Overview – Bandura et al (1961)

Bandura, A., Ross, D. and Ross, S. A. (1961) Transmission of aggression through imitation of aggressive models. Journal of Abnormal and Social Psychology, 63, (3), 575-582.

Bandura et al (1961) is a key study in the developmental area of psychology that investigated whether children could learn and reproduce aggressive behaviour by observing an adult model. The study is commonly known as the Bobo doll experiment. Children watched an adult either behave aggressively towards a large inflatable Bobo doll or behave in a quiet and non-aggressive way towards it. They were then placed in a different room with a range of toys including a Bobo doll. The researchers observed the children’s behaviour to see whether they reproduced the aggressive behaviour they had seen previously. The study found that children who had observed an aggressive adult model displayed significantly more aggressive behaviour than children who had observed a non-aggressive model or had not observed a model. This provided evidence that children can learn new behaviours through observing and imitating other people.

You can jump to different sections of this article via the links below:

• Aim
• Method (sampling and procedure)
• Results
• Conclusions
• Evaluation and methodological issues
• Relation to other core developmental psychology studies

Bandura et al (1961)

Aim

The main aim of the study was to investigate whether children would imitate aggressive behaviour after observing an aggressive adult model.

The underlying theory behind the study was social learning theory (SLT), which explains behaviour partly from the experiences of observing and imitating other people’s behaviour. According to SLT, people learn new behaviours by observing and imitating other people who act as models.

In line with SLT, Bandura argued that aggressive behaviour can be learned by observing aggressive models. Previous research had shown that children readily imitate behaviour demonstrated by an adult model when the model remains present. However, less was known about whether children would continue to show an observed behaviour when the model was no longer present and when the child was placed in a new situation. As such, this study aimed to find out whether children can reproduce the behaviour of a model even when that model is not present.

The researchers also investigated whether sex differences and similarity between the child and model affected imitation.

The study had four main hypotheses:

• Children exposed to aggressive models would show significantly more aggressive behaviour that imitated the model than children exposed to non-aggressive or no models.
• Children exposed to non-aggressive models would show significantly less aggressive behaviour than children exposed to aggressive or no models.
• Boys would show more imitative aggression than girls.
• Children would imitate the behaviour of a same-sex model more than the behaviour of an opposite-sex model.

Overall, the study was designed to investigate how observing other people’s behaviour can influence children’s subsequent behaviour.

Method

Bandura et al used a laboratory experiment with an independent measures but matched participants design. Plus, the researchers also used controlled and structured observations to measure the children’s behaviour in the final stage of the experiment.

The experiment used 3 conditions. The independent variable (IV) was the type of model behaviour children saw:

• Aggressive model condition: children observed an adult behaving aggressively towards a Bobo doll.
• Non-aggressive model condition: children observed an adult behaving quietly and ignoring the Bobo doll.
• Control condition: children did not observe an adult model before the test of their behaviour.

There were 24 children in each of these 3 conditions (72 total).

The dependent variable (DV) was the children’s aggressive behaviour, which was measured by observing whether and how much they reproduced aggressive acts towards the Bobo doll and other toys.

The experimental groups were further divided according to the sex of the child and sex of the model. So, children could observe either a male or female adult model, and the model would either be the same sex or opposite sex to the child.

Participants and sampling

The participants were recruited from the Stanford University Nursery School. This can be described as a form of opportunity sampling because the researchers used children who were readily available to them through the nursery school.

The characteristics of the participants were as follows:

• 72 children
• 36 boys and 36 girls
• Aged between 37 and 69 months (approximately 3 to 6 years old)
  • (Mean age of approximately 52 months)

Before the main experiment, the children were individually assessed for their existing levels of aggression using ratings from the experimenter and a nursery school teacher who knew the children well. The researchers then matched children with similar aggression scores before randomly assigning them to the experimental or control conditions. This helped reduce the possibility that pre-existing differences in aggression would skew the results.

Procedure

Part 1

During the first part of the experiment, each child was taken individually into a room.

The child was given an activity involving potato prints and picture stickers to keep them occupied. An adult model was on the other side of the room.

