Skip to main content icon/video/no-internet

Experiments test the causal relationship between at least one factor (or independent variable) and at least one potentially affected (dependent) variable. Like experiments in the natural sciences, experiments in the social sciences and in media research rely on the comparison of outcomes that follow from different starting conditions. These starting conditions constitute the experimental treatment and are designed to differ from each other in a systematic fashion and only in respect to the independent variable, whose impact on the dependent variables is to be assessed. Participants (e.g., media users) are confronted with these differing starting conditions, and one or several empirical methods are used to measure the dependent variables (e.g., questionnaires, observational methods, physiological measures). These data are then compared between participants who were exposed to different starting conditions. Substantial (i.e., statistically significant) differences in the dependent variables represent empirical evidence for a causal effect of the independent variable on the dependent measures. Alternative explanations for group differences in the dependent measures (e.g., gender, age, media literacy) are ruled out because participants are randomly assigned to starting conditions. This randomization is expected to form groups that are (in average values) very similar (or, given a sufficient number of individuals per group, equal) to each other. Therefore, alternative explanations for differences in the dependent variables cannot be valid because in the ideal case, any conceivable alternative factor is distributed identically across experimental conditions.

Example: Effects of Video Games

In research on adolescents and the media, the impact of violent media on aggression is frequently tested experimentally. Bushman and Anderson (2002) randomly assigned their participants to play either a violent video game or a nonviolent game for 20 minutes (experimental treatment). Subsequently, the participants were asked to read three stories that included a potential interpersonal conflict and were required to write down a completion for each story, addressing the actions, thoughts, and feelings of the protagonists. After the experiment, trained coders went through the participants' writings and counted the number of aggressive actions, thoughts, and feelings that were mentioned. This number reflects the salience of aggression in the participants' mind at the time they wrote the texts (i.e., immediately after the experimental treatment). Players of a violent video game produced more aggressive story elements than players of a nonviolent game, which the authors regard as experimental demonstration of the causal relationship between game violence and aggressive thinking (as a component of aggressive behavior tendencies).

Experimental Design

Experiments are the best choice to test short-term causal effects in media research. Careful planning is needed, however, to create methodologically sound experimental designs that are in line with one's theoretical assumptions. Video games, for instance, are interactive and—in contrast to films—open to participants' individual modifications, which is a potential problem for valid experimentation. Other limitations have to be kept in mind as well (e.g., differences between experimental laboratories and real life; neglected long-term effects); combining experiments with other approaches (e.g., panel studies) may amplify the evidence for theoretical assumptions.

  • research methods
  • experimental studies
  • dependent variables
ChristophKlimmt
  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading