Data publication: “Part & parcel” of the reproducible research process
Joost de Winter is an associate professor working in the Department of Cognitive Robotics at the Faculty of Mechanical, Maritime and Materials Engineering (3mE) at the TU Delft. Joost’s work focuses on Human-Robot interactions specifically in the field of automated cars.
“If you hear the name of the discipline Mechanical Engineering maybe you don’t immediately think about data, but in the section where I work it is all about the data. I work on the topic of human interaction with automated cars where we collect different types of data, for example, recorded data from sensors, models, and algorithms on how to process the data.”
Joost de Winter
In recent years, Joost has worked intensively to implement best practices for data sharing within his group and to raise awareness about the important role of data in ensuring reproducible research. His interest in reproducibility in his own field of research motivated Joost to apply for a replication study about ‘pupil size as related to interest value of visual stimuli’ financed by the Netherlands Organisation for Scientific Research (NWO), where he also collected valuable experience on the generation of better and reproducible research.
“Five years ago I started to realize that we need to be more careful with the data. When reading about all the reproducibility problems in psychology, for example, I thought that we could improve our workflows as well, from the collection to the publication of data.”
To achieve this, Joost established a set of principles for his group members on where to store the data to make it available for the whole group as well as how to organize and structure folders.
Joost and his group have archived a collection of data sets on 4TU.ResearchData. His main motivations to publish data are ensuring reproducibility and accountability of the research conducted by his group.
“If you publish something, you want to ensure that your research is reliable. As a minimum our group wants the results of our papers to be reproducible. The data used to create figures and all the analysis done have to be available for internal review as well. When you make the data available (as much as possible) it gives more confidence in the study.”
The rapid advancement of research on automating driving is resulting in the collection and publication of large amounts of data, bringing huge potential for new research. Joost and his group will continue contributing to the field publishing their data on 4TU.ResearchData.
“The publication of datasets is very easy and straightforward. 4TU.ResearchData provides personal support and customized help, the data are in-house (institutional servers), it complies with all requirements for a standardized repository and it’s free!”
At the end of our conversation, we asked Joost what he would tell other researchers to encourage them to publish their data. His response was:
“First of all, the research process is only fully completed if you can reproduce the study. Therefore, the publication of data is part of the research process. If there is a need for more encouragement, I can say that publishing your data ensures the continuity of projects (subsequent students have access to the data). If someone raises questions about your research they can go directly to the repository and check the data, multiple people can go through the data and check for mistakes and it gives you a good feeling to finalize the project having all the data compiled and findable.”
All of the published data sets generated by Joost de Winter and his group are available here.
Cover image by Chaitawat Pawapoowadon via Pixabay