DKZ.2R Data Competence College - Reproducibility in Research
Janz, Alicia; Hartman, Jonathan; Bossert, Lukas C.; Immel, Katharina; Mielke, Alexandra; Winter, Bodo; Kaithalikunnel Chandran, Anoop; Kunkel, Julian
- Publisher
- Zenodo
- Language
- en
Abstract
The second Data Competence College "Reproducibility in Research" was hosted from November 24th to 26th, 2025 at the IT Center of RWTH Aachen University in Germany. Based on the concept of the Wissenschaftskolleg in Berlin or the Institute of Advanced Studies in Princeton, we invited three individuals with high data competence from different scientific fields (“Data Experts”) to participate as part of the Data Competence College: Prof. Julian Kunkel (University of Göttingen, Germany; Gesellschaft für wissenschaftliche Datenverarbeitung mbH Göttingen (GWDG)) Prof. Bodo Winter (University of Birmingham, UK) Anoob K. Chandran (Forschungszentrum Jülich GmbH, Germany) For three days we aimed to create a space where not only local scientists, and especially early career researchers, learn from the data experts and each other regarding research data and methods but also data experts could inspire each other. The schedule included keynote presentations by all data experts, poster and group presentations by the participants, 1:1 sessions between data experts and early career researchers, as well as a method- and data-related workshop. We aimed foremost to create an environment in which everyone feels safe to give input, share their knowledge and learn from the other participants and experts.
Full text
9am 10am 11am 12pm 1pm 2pm 3pm 4pmm 5pm 6pm Day 1 (Monday, November 24th) Welcome and Introduction Poster Session Interdisciplinary Exchange Day 2 (Tuesday, November 25th) Day 3 (Wednesday, November 26th) Lunch Break Lunch Farewell Rising Star Presentation Rising Star Presentation Break Prof. Julian Kunkel: “Data Management Approaches and Challenges in Science and Research – These are my Voyages” Workshop Data Competence College Schedule of Events (November 24-26, 2025) Prof. Bodo Winter: „Designing for Diversity: Bayesian Methods and the Generalizability Crisis” Break Open to the Public 1:1 Sessions 1:1 Sessions Intro Anoop Chandran 1:1 Sessions 1:1 Sessions 1:1 Sessions 1:1 Sessions 1:1 Sessions 1:1 Sessions 1:1 Sessions 1:1 Sessions Break
Abstracts: Prof. Julian Kunkel: „Data Management Approaches and Challenges in Science and Research - These are my voyages“ Ensuring data reproducibility, usability, privacy, and performance in large-scale scientific computing remain critical challenges, particularly within high-performance computing (HPC) environments and research consortia. This talk elaborates on these challenges covering topics we aim to address including Trusted Research Environments (TREs), robust data management frameworks in HPC, secure data sharing via data pools, concepts for semantic data storage, optimized data ingress/egress mechanisms, the Göttinger Data Lake. The integration of these strategies is exemplified with some use cases. My professional journey has been deeply rooted in empowering researchers to effectively use data infrastructure. This includes active participation in international consortia such as IO500, where I contributed to benchmarking standards for high-performance storage systems, driving transparency and reproducibility in HPC data environments. As a PhD student, it is not about revolutionizing the world. My experience led to an understanding of technical, organizational, and ethical dimensions of data stewardship in large-scale research ecosystems with the problems that still need to be adressd by future generations and collaborations. Prof. Bodo Winter: „Designing for Diversity: Bayesian Methods and the Generalizability Crisis” The replication crisis has sparked much-needed scrutiny of methodological practices, but what has been called the “generalizability crisis” (Yarkoni, 2022) may be an even deeper problem: many findings, even when replicable, fail to generalize because they neglect key sources of variation (cf. Winter & Grice, 2022). In part, this results from the deeply ritualized tradition of Null Hypothesis Significance Testing (Gigerenzer, 2004), within which researchers seek homogeneous samples and focus on mean differences, thereby treating variation as mere noise. I argue that Bayesian methods offer a way forward: they enable more robust estimation of variation parameters, thereby supporting richer models that can incorporate the diversity present in real data. With more flexible statistical tools, we can revisit our design assumptions and encourage a new experimental ethos of “designing with variation in mind” (Yarkoni, 2022). This ethos asks us to move away from artificially homogeneous samples toward studies that deliberately embed and model heterogeneity, thereby producing findings that better generalize. We need a radical departure away from an exclusionary focus on mean differences towards a mindset that treats variation as an important source of information, and an important target of inferential statistics.
