The Werther Effect Revisited: Measuring the Effect of News Items on User Behavior

People are moved to act following exposure to media coverage of specific events. For example, the “Werther Effect” is the popular term for the observed increase in suicides following media coverage of suicides. Here we develop a finegrained method for assessing the effect of news stories on the intentions of internet users. Our method assesses the likelihood that a person was exposed to a given news story via the temporal and spatial distances between the location of the person and the location of the news story and/or the website where it was published. This analysis of likelihoods allows us to estimate the contribution of a particular news story to a person’s intent, as manifested in specific, intentdriven, search engine queries. Data were gathered over a ten-month period and cover both the search engine queries of a large population and the news stories to which this population was exposed. We estimated the contribution of news stories to negative effects (i.e. media coverage of suicides and their effect on queries indicating suicidal intention), and positive effects (e.g. media coverage of disease prompting queries into disease screening). We demonstrate that the contribution of news stories can be assessed at the level of individual users, and we analyzed titles and phrases therein for their effect. Finally, we propose a predictive model to be utilized by media outlets to predict the likely effect of specific stories prior to their publication.