Publications

Public health crises demand timely surveillance of disease and interventions. This was a substantial challenge during COVID-19 in the United States due to the federalized response. States varied widely in infection and vaccination rates. Continuous state-level probability samples for tracking were unavailable and would have been logistically and financially onerous. Administrative data eventually became inaccurate. We thus evaluate the accuracy of large-scale opt-in non-probability state and national samples (from the Covid States Project) that estimated infection and vaccination rates on a near-continuous basis. The estimates aligned closely with a high-quality national probability sample and state-level administrative data (when such data were reliable). We offer evidence that the success of the surveys, compared to other less accurate non-probability surveys, may have stemmed from the successful recruitment of respondents with low trust in health institutions (i.e., individuals who often avoid surveys). We conclude by discussing how the Covid States Project can inform further data collection efforts requiring extensive spatial and temporal coverage.

This study examines how mainstream Taiwanese newspapers portrayed sexual minorities from 2010 to 2021, a period marked by rising LGBTQ+ visibility and major legal change. Drawing on institutional mediation and legitimacy perspectives, we test three hypotheses: first, that coverage becomes more negative as LGBTQ+ issues gain visibility over time; second, that gay men receive more negative coverage than lesbian women; and third, that major legal and social milestones shift coverage toward more neutral portrayals and a broader range of topics. To evaluate these expectations, we analyze 18,558 articles from China Times, Liberty Times, and United Daily News using web-scraped data, AI (i.e. GPT-4 model) assisted classification of articles as gay with time series regression models. The findings reveal an overall increase in positive news coverage over the decade, but disparities in the frequency and sentiment of coverage between gay and lesbian subjects persist. The study concludes that while there has been progress toward inclusiveness, significant events and milestones have only partially influenced the portrayal of the LGBTQ+ community. The results highlight the ongoing need for efforts toward equitable media representation.

This study examines how mainstream Taiwanese newspapers portrayed sexual minorities from 2010 to 2021, a period marked by rising LGBTQ+ visibility and major legal change. Drawing on institutional mediation and legitimacy perspectives, we test three hypotheses: first, that coverage becomes more negative as LGBTQ+ issues gain visibility over time; second, that gay men receive more negative coverage than lesbian women; and third, that major legal and social milestones shift coverage toward more neutral portrayals and a broader range of topics. To evaluate these expectations, we analyze 18,558 articles from China Times, Liberty Times, and United Daily News using web-scraped data, AI (i.e. GPT-4 model) assisted classification of articles as gay with time series regression models. The findings reveal an overall increase in positive news coverage over the decade, but disparities in the frequency and sentiment of coverage between gay and lesbian subjects persist. The study concludes that while there has been progress toward inclusiveness, significant events and milestones have only partially influenced the portrayal of the LGBTQ+ community. The results highlight the ongoing need for efforts toward equitable media representation.

Follower ties play a major role in many social media platforms, representing users' choices on what content to pay attention to. This Element examines the role of geography and similarity by gender, age, race, and partisanship with respect to attention in social media by studying the follower ties among 1.1 million Twitter accounts matched to U.S. voter records. We find that geographic proximity is the dominant predictor of follower ties, and that demographic similarity by age and race/ethnicity are quite important. Surprisingly, given the prominence of political polarization in the contemporary US, partisanship plays a relatively minor role. In addition, our results indicate that the tendency to follow nearby users leads to following users of the same race/ethnicity and partisanship. Our findings highlight the enduring significance of physical geography in virtual spaces and that political preference is not a dominant determinant of online attention in social media.

Computational social science has expanded the capacity of scientists to study connected human behavior at previously unprecedented scales. Yet from its beginning, scientists expressed concern that its reliance on private companies might produce a body of work that cannot be critiqued or replicated. Such commercial determinants of science have been observed in other fields including public health where science has implications for corporate liability. In this meta-scientific report, we analyze the population of computational social science articles about technology platforms published in three general scientific journals to investigate commercial determinants of scientific replicability in computational social science. We find that only 26% of those papers can be replicated today, and that 34% of computational social science studies published in leading general scientific journals rely on special arrangements with corporations that are impossible to replicate without special permission. We find that articles relying on API access, scraping or access to nonprofit platforms have much higher potential for replication. These findings are consistent with broader literature on the commercial forces that determine the direction and reliability of science.

