Reproducibility is the cornerstone of science, and it can be compromised by insufficient data in peer-reviewed publications. Should scientists reveal everything?
Publishing Better Science through Better Data writing competition winner Emma Vander Ende.
One of the foundations of science is its reproducibility. Without it, results are not verifiable and are therefore not believable. But even if a published result is true, there is a chance it might not be reproducible, which introduces a plethora of problems for science.
Irreproducible experiments severely limit the ability of the scientific community to build on results and advance the field. This can happen when scientists don’t share enough data, or details of their experiments in papers, and it happens quite frequently.
So why might a scientist not share their data?
Science is an increasingly competitive and changing field. Fewer and fewer big discoveries are left for scientists to make, and funding agencies conservatively fund projects that are heavily built on previous knowledge, limiting the starting points for new endeavours. Science has consequently become a more subtle art; the average scientist focusses on smaller discoveries that tackle minute, but important, pieces of larger challenges. Cancer cures come in small steps: understandings of cell phenotypes and genotypes, the interaction of specific drugs with certain cell surfaces, the role nanoparticles may play, and so on.
To summarise, science used to mimic broad-brush painting, and modern science is more like pointillism. Every paper, dissertation, or even a principal investigator’s entire career makes up a small portion of a much larger picture, a tiny dot in a fantastically complex painting. So, to an individual scientist, every small step matters a great deal. The reality is that the chance of being “scooped” (someone else publishing first and therefore getting credit) is significant. Losing credit for a finding to a competitor, especially in the face of dwindling funding, may be devastating for a scientist’s career and their lab members’, beyond being personally disappointing.
Unfortunately, once a result is published and credit is assigned to the authors, that’s usually it. Game over. Furthermore, opportunities for scientists to share their negative data (showing that X does not cause Y) are rare and less impactful than publishing an exciting new result showing that A causes B. This is part of what places huge pressure on scientists to publish positive data, and to ensure that they alone can publish the next step in the pursuit of their ultimate goal.
Scientists therefore have to face an ethical and practical dilemma: should they share their results in a complete, data-inclusive way, or should they withhold some information to avoid being scooped?
The right choice is not immediately clear. On the one hand, science should be held to a higher standard than other endeavours, because it is, in theory, not about people but about ideas. The pursuit of knowledge should be a communal human effort. Especially when the knowledge pertains to things that affect all of humanity, like disease prevention or renewable energy technology.
But on the other hand, behind every paper there are human beings, and many papers are the stepping-stones in the bridge to a successful career for at least some of those people.
Should the possibility of an uncredited professional existence be the sacrifice of conducting scientific research that encourages as many contributions from the field as possible? Or should you hide a piece of data critical to reproducing the experiment, or seeing the obvious next step forward, to guarantee yourself the next, more prestigious paper?
Emma Vander Ende is a PhD candidate in chemistry at Northwestern University in Evanston, Illinois, USA, researching the applications of surface-enhanced Raman spectroscopy to detecting salivary human performance biomarkers. Outside of research, Emma heads the Northwestern Women in Science and Engineering Research (WISER) graduate student organization, which offers professional development opportunities geared to women, and studies science writing and communication.
This piece was selected as one of the winning entries for the Publishing Better Science through Better Data writing competition. Publishing Better Science through Better Data is a free, full day conference focussing on how early career researchers can best utilise and manage research data. The conference will run on October 26th at Wellcome Collection Building, London.
Suggested posts
Reproducibility: Seek out stronger science
Turning point: Reproducibility pro
Sharing data: Why it should be done
Transferable skills: What are scientists good at (other than science?)
Report this comment
Excellent article that highlights some important dilemmas in the arena of data sharing. In the project eTRIKS (eTRIKS.org) was put together to help facilitate the integration, sharing, exploration, and preservation of translational research datasets. The initial thinking was to have one big repository where all the data was shared and open. In working with projects we ran into a number of barriers to sharing data including the ones highlighted in this article. Turns out there is a lot of work to be done just to make datasets sharable regardless if you make then open or just want to use them again for a collaboration. This is where the value for the researcher can come back into the picture. Multi-stakeholder collaborations are how you can stil contribute to major advances without betting upon that one major breakthrough.