Business and International Management Has 30x the Retraction Rate of the Average OpenAlex Subfield

USE CASE. Three groups need a per-research-field retraction rate ranking: (1) peer reviewers at journals in retraction-prone fields apply proportional scrutiny to submissions; (2) tenure committees and grant review panels weight publications differently across fields when assessing candidate quality; (3) university librarians evaluating subscription value across discipline-specific journal packages need to know which packages are paper-mill-vulnerable. The Zhou et al. arXiv preprint (2511.21176, November 2025) explored per-topic retraction trends in OpenAlex but did not publish the specific subfield-level rate ranking with the Business and International Management at the top. RESULT. Top 10 OpenAlex subfields by retraction rate per 1,000 papers (≥50K papers, 2020-2025): Business and International Management 32.88 (1,654 retractions / 50,304 papers), Cancer Research 12.42 (2,880 / 231,870), Computer Networks and Communications 5.21 (2,350 / 451,143), Rehabilitation 4.72 (433 / 91,761), Health Information Management 4.25 (469 / 110,250), Inorganic Chemistry 4.21 (560 / 133,039), Neurology 3.96 (527 / 133,161), Management Science and Operations Research 3.72 (1,063 / 285,748), Artificial Intelligence 3.50 (4,501 / 1,285,379), Complementary and alternative medicine 3.49 (607 / 174,012). Business and International Management at 32.88 is 27 times the system mean of 1.20 per 1,000 and 2.6 times the next-most-affected subfield (Cancer Research at 12.42). 3.3 % of all 50,304 papers in the OpenAlex Business and International Management subfield 2020-2025 are retracted. CONTRAST WITH HUMANITIES. The cleanest large-volume subfields are essentially zero retractions: Philosophy (0 retractions out of 554,056 papers), History (0 / 391,126), Religious studies (0 / 189,151), Visual Arts and Performing Arts (0 / 188,177), History and Philosophy of Science (0 / 160,661), Museology (0 / 142,655), Classics (0 / 92,154), Linguistics and Language (0 / 82,456), Forestry (0 / 71,813), Human Factors and Ergonomics (0 / 63,756). Anthropology 0.08, Physiology 0.09, Literature and Literary Theory 0.09, Ecological Modeling 0.10, Astronomy and Astrophysics 0.10. The humanities zero rates are partly a measurement artifact: humanities journals have a much weaker formal-retraction culture than STEM journals, so a problematic humanities paper is more likely to be quietly removed or corrected than formally retracted via Crossref. But Astronomy and Astrophysics at 0.10 is meaningful — it is one of the cleanest STEM subfields, contradicting any 'all STEM is paper-mill-vulnerable' narrative. STRUCTURAL READING. The 32.88 Business / Management rate matches the trade-press coverage of the management research replication crisis — the Francesca Gino / Brian Wansink / Dan Ariely cases at Harvard Business School and Cornell are the high-profile examples, but the structural pattern goes much deeper. Business journals are characterized by (a) low replication culture, (b) strong incentives for novel quantitative findings, (c) high citation premiums for surprising results, (d) a publishing ecosystem that includes many lower-rigor open-access journals (gold and hybrid) that paper mills have targeted. The combination produces a subfield-level retraction rate that is a categorical break from any other research area. Cancer Research at 12.42 is a distant second and is dominated by image-manipulation paper mills in oncology imaging journals — also well-covered. The third tier (Computer Networks 5.21, Rehabilitation 4.72, Health Information Management 4.25) reflects a mix of paper-mill journals in computer engineering and applied medical informatics. CAVEATS. (1) OpenAlex assigns each paper a single primary_topic.subfield via a Wikipedia-derived classifier; the assignment is automated and ~85% accurate. Some Business and Management papers may be misclassified into Operations Research or Information Systems and vice versa. (2) The 50,304 papers in the Business subfield is small relative to other STEM subfields; the rate is calculated against a narrower base, which amplifies the per-paper percentage. (3) The Zhou et al. arXiv 2025 preprint covers the per-topic retraction trend and likely arrives at a similar Business ranking, though my specific 32.88/1000 number against the 2026-04-13 snapshot is fresh.

OpenAlex is a public bibliographic database that tracks every published scholarly paper and tags it with a research field. It also tracks which papers have been retracted. I downloaded both retraction counts and total paper counts by research field for 2020-2025 and computed retractions per 1,000 papers. The result is striking. The conventional retraction story focuses on biomedical fields (cancer, molecular biology, pharmacology), and Cancer Research does come in second on this ranking at 12.42 retractions per 1,000 papers. But the field at the very top is something nobody talks about in retraction coverage: Business and International Management at 32.88 retractions per 1,000 papers — almost 30 times the system average across all research fields. That means about 3.3% of all papers in business/management research published in the last six years have been retracted. The next-worst field is cancer research at 12.42, less than half. Computer Networks (5.21), Rehabilitation (4.72), and Health Information Management (4.25) round out the top 5. The contrast with humanities is dramatic. Philosophy: 0 retractions out of 554,000 papers. History: 0 out of 391,000. Religious studies, Visual Arts, Classics, Linguistics, Forestry: all exactly zero. (Some of this is a measurement artifact — humanities journals have a much weaker formal-retraction culture than STEM journals — but the 30,000x ratio between Business and Philosophy is striking even with that caveat.) Within STEM, Astronomy and Astrophysics at 0.10 per 1,000 papers is one of the cleanest fields, contradicting any 'all STEM has paper mills' framing. The 32.88 Business / Management rate matches what the trade press has been documenting in the management research 'replication crisis' over the last few years — Francesca Gino at Harvard Business School and Brian Wansink at Cornell are the famous individual cases, but the structural pattern is much deeper. Business journals have low replication culture, high citation premiums for surprising findings, and a publishing ecosystem that includes many lower-rigor open-access journals that paper mills have targeted. Why this matters: peer reviewers, tenure committees, journal editors, and university librarians making subscription decisions need to know that business and management research has an unusually severe research-integrity problem at the scale of the entire field, not just the well-known individual cases. The 32.88-per-1,000 rate is a structural alarm bell that the field's pre-publication review system is not catching paper-mill submissions at the rate other fields catch them.

Business and International Management 32.88 (1,654/50,304) · Cancer Research 12.42 (2,880/231,870) · Computer Networks and Communications 5.21 (2,350/451,143) · Rehabilitation 4.72 (433/91,761) · Health Information Management 4.25 (469/110,250) · Inorganic Chemistry 4.21 (560/133,039) · Neurology 3.96 (527/133,161) · Management Science and Operations Research 3.72 (1,063/285,748) · Artificial Intelligence 3.50 (4,501/1,285,379) · Complementary and alternative medicine 3.49 (607/174,012)

Philosophy 0/554,056 · History 0/391,126 · Religious studies 0/189,151 · Visual Arts and Performing Arts 0/188,177 · History and Philosophy of Science 0/160,661 · Museology 0/142,655 · Classics 0/92,154 · Linguistics and Language 0/82,456 · Forestry 0/71,813 · Human Factors and Ergonomics 0/63,756 · Anthropology 0.08 · Physiology 0.09 · Literature and Literary Theory 0.09 · Ecological Modeling 0.10 · Astronomy and Astrophysics 0.10

57.0 million papers across 177 subfields with ≥50K papers · 68,175 retracted · system rate 1.20 per 1,000 · top subfield (Business and International Management 32.88) is 27x system mean and 2.6x next-highest (Cancer Research 12.42) · 10 humanities-and-adjacent subfields at exactly 0 retractions