No study’s perfect: a cross-disciplinary analysis of published errata
Mistakes in research are inevitable, and publishing corrections is vital for the integrity of the literature. These errata rarely require a retraction, and are therefore considered a lesser concern. This perception might be wrong, however, because the actual prevalence, nature and impact of errors across disciplines are unknown. Indeed, while several large studies have looked at retractions, existing studies on errata are small, limited in scope and rather different in methods and aims.
We will conduct the first large quantitative analysis of errata published in all disciplines. These will be retrieved and sampled from the over 11,000 journals listed in the Essential Science Indicators database, which classifies journals in 22 broad disciplines. By combining quantitative and qualitative analyses, we will produce accurate data on the frequency of corrections issued in the various disciplines over the years, the types of errors that are most common, the impact of such corrections and we will identify characteristics of study and journal that are most strongly associated with the publishing of a correction.
These results will help answer important questions on the integrity of the literature and its preservation. They will point out strengths and weaknesses in the current publication system, and will draw attention to areas that might need improvement, hopefully stimulating new approaches to ensuring best editorial and research practices.
The project will be conducted by Dr Daniele Fanelli of the University of Edinburgh, with the help of a research assistant.
Retractions research project
The COPE Code of Conduct states that editors have a responsibility to ensure the reliability of the scientific record, implying that it will sometimes be necessary to retract an article, but it has never attempted to develop detailed guidance about this. Anecdotal evidence suggests that editors may be reluctant to retract articles because of concerns about litigation or uncertainty about the correct procedures. We are therefore examining retractions to understand journals’ current practices and any difficulties faced by editors. We are examining all retractions published on Medline in the last 10 years and categorising them according to the reasons for the retraction, who issued the retraction, etc. We are also doing qualitative research (using a semi-structured interview with journal editors) to learn about their experience of retracting articles, to discover what prompted the action, how editors decided what to do, and any barriers they faced in implementing their plans. We also plan to survey journal editors to gather further information about the retraction process. We plan to use these findings as the basis for developing practical guidelines about when articles should be retracted and how this should be carried out. This project is a collaboration between Liz Wager (freelance publication consultant) and Peter Williams of University College, London.
This study has now been published in the Journal of Medical Ethics (J Med Ethics doi:10.1136/jme.2010.040964). 'Why and how do journals retract articles? An analysis of Medline retractions 1988–2008' was published online first on 12 April 2011.