Two projects – one on who finances eviction and the other on using AI to study persuasive speech — have been awarded grants through the Dean for Research Innovation Fund for New Ideas in the Social Sciences.

The fund promotes explorations of society and human advancement through investigations of new or enduring questions. Administered by the Office of the Dean for Research, the funds are part of the University's commitment to fund new directions in research and scholarship. Outcomes may include new ideas, working groups, conferences, technologies, software, datasets, expanded access to research resources, and/or a major piece of scholarly work.

Projects are awarded funding following a competitive application process. The winning projects were selected on criteria including the soundness of the research design; how the proposal brings new ideas together to suggest novel outcomes; and potential impacts in terms of shifting thinking in new directions or addressing central questions in the field of scholarship.

Who finances evictions?

This project will investigate the financial structures that contribute to eviction hotspots, or regions within cities with a disproportionate share of displacement. Matthew Desmond, Maurice P. During Professor of Sociology, will lead a team to explore why such hotspots occur, with a focus on the loans and financing agreements that underlie ownership of multifamily buildings.

Every year in the United States, 2.7 million renter households face the threat of losing their homes to eviction. Previous work by the Desmond-led Eviction Lab at Princeton showed that a relatively small number of properties in each city are responsible for a disproportionate share of evictions. To understand these hotspots, the group will ask what types of lenders are more likely to finance high-evicting buildings, what sources of housing finance explain variations in eviction rates, and what kinds of financial instruments are most often used by high-evicting buildings. The answers will have major policy implications as the federal government evaluates actions to protect renters. The team will also create a related data product that will be made available to researchers in the field more broadly.

Assessing the elements of persuasion

Why are some arguments more powerful than others? What is it about an essay or a speech that convinces people to change their minds? A new project led by Brandon Stewart, associate professor of sociology and a researcher in the Office of Population Research, aims to learn what words, phrases and arguments successfully persuade people. The team will apply textual analysis tools and modern causal-inference designs to discover what features make a document persuasive. Using new large-language models, the team will create new machine-generated texts that possess these features, allowing the researchers to study systematically how specific attributes of the texts convince their audience. This ability to study the causal link between texts and outcomes will enhance our understanding of the relationship between words and resulting actions.

The team, which involves a collaboration with Ronnie Sircar, Eugene Higgins Professor of Operations Research and Financial Engineering and chair of the department, and work by graduate student Till Raphael Saenger, will develop statistical analysis and machine learning applications to create a new, social-science-driven methodology and toolset. The researchers aim to create technologies and user-friendly software so that others can replicate the strategy in their own studies.