As online search and recommendation platforms continue to permeate all aspects of human decision making, from how we navigate the web, to how we socialize and share information, studies detailing their impacts on society are urgently needed. However, identifying the impact of such platforms is challenging because they are often built around existing information ecosystems, trained on evolving behavioral data, and rarely an individual’s only source of information. Yet, as we’ve seen – from a man storming a pizzeria with an assault rifle in search of a secret pedophile ring, to a mass shooter citing a web search as the moment he was radicalized online, to an insurrection at the capitol – extreme consequences from these online environments can and do spill over into the real world. Given their ubiquity, research on these platforms should focus not only on the extent to which they have a sizable impact, but also on developing tools for identifying how, when, and why they produce biased, false, or unjust results. Among the most important of such platforms, is web search.In this thesis, I address three questions on the sociopolitical impact of partisanship in web search using an interdisciplinary mixed-methods approach that includes the psychological, behavioral, political, network, and computational social sciences. In Chapter 1, I provide a brief introduction on the importance of web search and each of the three research chapters presented in this thesis. In Chapter 2, I provide background on related topics, including the psychology of search engine influence, algorithm auditing techniques, and digital trace studies. In Chapter 3, I describe the results from a series of behavioral experiments that examine the impact of partisanship in web search rankings. In Chapter 4, I cover research in which I used surveys, real users, and simulated user activity to examine personalization and partisanship in web search rankings. In Chapter 5, I develop methods for collecting ecologically valid data on partisanship in web search, and report results from two studies. Last, in Chapter 6, I summarize the contributions and limitations of this thesis, and provide recommendations for future work in this vein, including the need to work with subject-matter experts, and the need for greater infrastructure and data sharing.
Christo Wilson (co-chair), Associate Professor, Khoury College, Northeastern University
David Lazer (co-chair, Professor, Network Science Institute, Northeastern University
Tina Eliassi-Rad, Professor, Network Science Institute, Northeastern University
Katya Ognyanova, Associate Professor, School of Comm. & Info.Rutgers University
Ronald is a fifth-year PhD candidate working with Drs. David Lazer and Christo Wilson. His research involves the design and application of computational tools, behavioral experiments, surveys, and qualitative interviews to measure user behavior, algorithmic personalization, and choice architecture in online platforms. Through his background in the social, behavioral, and network sciences, his goal is to foster a deeper and more widespread understanding of how humans and algorithms interact in digital spaces. Prior to joining Northeastern, Ronald graduated cum laude in psychology from the University of California San Diego and spent four years conducting experiments and surveys at the American Institute for Behavioral Research and Technology, a nonprofit research institute that he helped build and run. The research Ronald has been involved in has been published in journals including the Proceedings of the National Academy of Sciences, Proceedings of the ACM: Human-Computer Interaction, Journal of Technology in Human Services, Behavior Analysis: Research and Practice, and the Proceedings of the Web Conference, Web and Society.
No upcoming events