• In the non-aggressive condition, the adult model quietly played with construction toys and ignored the Bobo doll.
• In the aggressive condition, the adult initially played with the construction toys but then began behaving aggressively towards the Bobo doll. The model hit the Bobo doll with a mallet, punched the doll, kicked the doll, threw the doll in the air, sat on the doll, and made verbally aggressive comments.

Importantly, the children were not directly instructed to copy the model. They simply observed the adult while completing their own activity.

Part 2

After this, the child was taken to a different room. Before the final observation, the experimenter created a mild state of frustration by allowing the child to play with some attractive toys and then telling them that they could no longer play with them.

Part 3

The child was then taken into another room containing a range of toys. These included a Bobo doll, a mallet, dart guns, a tether ball, crayons and paper, dolls, cars and trucks, etc.

The child was observed for 20 minutes through a one-way mirror.

The researchers recorded different types of behaviour, including:

• Imitative physical aggression: reproducing physical actions performed by the model, such as hitting the Bobo doll with the mallet.
• Imitative verbal aggression: repeating aggressive phrases used by the model.
• Non-imitative aggression: other aggressive behaviour that had not been demonstrated by the model.
• Non-aggressive behaviour: playing quietly or using the non-aggressive toys.

The researchers recorded and scored children’s behaviour according to different categories of aggression. They then calculated mean aggression scores for each category and experimental group (see results below).

Results

The results showed that children who observed an aggressive adult model reproduced significantly more aggressive behaviour than children in the non-aggressive and control conditions.

The results also showed differences between boys and girls and between children who observed same-sex and opposite-sex models.

The table below shows the mean aggression scores for children in each experimental and control group. The scores represent the average amount of each type of aggressive behaviour shown by the children during the observation period. A higher score indicates that children showed more of that type of aggressive behaviour on average:

In short:

• Children who observed the aggressive model reproduced many of the model’s distinctive behaviours including hitting, kicking, and punching the Bobo doll. They also reproduced some of the model’s aggressive verbal comments.
• In contrast, children who observed the non-aggressive model and children in the control condition showed very little of this specific imitative aggression.

The researchers thus found that observing an aggressive model increased the likelihood that children would later imitate aggressive behaviour. The findings also suggested that children could be influenced by non-aggressive models because children who observed a non-aggressive model (particularly the non-aggressive male model) generally showed less aggressive behaviour than children in the control condition.

There were also differences between boys and girls:

• Boys showed more physical imitative aggression than girls after observing an aggressive model.
• There was less evidence of a difference between boys and girls for verbal aggression.

The sex of the model also appeared to influence behaviour. Boys generally showed more aggressive behaviour after observing the male aggressive model than after observing the female aggressive model. The researchers suggested that this may be because physical aggression was viewed as more strongly associated with males.

The study also found evidence that children could reproduce behaviour that was relatively unusual or novel. This was important because it suggested that the children were not simply becoming generally more aggressive. Instead, they were reproducing the specific behaviours they had observed.

Overall, the results showed that:

• Children would imitate aggressive behaviour after observing an aggressive adult model
• The imitation occurred even when the model was no longer present
• Boys showed more physical imitative aggression than girls
• The sex of the model appeared to affect imitation
• Observing a non-aggressive model could reduce some aggressive behaviour (compared to control).

Conclusions

Support for social learning theory

Overall, Bandura et al concluded that children can learn new behaviours through observing and imitating other people.

The fact that children reproduced specific aggressive actions that they had observed – such as hitting the Bobo doll with a mallet – provided evidence that learning can occur through observation. The children didn’t just become more aggressive in general, they reproduced the specific behaviours that had been demonstrated by the model. This suggests that they had learned specific behavioural responses through observing the model.

The children were also able to reproduce the behaviour later in a different situation and when the model was no longer present. This is important because they were not simply copying the adult’s behaviour as it happened. Instead, they had observed and remembered the behaviour, then reproduced it later when given the opportunity. This provides evidence that learning can occur through observation even when the behaviour is not immediately performed.

The results thus support the idea that behaviour can be learned through social observation and imitation. This became an important piece of evidence for Bandura’s social learning approach.

