Research

Technology ⋆Inequality ⋆ Activism ⋆ Information ⋆ Digital Data

You can find a full list of my publications on my Google Scholar page or on my CV.

Current Projects

Current or prospective students interested in working with me on these projects, or others, should get in touch via email.

Comparative Information Contexts: France and the U.S

In Progress postdoctoral work in collaboration with Jen Schradie (PI)

This series of papers takes a comparative approach to understanding information and civic participation in France and the United States. It combines data from a survey, digital trace data of online behavior, and a seven day daily diary.

Bringing the Wright Class Structures Forward:A Computational and Comparative Analysis of Labor in the U.S. and France

Abstract Applying a relational class analysis, how do workers in a welfare state with strong worker protections compare to those in a weaker welfare state with minimal worker protections in the current era? This study revisits and updates Erik Olin Wright’s class taxonomy to analyze contemporary labor structures in the United States and France. While much of the existing class stratification research relies on occupational categories, income, and education, Wright’s framework provides a class relational approach by identifying class positions based on control over production, organizational authority, and credential-based autonomy. Using survey data collected from 8,000 respondents in both countries, we construct a new computational classification model, CCOWS (Class Classifier Of Wright’s Scheme), to systematically apply Wright’s taxonomy in a comparative context. Our findings demonstrate that, despite differences in labor protections and welfare regimes, class relations remain deeply structured along the axes of capital ownership, workplace authority, and skill-based credentials. A key contribution of this study is the application of computational text analysis to classify jobs from open-ended responses, offering a scalable and replicable approach to modernizing class schema applications. We find significant national differences, with the U.S. displaying a higher prevalence of the bourgeoisie and small employers, likely reflecting weaker labor protections and an entrepreneurial culture. In contrast, France exhibits a higher proportion of semi-credentialed workers and lower job autonomy, highlighting the role of a stronger welfare state in shaping workplace structures. Our study bridges the gap between qualitative class theory and quantitative measurement, providing new insights into how class structures evolve in contemporary labor markets. We extend stratification scholarship’s class categories into class relations. This work underscores the continued relevance of relational class analysis and demonstrates the potential of computational methodologies for empirical class research.

Educated Men Aren’t Bowling Alone: Loneliness, Community, and Technology in France and the United States

Abstract The existence of a loneliness epidemic is widely debated, with some scholars suggesting that increased internet use may exacerbate loneliness, while others argue that online communities can foster social connection. Notably, claims of heightened loneliness among men, especially youth, have gained attention, though the role of social class remains underexplored. Classic works such as Bowling Alone highlight the decline of social bonds in the United States, but comparative analyses across countries remain limited. Although global measures of loneliness exist, few account for both digital communities and inequalities, as well as the intersection of gender and class. In this paper, we leverage three connected data sources and computational methods to provide an in-depth cross-national comparison of the U.S. and France to investigate the question: How do people in countries with differing state policies and values around collectivism and individualism experience the so-called loneliness epidemic? Furthermore, what relationship do internet communities and information sharing practices have with community strength or isolation? Our data connects a survey fielded in both countries (N=8,000) with a 7-day diary (N=23,982 entries from 4,204 respondents) and 9 million digital trace observations (from over 1,800 respondents). Across three levels of community structure, focused on (i) family and friends, (ii) acquaintances, and (iii) online only connections, we find that in both countries, college educated men routinely report the highest levels of community and U.S. respondents consistently report higher levels of community than those in France. Contrary to claims of a loneliness epidemic among men or within more collectivist societies, we found the opposite. Analysis of the diary data suggests that this strength of community is not necessarily reflected in informational discussions, with both isolated and connected individuals reporting frequent conservations with friends and family members.

Intersectionality and inclusivity in a cross-country trilingual survey in the age of AI (also with Alexia Vallenas Weisse)

Abstract In the 20th century, feminist and intersectional scholars began to critique how researchers excluded women and other marginalized groups from the construction of survey variables by failing to pose questions relevant to their experiences and using language that “othered” identities of people who were not white men. Today, inclusive language practices are more common, but are often heavily politicized. This presents a challenge for survey researchers who need to recruit representative samples across a variety of ideological groups while not repeating the exclusionary survey practices of the past. This only gets more difficult in comparative work, as different countries approach inclusive language in various ways. For example, American feminists have struggled for more gender-neutral language while in France, “feminized” language is seen as more inclusive. Drawing on the process of designing and fielding a trilingual representative panel survey in France and the United States, we detail the challenges of writing consistent surveys across these two countries. Specifically, we document the challenges of balancing gender-inclusivity, respecting cultural differences around racial categories, and standardizing social class measures across these countries’ distinct economic and political systems, all while maximizing response rate and quality. In so doing, we analyze the incorporation of AI translation tools into the TRAPD workflow and the implications for creating diverse and inclusive surveys. In particular, while AI makes the process of translation more efficient, it is impossible to create surveys responsive to diverse gender identities and social contexts with these tools alone. We find how the well documented gender and racial biases within AI, therefore, also create bias in the adoption of AI tools in survey translation.

