“Reporting on climate change: A computational analysis of U.S. newspapers and sources of bias, 1997–2017”

Abstract

News organizations constitute key sites of science communication between experts and lay audiences, giving many individuals their basic worldview of complex topics like climate change. Previous researchers have studied climate change news coverage to assess accuracy in reporting and potential sources of bias. These studies typically rely on manually coding articles from a handful of prestigious outlets, not allowing comparisons with smaller newspapers or providing enough diversity to assess the influence of partisan orientation or localized climate vulnerability on content production. Making these comparisons, this study indicates that partisan orientation, scale of circulation, and vulnerability to climate change correlate with several topics present in U.S. newspaper coverage of climate change. After assembling a corpus of over 78,000 articles covering two decades from 52 U.S. newspapers that are diverse in terms of geography, partisan orientation, scale of circulation, and objectively measured climate risk, a coherent set of latent topics were identified via an automated content analysis of climate change news coverage. Topic model results indicate that while outlet bias does not appear to impact the prevalence of coverage for most topics surrounding climate change, differences were evident for some topics based on partisan orientation, scale, or vulnerability status, particularly those relating to climate change denial, impacts, mitigation, or resource use. Overall, this paper provides a comprehensive study of U.S. newspaper coverage of climate change and identifies specific topics where outlet bias constitutes an important contextual factor.