Predictive Journalism: On the Role of Computational Prospection in News Media

Abstract

Whether it’s who is going to win an upcoming election, how a pandemic will unfold next month, or just what we plan to have for lunch, human beings are cognitively wired to think about—and want to know about—the future. Predictive journalism is a genre of data journalism focused on generating evidence-based forecasts that can help to fill this audience need. This report develops this idea of predictive journalism, describing and synthesizing observations from the domains of elections, COVID, and beyond. Based on interviews with some of the genre’s key players as well as a content analysis of dozens of published examples, I elaborate on its journalistic value such as communicating predictions and explanations, enabling analysis through counterfactuals and critiques, and informing downstream news coverage, as well as describe how it works in practice in terms of interactivity, uncertainty communication, and transparency disclosure. I then argue for the development of a responsible journalistic practice of prediction, work toward articulating what that might entail, and make recommendations for how the field can further develop predictive journalism.

Keywords: predictive journalism, news media, computational journalism, data journalism