TY  - GEN
TI  - Transparently Safeguarding Good Research Data Management with the Lean Process Assessment Model
Y1  - 2021///
EP  - 1
ID  - heidok29719
CY  - Heidelberg
A1  - Geßner, Hendrik
N2  - In the last years, research data management moved into the spotlight of the scientific community. Organizations like the DFG and projects like FDMentor updated their guidelines to include current research software and data developments, while concepts like FAIR publishing gained traction interdisciplinarily. However, research guidelines often either take an abstract policy-driven perspective or solely focus on practices that, by omitting the underlying principles, become obsolete as the state-of-the-art advances. When looking at quality and evaluation methods in the industry, especially in systems and software development, models like CMMI, SPICE, or Six Sigma take a holistic approach by combining a process or life cycle perspective, clear goals and, target-oriented practices. These models were created with industrial processes in mind, and applying them to research projects directly is counterintuitive. We developed a Lean Process Assessment Model (LPAM) for research software and data that adheres to the CMMI framework. Following the lifecycles for research software and data, guideline practices from existing literature were analyzed and grouped into processes and goals. LPAM was developed with regular researcher feedback to ensure suitability for research projects. This procedure resulted in a discipline-agnostic model to manage and assess research projects, chairs, or organizations. The different processes were assigned to CMMI's Maturity Levels, which rank each process's priority and give a clear improvement path. CMMI follows the idea that unplanned processes are chaotic so that one project's success may not be repeated in another. While one project may follow agreed-upon community practices such as FAIR publishing, the next project could fail to meet quality standards due to time pressure while overachieving in other areas. With clear priorities, the model helps researchers in balancing goals and practices in their work. For assessing the state of a research project with LPAM, we propose a peer-review based procedure that is intuitive and well-established for researchers. CMMI knows three assessment method levels, which reflect different granularities of reliability, correctness, and effort. Researchers can choose a suitable assessment class based on assessment frequency. LPAM consists of three process areas: software, data, and project management/support, each with specific goals and practices. Goals and practices contain extensive hints that refer to published materials and guidelines. It also contains comments on maturity levels, generic goals, and assessments. CMMI allows individual practices to be replaced by equivalent ones if they are suitable for achieving the overall objective. The framework allows LPAM to stay up-to-date, even when the state-of-the-art advances. We are convinced that LPAM narrows the gap between goals, principles, and practices and is a suitable tool to safeguard good research data management transparently.
UR  - https://archiv.ub.uni-heidelberg.de/volltextserver/29719/
AV  - public
ER  -