We review the literature that identifies selection bias in media-based data and propose a theoretical model of the sources of these biases. Given the implications for validity and reliability, we conclude that newspaper data often do not reach acceptable standards for event analysis and that using them can distort findings and misguide theorizing. Furthermore, media selection biases are resistant to correction procedures largely because they are unstable across media sources, time, and location. We end with a plea for more circumspect approaches to media data that fully and openly consider the implications of their inherent limitations.

This content is only available as a PDF.
You do not currently have access to this content.