Pseudo-events: Tracking mediatization with machine learning over 40 years

Abstract

Using automated content analysis, this research explores the phenomenon of pseudo-events coverage in The New York Times (N = 70,370 articles) from 1980 to 2019. By clarifying the operationalization of pseudo-events, this study introduces pseudo-events as a valuable tool to index how different social subsystems perpetuate mediatization (which is when institutions absorb and abide by media logic). Machine-learning classifiers were constructed to measure pseudo-events, which provides historicity, specificity, and measurability — three tasks set forth for new mediatization research. We found a significant increase in pseudo-event coverage, expressing a more positive tone than genuine event coverage. Moreover, political pseudo-event coverage shows quadrennial cycles with peaks in each presidential election year. Our findings reveal the expansion of mediatization since 1980 and show how media logic has been internalized in different ways by the social subsystems of politics, culture, and economics. Institutions and their social actors need efficient tools to abide by media logic in seeking publicity and commanding authority, and pseudo-events have matured into one of the most dominant tools, especially for political actors. This study offers an innovative approach to capture complex phenomena and shows promises of broader application of machine learning to empirically quantify and identify patterns using theoretical concepts.