Disentangling media bias

According to a 2023 Gallup poll, Americans’ confidence in the mass media to report the news fully, fairly and accurately is at its lowest point since 2016.

In their working paper, “Disentangling Demand and Supply of Media Bias: The Case of Newspaper Homepages,” economics professors Tommy Leung and Koleman Strumpf introduce a new economics-based method for detecting media bias that operates independently of external influences, such as ownership or the ideological leanings of editors. Leung and Strumpf examined more than 100,000 articles from The New York Times (NYT) and The Wall Street Journal (WSJ) alongside data from 22 million tweets to see if editorial bias or consumer demand determined the length of time an article remained on the front page.

Economics professor Tommy Leung

Leung shares takeaways from the research in this Q&A.

What kinds of data did you use for your research and what were your findings?

Drawing from an expansive dataset of articles from both NYT and WSJ, juxtaposed against a backdrop of Twitter interactions, we assessed the factors influencing article duration on their digital homepages and found several key insights.

Firstly, even after controlling for an article’s popularity, measured by the number of tweet shares it receives, articles that aligned more with the newspaper’s political stance tended to stay longer on the homepage. Specifically, more liberal-leaning articles in the NYT and more conservative-leaning articles in the WSJ exhibited prolonged homepage presence. This finding provides robust evidence of supply-driven media bias, indicating that editorial preferences significantly shape the visibility of news content.

Secondly, competition appears to reduce media bias. When both newspapers covered the same news, the tendency for liberal-leaning NYT articles and conservative-leaning WSJ articles to stay longer on their respective homepages diminished. This observation aligns with economic theory, suggesting that increased competition in the media landscape can mitigate supply-driven biases, leading to a more balanced presentation of news.

Lastly, our study implies that media consolidation observed over the past decade might exacerbate media bias. With fewer competitors in the media market, the reduced competition likely contributes to the heightened media bias prevalent today. This underscores the importance of a competitive media environment in fostering diverse and balanced news reporting.

How did you measure editorial decision-making and potential bias through an economic lens?

Koleman and I were interested in how NYT and WSJ editors made decisions on front-page content and why. Were editors making choices due to personal or organizational bias or because they were responding to consumer demand? We looked at which articles were on the front page, for how long, and evaluated social sharing. 

We used a web scraper to automatically extract specific data on front-page headlines in the NYT and WSJ, while also collecting data from Twitter’s application programming interface, or API, to see if consumer demand (measured by tweets) or editorial inclinations might be influencing decisions on an article’s homepage position and the length of time it remained on the digital front page.

For example, a NYT article on events at the White House on January 6, 2021, described the incident as an attack on American democracy. It was published at 3 a.m., featured on the NYT front page by 6 a.m. and had approximately 200 tweet shares by noon. The article remained on the digital front page for more than two days. 

Another article covering the news of migrants who had died in a tractor-trailer in San Antonio was published on June 27, 2022. The piece quoted the state’s Republican governor saying that the deaths were the result of Biden’s open border policies. This piece was published at 9 p.m., immediately featured on the front page, had more than 600 tweet shares and was removed within two hours.

Is there a way for readers to recognize and balance these biases when seeking fair and accurate news coverage?

Based on our study, media bias is a prevalent issue in most newspapers, including prominent ones like NYT and WSJ. This bias is often aligned with the political inclinations of the newspaper’s editorial board. 

Competition among news outlets can help reduce these biases, as seen when both NYT and WSJ cover the same news. However, to truly recognize and balance these biases, readers should diversify their news consumption. Relying solely on one source or one side of the political spectrum can skew perceptions and understanding of events. By reading from a variety of sources, cross-checking information, and engaging with different viewpoints, readers can achieve a more balanced and fair perspective on news coverage.

Leung is available to talk with media. Contact media@wfu.edu to request an interview.