Project on Democracy and the Internet
This work seeks to categorize the degree to which web search results for U.S. candidates for political office are affected by hyperpartisan results. Using a data set of weekly search results for several hundred candidates for political office in the months leading up to the 2016 U.S. elections, we analyze trends in the kinds of results shown to users. In particular, we annotate results by their partisan bias in order to get a sense of the search result ecosystem for each candidate over time, prior to the election. In addition, we have begun collection of a similar data set in anticipation of the 2018 election season.