Error Culture in Science

At the NFDI4Chem Consortium Meeting 3.0, October in Hannover a panel discussion was held on cultural change and error culture.

This addresses an important fact: researchers are afraid of publishing their data „because others may discover mistakes in their data or research which in turn could detrimentally impact their career.“ This underlines the absence of a systematic approach to dealing with errors in academia. High Risk Organisations have already implemented an error culture, so they can easily find and fix problems, as Joachim Richert (BASF) said: „The industry is working with errors. This is part of our quality management.“

The goal of the panel discussion was to create in a microcosm the kind of discussions we want to see being held more openly within the wider academic community. Therefore, it was essential that important stakeholder groups were represented. The industrial point of view was represented by Joachim Richert (BASF), academic researchers by Sonja Herres-Pawlis (RTWH-Aachen University), the NFDI by Cord Wiljes, NFDI4Chem by John Jolliffe who also moderated the discussion, IUPAC by Leah McEwen and publishers by Laura Woodward-Heni (Wiley-VCH). We felt it was also very important to have a perspective from a psychosocial point of view and therefore we were very fortunate to have psychologist Maximilian Frank (LMU) join the discussion.

The opening main question to the participating scientists was, what kind of change in approach to dealing with errors and scientific practices would they like to see?

A first consideration from Joachim Richert was to motivate scientists to use open data despite the lack of an error culture: “We should look at an incentive such as a citation index, for instance the reuse of your data.” The discussion brought up several possible directions. One was how to reduce errors in the first place. In terms of data, this could for example be achieved by automated quality control of data. 

Importantly, during the discussion it was asked why poor error cultures exist in the first place in academia. The psychologist of the panel discussion, Maximilian Frank (LMU), said: „This also is a question of good leadership.“ In industry, leadership qualities are selected for and further nurtured, whereas in academia little to no leadership training takes place. Furthermore, a culture of publish or perish has led to good research practice not always being adhered to and that the pressures of publish or perish and poor error cultures within academic working groups and academia as a whole are clearly linked.

Following the opening questions it was asked: ”How as a community can we better deal with errors. E.g. that retractions are no longer feared as potential career-enders by early career researchers?” Laura Woodward-Heni (Wiley-VCH) provided some very valuable insights: “The purpose of a retraction is not to punish anybody but to correct the scientific record”. She noted that it was important to distinguish between retractions where the intentions of the authors were honest and ethical and retractions due to deliberate ethical misconduct. To remove the fear about retractions a more open discussion within the community about retractions is needed. The majority of retractions are really just due to honest errors. If more academics shared their experiences with retractions and talked about these openly, other researchers would be able to see these are not necessarily career enders.

During the course of the panel discussion various points were brought up that are worth highlighting here.

Cord Wiljes (NFDI-GS) pointed out that in the end, open data is not only a threat, but delivers a chance as well: “Open data helps to find errors. And there are different levels of errors. And there are errors that help to advance science.” 

All panellists agreed on Good Research Practice (GRP) having to be more prevalent in undergraduate curricula and guidelines for postgraduates. If GRP was seen as an unquestionable code of honour, fewer questionable research practices would take place that might be holding people back from publishing their data (as these would then become apparent). 

Leah McEwen (IUPAC) said that the positive impact of data needs to be better demonstrated and remove the artificial separation that largely exists between data producers and -reusers. 

The panel discussion was closed with the following thoughts. These are all complex issues that will not be fixed overnight. Especially the pressures that arise from the publish or perish paradigm. It is incredibly important that awareness is raised for this topic to get people talking about it so that the community more openly acknowledges that we have a problem. Therefore, the NFDI will continuously raise awareness for these issues but we also invite each and everyone of you to discuss the topic more openly with your peers. Furthermore, an interdisciplinary working group within the NFDI on error cultures has been created as these are not issues just limited to chemistry. As well as raising awareness and reaching out to the community, the working group aims to develop guidelines for how academic research groups might establish better error cultures.