Publications
Journal Articles
-
Kantor, A., & Rafaeli, S. (2026). Fewer but more elaborate: How information accessibility filters participation and shapes online news comments. Journalism Studies, 1–24. https://doi.org/10.1080/1461670X.2026.2700623 · OSF
Conference Proceedings
-
Kantor, A., & Rafaeli, S. (2025). Data, sense, and sensibility: How data journalism style shapes interactivity. Selected Papers of Internet Research (AoIR). https://doi.org/10.5210/spir.v2024i0.15187 · OSF
-
Kantor, A., & Rafaeli, S. (2025). Conversations with data: How data journalism affects online comments in the New York Times. Proceedings of the 58th Hawaii International Conference on System Sciences, 8, 2386–2395. ScholarSpaceAcceptance rate: 37%
-
Kantor, A., & Rafaeli, S. (2021). Independence through data journalism. Selected Papers of Internet Research (AoIR). https://doi.org/10.5210/spir.v2021i0.12191 · GitHubAcceptance rate: 52%
-
Kantor, A., & Rafaeli, S. (2020). Open PISA: Dashboard for large educational dataset. Proceedings of the Seventh ACM Conference on Learning@Scale, 253–256. https://doi.org/10.1145/3386527.3406721
Submitted
-
Kantor, A. Choosing content is choosing a mechanism: How information features shape who joins online discussions and how much they elaborate.Separates two ways a news article’s information shapes discussion: changing who takes part, and changing how the same people engage. Using within-commenter variation across about 4,600 NYT stories, it finds that who takes part matters more, and that a single feature can work through both pathways in opposite directions.
Working Papers
-
Kantor, A. The narrative ratchet: Online information environments and the asymmetric adoption of inflation explanations.Classifies the causal explanations for inflation in New York Times coverage and about 620,000 reader comments. Explanations that blame identifiable actors are adopted quickly and decay slowly, regardless of the evidence against them.
-
Kantor, A. The last mile: Information design after toxicity removal.Shows that the information features of a news article predict the quality of the discussion that remains after moderation, across 29 Perspective API dimensions and about 620,000 NYT comments.
-
Kantor, A. Open data that cannot be told: Essential statistical opacity in international assessments.PISA’s released scores preserve gaps between groups but understate differences in who is affected, by about a fifth in the Israeli 2022 and 2025 samples. Claims that exceed the data come mainly from those who interpret the results for the public, not from news reports. The companion site opisa.org presents the 2025 results with their margins of error.
-
Kantor, A. Diagnosing and guiding students’ graph interpretations with a generative AI agent.Presents and evaluates a generative AI agent that diagnoses the data claims in students’ posts on The New York Times’ “What’s Going On in This Graph?” and drafts guiding replies. On 200 posts, it located 86 percent of seeded errors, but flagged misreadings about twice as often as expert moderators.
-
Kantor, A., Elbaz, Y., & Burk, L. Preserving growth rates in additive reconciliation of national accounts: A smoothed-Denton approach and the limits of weighting.Proposes a reconciliation method for seasonally adjusted national accounts that anchors levels to proportional raking while preserving each component’s growth rates.
Applied Statistics
-
Mor Shalom, M., Kantor, A., et al. (2025). Changes in trends of visits and service utilization by mental health patients in the community: A twelve-year study in Israel. Israel Journal of Health Policy Research.
-
Shapiro Ben David, S., Kantor, A., et al. (2025). Reported penicillin allergy in Israel: Clinical outcomes and antibiotic costs in a nationwide population-based cohort study. Journal of Allergy and Clinical Immunology: Global.
-
Kantor, A., & Kantor, D. What are the odds, Dr. Quinn?Uses a blood transfusion scene from Dr. Quinn, Medicine Woman and the real debate over donor sex and recipient mortality to teach base rates, confounding, the evidence hierarchy, Bayesian updating, and subgroup analysis. Submitted.
-
Kantor, A. One shot, twenty-one balls: Can a single stroke of snooker clear the whole table? GitHubProves that a single cue stroke can, in principle, pot all 21 balls on a snooker table, and estimates how rare such a configuration is. Submitted.
Popular Writing
-
Kantor, A. (2026). We analyzed 6,400 New York Times stories to find out how comments change when you give readers more information. Nieman Lab. Read the piece