Disentangling Demand and Supply of Media Bias: The Case of Newspaper Homepages

In this study, we propose a novel approach to detect supply-side media bias, independent of external factors like ownership or editors' ideological leanings. Analyzing over 100,000 articles from The New York Times (NYT) and The Wall Street Journal (WSJ), complemented by data from 22 million tweets, we assess the factors influencing article duration on their digital homepages. By flexibly controlling for demand-side preferences, we attribute extended homepage presence of ideologically slanted articles to supply-side biases. Utilizing a machine learning model, we assign "pro-Democrat" scores to articles, revealing that both tweets count and ideological orientation significantly impact homepage longevity. Our findings show that liberal articles tend to remain longer on the NYT homepage, while conservative ones persist on the WSJ. Further analysis into articles' transition to print and podcasts suggests that increased competition may reduce media bias, indicating a potential direction for future theoretical exploration.