In 1961 Albert Bandura, Dorothea Ross and Sheila Ross showed nursery-school children an adult behaving aggressively toward a 5-foot inflatable Bobo doll, then watched how the children played alone. Children who had seen the aggressive model reproduced much of that behavior, while children who saw a quiet model or no model at all did not. The result became a landmark for social learning theory.
How the experiment was set up
The paper reports 72 children, 36 boys and 36 girls, from the Stanford University Nursery School, aged 37 to 69 months with a mean of 52 months. Two adults, one man and one woman, acted as models, and one female experimenter ran every session. There were eight experimental groups of six children and a control group of 24. Half of the experimental children saw aggressive models and half saw subdued, nonaggressive ones, and each child saw a model of either the same or the opposite sex. The control group saw no model.
Before the sessions, the experimenter and a nursery school teacher rated every child on four five-point scales of aggression (physical, verbal, toward objects, and aggression inhibition). The composite rating had an interrater reliability of .89, and children were matched into triplets on it before random assignment.
Each child was seated in a corner with potato prints and stickers while the model played in the opposite corner with a tinker toy set, a mallet and the Bobo doll. In the nonaggressive condition the model assembled the tinker toys and ignored the doll. In the aggressive condition the model began with the tinker toys and after about a minute turned to the doll for the rest of the period, performing distinctive actions and phrases so that copying could be recognized. The exposure lasted 10 minutes.
The test session
Every child then received what the authors call mild aggression arousal. They were shown attractive toys, and after about two minutes were told those toys were reserved for other children, so they could play only with the toys in the next room. In that room each child played for 20 minutes, with observers scoring behavior in 240 five-second units, with no model present.
What the authors found
- Children who saw the aggressive model reproduced a good deal of physical and verbal aggression resembling the model's. The nonaggressive-model and control groups showed virtually no imitative aggression.
- About one-third of the aggressive-model group repeated the model's nonaggressive remarks, and none of the children in other groups did.
- Boys showed more imitative physical aggression than girls, and the two sexes did not differ on imitative verbal aggression.
- The male model's influence was stronger than the female model's, and children who saw the subdued male model were less aggressive than controls.
- The authors cautioned that some interactions were not statistically significant because group sizes were small.
Why it mattered
Stanford's obituary for Bandura says the experiments "upended the established behavioral doctrine" that learning is a conditioned response to external rewards and punishments, and challenged the view that watching violence on television relieves aggressive impulses in children. OpenStax presents Bandura's alternative, social learning theory, as one that takes cognitive processes into account. It describes four steps in modeling (attention, retention, reproduction and motivation), three kinds of models (live, verbal and symbolic), and vicarious reinforcement and punishment, meaning that what happens to a model affects whether observers copy them.
A common mix-up
OpenStax's short account of the Bobo doll work says children's behavior depended on whether the adult was punished. The 1961 paper as reprinted by York University compares aggressive, nonaggressive and no-model groups and does not include a punished-model condition. EBSCO's Research Starters credits the punishment and reward finding to a separate 1965 study. When a claim says what the Bobo doll children did, check which study it describes.
Limits and criticisms
EBSCO lists two criticisms often raised: the doll rights itself when hit, so children may have viewed it as something meant to be struck, and striking a doll does not show how a child would treat a real person. It also says other psychologists have raised ethical concerns about the original design. The sample was one nursery school, so the paper cannot say how far the numbers carry to other children, and the authors themselves flagged the small cells.
Study checklist
- Name the independent variable (type of model, model's sex) and the dependent variable (imitative physical and verbal aggression).
- State the finding in one sentence, then add one limit.
- Keep 1961 (live model, three groups) apart from later studies of film models and reinforcement.
Take this further with Oakgen Agent
Agent can search the web and fetch the paper and OpenStax pages, write the study guide and quiz, and save it as a docx or pdf.
Coming soon: Pasting a YouTube link and getting a summary of the video inside Agent is not available yet.
Looking for a recap of one specific YouTube video? The YouTube Summarizer does that today.
Open this task in Oakgen Agent
The prompt is filled in for you. Nothing is sent until you press send.
Frequently asked questions
How many children took part in the 1961 Bobo doll study?
According to the original paper by Bandura, Ross and Ross, the subjects were 36 boys and 36 girls from the Stanford University Nursery School, aged 37 to 69 months with a mean age of 52 months. They were divided into eight experimental groups of six and a control group of 24, and one female experimenter ran the study for all 72 children.
What did the 1961 study find?
Children who saw an aggressive adult model reproduced a good deal of physical and verbal aggression later, in the absence of the model, while children in the nonaggressive-model and control groups showed virtually no imitative aggression. Boys showed more imitative physical aggression than girls, and the effect of the male model was stronger. The authors noted that not every interaction reached statistical significance given small group sizes.
Did the original 1961 study include a punished or rewarded model?
The 1961 paper as hosted by York University's Classics in the History of Psychology compares aggressive-model, nonaggressive-model and no-model groups. EBSCO's Research Starters attributes the finding that children who saw aggression punished were less likely to copy it to a 1965 Bandura study. OpenStax's one-paragraph summary of the Bobo doll work mentions punishment, so check which study a claim comes from.
What are the main criticisms of the Bobo doll experiment?
EBSCO's Research Starters lists two points often raised: the Bobo doll rights itself when hit, so children may have treated it as a target meant to be struck, and drawing conclusions about striking a real person from striking a doll is problematic. It adds that other psychologists have raised ethical concerns about the structure of the original experiment.
Sources
- Bandura, Ross and Ross (1961), Transmission of aggression through imitation of aggressive models, Journal of Abnormal and Social Psychology 63, 575-582, reprint at Classics in the History of Psychology, York University (read 2026-11-19)
- OpenStax, Psychology 2e, 6.4 Observational Learning (Modeling) (read 2026-11-19)
- OpenStax, Psychology 2e, 11.4 Learning Approaches (read 2026-11-19)
- Stanford Report, Psychology Professor Albert Bandura dead at 95, 2021-07-30 (read 2026-11-19)
- EBSCO Research Starters, Bobo doll experiment, Dillon and Dillon, 2024 (read 2026-11-19)

