Assessing Algorithmic and Human Selective Exposure to Political Information: Behavioral Evidence from Online Search and Video
Political scientists Brendan Nyhan and Andrew Guess and computer scientist Christo Wilson will analyze the extent to which selective exposure online is the result of algorithms versus human behavior. They propose to test four hypotheses. First, that personalization algorithms significantly change recommended content and search results on online platforms. Second, that observed differences in search results due to these algorithms will be correlated with demographic and attitudinal factors (such as ideology and party identification). Third, those Americans who differentially consume politically congenial content will consume more political news and information than others, will be more politically knowledgeable and engaged, and will be more likely to be extreme ideologically. Finally, they hypothesize that selective exposure online will be due more to users’ selective searching and choice than algorithmic personalization. These studies will be conducted during two important political events—the 2020 primaries and general election for president—and will examine how algorithmic personalization shapes the information to which people are exposed.