Combining Content Analysis and Assessment of Exposure through Self-Report, Spatial, or Temporal Variation in Media Effects Research

An interest in media effects is in most cases an interest in the effects of various types or instances of media content. The more information that researchers capture about such content, the more meaningful inferences about effects can be. Therefore, approaches to media effects research in which researchers first analyze the content of interest and then study impact of exposure to instances of such content has much to recommend them. One such approach is to analyze content to which a given population is likely to be exposed and then, in a survey, identify which instances of that content the respondent has (or likely has) seen. Another approach is to identify differences in media content (ideally identified via content analysis) by geographic unit, and use multi-level analyses to look at associated or prospective differences in outcomes. Another approach is via experiment: experimenters can characterize populations of messages via content analysis, and can sample either message exemplars of particular interest to manipulate in fixed effect designs, or create random samples of types of messages to represent message differences using random effect, multi-level analyses. Such approaches can substantially increase strength of inference, support rigorous test of theory, and increase external validity and policy relevance.