When Swedish National Data Service (SND) held its network meeting in October, participants gained insight into how Stockholm University (SU) are working towards Open Science. We met with Margareta Ödmark and Merlijn de Smit from SU’s Research Data Management Team, as well as Maria Almbro and John Fitzpatrick, who shared their experiences and perspectives on making data accessible from a researcher’s point of view.
Open Science at SU
The SND network meeting began with Margareta Ödmark, member of SU’s Research Data Management Team, describing SU’s journey toward Open Science. She outlined the transition to what the team calls “the new normal” in four steps: a policy to make sharing research data normative and rewarding, a plan with goals and responsibilities, an internal organization and infrastructure, and support and training in how to do Open Science right.
“Drafting a policy and letting it be the end goal would make it just a paper product. A policy must be implemented in reality, and that’s when the real work begins. It must be applied across the entire university, and we need to talk to researchers about the challenges they face. If we work in isolation, it will never have a ripple effect. We need to have this continuous dialogue to collaborate on the researchers’ needs,” explains Margareta Ödmark.
As early as 2015, SU began working on Open Science with an informal initiative, and in 2018, the Research Data Management Team was set up, coordinated by Merlijn de Smit, Data Steward at SU. SU’s strategy states that the university aims to be a leader and at the forefront in the transition to Open Science.
“The larger universities must take on greater responsibility, because we can’t expect smaller institutions to manage these complex issues,” says Margareta Ödmark.
“I agree, but SU is quite a broad university. We can see that some smaller institutions, which focus on specific subjects, come up with excellent solutions. At the same time, we need to find solutions that work for both the natural sciences and the humanities. So, we have some challenges,” adds Merlijn de Smit.
Mapping the current situation
To gain a clearer understanding of how the transition to Open Science is progressing, an investigation has been initiated, which Maria Almbro, an Analyst at SU, presented during the network meeting.
“We can see what is being submitted to our own repositories, but that doesn’t provide a very comprehensive picture as researchers are free to share data wherever they wish. We wanted to gain a better understanding of where researchers are making their data accessible and how they are doing it,” explains Maria Almbro.
In 2022, researchers at SU published approximately 3,700 articles. The investigators reviewed a 10% sample of these. After excluding articles that were either not accessible or deemed not to be based on data, 301 articles remained. The results of the investigation showed that a total of 39% of these articles included at least one reference to supporting data published with a persistent identifier, data made accessible in appendices, or with links to another data source.
“The positive takeaway is that we believe many researchers genuinely want to be transparent or share their data materials, and good efforts are being made. However, they might not be aware that they could do it in a somewhat better way by using a repository to make their materials more visible and persistent over time,” says Maria Almbro.
“It’s also likely many that there are many who don’t think this is relevant, who find it too difficult, or who come from disciplines where it’s not standard practice. So for us, this investigation was very enlightening,” Maria Almbro adds.
Positive effects from data sharing
Someone who knows firsthand the benefits of sharing research data is John Fitzpatrick, Professor of Ethology at the Department of Zoology at SU. When publishing an article in 2014, he made the supporting data of the study accessible to others. While the article was on the journal’s website but not yet published, John Fitzpatrick was contacted by a researcher in the same field who pointed out that they couldn’t replicate the results. Fitzpatrick discovered a calculation error and, thanks to this, was able to add a correction before the journal was published.
“This was the first article where I made all the raw data and code accessible, and someone on the other side of the globe could identify the error so quickly. For me, this is one of the great examples of how well this can work,” says John Fitzpatrick.
Despite the advantages of data sharing, Fitzpatrick sees a significant downside:
“There’s no reward for this. I’ve yet to see anyone get hired because they have better open access to their research. We haven’t reached a point where there are clear, tangible benefits beyond the overarching goal of making research data accessible to the public. Yet so many people do it, and they do it really well, which is phenomenally cool!
Challenges in encouraging more researchers to share research data
When asked how to get more researchers to share data, all four experts agree: information and training, support and assistance, a straightforward process, and a functioning infrastructure are essential. However, they differ in their views on whether sharing research data should be rewarded with merits and how incentives should be implemented:
“In many ways, it’s already rewarding, as evidence suggests that articles with published datasets are cited more often. But the challenge lies in making this clear to researchers since publishing data involves extra work,” says Merlijn de Smit.
“For early-career researchers, it can be a roadblock to spend significant time in working transparently and making materials accessible. It simply takes a lot of time, which has to come from somewhere. If that means you can’t publish another article because you’ve done everything right, you risk falling behind. Grant providers should consider this when allocating funding,” adds Maria Almbro.
John Fitzpatrick, however, disagrees: “I’m quite sceptical of a merit-based system for the simple reason that there are already so many incentive systems like this. I’d worry that if we created metrics for open science practices, we’d spend a lot of energy developing something that would never be used.”