Saudi Arabia Has the World's Highest Country-Level Retraction Rate, and Ethiopia and Uzbekistan Are in the Top 3

USE CASE. Three groups need a country-level retraction rate ranking: (1) international research funders (Wellcome Trust, Bill & Melinda Gates Foundation, EU Horizon Europe, World Bank Knowledge for Change Program) deciding which country research-capacity programs need integrity training intervention; (2) journal editors and peer reviewers deciding how to weight submissions from specific country authors when assessing self-citation, paper-mill, and fabrication risk; (3) US NIH ORI / NSF OIG and the Committee on Publication Ethics (COPE) prioritizing international research-integrity coordination. The conventional country retraction ranking is by RAW count, which is dominated by China and India because they publish the most papers. Per-paper rate is the meaningful normalization. The Research Integrity Risk Index (RI^2) published in Scientometrics 2025 covers institutions and journals but not the per-country rate aggregation in current published form. RESULT 1 (the rate ranking). Top 15 countries by retraction rate per 1,000 papers (≥10K papers in 2020-2025): Saudi Arabia 7.96 (3,342 retractions / 419,833 papers), Ethiopia 6.10 (636 / 104,206), Uzbekistan 5.55 (475 / 85,527), Iraq 5.53 (1,083 / 195,753), Pakistan 5.47 (2,071 / 378,335), Macao 5.14 (245 / 47,708), Egypt 4.29 (1,576 / 367,778), China 4.17 (27,478 / 6,590,178), Oman 4.07, Jordan 4.02, India 3.96 (9,420 / 2,379,970), Yemen 3.94, Kyrgyzstan 3.74, Bahrain 3.54, Nepal 3.45. The system mean is 1.45 — Saudi Arabia is 5.5x the system mean and 18.9x the cleanest large-volume country (Indonesia at 0.42x system). RESULT 2 (the conventional ranking flipped). Top 5 by raw count: China 27,478 (rate 4.17, rank #8), India 9,420 (rate 3.96, rank #11), United States 3,504 (rate 0.43, rank ~70th), Saudi Arabia 3,342 (rate 7.96, rank #1), Pakistan 2,071 (rate 5.47, rank #5). The US has the third-highest raw count but the rate is 0.43 per 1,000, far below the system mean — typical for high-volume publishing nations with aggressive journal retraction policies. The conventional 'China dominates retractions' framing is correct on raw counts but misses that Saudi Arabia, Ethiopia, Uzbekistan, Iraq, and Pakistan all have higher per-paper rates than China. RESULT 3 (the cleanest countries). 15 countries with the lowest retraction rate (≥10K papers): Panama 0.04 per 1,000, Paraguay 0.07, Costa Rica 0.12, Cuba 0.12, Bolivia 0.16, Indonesia 0.17, Brazil 0.20, Luxembourg 0.20, Austria 0.21, Slovenia 0.22, Cote d'Ivoire 0.24, Venezuela 0.25, Moldova 0.25, Argentina 0.26, Croatia 0.26. The Latin American cluster (Panama, Paraguay, Costa Rica, Cuba, Bolivia, Brazil, Venezuela, Argentina) is striking — the entire region has retraction rates below 0.30 per 1,000, less than 21% of the system mean. Indonesia at 0.17 with 2.3 million papers is the single most-anomalous low-rate country given its publishing volume. STRUCTURAL READING. The country-level retraction rate is a JOINT measure of (a) the underlying research-integrity rate at the country's institutions and (b) the willingness of the journals where those researchers publish to issue retractions. High-rate countries (Saudi Arabia, Ethiopia, Uzbekistan) tend to publish in international journals (Wiley, Springer, Elsevier, Hindawi, MDPI) that DO retract. Low-rate countries like Indonesia (with 2.3M papers but only 381 retractions) likely publish more heavily in domestic and predatory journals that don't issue formal retraction notices even when papers are problematic. So Indonesia's 0.17 should not be interpreted as 'Indonesian research is the cleanest in the world'; it should be interpreted as 'Indonesian researchers publish in venues that don't retract, so the retraction rate measures journal policy more than research quality'. Conversely, the Saudi Arabia, Ethiopia, and Uzbekistan numbers ARE meaningful because their researchers publish in venues that DO retract — so the high rate reflects an actual signal about their publishing systems, not a measurement artifact. CAVEATS. (1) OpenAlex authorships.countries reflects the country of any author affiliation; multi-country papers count toward both countries. (2) The is_retracted flag is set by Crossref / publishers and propagates to OpenAlex with some lag; very recent retractions may be undercounted. (3) The 'small countries clustering at the high end' (Macao 5.14, Oman 4.07, Yemen 3.94, Kyrgyzstan 3.74) reflects small-N variance in the retraction count. (4) The per-paper rate doesn't normalize for paper QUALITY — a country publishing primarily in low-impact venues may have a lower retraction rate than one publishing in high-impact venues for reasons unrelated to research integrity.

