Data stewardship: case studies from North-American, Dutch, and Finnish universities

Antti Mikael Rousi, Reid I. Boehm, Yan WangarXiv 2023 · DOI · arXiv

Introduction

As global research shifts toward Open Science and FAIR (Findable, Accessible, Interoperable, and Reusable) data principles, universities face a significant logistical challenge: how to effectively support researchers in managing their data. While the necessity of data stewardship is well-documented, the optimal organizational structure for these programs remains a subject of debate. This paper compares three distinct data stewardship models at Purdue University (USA), Delft University of Technology (Netherlands), and Aalto University (Finland), revealing how institutional culture, national legislation, and funding structures dictate whether a program succeeds as a centralized library service or a decentralized network of embedded researchers.

Methodology

The researchers employed a qualitative case study design, utilizing two primary techniques:

  1. Autoethnography: The authors, managers of the respective programs, provided critical reflections and internal case notes based on their professional experience.
  2. Thematic Analysis: An analysis of 45 publicly available documents, including institutional policies, national legislation, and service descriptions, was conducted using the "four dimensions" framework: policy and governance; service establishment and funding; advocacy and support; and technical research data services.

The study's dataset includes over 2,300 words of autoethnographic notes and a curated list of policies such as TU Delft’s Research Data Framework Policy and national mandates like the US HIPAA or the EU’s Open Data Directive.

Comparative Analysis: Drivers and Models

Data stewardship programs exist on a spectrum between "Coordinator" and "Embedded" models. At one end, the Coordinator model (e.g., TU Delft) provides broad support across domains. At the other, the Embedded model (e.g., Aalto) involves stewards working closely within specific research teams.

A primary driver for these differences is the legislative environment. In the United States, stewardship is largely propelled by federal funding mandates and specific privacy laws like HIPAA and FERPA. In contrast, European institutions are heavily influenced by the General Data Protection Regulation (GDPR) and the European Open Data Directive.

Table 3: Summary of main legislation and policies affecting the case study institutions.

Furthermore, the technical infrastructure plays a role. Purdue’s program is tightly integrated with its in-house repository (PURR). Aalto, lacking a dedicated institutional repository, relies on national and European infrastructures, shifting their stewards' focus toward advisory and peer-to-peer support roles.

Institutional Outcomes and Staffing

The study highlights a significant variance in staffing levels and professionalization across the three institutions.

  • TU Delft (The "Heavyweight"): Features the highest investment with 13 Full-Time Equivalents (FTEs). These are full-time professionals embedded directly within faculties.
  • Purdue (The "Centralized"): Operates with 3 FTEs based within the University Libraries, focusing heavily on the administration of the PURR repository.
  • Aalto (The "Hybrid"): Utilizes a network of part-time researchers—"Data Agents"—totaling 2.5 FTEs. These are active scientists who provide peer-to-peer support in exchange for salary compensation.
Table 2: Summary of the data stewardship program types, tasks, and used FTE numbers per case institution.

Despite these structural differences, all three institutions converged on four core service themes: Research Data Management (RDM) training, policy development, community engagement, and providing multimodal support systems (combining face-to-face, email, and webinars).

Conclusion and Future Outlook

These case studies provide a snapshot of a rapidly evolving field. While the findings are primarily focused on technical and natural sciences, they offer a roadmap for other institutions by highlighting the trade-offs between different organizational models.

Future research should focus on developing more sophisticated metrics and cost-benefit analyses to determine the long-term sustainability of these staffing models. Additionally, further investigation into the career paths of data stewards is necessary to ensure the professionalization and viability of the field.

Summary updated

BibTeX
@misc{rousi2023data,
  author = {Antti Mikael Rousi and Reid I. Boehm and Yan Wang},
  title = {{Data stewardship: case studies from North-American, Dutch, and Finnish universities}},
  year = {2023},
  eprint = {2312.04092},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2312.04092},
}
Figure