Errare Humanum est: Concrete Steps to Establishing a Positive Error Culture

Errors are human and thus inherent in everything including research and the data it generates. While some errors lead to unforeseen scientific progress like the discovery of penicillin by Alexander Flemming, the overall perception of errors in science is a negative one. Yet, the absence of a positive error culture has far-reaching consequences for Research Data Management (RDM): Undiscovered or unpublished errors in research data can not only affect the results of an individual research project, but also have long-term negative consequences for the future reuse of these data. They are thus a challenge to RDM. Authors may fear that the transparency that comes with publishing their data could expose errors, making them vulnerable and potentially jeopardizing their reputation. Even if "honest errors" do not involve deliberate falsification or fraud, a positive error culture is needed so that the fear of errors being discovered does not stand in the way of data publication. The contribution is based on an explorative analysis of error types and the handling and perception of errors in science [1]. To create awareness among researchers and RDM professionals and collect real-life examples of errors from across scientific disciplines the authors designed a series of five workshops. In the paper and in the workshops we focused on existing shortcomings in the scientific quality assurance process also beyond peer review and illustrate how an established error culture in industrial domains can represent a blueprint for a transparent approach to errors in academia. Beyond these examples this contribution presents an approach in NFDI's section EduTrain's work package 8 to develop actionable strategies. These strategies aim to establish a more constructive and transparent way to identify, report, rectify and document errors. These strategies might become a standard and thus help NFDI pioneer a new and promising way to drive forwarded quality assurance and re-use for research data.