Research Methods, Content Analyses

The method of content analysis extends back into the 18th century, when the method of counting words in hymns and sermons was used to uncover heresy. It has since been used for a variety of purposes. Historically, although content analysis is embedded in the social sciences, communications researchers have focused primarily on the manifest content of communications exchanges; these analyses typically have not been concerned with behaviors beyond those strictly related to message producing and handling. In terms of television, the primary purpose of content analysis has traditionally been to identify, document, and trace major dimensions of the programming. For example, researchers have assessed the nature and measured the extent of violent acts on broadcast television programs. Content analysis has been used to measure and assess content of other types of media, such as newspapers, magazines, and, more recently, video games and the Internet. Recently, data gathered through content analyses have also been used in combination with other data, such as survey data, to relate content variables to other variables, such as personality characteristics, attitudes, and violent or sexual behaviors.

The major steps of a content analysis involve identifying the variables to be counted (e.g., physical violence against women), determining the unit of analysis (e.g., per hour or scene, page, song, or game level), deciding on a sample and sampling procedure (e.g., one week of prime-time television or one Internet site), training coders and assessing reliability (i.e., via percentage agreement or kappa calculation), and coding the sample. The process can be complex; variables and units of analysis must be well-defined to achieve acceptable reliability among coders—something as simple as counting the number of red shirts in a single television program can be surprisingly difficult (e.g., is a tank top a shirt? Where is the line between maroon and red? If the shirt is on a chair, but not presently being worn, should it still be counted?). Reliability and validity are important aspects of content coding. Researchers juggle defining a variable narrowly enough that coders will consistently make the same judgments about the material (i.e., be reliable) and making sure that the broader viewing audience would be likely to interpret the data in the same manner (i.e., be valid).

As a research method, a content analysis can evolve according to the complexity of research questions. Some analyses have now begun to include contextual elements surrounding the depiction or activity in question as a way to explore whether certain attributes of the message tend to be more meaningful to viewers than others. Other studies grapple with more philosophical concerns, such as how to accurately consider the ways individual interpretations of the material in question may vary by age, class, gender, or culture. Because the process involves counting, content analyses are most often regarded as a quantitative research method; however, content analysis is also closely associated with qualitative research, particularly narrative methods.