Gabriele Pinto (University of Roma): “Data, Trends, and Biases of Television Political Talk Shows: Evidence from Italy”

Even though we live in the golden age of the Internet and social networks, the idiot box remains the primary and undisputed source of political information. At the same time, the production and consumption of videos are becoming ubiquitous (e.g. tik-tok and reels). Currently, television and videos are seldom employed as primary data sources by social scientists utilizing a quantitative approach. This is primarily attributed to the costs and technical challenges inherent in the data collection process, particularly when contrasted with more accessible mediums such as text (e.g., “text-as-data”).

In the last couple of years, recent advances in machine learning tools for computer vision and audio analysis have finally paved the way to the possibility of extracting meaningful data from large video archives, at relatively little cost. We show how to apply these tools through a case study involving a web-scraped collection spanning five years (2018-2023) of episodes from two prominent Italian Political Talk Shows (Otto e Mezzo & Stasera Italia) that we match with minute-by-minute TV viewership data from Auditeltm. We provide a descriptive analysis of the trends we find in our case study concerning political representation, topics discussed, and emotional behaviors.