Stefan Geiß

How Content-User Data Linking Decisions Affect Media Effects Estimates: A Model Comparison Approach

This study presents a framework for linking data about media content with data about media use of participants to estimate the extent of exposure to the contents as measured in the content analysis. Decisions in content–user linking involve (1) the time frame, (2) the effect curve, (3) the specificity and resolution of media use measures, and (4) the treatment of salience and duration of news stories. Comparing 16 different constellations of content–user linking decisions demonstrates that the choices researchers make can strongly affect the results of hypothesis tests and estimates of effect sizes. The decisions involved in content-user linking should therefore be explicated in any study in a standardized fashion. The decision about the time frame was the most important one, followed by news consumption measures, news story emphasis and length, and the shape of wear-out curves. The code used for linking the models is available from https://github.com/stefangeiss/content-user-link.