It’s rare, but it happens. A very small fraction (about 4 in 10,000) of peer-reviewed manuscripts are retracted – papers that are withdrawn from their original publication. Some are retracted because of honest errors – an error in a modeling equation, a mistake in patient data entry – and some are not, for instance intentional manipulation of data.
When a manuscript is retracted, the publisher removes the paper from the website (presumably there are still print copies in existence, if the journal offers a print format). Until recently, there was no systematic way to find retracted papers or comb the data on retractions.
Now, there is: the Retraction Watch Database, which contains information on over 18,000 retracted manuscripts, including the reasons for retraction.
The folks at Retraction Watch teamed up with colleagues at Science to analyze the retraction data and they found some interesting trends:
- Relatively few authors (about 500) are responsible for a disproportionate number of retractions.
- The majority of retractions have involved scientific fraud or other kinds of misconduct.
- The rate of retraction due to plagiarism looks to be stabilizing and possibly declining over the last 7 years.
- Retraction due to fake peer-review has increased steadily, and is the reason for about 20% of all retractions (as of 2015 data).
Before you submit your next manuscript for peer-review, double check the Retraction Watch database to ensure you’re not unknowingly citing a retracted paper.