He wanted to explore and explain antisocial aggression in boys who came from intact homes in advantaged residential areas rather than simply demonstrating that multiple adverse conditions tend to spawn behavioral problems. This research led to a program of laboratory research into the determinants and mechanisms of observational learning (Isom, 1998). Bandura and his assistants found that hyper-aggressive adolescents often had parents who modeled hostile attitudes. Although the parents would not tolerate aggression in the home, they demanded that their sons be tough and settle disputes with peers physically if necessary, and they sided with their sons against the school. The youngsters modeled the aggressive hostile attitudes of their parents. According to Frank Pajares, a professor at Emory University, "for these aggressive adolescents, the vicarious influence of seeing a model meting out punishment outweighed the suppressive effect of receiving punishment directly for aggressive acts." Bandura's findings conflicted with the findings of his mentor's colleague, Clark Hull of Yale University, which assumed that direct parental punishment would internally inhibit children's expression of aggressive drives (Pajares, 2002). The results of this study led to his first book, Adolescent Aggression (1959) and to a subsequent book several years later, Aggression: A Social Learning Analysis (1973). .
Bandura's next research project was with Dorrie and Sheila Ross on social modeling involving the now famous inflated plastic Bobo doll. At that time, it was widely believed that modeled violence would drain observers' aggressive drives and reduce such behavior. The children in these studies were exposed to social models who demonstrated either novel violent or nonviolent behaviors toward these rebounding dolls. Children who viewed violent models subsequently displayed the novel forms of aggression toward the Bobo doll whereas control children rarely, if ever, did so.