Are Women Treated Differently Than Men When Presenting Research?

Dupas et al. (2025) examine whether men and women are treated differently when presenting research in economics seminars. They ask whether presenter gender affects how often and how negatively audience members interrupt speakers. They analyze detailed data on thousands of seminars, job talks, and conference presentations, combining human-coded observations with machine learning analysis of audio recordings. The authors find that women receive about 10–20 percent more interruptions than similarly situated men, even after controlling for topic, presenter characteristics, and audience size. Women also experience more interruptions rated as negative or mid-sentence, suggesting systematic differences in seminar treatment rather than chance or topic-driven effects.

Why This Article Was Selected for The Policy Scientist

This article addresses a broadly important policy issue: how gender dynamics within elite professional institutions shape participation, career advancement, and the production of knowledge. Seminar culture is a central arena in which authority, credibility, and expertise are publicly assessed, making gender-based differences in treatment consequential for hiring, promotion, and retention in economics and related fields. The study is timely given renewed attention to gender representation and professional conduct across academia. Dupas and coauthors have written extensively on related topics, and this paper advances the literature using unusually large, high-quality data across settings. While the observational data and regression framework are carefully implemented, causal inference remains limited; future experimental or quasi-experimental designs would strengthen conclusions about gender-based mechanisms.

Full Citation and Link to Article

Pascaline Dupas, Amy Handlan, Alicia Sasser Modestino, Muriel Niederle, Mateo Seré, Haoyu Sheng, Justin Wolfers, & The Seminar Dynamics Collective. (Forthcoming 2026). Gender differences in economics seminars. American Economic Review. https://doi.org/10.1257/aer.20241718.

Central Research Question

This article examines whether men and women are treated differently when presenting research in economics seminars. The central question is whether presenter gender affects the frequency, nature, and tone of interruptions by audience members within otherwise similar academic settings. The authors focus on differential treatment rather than differential outcomes, asking whether women receive more interruptions than men after accounting for seminar context, research topic, presenter characteristics, and audience composition. A related but secondary question concerns the character of these interruptions: whether they differ in tenor, tone, timing, or source in ways that may plausibly reflect differences in how authority, credibility, or expertise are perceived during professional interactions. The study also explores whether higher engagement with female presenters reflects benign interest, bias, or a mixture of both.

Previous Literature

The paper builds on a large and growing literature documenting gender disparities within the economics profession. Prior research shows that women receive less credit for coauthored work, face higher publication barriers, experience longer review times, and are cited less frequently in top journals. Other work finds gender differences in promotion, recognition, and professional advancement, even among similarly qualified economists. Outside economics, studies in fields such as engineering and astronomy have found that women are asked more questions or experience more audience engagement during presentations. Related work in political science, law, and public institutions documents higher interruption rates for women in formal deliberative settings.

What distinguishes this article from earlier studies is its focus on seminars as a central professional institution where reputations are formed and labor market outcomes are shaped. While earlier work documents disparities in outcomes such as promotion or publication, this study examines the interactional processes that may contribute to those outcomes. The authors, many of whom have written extensively on gender, labor markets, and professional norms, position the paper as filling an important gap by systematically measuring seminar dynamics across institutions, formats, and years.

Data

The study draws on five distinct datasets covering thousands of presentations between 2019 and 2023. These include in-person job market talks, regular departmental seminars, conference presentations at the NBER Summer Institute, online seminars, and hybrid conference presentations. In total, the data span over 2,300 talks across a wide range of institutions and fields within economics.

Three datasets rely on real-time human coding by trained graduate students who recorded every interaction between presenters and audience members, including the timing, duration, type, and perceived tenor of interruptions. Two datasets rely on machine learning techniques applied to audio recordings, using speaker diarization to identify interruptions, neural networks to classify speaker gender, and audio analysis to measure tone. The combination of human-coded and machine-coded data allows for triangulation across methods, reducing reliance on any single measurement approach.

The datasets are unusually rich, capturing not only counts of interruptions but also their characteristics and the gender of both presenters and audience members. While the samples are not randomly drawn from all seminars, they cover a broad and influential segment of the economics profession, including top departments and major conferences.

Methods

The authors use multivariate regression models to estimate the association between presenter gender and various measures of interruptions. The primary specification compares men and women presenting within the same seminar series or conference context, using fixed effects to control for field, institution, and seminar norms. Additional controls include presentation duration, presenter seniority, citation counts, institutional rank, and research characteristics such as topic and method, when available.

The main outcome variables include the total number of interruptions, interruptions per hour, and counts of interruptions by type, tenor, and tone. Separate analyses examine who interrupts, whether interruptions overlap with the presenter’s speech, and whether they are rated as negative, neutral, or positive. The authors emphasize robustness, showing that results are stable across alternative specifications, samples, and coding approaches.

The empirical strategy is observational. While the models are carefully specified and make extensive use of fixed effects, they do not rely on randomized or quasi-experimental variation. As a result, the findings speak to differential treatment rather than causal mechanisms, a limitation the authors acknowledge explicitly.

Findings/Size Effects

Across most seminar settings, women receive significantly more interruptions than men. In job market talks and regular departmental seminars, women experience roughly 10 to 20 percent more interruptions than similarly situated men. In absolute terms, this corresponds to approximately three to four additional interruptions per presentation in longer, highly interactive settings. These differences persist after controlling for research topic, presenter characteristics, and audience size.

The gap is particularly pronounced in recruitment contexts, such as job talks, where professional stakes are high. In contrast, the gender gap is smaller and statistically indistinct from zero in some conference settings, particularly where presentations are shorter or more structured. Online seminars show fewer interruptions overall, but still exhibit a sizable relative gender gap.

Beyond frequency, the nature of interruptions differs. Human-coded data indicate that women receive a disproportionately larger share of interruptions rated as negative in tenor, including patronizing or hostile remarks. Machine-coded audio data show that women experience more overlapping speech, particularly from male audience members. At the same time, women also receive more interruptions of all types, including suggestions and clarifying questions, making interpretation ambiguous.

The authors also find that seminars with female presenters tend to have larger and more gender-diverse audiences, suggesting higher engagement or a potential role model effect. Importantly, controlling for audience size does not eliminate the interruption gap, indicating that higher attendance alone does not explain the results.

Conclusion

This article provides the most comprehensive empirical analysis to date of gender differences in economics seminar dynamics. By documenting consistent differences in how men and women are interrupted across settings, it highlights interactional mechanisms that may contribute to broader gender disparities in the profession. The study’s strengths lie in its scale, innovative data collection, and careful attention to institutional context.

At the same time, the observational nature of the analysis limits causal interpretation. The results would be strengthened by future work using experimental or quasi-experimental designs, such as randomized seminar rules or natural experiments affecting interaction norms. Despite this limitation, the findings generalize plausibly to other high-skill, discussion-intensive professional settings where authority is publicly negotiated. Overall, the paper makes a substantial contribution by shifting attention from outcomes to processes, offering a clearer empirical foundation for understanding gender dynamics in elite academic environments.