The Main Methods of Sociological Research

In the mid-1960s, the first generation of North American children exposed to high levels
of TV violence virtually from birth reached their mid-teens. At the same time, the rate
of violent crime began to increase. Some commentators said that TV violence made violence
in the real world seem normal and acceptable. As a result, they concluded, North
American teenagers in the 1960s and subsequent decades were more likely than pre-
1960s teens to commit violent acts. The increasing prevalence of violence in movies,
video games, and popular music seemed to add weight to their conclusion.
Social scientists soon started investigating the connection between media and realworld
violence using experimental methods. An experiment is a carefully controlled
artificial situation that allows researchers to isolate presumed causes and measure their
effects precisely (Campbell and Stanley, 1963).
Experiments use a procedure called randomization to create two similar groups.
Randomization
involves assigning individuals to groups by chance processes. For example,
researchers may ask 50 children to draw a number from 1 to 50 from a covered box. The
researchers assign children who draw odd numbers to one group and those who draw
even numbers to the other group. By assigning subjects to the two
groups using a chance process and repeating the experiment many times, researchers ensure that
each group has the same proportion of boys and girls, members of different races, children highly
motivated to participate in the study, and so on. After randomly assigning subjects to the two groups, the researchers put the groups in separate rooms and give them toys to play with. They observe the children through one-way mirrors, rating each child in terms of the aggressiveness of his or her play.
This is the child’s initial score on the “dependent variable,” aggressive behavior. The dependent
variable is the effect in any cause and-effect relationship. Then the researchers introduce
the supposed (or “hypothesized”)

cause to one group—now called the experimental group. They may show children in the
experimental group an hour-long TV program in which many violent acts take place. They
do not show the program to children in the other group, now called the control group. In
this case, the violent TV show is the “independent variable.” The independent variable is
the presumed cause in any cause-and-effect relationship.
Immediately after the children see the TV show, the researchers again observe the
children in both groups at play. Each child’s play is given a second aggressiveness score.
By comparing the aggressiveness scores of the two groups before and after only one
of the groups has been exposed to the presumed cause, an experiment can determine
whether the presumed cause (watching violent TV) has the predicted effect (increasing
violent behavior; see Table 1.1).
Experiments allow researchers to isolate the single cause of theoretical interest and
measure its effect with high reliability, that is, consistently from one experiment to the
next. Yet many sociologists argue that experiments are highly artificial situations. They
believe that removing people from their natural social settings lowers the validity of
experimental results, that is, the degree to which they measure what they are actually
supposed to measure.
Why do experiments on the effects of media violence lack validity? First, in the real
world, violent behavior usually means attempting to harm another person physically.
Shouting or kicking a toy is not the same thing. In fact, such acts may enable children
to relieve frustrations in a fantasy world, lowering their chance of acting violently in the
real world. Second, aggressive behavior is not controlled in the laboratory setting as it
is in the real world. If a boy watching a violent TV show stands up and delivers a karate
kick to his brother, a parent or other caregiver is likely to take action to prevent a recurrence.
In the lab, lack of disciplinary control may facilitate unrealistically high levels of
aggression (Felson, 1996).
Many experiments show that exposure to media violence has a short-term effect on
violent behavior in young children, especially boys. However, the results of experiments
are mixed when it comes to assessing longer-term effects, especially on older children
and teenagers (Anderson and Bushman, 2002; Browne and Hamilton-Gilchrist, 2005;
Freedman, 2002).