Tina Law

Logo

Computational Sociologist

GitHub

Google Scholar

Email

Bluesky

about

I am an Assistant Professor of Sociology at the University of California, Davis. I study inequality, race and ethnicity, democracy, and computational and quantitative methodology. My primary area of research focuses on understanding how racially minoritized and low-income Americans define and advance their goals for housing, safety, and political self-determination, particulary in cities that are often highly unequal and anti-democratic. Another growing area of my research focuses on adapting and applying generative AI for text- and image-based analysis tasks in social science research. I received my Ph.D. in sociology from Northwestern University in 2022.

I am also the co-founder of varyCSS, a collective of scholars who seek to advance and improve the field of computational social science by centering the work of scholars from historically marginalized communities. We maintain a Scholars Database and a Literature App on our website, varyCSS.org.

recent papers

Law, Tina and Leslie McCall. 2024. "AI Policymaking: An Agenda for Sociological Research." Socius, 10. [paper]

This paper calls for more robust and coordinated policy-oriented research on AI among sociologists. To support sociologists in this endeavor, we provide an overview of the rapidly evolving AI policy landscape, contrasting two leading approaches to AI governance: a safety-based vs. an equity-based approach. We argue that the safety-based approach is the predominant approach but is inadequate, and suggest that sociologists can shift AI policymaking to prioritize equity by examining corporate political power in AI policy debates and by organizing research around four sociological questions centered on equity and public engagement.

Kesari, Aniket, Jae Yeon Kim, Sono Shah, Taylor Brown, Tiago Ventura, and Tina Law (equal authorship). 2024. "Training Computational Social Science Ph.D. Students for Academic and Non-Academic Careers." PS: Political Science & Politics 57(1):101-106. [paper]

This paper shares the “hidden curriculum” of computational social science (CSS) training and professionalization for social science Ph.D. students. We outline core competencies and provide resources as it relates to three primary areas of CSS training and professionalization: (1) learning data science skills, (2) building a portfolio that uses data science to answer social science questions, and (3) connecting with computational social scientists. We also provide practical recommendations for departments and professional associations to better support Ph.D. students interested in CSS.

Law, Tina and Elizabeth Roberto. "Generative Multimodal Models fo Social Sceince: An Application with Satellite and Streetscape Imagery." revise and resubmit [preprint]

This paper introduces an adapted framework for analyzing images with generative multimodal models (gen MMs), artificial intelligence models that can engage wtih multiple modalities (e.g., text, image, audio). We demonstrate this framework with an empirical application that uses OpenAI’s gpt-4o model to analyze satellite and streetscape images to identify built environment features that contribute to contemporary residential segregation in U.S. cities. We find that model-generated labels are more reliable than research assistant-generated labels and comparably valid to expert-generated labels. We also provide recommendations for ensuring that the future of large-scale image analysis in the social sciences is not only rigorous and reproducible but also ethical and sustainable.

Than, Nga, Leanne Fan, Tina Law, Laura Nelson, and Leslie McCall. "Updating "The Future of Coding:" Qualitative Coding with Generative Large Language Models." revise and resubmit [preprint]

This paper tests the ability of generative large language models (LLMs) to replicate and augment traditional qualitative coding with an empirical application focused on identifying U.S. news articles discussing the complex concept of economic inequality. We generally find that generative LLMs can perform nearly as well as supervised machine learning models in accurately matching hand-coding output. We also find that propietary generative LLMs do not perform substantially better than open-source generative LLMs.

Ochoa, Erin, Tina Law, and Andrew Papachristos. "A Spatialized Synthetic Control Analysis of a Multi-Site Community Initiative to Reduce Gun Violence." revise and resubmit

This paper uses an innovative extension of synthetic control methods that accounts for spatial dynamics to evaluate the effects of a community violence intervention (CVI) led by a collaborative of organizations in 13 communities in Chicago. CVIs are community-led approaches to reducing gun violence which have proliferated in recent years but are difficult to assess due to their non-random and varied design and implementation. We estimate that the CVI was associated with significant decreases in victimization rates in six treated areas over 54 months. We call for more quasi-experimental evaluations that appropriately model the on-the-ground realities of program implementation in order to advance the growing science of CVIs.

recent and upcoming talks

I am currently available for in-person and remote talks during the Winter and Spring 2025 quarters

teaching

Social Relationships

This course introduces undergraduate students to the sociological study of social networks, or how social networks can serve as, what sociologist Charles Kadushin calls, “problem-solving tools.” Social networks consist of a set of actors and relationships within a given context, and social network analysis is a theoretical and methodological approach that examines patterns of relationships between actors in order to understand pressing social issues. We will learn about social network theory and analysis and how to apply these concepts and tools to design a network-based research project. In particular, we will discuss examples of how social network theory and analysis can be used to address three important social topics: inequality, political behavior, and artificial intelligence.

Introduction to Social Statistics

This course introduces undergraduate students to quantitative sociological research, or the use of quantitative methods (i.e., tools based on numbers and involving software) to ask and answer research questions in order to better understand pressing social issues and potential interventions. In particular, we will focus on applying quantitative methods to examine poverty, inequality, and social mobility in the U.S. We will cover: research question formulation, sampling and measurement, probability and uncertainty, basic statistical inference, and hypothesis testing.