Online publication

Our main analysis investigated a variety of channels through which individuals read the news, but it was limited to a particular opt-in sample of individuals. In this supplementary section, we augment our analysis by examining the news consumption habits of a nearly complete set of users on one specific social information channel, Twitter, one of the largest online social networks, and arguable the largest designed primarily for information discovery and dissemination, as exemplified by their instructions to users to “simply find the accounts you find most compelling and follow the conversations.” The Twitter and toolbar datasets differ on two additional substantively important dimensions. First, Internet Explorer and Twitter users are demographically quite different. For example, whereas Internet Explorer users are believed to be, on average, older than those in the general Internet population, Twitter users skew younger. In particular, 27% of 18–29 year-olds use Twitter, compared to 10% of those aged 50–64 (Pew Research, 2013). Second, because of differing levels of information in the two datasets, in the toolbar analysis we examine the articles that an individual viewed, whereas with Twitter we look at the articles that were merely shared with that individual, regardless of whether or not he or she read the story. Thus, given these differences, to the extent that our results extend to this setting, we can be further assured of the robustness of our findings. To generate the Twitter dataset, we start with the nearly complete set of U.S.located individuals who posted a tweet during the two-month period March–April, 2013. We focus on accounts maintained and used by an individual (as opposed to corporate accounts), and so further restrict to those that receive content from (“follow”) between 10 and 1,000 users on the network. This process yields approximately 7.5 million individuals. Finally, similar to our restriction in the toolbar analysis, we limit to active news consumers, who received (i.e., followed individuals who posted)