Experimental Evidence on the (Limited) Influence of Reputable Media Outlets

High quality news outlets are widely regarded as essential to responsive, uncorrupt democratic governments. However, experimental validation of the mechanisms of this claim, whereby outlets influence citizen knowledge and views, has proven elusive because reputable outlets try to publish the truth (and so valid control groups are hard to find), do not randomize news content, and have business models that generate massive endogeneity for researchers. We worked with a major media outlet to overcome these problems and meet journalistic and scientific standards. The results of four experiments covering crime, the economy, the environment, and gender equity indicate that editorial decisions have large effects on readers’ factual knowledge, as implied by claims about the importance of the press, but they are only modestly larger than the effect of sponsored content on the same sites, which anyone can buy without editorial oversight. Moreover, at least in the short term, editorial decisions are no different from sponsored content purchases for other outcomes: Effects on political attitudes and policy preferences are statistically indistinguishable from each other, approximately zero, and the same across policy areas. Our results suggest that the traditional news media provides a clear but tenuous foundation for democratic citizen education.