Open dataset highlights researchers’ perspectives on managing research data

As part of the KAFFE project (kartläggning av förmågor inom forskningens e-infrastruktursystem – Mapping Capabilities in the Research e-Infrastructure System), researchers and doctoral students were interviewed to gain a better understanding of their practices and needs when working with research data. SND is now making openly available a dataset containing the 40 qualitative interviews that members of the SND Network helped to collect. Madeleine Dutoit, Training Coordinator at SND and one of the project’s team leaders, explains why it is so valuable to bring researchers’ voices to the fore. 

How did you go about conducting the interviews?  
“We wanted to learn more about what it is like for researchers to work with research data today – what works and what does not. If we do not talk to researchers, we can't develop services and support that work well for them.

“At an SND Network meeting, everyone in the network was invited to conduct interviews for the project at their own institution, and several people volunteered. A methodology had already been developed, with prepared interview questions and practical instructions for conducting the interviews.

“It was important for the interviews to be conducted consistently, so we held regular meetings with the group to ensure that the data collected were of good quality and to provide support. Not everyone had previous interviewing experience.

“To reach as broad a group of researchers as possible, the interviewers were encouraged to recruit participants from a range of disciplines and at different career stages. In total, 40 doctoral students and researchers at seven higher education institutions were interviewed,” says Madeleine Dutoit. 

Why are you making the interviews available on Researchdata.se? 
“It would have been difficult to justify SND, an organization that promotes good research data management, not making the interviews available! Forty qualitative interviews constitute a substantial body of material that would take an individual researcher a long time to collect.

“Our interview data are now completely open and only a few clicks away for anyone interested in researchers’ own experiences and perspectives on data management. For someone who wants to explore the subject in greater depth, having the data already available can save a great deal of time. But the data may also be of interest to anyone curious to gain an insight into individual researchers’ experiences – for example, people working with research data support at higher education institutions.” 

What is involved in making interview data available? What challenges have you encountered, and how have you dealt with them? 
“We needed to correct all the errors introduced by the AI service used to transcribe the audio recordings into text – for example, phase becoming face, the Swedish word behjälpliga [helpful] becoming begärda [requested], and happy becoming unhappy.

“But the greatest challenge was pseudonymizing the data. It was difficult to determine what information might lead to the participants being identified. Names and institutions mentioned during the interviews were obvious examples. But it was much harder when it came to information such as details of unique doctoral or research projects, or uncommon systems.

“Information like that can act as an indirect fingerprint, so the group had to discuss the appropriate approach on a case-by-case basis. There are no ready-made answers. We made countless micro-decisions about individual words and details, weighing the risk that participants might be identified against the risk of ultimately producing watered-down texts that would be of little use to anyone.

“We also wanted the written transcripts to reproduce the researchers’ recorded accounts in as much detail as possible, capturing emphases and non-verbal expressions such as laughter, irony, and hesitation. We didn’t want to lose these expressions, because they can convey a great deal. And since the audio recordings ultimately had to be deleted, this was our last opportunity to preserve them.

“Taken together, all of this made the work very time-consuming – we are talking about 650 pages of text.” 

Was there anything in the interview findings that stood out, or that you found particularly useful or interesting? 
“If I had to choose one thing, I find it interesting that many researchers still experience difficulties sharing data with others in their day-to-day work.

“It's perhaps not surprising that sharing data with researchers in other countries can be problematic when they are working under different legislation or using different systems. But the researchers also describe difficulties sharing data within Sweden, between higher education institutions and even within individual institutions.

“Researchers are resourceful and find solutions – although these may not always be particularly secure – and sometimes even meet in person to transfer data. This is happening at a time when all forms of research collaboration are being encouraged. Problems with data sharing take time away from research, so a coordinated effort is needed in this area.”

What will you take away from the project, and what happens next? 
“The KAFFE project has been rewarding in many ways. It has involved close collaboration with the SND Network and given us many new experiences to reflect on.

“SND is now supporting other higher education institutions and organizations that are conducting interviews using the KAFFE project methodology, including Chalmers and SciLifeLab.

“Anyone who would like to carry out interviews at their own institution can find all the methodological material they need in DORIS alongside the data, and we are here to answer any questions. We will also develop an updated version of the methodology based on what we have learnt from the project, which will eventually be published in SND’s community on Zenodo.”