Abstract

With news increasingly distributed through social media platforms, journalists have become the targets of online harassment. This harassment can take many different forms — ranging from name-calling to threats of violence — and preliminary evidence indicates it disproportionately affects women, people of color, and those early in their careers. We seek to refine existing models of online harassment, narrowing in on harassment towards journalists, specifically. First, we will explore how a diverse pool of distributed  workers evaluate different responses to journalists’ posts, investigating whether perceptions of harassment are influenced by contextual factors (e.g., the content of the original post, the identity of the journalist) and whether these perceptions vary across demographic categories. Next, these hand-coded responses will be used to train a classifying algorithm, to be applied to a larger set of unseen tweets and responses. Here, our objective will be to model harassment as a function of journalist attributes, including their demographic characteristics, and the level of political polarization in their locality. Ultimately, our work will speak to how online harassment is both perceived by a diverse audience and actively experienced by journalists working today.

Principal Investigators