News habits and broad patterns in coverage can remain highly stable while the role of media, their portrayals of social issues and people’s understanding change. This is a central finding of Johannes Johansson’s doctoral thesis at JMG, University of Gothenburg.
The thesis combines surveys conducted over time with experiments and computational text analysis. The analyses show that stability in overall coverage can conceal changes within individual news outlets and differences in how they portray the same issues. Which aspects receive attention and how they are presented therefore need to be distinguished.
In one of the panel studies, relationships between news content and what people considered important varied across different phases of a crisis. These relationships were often stronger early on and could strengthen again during renewed attention. At the same time, people’s perceptions became more persistent, with later responses increasingly related to their earlier perceptions.
The experiments show that news can change how people understand and reason about an issue while earlier perspectives remain. New perspectives could emerge alongside existing interpretations, and the same aspect of a problem could acquire a different meaning. These changes became visible by comparing responses with what participants had expressed before reading the news articles.
This matters for democratic debate and accountability. How a problem is understood has implications for who is considered responsible and which solutions seem reasonable. Interpreting public perceptions therefore requires more than tracking which issues people prioritize. An unchanged priority can coexist with a changed understanding, and the role of media during one phase of a crisis cannot be assumed to remain the same throughout.
Facts
Work on Media Effects in Motion: Temporal Processes and Contextual Dynamics formed part of Varieties of Media Effects (VARME), funded by the European Research Council (ERC).
The studies comprise a four-wave survey panel beginning with 2,291 participants and a five-wave survey panel with 2,238 participants. The second links survey responses to approximately 55,000 news articles and broadcast items from web-scraped, print and manually coded datasets. The research also includes a ten-year analysis of approximately 120,000 print news articles and two experiments with three waves each, comprising approximately 8,000 open-ended responses in total.
