Data Stewardship – addressing disciplinary data management needs
In 2016 TU Delft embarked on a new project aiming to comprehensively address research data management needs in a disciplinary manner. Rector Magnificus Karel Luyben announced the TU Delft Open Science programme with these words: “The world is facing challenges that our university of technology alone cannot meet.” Good research data management is a necessary prerequisite for effective data sharing and greater openness. Therefore, research data stewardship was recognised at TU Delft as a key component of its Open Science programme.
In research data management there tend to be very few (if any) one-size fit all solutions. At TU Delft data management concerns and recommended workflows will be different for researchers studying social behaviours of people in cities and for researchers collecting and analysing live-time weather data or working on 3D printing projects. In order to be truly relevant to diverse types of research, data management advice needs to be discipline-specific and thus those advising researchers need to have discipline-specific expertise.
Additionally, ensuring any lasting cultural change is not just about technology and expertise but, perhaps more importantly, about communication and trust. Relationships between researchers and those who advise them on data management practice need to be developed over time and by allowing people to get to know each other and to work closely together.
As a result, we came up with the idea of a one year project where discipline-specific Data Stewards will be appointed at each one of the eight TU Delft faculties. Alignment between activities of individual Data Stewards will be ensured by a dedicated Data Stewardship Coordinator leading the project from the Library, which was outlined in their roles and responsibilities.
What are the objectives of the Data Stewardship project?
A long-term objective of the project is to ensure that researchers across all the disciplines supported at TU Delft adhere to good research data management practice in their day-to-day work. This is, of course, a difficult task to achieve and quite unrealistic during the period of a one year project.
We have therefore split the project into several short, medium and long-term objectives, which would allow us to realistically assess the progress of the project and, at the end of the one year period, decide whether it indeed helps improve research data management practice and provides value to the research community at TU Delft. Some of these goals are discussed below.
Short-term goals
There are several short-term goals of the project, which we would like to achieve by the end of 2017. First, there needs to be a mechanism which will create active links between the researchers, Data Stewards and the Library in order to ensure consistent and aligned messages. In addition, we need to develop a framework within which Data Stewards could effectively work together. Some of the challenges which will need to be addressed are actually not so much different from challenges faced by researchers. Data Stewards who will be based at different faculties will have to be able to effectively exchange their files, they need to develop a robust system for version control and data backup, as well establish effective ways of communicating with each other and with the research community.
Second, Data Stewards will join the project with a various degree of expertise of the different aspects of research data management. Therefore, all Data Stewards will need to complete an intense training programme. Some of that training will be externally provided (Essentials 4 Data Support), whereas other sessions will be delivered by local experts (about 4TU.Centre for Research Data or on the local Data Management Plan support service). Completion of the training programme will ensure that all Stewards are equipped with the necessary knowledge and skills to advise their faculty researchers.
Third, in order to judge the progress of the project, it is necessary to develop and agree on an effective set of metrics, which would allow deciding whether the project is moving towards its goal of improving good research data management practice. We will write more about this in future blog posts.
Finally, and crucially, Data Stewards need to get to know the research community they are supporting, not only to get a thorough understanding of the discipline-specific needs of their communities, but also to start building trust necessary for them to establish themselves as the ‘go to people’ and the first points of contact whenever advice on data management is needed.
Mid-term goals
There are two main mid-term goals (beginning – mid-2018). First, we would like to develop faculty-level policies on research data management, as well as an overarching TU Delft’s research data policy framework. Many institutions, in particular in the UK, have central institutional policies on research data management. However, our rationale for deciding on a different approach was similar to the rationale behind the whole Data Stewardship project: we believe that one size fits all solutions for research data management are difficult to implement in practice and come with a risk of becoming too aspirational and detached from the day-to-day practice. Our aim is to create tools which would give faculties sufficient flexibility to decide themselves on their most practical local solutions and expectations and to create policies, which could be truly implemented by the research community. We also thought that ensuring faculty’s leadership in policy development will result in a greater ownership and engagement with research data management.
Our second mid-term goal is to put our set of agreed assessment metrics to test: to obtain first meaningful data on the development of the project and to enable the assessment of its progress. This should happen by mid-2018 and this would also allow any necessary iterations to be devised and implemented.
Long-term goals
As explained before, the ultimate goal of the project is to ensure that researchers at TU Delft adhere to good data management practice on a day-to-day basis (2018 and beyond). However, a more realistic goal for a one-year project is to assess whether the proposed solution of having dedicated Data Stewards network is indeed working and whether it leads to improvements in data management practice with the use of agreed metrics.
The ultimate evidence of the success of the project would be however the judgement of the research community itself. Will researchers perceive Data Steward as their trusted source of necessary data management expertise? If so, one would expect to see Data Stewards to become permanent, key faculty staff members. And if subject-specific expertise is the solution for good data management practice, perhaps one day the presence of a dedicated Data Steward in every research group will become the normal thing, similarly as the presence of Lab Managers or Project Managers nowadays.
We will post regular updates on this project here, so watch this space!
Open Working 
That’s a very useful overview that can share with people asking about data stewardship. Much appreciated.
LikeLike
This is very interesting, and I look forward to following developments. Sabina Leonelli at Exeter and colleagues including Robert Davey, Elizabeth Arnaud, Geraint Parry and Ruth Bastow recently published a fantastic paper in Nature titled, ‘Data management and best practice for plant science’ (https://www.nature.com/articles/nplants201786) which among other things explores in interesting detail the balance between subject-specific and generic data management in a specific discipline, i.e. plant science. It’s highly relevant to the issues you will be exploring and a good reference point.
LikeLike