When a scientific paper is found to be wrong (faked data, image manipulation, plagiarism, paper-mill submission), the journal can issue a 'retraction'. A bibliographic database called OpenAlex tracks every retracted paper. The conventional retraction story is 'China and India dominate' because in raw count China retracted 27,478 papers in 2020-2025 and India retracted 9,420. But raw count is misleading — China and India also publish the most papers in the world, so high counts are partly just a function of scale. I downloaded both the retracted-paper count and the total-paper count by country, divided one by the other, and asked: which country has the highest retraction RATE per 1,000 papers? The answer flips the headline. The world's highest country-level retraction rate is Saudi Arabia at 7.96 per 1,000 papers — almost 1% of all Saudi-authored papers in the last six years have been retracted. China is 8th at 4.17. India is 11th at 3.96. The two countries the conventional narrative ignores entirely are Ethiopia (6.10, 2nd) and Uzbekistan (5.55, 3rd) — both of which have research integrity stories that almost never make it into mainland integrity coverage. Iraq, Pakistan, Macao, and Egypt round out the top 7. Saudi Arabia's rate is more than 5 times the world average of 1.45 per 1,000, and 18 times the cleanest large-volume country (Indonesia, at 0.17). Why this matters: international research funders (Wellcome Trust, Gates Foundation, EU Horizon Europe) decide which countries to target with research-integrity training programs, and journal editors decide how to weight peer review for submissions from specific countries. The standard conversation focuses on China and India because those countries publish the most. But the per-paper rate ranking shows that Saudi Arabia, Ethiopia, and Uzbekistan are higher-leverage targets per dollar of intervention. There's an important caveat in the other direction: Indonesia's 0.17 rate (with 2.3 million published papers) does NOT mean Indonesian research is the cleanest in the world. It probably means Indonesian researchers publish more heavily in domestic and predatory journals that don't issue formal retractions, even when papers are problematic. The country-level retraction rate measures journal-policy effects as much as research quality. So the right framing is: high-rate countries (Saudi, Ethiopia, Uzbekistan) ARE producing problematic papers that get caught; low-rate countries with high publication volume (Indonesia, Brazil) may be producing problematic papers that don't get caught because their venues don't retract.

Saudi Arabia 7.96 (3,342/419,833) · Ethiopia 6.10 (636/104,206) · Uzbekistan 5.55 (475/85,527) · Iraq 5.53 (1,083/195,753) · Pakistan 5.47 (2,071/378,335) · Macao 5.14 (245/47,708) · Egypt 4.29 (1,576/367,778) · China 4.17 (27,478/6,590,178) · Oman 4.07 (132/32,427) · Jordan 4.02 (314/78,163) · India 3.96 (9,420/2,379,970) · Yemen 3.94 · Kyrgyzstan 3.74 · Bahrain 3.54 · Nepal 3.45

Panama 0.04 · Paraguay 0.07 · Costa Rica 0.12 · Cuba 0.12 · Bolivia 0.16 · Indonesia 0.17 (with 2.3M papers — likely a journal-policy artifact, not research-quality signal) · Brazil 0.20 · Luxembourg 0.20 · Austria 0.21 · Slovenia 0.22 · Cote d'Ivoire 0.24 · Venezuela 0.25 · Moldova 0.25 · Argentina 0.26 · Croatia 0.26

49.1 million papers across 127 countries with ≥10K papers · 71,097 retracted papers · system rate 1.45 per 1,000 · top rate (Saudi Arabia 7.96) is 5.5x system mean · top raw count (China 27,478) is 18.6x the second-largest raw count (India 9,420) but only 1.05x India's per-paper rate