Promising news items: headline characteristics predicting clicks

Headlines of newspaper articles serve both to summarize the article and to entice readers to read more. Especially online newspaper readers use headlines to select news articles. Some typical headline formulations may be recognized as ‘clickbait’, as they are widely used in new and traditional media. However, research of the mechanisms and outcomes of such linguistic devices is scarce. In this paper, we analyze formulation features on a Dutch news website, using a data collection of A/B tests. We analyzed how formulation features predicted clicks on a headline. The research replicates and extends earlier research, with additional data and an alternative verification method. Previous findings were largely confirmed: headline formulations with more acting characters, containing forward references to the news article, and with intensifying language predicted more headline clicks. However, the additional data did not replicate effects of the news value negativity.