Today, almost most news outlets such as the New York Times, the Guardian, Science News and the Scientific American report the results of a recent publication in the Proceedings of the National Academy of Science. The study explored the various reasons for retraction of scientific articles from publication since 1975. The most common reason (67%) was scientific misconduct in the form of fraud/suspected fraud (43.4%), plagiarism (9.8%) and double publication (14.2%). Retraction due to honest errors of researchers amounted only to 21.3% of the total.
According to the US National Science Foundation, there are 3 types of scientific misconduct; fabrication, falsification and plagiarism. In addition, authorship irregularities and violation of human and animal research ethical guidelines are considered a form of research misconduct. Recently, with the excessive expansion of scientific publications, another of misconduct was named, duplication of publication (Wikipedia).
The authors identified around 2000 retracted articles in the Pubmed database with the first reported retraction taking place in 1977. They uncovered a tenfold increase in retractions due to fraud/suspected fraud since 1975, which cannot be attributed entirely to the rise of scientific publications. Retractions due to plagiarism and duplicate publication are but recent findings and have existed since 2005 only.
The reasons for retraction differed according to the geographical location of research authors. For Americans, Japanese and German authors, the most common cause of retraction was fraud/suspected fraud whereas plagiarism and duplicate publication were the more common among Indian and Chinese authors.
The impact factor of the publication was also correlated to the cause for retraction. Journals of high impact factor commonly retracted studies with errors and fraud while low impact ones mostly withdrew plagiarized articles and double publication.
To complicate things further, not all studies suspected of fraud have been withdrawn from literature. and some of the retracted studies continue to be cited by other authors for years after withdrawal. on average, it takes up to 47 months to remove a fraudulent study and 26 months to retract an erroneous study. In addition, the issued retraction notices are more commonly than not ambiguous and non specific. The authors report 119 cases where no reasons were stated for retraction. This could be attributed to the fear of stigma associated with possible scientific misconduct and the need of publishers to dig deep to identify the cause of retraction which may need considerable time and effort to prove.
There are proposed solutions to address the problem which include but not limited to; better software to spot plagiarism and duplicate publication, clearly defined terms of scientific misconduct which should be made universally available for researchers, intensive training of researchers on statistics, research methods and ethics, encouraging the reporting of misconduct or errors while diminishing the stigma associated with them and lastly but not the least finding alternative means of reward systems for science.