Disentangling the Influence of Recommender Attributes and News-Story Attributes
A new publication by Erik Knudsen and Mikael Poul Johannesson used data collected through the Norwegian Citizen Panel to explore what lies behind our decisions to read and share news stories on social media.
The news we consume is increasingly determined by what others share on social media such as Facebook. Our decision to read and/or share a news story is influenced by the characteristics of the story (e.g., its topic, news angle etc.) and the characteristics of the person who shared it. In this study, Johannesson and Knudsen separate and compare these two reasons for sharing and/or reading news on social media.
To do so, they conducted a conjoint survey experiment in which respondents were randomly shown information about a person who shared a news story, information about a shared news story, or both, before being randomly asked about sharing or reading decisions.
The study finds that although sharing and reading behavior is different (e.g., sharing a news article has a higher social cost than reading a news article), they are influenced in similar ways by the characteristics of news stories and the characteristics of the people who share them (with some notable exceptions).
In addition to this, the study finds that although the party mentioned in a news story is important for sharing decisions (we are more likely to share a news story featuring a party we like than a party we dislike), this is contingent on the party preference of the person who shared the story. If the recipient and the sharer agree politically, the party mentioned in the news story is less important for the recipient's sharing decision.
As a final contribution, the authors test the economists' solution to the issue of how to order coauthor names in academic publications. This is normally done alphabetically or to signal the contribution of each author. Johnnesson and Knudsen instead randomize the order in which their names appear, and separate them with an ⓡ symbol rather than the conventional "&" or "and".
The open access article is available here: https://doi.org/10.1080/21670811.2020.1805780