AIMOS 2026 · Wellington, New Zealand · 5-minute lightning talk The Association for Interdisciplinary Meta-Research and Open Science, annual conference, 30 November to 2 December 2026
at least 44,046 times in 2024 alone
the data, citable and permanent doi:10.7910/DVN/WT8VOA Harvard Dataverse · CC BY 4.0 · 12 filesRetraction is supposed to stop a paper's influence. This counts how often it doesn't.
Every retraction in the Retraction Watch database, joined against the OpenCitations Index. The scripts run on public inputs, so you don't have to take my word for any of it.
the measurement
Counted as citing works: one citing work against one retracted paper, once. The OpenCitations dump writes a row per identifier, and 119 retracted papers sit in it under two or three identifiers, so its raw rows run higher: 999,842 in all and 330,834 after retraction. Both counts come out of the same deposited file.
a name in the data
The hit show Dr. Death put Paolo Macchiarini in front of a lot of people who do not read The Lancet. His papers are in this dataset.
| who | cited after | in 2024 | retracted | cited before |
|---|---|---|---|---|
| Andrew Wakefield1 paper, The Lancet 1998 | 1,519 | 81 | 2010 | 884 |
| Paolo Macchiarini13 papers in the Retraction Watch set | 281 | 35 | 2012–2024 | 2,430 |
↔ more columns off to the right
His 13 retracted papers have been cited by 281 works since the dates they were retracted, 35 of those in 2024 and 32 in 2025. The median gap is 619 days. Wakefield is in the table for scale, because that is the case this audience already has a number for.
Both show why the counting unit matters. The index holds Wakefield's paper under two identifiers and writes most of its citations twice, so a raw count of the dump's rows gives 2,941 after retraction. Macchiarini's papers give 522.
macchiarini, three most-cited since retraction
| paper | cited after | retracted | published | cited before |
|---|---|---|---|---|
| Tracheobronchial transplantation with a stem-cell-seeded bioartificial nanocomposite: a proof-of-concept study10.1016/s0140-6736(11)61715-7 | 76 | 2018 | 2011 | 344 |
| Engineered whole organs and complex tissues10.1016/s0140-6736(12)60073-7 | 67 | 2018 | 2012 | 316 |
| Clinical transplantation of a tissue-engineered airway10.1016/s0140-6736(08)61598-6 | 41 | 2023 | 2008 | 1,315 |
↔ more columns off to the right
The 2008 paper was cited by 1,315 works before anyone retracted it, and that took until 2023.
figures
On a narrow screen a chart pans sideways. The caption carries the same numbers.
↔ drag the chart sideways, or just read the caption
↔ drag the chart sideways, or just read the caption
method
the part that matters
What this count can't see.
RetractionNature has four values. The 8 September export is
66,764 Retraction, 3,713 Expression of concern, 1,509 Correction and
160 Reinstatement. Only the Retraction rows enter the set behind this page.
2,471 DOIs on the list never had a Retraction row. Left in, they'd take the post-retraction count from 323,756 to 439,489, up 35.7%. The biggest single paper is Corman et al. 2020 in Eurosurveillance, the COVID PCR protocol. It was corrected and never retracted, and an unfiltered list counts 6,379 of its citations.
Retraction Watch defines Reinstatement as a replacement published under the original DOI, so that the replacement doesn't get tagged as retracted in other tools. Keep those rows and your list calls the replacement retracted.
what a count depends on
The same count moves with five choices made while counting. Ranked by how far each one moved it on this data:
| choice | what it did here |
|---|---|
| Where the line between before and after falls | 18.7% to 45.0% of dated citations count as after, depending on the rule |
| Which Retraction Watch records count as retracted | +35.7% if the 2,471 DOIs that were never retracted stay in |
| Which release of the index | +14.2% in one release step, and 2025 alone +51.6% |
| What counts as one citation | 2.1% fewer counted as citing works than as index rows. Wakefield: 1,519 against 2,941 |
| When the retraction list was pulled | +0.13% over 20 days |
where the line falls
| rule | after | share | 2024 | index rows |
|---|---|---|---|---|
| citing date after the retraction date (this study) | 323,756 | 33.4% | 50,629 | 330,834 |
| citing year after the retraction year | 272,082 | 28.1% | 42,367 | 278,216 |
| citing year the same as the retraction year or later | 436,223 | 45.0% | 66,458 | 444,768 |
| citing year more than one year after the retraction year | 180,885 | 18.7% | 20,479 | 185,655 |
| citing date more than 365 days after the retraction date | 207,410 | 21.4% | 25,185 | 212,603 |
↔ more columns off to the right
Share is of the citations that carry a usable date. The title was written from the first run, on the March 2026 index, where 2024 came to 44,046 index rows. On the July release it's 50,629 citing works, so "at least" holds under the rule this study uses.
one release later, same retracted papers
| citing year | March 2026 | July 2026 | change |
|---|---|---|---|
| every citing year to 2022 | 180,190 | 186,642 | +3.6% |
| 2023 | 29,408 | 34,108 | +16.0% |
| 2024 | 43,867 | 50,440 | +15.0% |
| 2025 | 28,995 | 43,944 | +51.6% |
| all post-retraction | 282,764 | 322,962 | +14.2% |
OpenCitations posted its July 2026 release the morning the talk was accepted, so the join was re-run on it. The 54,534 retracted papers present in both runs are held fixed here. Every year went up, because indexing fills in backwards. The release comparison and the unfiltered-list figures come from two further scans that aren't in the deposit yet.
reuse
The deposit is the citable thing and it will outlast this page. Cite that, not this URL.
CC BY 4.0. Reuse it, including commercially, with credit.
sources
Nothing here is my data. It is a join of other people's, and this is who they are.
get in contact
Email reaches me directly. The rest are profiles, and nathanthornhill.com carries the full publication list.
email research@nathanthornhill.com Questions about the method or the data.