Should you share your data?
The opportunities and challenges of data sharing
There are many benefits to sharing data as a researcher, such as getting credit for your valuable data findings, making research findings more reproducible and replicable, and supporting the preservation of data long-term. There are also certain challenges – such as how to share data when they’re confidential or sensitive, or what if my data set is scooped?
The Journal of Social Psychology, rewards data sharing with open science badges, which are given to authors contributing to scientific transparency and their efforts to make their research more open. We spoke to the journal editor, Jon Grahe to hear his views on why sharing data is worth rewarding, some of the challenges involved, and how to overcome these as an author.
From Jon Grahe
Scientific transparency helps encourage scientific reproducibility.
I trust authors to be honest researchers, but we also know that individuals are biased towards finding evidence that supports their assertions. Decisions we make about exclusion criteria for data, scaling variables, even model parameters can have major impacts on the conclusions from our findings. By sharing the data, we invite readers to challenge the authors’ assertions directly. This has a benefit of pressing authors to make sure their data and analyses are correct. More than once, an author earning an open data badge has contacted me and alerted me to a data error of some sort as they prepared to make their research open. If not for the data sharing process, these errors would never be found or would have been found post publication.
More public data also means more secondary research opportunities. I don’t trust any single study as making strong claims of evidence. It is only after many studies conducted on the same question can be combined and evaluated in a meta-analysis that research becomes convincing. To make these meta-analyses studies more effective, we need to make it easier to conduct them by sharing data.
I think most reservations about sharing data and materials reflect inertia and lack of understanding rather than actual conflict with the idea of openness.
Our attitudes about scientific methods and sharing are driven not only by our own beliefs, but also our experiences during learning how to conduct research. Until recently, technology made data and materials sharing challenging and we developed the present status quo model in that technological era. The recent push for increased scientific transparency is partly driven by concerns about methodological practices, but also driven by changes in technology that make it easier to do better science. However, there are some real challenges that authors must address before they share their data.
Some authors express inability to share their data because they don’t have their Institutional Review Board’s approval to do so.
Institutional Review Boards (IRB) that regulate research ethics are not regulated nationally, they are local entities. Some argue that IRBs don’t control anonymized data outcomes, but authors must live in their local environments and this challenge should be addressed before sharing data. My solution was to add a line in my informed consent forms that stated the data would be shared publicly. That way, there are no IRB concerns when I share data and in my own research there is a clear demarcation between the present where I share data publicly and the past where I would share data upon request.
Some data contain personal, private, sensitive information that might identify or otherwise put participants at risk if the data were published.
This is not easily resolved with a new consent form because the participant’s risks should be our primary concern as researchers, and sometimes participants might not recognize those risks. One solution is to house the data with a respected 3rd party repository who will share the data following a request from a qualified researcher. This way, the author does not control access, but the requester must meet certain standards and acknowledge that they are accepting responsibility for the welfare of the participants.
Another concern is that someone will “scoop” an author’s future paper.
The fear is that a reader will take the data set and publish another paper before the author can write and submit the manuscript. This is often addressed by reminding authors that we request only they publish the data included in the manuscript. If there is a larger data set, they can withhold the rest of the data as long as they fully describe the data set. It is also good to remember that if someone uses the data, they need to cite that prior work. Citations are the metric of impact in science; more citations promote the research, even if some of those citations are contrary to the author’s position.
In general, authors should consider data sharing as an opportunity to connect a reader of that single study to the larger research agenda.
If data are published on a project that also directs readers to a main page where other study data sets are kept, the research can have even greater impact.
Want to find out more about data sharing at Taylor & Francis? Read our guidelines.
See also: Using Open Science badges: how a journal is recognizing and rewarding open science on Taylor & Francis Editor Resources.
Jon Grahe is Professor of Psychology at Pacific Lutheran University. He received his B.A. in Psychology from Shippensburg and his M.A. and Ph.D. from University of Toledo. He is the Past-President of Psi Chi and the Managing Executive Editor of The Journal of Social Psychology. He is the principal investigator for the Collaborative Replications and Education Project (CREP) and the Emerging Adulthood Measured at Multiple Institutions (EAMMi2) project and recently coauthored a book, “Designing and Teaching Undergraduate Capstone Courses.” He serves as a Center for Open Science Ambassador and is chair of the Open Science Badges committee.