Measuring the media agenda: How to research news coverage trends on topics
The amount of media coverage that different subjects receive can affect public knowledge and opinion on that issue, which in turn can influence policy makers and legislators, both in terms of setting the agenda and influencing its direction. Likewise, the more politicians and the public are concerned with certain subjects, the more likely they are to be covered in the media. The media’s ability to shape opinion in this way makes patterns of coverage a key variable in determining how the democratic system functions.
But what is a scientifically valid approach to discerning these patterns with precision? Many researchers point toward factors that encourage a single national media agenda, such as shared norms for news-worthiness, limited resources for newsgathering, and copycat propensities among journalists. If a single national news agenda exists for a topic, then studying one or two news sources to understand the media’s coverage of that topic is a valid approach. However, critics point out that a single source may have a particular set of influences that alter the amount of attention it gives to different issues, making it a poor proxy for the wider news agenda. Differences in print and broadcast media, for example, mean the former can cover more topics than the later, to which the availability of good visual images is also of greater importance. Other factors such as regional salience, and the need to appeal to different target audiences that advertisers wish to reach, may act against the formation of a single media agenda.
How to measure the tenor of media coverage on a particular issue — pinpointing consistent features that appear across news outlets — is a difficult task. One common technique used by university scholars and popular columnists alike is to employ keywords to search for a representative sample of articles in a database and then present the trends in the sample as representing the general media agenda. A 2014 study published in Political Communication, “Measuring the Media Agenda,” examines the conditions under which a single national media agenda emerges and those under which it does not. To do this, Mary Layton Atkinson of the University of North Carolina at Charlotte and John Lovett and Frank R. Baumgartner of the University of North Carolina at Chapel Hill conducted 90 different keyword searches covering a wide range of policy topics, on 12 national and regional media sources going back as far as the 1980s. Policy topics ranged from domestic issues such as Social Security, education and crime to international issues such as war and trade negotiations. The searches also included non-policy topics such as sports and natural disasters. The study is “empirically the most extensive study to date,” the researchers write.
The study’s findings include:
- Across the 90 different search topics, the presence of a single national news agenda could explain between 12 percent and 93 percent of the variance in the news coverage of that issue. In other words, on any given topic a consistent national media agenda can be obvious — or barely discernible.
- For each topic, across all sources, the average number of articles a month ranged from 1.7 to 687, with a median of around 60 stories a month.
- Topics with low levels of coverage, a maximum of around 23 articles per month, were not necessarily trivial or unimportant. They included subjects such as such as water pollution, farm subsidies, the unemployment rate, and racial discrimination.
- Issue salience is a key determinant of whether a national news agenda emerges on that topic. This can be understood in two ways: the amount of coverage a topic receives, on average over time, and when a subject receives an “attention spike” of high coverage for a short time.
- These two factors — average levels of coverage, and whether there is a spike in coverage — account for around 90% of the factors that predict whether a national media agenda exists for a particular subject or not.
- “Researchers studying low-salience issues, such as school prayer or food safety, for instance, should be wary of creating time series measures of media attention.”
If an issue is of high salience and receives consistently high levels of coverage or attention spikes, almost any major news source will show similar patterns in coverage. In such circumstances, using one or two news sources as a proxy for the wider media agenda would be appropriate, conclude authors Atkinson, Lovett and Baumgartner. For topics where there is low coverage and no spikes in attention, it is much less likely that there is a cohesive news agenda. In such cases not only would the use of a small sample of media as a proxy for a wider agenda be inappropriate, the construction of a wider news index would be unlikely to work either. Such an index may smooth differences across publications, but would be unlikely to represent an accurate picture of how any single person would consume such media, the study’s authors note.
One suggestion for further research relates to the construction of the keyword searches themselves, which can have a significant effect on results. Low hit rates produce data sets that are too small to be reliable. The study’s authors found surprisingly low numbers of results for issues that they would expect to be more prominent. Experimenting with the search terms may do more for the reliability of researchers’ results than the selection of media to use as a source: “Greater error likely creeps into any analysis based on poorly thought out or verified keyword searches than from the inclusion of any individual media source. For robust searches based on many hits, our analysis shows that a single media agenda is the rule.”