7 minutes reading time (1313 words)

A library’s learning from DMP feedback

By Signe Gadegaard and Jitka Stilund Hansen, Technical University of Denmark

The Technical University of Denmark (DTU) has 11.000 students and 6000 employees among which 3500 are researchers and Ph.D. students.

At DTU Library, we are a small group of people dedicated to facilitate and coordinate development and activities within Research Data Management (RDM), as well as to provide and maintain RDM services and tools.

As a result of DTU`s implementation of a Policy of the Retention of Primary Materials and Data from 2016, a data management plan (DMP) must now be prepared for all research projects at DTU, including all Ph.D. projects. Details for data management vary in the different departments and are therefore elaborated in local guidelines and procedures.

Information about RDM at DTU and our services are described in our pamphlet and on our library`s external website. Guidance to writing a DMP and using the DMP tool are found at the DTU intranet, and beside that, we offer a personal DMP feedback service.

Feedback service from the library
In the process of developing our feedback service, we are creating a schema with example suggestions and inputs for feedback on each aspect of a DMP customized to our own context and DMP template. In addition, we develop guidelines for ourselves so we can enhance the quality of the feedback we give. For instance, we found inspiration in the DMP review evaluation rubic grid .

We deliberately articulate that we offer feedback and not a review, as we wish to avoid the misconception that a feedback is providing the final answers to aspects covered in the DMP.

DMPOnline and the DTU Template
DTU Library uses DMPOnline as a tool for writing DMPs. DMPOnline is provided by the Danish e-Infrastructure Cooperation and offered free of charge to all Danish research institutions.

We have developed a DTU-specific template, which is divided into five sections: data collection, data storage, documentation, sharing and long-term storage. Example questions and DTU specific guidance are provided for each of these sections. Included are links to our RDM key concepts pages on the intranet, information about our data repository, links to storage systems at DTU as well as to GDPR guidelines and relevant DTU policies.

Example questions should guide the author of a DMP to cover key practices of data management, for example; "Which file formats are the data in"?, "What metadata will be included"? or "How will the data be made discoverable?".

Minimizing barriers
The core of the DTU DMP template is to aid the authors in describing their data and how they will work with data to follow good scientific conduct and to obtain FAIR data. However, complex FAIR data terminology are avoided as it could be perceived as an obstacle for getting started.

The template is offered to all projects and in cases where a template from the funder doesn`t exist. In terms of templates and funders, researchers often ask which template to select in DMPOnline. They are also interested in knowing how H2020 DMP`s will be evaluated as a deliverable and what sort of feedback to expect from the European commission.

Not all questions given in the template are equally relevant for all areas of research and a few research groups are adapting it in order to integrate domain specific data handling workflows in the their DMPs.

In most cases authors of DMP`s are requesting feedback by using the function "request expert feedback" in DMPOnline, but some authors prefer to e-mail a copy of the DMP to our mailbox, and both ways are accepted in order to minimize barriers for using the service.

Feedback emphasizes practicalities
Quite often, our users accidentally click on the feedback button in DMPOnline while browsing the tool; therefore, we have started asking the author if this was an intentional request.

Before submitting the DMP to us for feedback, we ask Ph.D. students to have their supervisor revise the plan. In general, we suggest having a peer to read and comment on the plan as well to check it is understandable to others.

Giving feedback is a learning process and in order to standardize and ease the task we are continuously working on our procedure. When giving feedback, we suggest steps for making data FAIR and best practices that can easily be implemented in the research data workflows.

Some DMPs are characterized by containing declarations of intent rather than describing concrete actions. In these DMPs, we emphasise that by answering the example questions of the template, a lot is achieved in terms of thinking practices through, write them down in a structured way and transforming the DMP into an active and living document.

We suggest small concrete measures that can transform a DMP into a basis for efficient and transparent workflows: e.g. to publish underlying data in our repository, DTU Data, with metadata or metadata only, a license and DOI. We explain that this makes data openly available and ensures that others can discover, access and re-use the data.

Often, our feedback addresses very practical content: what are the names of those who can access data, on which drive are data stored? We try to create a picture of the DMP as an asset for the research project and not an administrative burden.

Creating awareness
Researchers are often in doubt of data management requirements from DTU and funders. Such questions are inevitable in a dynamic research environment with researchers continuously entering DTU. Thus, creating awareness about responsibilities that apply for research activities at DTU as well as guidelines, infrastructures and tools is an ongoing task.

Our response to questions about the necessity of writing a DMP is that a DMP is a tool to identify and establish best practices and document this for others as well as one's future self. We aim to implement the DTU policy by sustaining a cultural change in RDM practices.

What is data sharing?
The sections in the DMP on sharing and publishing data are causing confusion. Researchers frequently describe that data will be published in scientific articles instead of dealing with the public sharing of research data. Statements like "no data will be shared in the project" are common and the concept of "sharing" is often perceived as sharing of data internally in the project.

Metadata and standards for metadata is another aspect that is usually insufficiently addressed. Sometimes it is stated that the project doesn`t use or generate metadata. Researchers tend to skip questions in the DMP, for example questions regarding file formats and versioning, which all reflects that the familiarity with the FAIR principles are limited.

Legal issues in terms of ownership to data and licenses for re-use of data are often not touched upon. Gradually, we are taking steps to rethink the template as we see emerging needs for closer GDPR reviews and support in legal aspects - especially in terms of the publication of code and software.

In our role as supporting data management, we need to consider if the language and terminology we use in the DMP is helping to clarify the concepts.

Change needs time
Completing a feedback is a significant investment of time. However, we experience the feedback service is necessary and appreciated. On the long term, we wish to have appointed data managers at the departments instructed in the DMP procedures. Currently, we are the stepping stones for such a process and our prevailing focus is supporting the Ph.D. students and researchers, who are motivated. The DMP feedback helps to create awareness on RDM and build bridges between our infrastructure, tools and services - and those who need it.

The researchers we support do see benefits of writing DMPs, but in practice, they lack time. Understanding and being in line with the FAIR principles requires training, resources and skills management. Encouragingly, we receive an increasing number of DMPs for feedback reflecting an attentive approach to data management.

Tags: