This document analyzes the relationship between comments on PubPeer and the speed of article retraction. It finds that articles without PubPeer comments had an average time to retraction of 655 days, while those with comments had a longer average of 939 days. However, when measuring from the date of the first PubPeer comment, the average time to retraction was much shorter at 278 days. This suggests that public commentary on platforms like PubPeer may help improve retraction processes and speed up corrections to the literature.

1. The relationship between PubPeer
comments and the speed of retraction
Alison Abritis1,2, Brandon Stell3,4, Boris Barbour3,4, Ivan Oransky1,5,6
1Retraction Watch/The Center for Scientific Integrity, USA
2College of Public Health, University of South Florida, USA
3PubPeer Foundation, USA
4Centre National de la Recherche Scientifique (CNRS), France
5Arthur Carter Journalism Institute, New York University, USA
6Simons Foundation, USA

2. Disclosures
• I am the volunteer co-founder of Retraction Watch, and volunteer
executive director of The Center For Scientific Integrity, our parent
501(c)3 nonprofit organization.
• I am on the board of directors of the PubPeer Foundation, a 501(c)3
nonprofit organization.
• I am an employee of the Simons Foundation, where I am editor in
chief of Spectrum.
• I am an employee of New York University’s Arthur Carter Journalism
Institute, where I serve as Distinguished Writer in Residence.

3. Why study time to retraction (TTR)?
• Provides measure of the efficiency of correction of the literature
• Long TTR allows for continued spread of poor or misleading information
• Long TTR allows for continued citation of problematic research, and waste of resources
What does previous research show?
• A 2021 study by Guadino et al. gives median TTR of 657 days (1.8 years)2
• A 2012 study by Steen et al shows an average TTR of 987 days (32.91 months)1
• A 2021 study by Serghio et al. gives a median TTR of 457 days (1.3 years)3

4. “due to concerns with an
undeclared splice identified in
Figure 5(a), p-MYPT1, as
originally raised on PubPeer
[2].”
PubPeer: Clearly part of the impetus for literature cleanup
“As originally reported in
https://pubpeer.com/publicatio
ns/F4F79D78FA1249D847FBB6E
A4B3381, Fig. 4E shows evidence
of data manipulation,…”

5. Retraction Watch Database4: a database of retractions going
back to 1756 and launched in 2018, created and maintained by
Retraction Watch, a blog started in August 2010 addressing
retractions and other issues in publishing and misconduct. Both
are projects of The Center for Scientific Integrity, a U.S.-based
501(c)3 nonprofit organization.
PubPeer5: a 501(c)3 nonprofit organization founded in October
2012; provides an online forum for readers and authors to discuss
published articles.
Sources of Information

6. Methods:
Time Frame: Original date of article publication from January 1, 2015 through December 31, 2020, and
entered into RWD by October 11, 2021
Original Data Source: Retraction Watch Database (RWD)
Article Filters: Article Types (RWD Category) restricted to Research Articles, Clinical Studies, Case
Reports, Review Articles (excluding book reviews)
Subtotal Articles in Sample = 8,840
Eliminated Articles: inability to ascertain date of
retraction: n = 466
Final Sample Article Total – 8,374

7. Methods
PubPeer (PP) entries matched via DOIs: 1,923
Final Sample of PubPeer Entries for comparison: 1,174
PP entries made on or after article retraction: 749
PP entries made with critical comments: 1,923

8. Articles without PubPeer
comments; retraction
from date of publication
(n = 7200)
Articles with PubPeer
comments; retraction
from date of
publication
(n=1174)
Articles with PubPeer
comments; retraction
from date of first
PubPeer comment
(n = 1174)
Average TTR
(days)
655.09
(sd = 461.53)
939.51
(sd = 446.30)
277.83
(sd = 171.88)
Results

9. Findings:
 The TTR for articles without PP comments (655 days) is in the range of two more
recent studies but shorter than study from 2012, when PubPeer first launched
 The overall TTR for articles with PP comments is on the higher end (939 days) of
previous studies
 The TTR starting at the time of the PubPeer comment (278 days) is much shorter than
the overall TTR of any previous studies

10. Conclusion:
Limitations
• As with any study, not every retracted article may have been captured in the search
• Unable to determine TTR from non-published allegations to retraction
 Findings suggest that Internet platforms including PubPeer and social media create space for critical
commentary not made previously visible
 PP commenting may have grown due to frustration with a lack of editorial responses, as suggested by
lengthy overall TTR for papers with PP
 Findings suggest that journal and publishers have improved or made better use of their retraction
protocols

11. References
2Guadino M, Robinson NB, Audisio K, Rahouma M, Benedetto U, Kurlansky P, Fremes SE. 2021.
Trends and Characteristics of Retracted Articles in the Biomedical Literature, 1971 to 2020. JAMA
Internal Medicine. 181(8):1118-1121.
1Steen RG, Casadevall A, Fang FC. 2012. Why Has the Number of Scientific Retractions
Increased? PLoS One. 8(7): e68397.
4Retraction Watch Database: http://retractiondatabase.org/
5PubPeer: https://pubpeer.com/
3Serghiou S, Marton RM, Ioannidis JPA. 2021. Media and social media attention to retracted
articles according to Altmetric. PLoS ONE 16(5): e0248625.