Background: Generative artificial intelligence (AI) use has been suggested to have adverse mental health consequences but a causal relationship has not been examined.

Objective: To simulate a randomised controlled trial of AI use in a work, school or personal context by applying target trial emulation to multiple waves of data from a nationally representative survey.

Methods: We conducted a target trial emulation using non-probability survey data from three waves of a nationally representative survey conducted between 18 June 2024 and 8 January 2025. Participants aged ≥18 years reported generative AI use frequency at baseline. High-frequency use was defined as multiple times per week or more. The primary outcome was depressive symptom severity measured using the Patient Health Questionnaire 9-item (PHQ-9) at follow-up. Generalised causal forests assessed heterogeneity of treatment effects.

Findings: Among 19 099 participants assessed at baseline, 2862 (15.0%) reported AI use at least multiple times per week. A subset of 3109 (16.3%) returned for follow-up. In the primary weighted analysis, high-frequency use was not significantly associated with change in PHQ-9 score at follow-up (mean difference -0.18, 95% CI -0.94 to 0.59; p=0.65). Multiple sensitivity analyses using alternate outcome definitions also did not identify significant causal effects. Generalised causal forests yielded no significant evidence of heterogeneity of effect (p=0.81).

Conclusions: In an emulated randomised trial among US adults, generative AI use was not associated with subsequent depressive symptoms. This result does not support the premise that AI use causes greater depressive symptoms, although adverse outcomes among vulnerable individuals cannot be excluded.

Clinical implications: AI use is unlikely to cause increased depressive symptoms among most US adults. Continued monitoring should clarify potential risks among vulnerable populations.

Keywords: Mental Health; Patient Health Questionnaire; Psychiatry.

‍

Importance  Generative artificial intelligence (AI) has rapidly entered mainstream use in the US, but its association with mental health has not been characterized.

Objective  To examine the associations of the extent and type of generative AI use among US adults with negative affective symptoms in a large, nationally representative sample.

Design, Setting, and Participants  This survey study used data from a 50-state US internet nonprobability survey conducted between April and May 2025. Survey respondents were aged 18 years and older. Data were analyzed in August 2025.

Exposure  Participants self-reported generative AI and social media use.

Main Outcomes and Measures  The outcome of interest, negative affect, was measured using the Patient Health Questionnaire 9-item (PHQ-9).

Results  There were 20 847 unique participants, with mean (SD) age 47.3 (17.1) years and 10 327 (49.5%) female, 10 386 (49.8%) male, and 134 (0.6%) nonbinary participants; 2152 participants (10.3%) reported using AI at least daily, including 1053 participants (5.1%) who reported daily use and 1099 participants (5.3%) who reported use multiple times per day. Among participants who used daily or more frequently, 1033 (48.0%) reported use for work, 246 (11.4%) for school, and 1875 (87.1%) for personal applications. In survey-weighted regression models, daily or more frequent AI use was significantly more common among men, younger adults, those with higher education and income, and those in urban settings. Greater AI use was associated with greater levels of depressive symptoms in sociodemographic-adjusted regression models: (daily use: β = 1.08 [95% CI, 0.55-1.62]; multiple times per day: β = 0.86 [95% CI, 0.35-1.37]) compared with nonuse, and with greater likelihood of reporting at least moderate depressive symptoms (odds ratio [OR], 1.29 [95% CI, 1.15-1.46]); similar patterns were observed for anxiety and irritability. The highest estimates were observed among individuals using AI for personal use (β = 0.31 [95% CI, 0.10-0.52]) and those aged 25 to 44 years (β = 1.22 [95% CI, 0.70-1.74]) or 45 to 64 years (β = 1.38 [95% CI, 0.72-2.05]).

Conclusions and Relevance  This survey study found that AI use was significantly associated with greater depressive symptoms, with magnitude of differences varying by age group. Further work is needed to understand whether these associations are causal and explain heterogeneous effects.

‍