‘MULTIDIMENSIONAL SLANT’: Complementarities in news reporting by US newspapers

16 Apr 2019

New evidence from major printed US newspapers shows that front pages are biased to certain combinations of topics on top of being biased to certain topics, given overall market topic coverage. The research by Sandra García-Uribe will be presented at the Royal Economic Society's annual conference at the University of Warwick in April 2019.

 

Front-page news not only plays a special role in determining public awareness of events, but is also a clear-cut, observable outcome on a daily basis that is amenable to systematic scrutiny. The new study analyses front-page coverage of topics and combinations of topics across time, making use of new measures of topic relevance constructed with data from the News Coverage Index of the Pew Research Centre for Journalism for the period from 2007 to 2012.

 

Previous research has focused on measures of slant associated with political ideology along the one-dimensional left-right divide. But the notion of media bias may be extended to the choice of leading topic or combination of topics that a media outlet decides to emphasise among the relevant news of the day. It is also important to understand such reporting patterns because of their potential effects on lifestyles, including values, worldviews or health.

 

This study documents the presence of ‘multidimensional slant’ in the front pages of major US newspapers. In order to do so, the author builds an empirical framework for the measurement of this more general form of slant, which requires the aggregation of news items into topics that are consistent across time.

 

The study formalises daily front-page choices in a model where the newspaper editor faces a target population of readers and chooses the bundle of news that maximises aggregate readership.

 

Through the lens of the model, the measures of slant are interpreted as preferences of the average reader and, in particular, biases for combinations of news are referred to as complementarities between news items. The results depict a map of complementarities where each newspaper has a position, as happens in other product-differentiating decisions.

 

Finally, the study finds that complementarities between news items contribute in a considerable portion to the probability that news on a topic appears on the front page. For example, complementarities to economic news in the New York Times imply that half of the probability that the paper publishes economic news is due to the satisfaction of combining it with other topics.

 

Alternatively, the probability that legal news makes it to the front page of the Wall Street Journal is half of what it could be due to the dissatisfaction of publishing it alongside other topics.

 

This work provides new evidence on media bias and opens a set of questions about the determinants and effects of co-existing different biases and their interactions in the media.

 

Sandra García-Uribe

Banco de España