Factors affecting social learning

The researchers identified several factors that appeared to affect the extent to which children reproduced behaviour:

• Aggressive or non-aggressive model: Children who observed an aggressive model were much more likely to reproduce aggressive behaviour, while children who observed a non-aggressive model generally showed less aggression.
• Sex of the model: The male model appeared to have a particularly strong influence on physical aggression. This may have been partly because aggression was viewed as more strongly associated with males.
• Sex of the child: Boys showed more physical imitative aggression than girls. However, this difference was not found in the same way for verbal aggression.
• Type of behaviour: Children were more likely to reproduce behaviour that was clearly demonstrated by the model. The study therefore provided evidence that observation can lead to the learning of specific new behaviours.

Gender differences

The study found some differences between boys and girls in the aggression they copied.

• Boys showed more physical aggression than girls, especially after watching a male model.
• Verbal aggression was more likely to be copied from a same-sex model. Boys were more likely to copy a male model, while girls were more likely to copy a female model.

This suggests that group differences between boys and girls can affect which behaviours are most likely to be learned through observation. However, these differences were not the same for every type of aggression – the differences between the two genders depended partly on the type of behaviour being observed.

Evaluation

Research methods and techniques

Bandura et al used a laboratory experiment with a controlled observation of children’s behaviour.

• High control and standardisation: The researchers carefully controlled the environment, the behaviour of the models, the toys available, and the observation period. This makes it more likely that differences in children’s behaviour were caused by the IV (i.e. the model they observed).
• Control of individual differences: Children were assessed for their existing levels of aggression and matched according to their aggression scores before being assigned to conditions. This helped reduce the effect of pre-existing differences in aggression between participants.
• Controlled observation: Children’s behaviour was observed through a one-way mirror using clearly defined categories of behaviour. This allowed the researchers to record behaviour systematically.
• Artificial setting: The situation was highly controlled and unusual. Watching an adult attack a Bobo doll and then being placed in a room full of toys is not necessarily representative of how children normally learn aggressive behaviour in everyday life. This may reduce ecological validity.
• Demand characteristics: The children may have interpreted the model’s behaviour as showing them a new way to play with the Bobo doll. This means some of the behaviour may have been imitation of a game rather than evidence that observing aggression causes children to become more aggressive generally.

Population and sampling

The study involved 72 children from Stanford University Nursery School.

• Equal numbers of boys and girls: The sample contained 36 boys and 36 girls, allowing the researchers to investigate possible sex differences in imitation.
• Opportunity sample: The children were recruited from a single nursery school. This means the sample was not randomly selected from the wider population.
• Limited age range: All participants were young children aged approximately 3 to 6 years old. The findings may therefore not generalise to older children or adults.
• Limited population diversity: The participants came from one nursery school in the United States. This limits how confidently the findings can be generalised to children from different cultural, social and educational backgrounds.

Types of data

The study mainly produced quantitative data because the researchers counted different types of aggressive and non-aggressive behaviour.

• Quantitative data: Researchers recorded the frequency of different behaviours, including physical aggression, verbal aggression, imitative aggression and non-aggressive play.
  • Objective and easy to compare: Behaviour was recorded using clearly defined categories, allowing the researchers to compare the different experimental conditions.
  • Statistical analysis: The numerical data allowed the researchers to test whether differences between the groups were statistically significant.
• Some qualitative information: Observers also recorded examples of the children’s behaviour, including the specific aggressive actions and phrases they reproduced.
  • More detailed understanding: The observations provided information about how the children behaved, not just how much aggression they showed. This helped researchers identify whether children were reproducing specific behaviours demonstrated by the model.

Representativeness and generalisability

The sample consisted of 72 young children from one nursery school.

• Limited population: The participants were all young children from the same nursery school, so the findings may not generalise to older people or children from different backgrounds.
• Cultural limitations: The study was conducted in the United States and does not tell us whether children from different cultures would respond in exactly the same way.
• Equal representation of boys and girls: The sample contained equal numbers of boys and girls, allowing the researchers to examine differences between the sexes.
• Controlled matching: Matching children according to their existing aggression levels helped ensure that differences between groups were not simply caused by some groups containing more aggressive children at the beginning of the study.

