How Long Does a Plagiarism Retraction Take?

Plagiarism, along with fabrication, data manipulation and other misdeeds, is considered to be one of the biggest sins an academic researcher can commit.

As such, it makes sense that journals, editors and others overseeing scholarly publishing take plagiarism very seriously. They have protocols in place for receiving plagiarism complaints and investigating allegations and routinely retract plagiarized works.

But how long does it take to retract a plagiarized paper? How long does it remain published before the retraction is issued?

A recent study published in the journal Accountability in Research aims to answer those questions. The study’s authors, Yulia Sevryugina and Yifan Li at the University of Michigan, examined over 10,000 plagiarism-related retractions over a 20-year period.

They found that, while the process took multiple years on average, certain types of plagiarism remained unretracted for nearly twice as long.

All in all, it’s a fascinating examination of how long plagiarized works remain in journals before eventually being pulled.

Overview of the Study

The two authors used the Retraction Watch database for the study. They examined the period between 2001 and 2022 to find all retracted articles during that time.

They found that 5,924 articles were retracted due to self-plagiarism. Another 4,579 were retracted due to plagiarism of outside works. The authors then subdivided those retractions into article types, the field the article is in, the country it is from and the duplicated elements (article, data, images, etc.).

According to their findings, journals retracted plagiarized articles an average of 1.7 years after publication. While this is long, it’s significantly shorter than the 3.2-year average for self-plagiarized articles.

To make matters worse, self-plagiarized studies that featured duplicate images had a median retraction time of 4.0 years. This is worrying because, according to the authors, image duplication is a significant issue in biomedical sciences.

Still, journals retracted self-plagiarized articles more slowly across all fields and types.

The study also examined the number of authors on retracted papers. It found that journals were more likely to retract studies with multiple coauthors due to self-plagiarism than plagiarism.

Finally, the authors also examined the papers’ country of origin. They found Russia had the highest number of retractions per 1,000 papers (0.675), and Iran was second (0.475). For comparison, the United States had a rate of 0.04.

None of this should surprise anyone who has been following academic integrity. In fact, the authors highlight how similar their findings are to past research. To that end, the study’s normalcy may be its most important feature.

What the Study Means

What’s most interesting about the study is the time to retraction. Though the study doesn’t examine how long the journal took to decide, it does track the time between publication and retraction.

To that end, 1.7 years may seem like a long time. However, most of these cases were not detected immediately. Since these articles passed the peer review process, it takes time for someone to spot the plagiarism and report it. It then takes time to investigate and retract the paper.

So, neither the 1.7-year nor the 3.2-year statistics are particularly noteworthy. What is notable is that the 1.7-year number is for traditional plagiarism, and the 3.2-year one is for self-plagiarism.

Several factors likely contribute to this. First, self-plagiarism, sometimes referred to as duplicative publication, is not as clearly defined as regular plagiarism. There is a massive debate about what authors can reuse and under what terms.

While most agree that duplicative publication, salami slicing and self-plagiarism are all academic integrity violations, defining them has been elusive. This is partly because researchers often repeat experiments with slightly different variables. Editors and readers expect there to be some similarities.

It also makes sense that studies with more authors would have greater issues with self-plagiarism. This not only fits the narrative of guest/gift authorship, but more authors means more people to detect plagiarism issues.

During the recent spate of DEI-based plagiarism allegations, Sherri Charleston, LaVar Charleston and Jerlando Jackson faced allegations of this exact behavior.

Still, this study doesn’t doesn’t have any major surprises. However, there is a big warning.

The Big Warning

As the authors point out, the idea of self-plagiarism is in an unusual space. It is poorly defined, often minimized and difficult to nail down terminology for.

This means self-plagiarism is difficult to spot, report, investigate and retract. As such, any duplicative publication that gets past the peer-review stage will remain undetected for a very long time.

The issue worsens when you tack on image duplication (both self-plagiarism and regular plagiarism). Since image duplication is more difficult to detect than text duplication, it will remain in the academic sphere longer.

However, this only addresses the cases that are eventually caught. The extra time to retract these articles strongly indicates that we only detect a smaller percentage of them. This means that, most likely, there are far more of these cases that we don’t know about.

As such, this sends a simple but clear message: We need to do a better job of defining and detecting these issues. The current system is not working; these are clear weaknesses that need addressing.

While journals need to improve in all areas of research integrity, this study highlights areas that require special attention.

Bottom Line

In the end, there aren’t many surprises in this study. Plagiarized works that have passed peer review tend to stay unretracted for years, and types of plagiarism that are more difficult to detect take longer.

Still, the numbers should give academics pause. Most published works receive most of their attention shortly after publication. While retracting a paper 1-4 years after publication is a necessary step, it likely has received the majority of the attention it will by that point.

So, while the retraction process is essential, the focus must be prevention. Relying on the retraction process to catch bad actors after the fact is not working. Even the best numbers are still too long.

To be clear, there’s no easy answer here. But I think the authors have it largely correct. First, we must work on universal and clear definitions of self-plagiarism and duplicative publication. Once we have those standards, we can build better tools for detecting it.

That will help journals avoid publishing such articles in the first place.