Digital platform companies are taking increasingly important roles in the contemporary information life of citizens. Together with social media giants like Facebook and Twitter, Internet search engines are vital information intermediaries that could enable, channel, or inhibit exposure to diverse media. While social media use and their democratic impact have gained meticulous academic attention, relatively less research in the domain of mass communication has been conducted on the effects of search personalization on citizens’ access to news and political information. Search engine personalization effects have not been widely studied mainly because of the heterogeneity in individuals’ information-seeking habits and the opaque mechanisms of algorithmic search results.

This dissertation contributes to a growing line of research on the democratic role of search media, in particular the use of search engines for political information and the potential personalization in search results, which carry far-reaching implications for contemporary issues including access to legitimate content, exposure and susceptibility to misinformation, and digital information disparities.

The dissertation examines three research problems: first, the political and informational factors that explain individual differences in search perceptions and attitudinal outcomes, which are crucial in understanding the extent to which people rely on search engines for political information; second, the extent to which information-seeking behaviors among politically engaged individuals, i.e. political partisans, reflect ideological bias and news use habits; and third, the extent to which biased search queries and political ideology lead to differences in search results.

Using a combination of methods including secondary survey analysis, original survey design and a series of crowdsourced experiments, the dissertation reports three main findings: First, the breadth and variety of Internet use, political media use, and media trust were important determinants of search-related outcomes and political search behavior. Knowledge of search engine result determinants significantly predicted political search frequency, and significantly mediated the effects of Internet use, political interest, political media use, media trust, and search ability on political search frequency.

Second, political partisans indicated preference towards specific biased terms as search queries and reported search queries that reflected issue positions under accuracy and directional motivated information-seeking goals.

Third, Google search results returned from biased queries led conservatives and liberals to different sets of information, but search result differences were driven largely by specific search queries than by the political ideology of the searchers.

Read my dissertation here