Learning through observing the consequences of the actions of another person - individuals don't need to experience the reward or punishment themselves to learn.
If such behaviour has been rewarded, it's more likely to be imitated.
When a learner observes the behaviour on a role model, especially someone they identify with (respect and admire), the learner is more likely to imitate the behaviour.
Children are more likely to identify with and preferentially learn from models similar to themselves - eg: same-sex models.
Children in Group 1 showed far more imitative aggressive responses than other groups. They also showed more non-imitative aggressive behaviours.
Girls in the aggressive model condition showed more physically aggressive responses if the model was male, and more verbally responses behaviours if the model was female.
Boys imitated more physically aggressive actions than girls.
Bandura's ideas are mostly developed through observing children's behaviour in lab settings - which limits the ecological validity.
In Bobo Doll Experiment, children witnessed the adults displaying aggressive behaviour on a television - so findings can't be directly extrapolated to children's behaviour in real life.
Also, there's the problem of demand characteristics.
The children may not have actually learned vicariously - but instead were just displaying behaviour they though Bandura wanted to do. One child even said 'that's the doll they want us to hit. This reduces the internal validity of the experiment.
Also, a doll can't feel pain / sadness - as an animate human would. This reduces the ecological validity of the experiment as we can't presume children imitate aggression directed towards an animate object who would experience hurt.
Therefore, this research lacks mundane realism, decreases applicability and support. Experiments like Bobo Doll tell us little about how children learn aggression in everyday life.
This is useful in providing understanding in a range of behaviours, like how children come to understand gender roles by imitating their parents.
In contrast, the biological approach doesn't account for the influence of culture and the environment of upbringing - takes a biologically deterministic stance.
This increases the population validity and applicability as this approach, as it can be applied to a larger target audience, thus increasing its support.