Data Access in the Post-API Age

The Administrative Burden of Social Media Data Access in the Post-API Age (With Jen Schradie)

Working paper available soon

Social Science Research in the Post-API Age : How data access shapes methods and questions

In Progress

Abstract In April 2018, Facebook closed the Pages API, a tool designed for advertisers but used by social science researchers for data collection. This sudden restriction threw research designs and protocols into disarray: halting ongoing data collection efforts, making replications impossible, and scrambling the ecosystem of software built off of the API. What can this event – the first marker of the “Post API Age” (Freelon, 2018) – tell us about the relationship between data, research questions, and findings? Classifying over 70,000 social science abstract between 2006 and 2022 by their method and topic, this project empirically demonstrates how dependent research is on easy, robust, and reliable data access. Preliminary results show that the closure of the Facebook Pages API reduced the computational study of Facebook by more than 50%, attesting to the role of company APIs in shaping social science research.

Gender Inequality in the Open Science Movement

Archive of Our Own: A Model for Gender Equity in Open Source Software

In Progress

Abstract Open Source Software, foundational to Internet technologies, has long had a gender problem. Women participate in open source software communities at rates far lower than their participation in wider technology spaces, which themselves are well known for persistent gender gaps. Here, I use the exceptional case of the open source project Archive of Our Own (Ao3), a site with feminist origins and design principles, to demonstrate that gender equality can exist, and continue to succeed, within open source software projects. Using digital trace data of all the contributions and contributors to the source code for Archive of our Own, I look at how the work itself is organized in ways that might support the feminist design principles and resilience of the larger site. Furthermore, this case study provides a third organizational format within open source software, that of cyclical and coordinated teams, instead of wide non-hierarchical participatory governance or oligarchic bureaucracy. While the specific practices of the Ao3 open source contributors are hardly generalizable to all other open source projects, it challenges the majority view of open source as a male domain.

Selected Publications and Projects

Book Bans in Political Context: Evidence from US Public Schools In PNAS NEXUS (2024) with Katie Spoon (U Colorado Boulder), Jack LaViolette (Columbia), and Marcelo Silva (Duke)

Paper now available!

Abstract In the 2021–2022 school year, more books were banned in US school districts than in any previous year. Book banning and other forms of information censorship have serious implications for democratic processes, and censorship has become a central theme of partisan political rhetoric in the United States. However, there is little empirical work on the exact content, predictors of, and repercussions of this rise in book bans. Using a comprehensive dataset of 2,532 bans that occurred during the 2021–2022 school year from PEN America, combined with county-level administrative data, multiple book-level digital trace datasets, restricted-use book sales data, and a new crowd-sourced dataset of author demographic information, we find that (i) banned books are disproportionately written by people of color and feature characters of color, both fictional and historical, in children’s books; (ii) right-leaning counties that have become less conservative over time are more likely to ban books than neighboring counties; and (iii) national and state levels of interest in books are largely unaffected after they are banned. Together, these results suggest that rather than serving primarily as a censorship tactic, book banning in this recent US context, targeted at low-interest children’s books featuring diverse characters, is more similar to symbolic political action to galvanize shrinking voting blocs.

This project is funded by a grant from the Social Science Research Council.

The Gender Divide in Wikipedia: Quantifying and Assessing the Impact of Two Feminist Interventions In Journal of Communication (2022) with Sandra González-Bailón (Upenn)

Available here

Video presentation here

Abstract Wikipedia has a well-known gender divide affecting its biographical content. This bias not only shapes social perceptions of knowledge, but it can also propagate beyond the platform as its contents are leveraged to correct misinformation, train machine-learning tools, and enhance search engine results. What happens when feminist movements intervene to try to close existing gaps? Through a quantitative analysis of over 11,000 Wikipedia articles, we provide an evaluation of two popular feminist interventions designed to counteract gender inequality within digital information projects. We find that the interventions are successful at adding content about women that would otherwise be missing, but they are less successful at addressing structural biases that limit the visibility of that content. This leads us to argue for a more granular and cumulative analysis of gender gaps in collaborative environments. We also discuss the implications for future scholarship on digital inequalities.

Protest Networks, Mobilization, and Resilience Book Chapter in Social Networks & Social Resilience edited by T.A.B.Snijders, E.Lazega & R.Wittek. (2023) – with Sandra González-Bailón (UPenn)

Available here

The movement Black Lives Matter has spurred massive, international protests since it first emerged around the eponymous Twitter hashtag in 2013. Social network analysis is well primed to answer questions around how movements like BLM gain traction, create or change story frames, enact policy change, and mobilize supporters, both online and offline. Using BLM as an exemplar, this chapter reviews research about the movement that uses network approaches to understand its emergence, growth, and resilience, especially as it enters its second decade and contends with counter movements. More generally, the chapter offers a discussion of the opportunities and challenges associated with incorporating social media data to the analysis of resilience as it manifests in the growth of networked social movements.