Hock. Each is given in plain words, with the reason the authors gave for it.

1. Watching aggression produces copied aggression. Children who watch an aggressive adult will later copy that adult's aggressive acts. They will differ clearly from children who watched a calm adult and from children who saw no adult at all. The authors' reason: children have been rewarded for copying adults many times in the past, so a general habit of imitation should carry over to new adults. This reasoning comes from Miller and Dollard (1941).

2. Watching a calm adult reduces aggression. Children who watch a calm, non-aggressive adult will be less aggressive than children who saw no adult. The calm model would act as a brake. Psychologists call this an inhibiting effect.

3. Children copy their own sex more. Children will copy a model of their own sex more than a model of the other sex. The authors' reason: parents reward boys and girls for "sex appropriate" behaviour and discourage the opposite, so children build up a stronger habit of copying adults of their own sex (p. 575).

4. Boys will copy aggression more than girls. In the paper's words: "Since aggression, however, is a highly masculine-typed behavior, boys should be more predisposed than girls toward imitating aggression, the difference being most marked for subjects exposed to the male aggressive model" (p. 575).

Notice two things. First, the reasons are written in the language of reinforcement and habit, the behaviourist language of the time. The paper's bolder claim, that copying happened without any reward, comes later, in its discussion of the results. Second, hypotheses 3 and 4 rest on the gender roles of 1950s America, which the authors treated as a fact about society. Later pages ask whether that was a fair starting point.

The theory behind the predictions

The theory being tested is an early form of social learning theory. It says that much of what people do is learned from other people, and that watching is one of the main ways this happens. The paper uses three ideas that you will meet again.

• Learning without reward. The children were not rewarded for copying, and the model was not rewarded for being aggressive. If the children copied anyway, then reinforcement is not necessary for learning a new behaviour. This was the main challenge to the behaviourist account.
• Disinhibition. Children are usually told not to hit or kick. Watching an adult do it freely may send a message that it is allowed here. Removing a restraint in this way is called disinhibition. It means the adult's example lowers the child's self-restraint, rather than teaching a new act.
• Imitation versus general aggression. If children only became more aggressive in general, disinhibition alone could explain it. If they reproduced the model's exact acts and exact words, that shows specific new learning. This is why the researchers counted copied acts and new aggressive acts separately.

The variables

An experiment changes some things and measures others. The thing the researcher changes is the independent variable. The thing measured is the dependent variable.

The sex of the child is not something the researchers could assign at random. A variable like this is sometimes called a participant variable or quasi-independent variable. It matters later, because differences between boys and girls in this study cannot be traced to one cause.

Try this

A student writes: 'Bandura showed that children who watch aggressive adults become more aggressive people.'

What is wrong with this sentence, given what the study was set up to test?

Show the reasoning

The study tested whether children would copy specific aggressive acts in a new room shortly after watching. It did not measure whether the children became more aggressive people, which would mean lasting change across many situations and many weeks.

A more accurate sentence is: "Children who had just watched an adult attack a Bobo doll were more likely to copy those acts towards the same kind of doll a few minutes later, in a different room." The difference matters, because claims about lasting personality change need long-term studies, which this was not.