Out of sight out of mind: the (in)visibility of corrections in bibliographic databases, repositories, and literature tools

<p xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" class="first" dir="auto" id="d155423529e118">In scholarly communication, there are several ways to communicate that scientific articles are, or may be, problematic: retractions, expressions of concern, and corrections (sometimes also called errata, corrigenda or addenda). If these notices are to alert the reader, all falls or stands with these notices being visible to the reader. While the visibility and informativity of retraction notices has been widely studied – and criticized, much less is known about corrections. This study investigates: How and to what extent are corrected papers visibly labelled as such in bibliographic databases, repositories, literature tools? Using a set of recently corrected papers with a non-trivial correction (as indicated by two or more PubPeer comments) we investigate to what degree corrected papers are visibly labelled as such in various bibliographic databases, repositories, and literature tools. The visibility of corrections varies a lot, from not notifying the reader at all (Web of Science), to having a clearly visible warning label that links to the correction notice for the majority of the articles (PubMed and Scopus). The labels used for corrections vary widely and are used inconsistently. What could be labelled as a correction notice in one database could be editorial material in another. This was further confused by inconsistent labelling of retractions, not only because retraction notices can get corrected and correction notices can be retracted, but also because articles that are corrected according to Crossref turn out to be retracted. These inconsistencies and lack of clarity call for better correction handling procedures.