Abstract

Misinformation has the power to distort normal political processes and erode democratic values. Scant research combines individual level correlates of misinformation sharing with aggregate network features. In this work, we propose a parallel methodology where we actively observe the behavior of a large panel of active Twitter users (N=8,150) during the 2022 US midterm elections, and directly survey them in real time as they share election-related misinformation and authentic news. Both the sampling frame and misinformation definition will be reactive: we will update our  definition of misinformation and recruit more panelists from contested regions as the election unfolds. We will test for differences across a range of demographic, psychological, and network measures between users that posted misinformation as opposed to those that shared authentic news. This approach allows us to collect the most comprehensive election-related misinformation dataset that we know of, enriched by detailed individual-level correlates of news spread and collected in real time. Finally, we anticipate the data collected serving as the basis for multiple peer-reviewed articles across disciplines.

Principal Investigators