It is true that the argument of violent media causing violence in society has gone on for years, and that this social trend of thought has permeated even the music scene, the video game (gaming) industry, and many other ramifications of global media. The argument has resulted in the labeling of material, most commonly seen in music releases and in video games which, depending on the level of violence that they portray, are classified as "M", for "mature content". If society were not concerned about the influence of violence in the media, surely there would not be a preocupation to classify and categorize it.
The theory that supports the idea of media violence influencing the behavior of viewers is Albert Bandura's social learning theory. Although there are current sub-theories that attempt to explain behavioral changes, it is Bandura who first proposed the study of how observing others (observational learning) affects our processes of choice-making.
Conducting experiments involving adults behaving violently, Bandura argued that his findings show that children are more likely to mimic the behaviors of those whom they look up to, whether those behaviors brought in positive or negative consequences. In the case of children, it is not only behavioral, but also cognitive and developmental factors which motivate them to follow the lead of someone in control.
In a modern scenario, Bandura would argue that the devotion that adolescents have toward their social "role models", toward the ones imposed by the media, and their want for entertainment, lead them to reproduce the behaviors that they most consistently exposed to.
This is not only violence, but it can also be sexual behaviors, personal demeanor, and social identification.
Therefore, Bandura's Social Learning theory supports the notion that behavior is directly affected by the media in many different manifestations.