Ethical issues

• Informed consent: The participants were very young children so they could not provide the same level of informed consent as adults. Permission from parents or guardians would therefore have been important but the paper does describe the consent procedure.
• Deception: The children were not told that the study was specifically investigating whether they would imitate the adult’s behaviour. However, this was necessary to prevent them deliberately changing their behaviour because they knew what the researchers were studying.
• Protection from psychological harm: The children were exposed to an adult behaving aggressively. Although the aggression was directed towards a toy, this could potentially have influenced the children’s behaviour or attitudes towards aggression.
• Debriefing: Because the participants were young children, it was important that they were properly debriefed after the experiment. However, the original paper does not describe any debriefing procedure.

Overall, the study raises ethical concerns because young children were exposed to an aggressive adult model and were then given an opportunity to reproduce the behaviour.

Validity

• Internal Validity: Are the findings caused by the experimental variables rather than another variable?
  • High control: The researchers carefully controlled the behaviour of the adult models, the environment, and the toys available to the children. This makes it more likely that differences in behaviour (DV) were caused by exposure to the different models (IV) and not some other extraneous variable.
  • Matching participants: Children were matched according to their existing aggression levels before being assigned to the different conditions. This reduced the effect of individual differences in aggression.
  • Objective behavioural measures: The researchers used clearly defined behavioural categories (e.g. ‘hitting the Bobo doll with a mallet’,’ punching or kicking the doll’, etc.) and observers recorded the children’s actions systematically. This improves internal validity because the researchers were measuring specific behaviours rather than relying on subjective judgements about how aggressive the children appeared. This makes it more likely that differences between groups reflected genuine differences in aggressive behaviour.
• Ecological Validity: Do findings translate to the real world?
  • Artificial task: Watching an adult attack an inflatable doll in a laboratory is very different from the situations in which children normally observe aggression.
  • Unusual behaviour: The aggressive actions performed by the model were highly distinctive and exaggerated. Children may therefore have copied them because they appeared interesting or unusual rather than because they had learned aggression in the same way that it occurs in everyday life.
  • Real behaviour was observed: Despite the artificial setting, the researchers observed children’s actual behaviour rather than relying on questionnaires or asking them what they would do. This provides useful evidence about how children may behave after observing a model.

Reliability

• Internal Reliability: Is the procedure consistent across all participants in this particular study?
  • Standardised procedure: The children were exposed to carefully controlled model behaviour and were tested using the same general procedure.
  • Clear behavioural categories: The researchers used specific categories for recording aggression and imitation. This made it easier for observers to record behaviour consistently.
  • High inter-rater reliability: The researchers reported high agreement between observers when scoring the children’s behaviour. This increases confidence that the behaviour was recorded consistently.
• External Reliability: Can the study be replicated at a different time and across different groups to produce consistent results?
  • Easy to replicate: The procedure was clearly described and involved standardised model behaviour, toys and observation periods. This makes it possible for other researchers to repeat the study.
  • Clear operationalisation: Aggression and imitation were defined in terms of specific observable behaviours. This makes the variables easier for other researchers to measure in a similar way.

Ethnocentrism

The study was conducted in the United States and involved children from Stanford University Nursery School.

• Weaknesses:
  • Limited cultural representation: The study provides limited evidence about whether children from different cultures would learn aggressive behaviour through observation in exactly the same way.
  • Single setting: All participants came from one nursery school, so the sample may not represent children from different social, cultural or educational backgrounds.
  • Cultural assumptions about aggression: The researchers interpreted physical aggression partly in relation to gender expectations. These expectations may themselves have been influenced by the culture in which the study was conducted.

However, the basic finding that children can learn behaviour by observing other people may reflect a general psychological process rather than something specific to American culture. Later research would be needed to establish how consistently observational learning occurs across different cultures.

Evaluation summary table

Relation to developmental psychology more broadly

Bandura et al (1961) contributes to developmental psychology by demonstrating that people can learn behaviour by observing other people. The study challenged the idea that learning must always occur through direct experience and reinforcement.

The table below summarises how this study compares with other core studies and evaluates its contribution and current relevance: