Beyond ISO 5725-2: robust statistical methods for precision and reproducibility in materials testing
Publication date
2026-05-30
Document type
Konferenzbeitrag
Author
Moro, Fabrizio
Organisational unit
Conference
RILEM Spring Convention and Conference 2026 (RSCC 2026) ; Ghent, Belgium ; April 13-17, 2026
Publisher
Springer
Series or journal
RILEM Bookseries
ISSN
Periodical volume
69
Book title
Proceedings of the RILEM Spring Convention 2026
Volume (part of multivolume book)
1
First page
614
Last page
622
Peer-reviewed
✅
Part of the university bibliography
✅
Language
English
Keyword
Performance-based testing
Measurement precision & reproducibility
Interlaboratory validation (ISO 5725)
Robust mixed-effects models
Abstract
Precision and reproducibility are critical for ensuring the reliability of measurement methods, especially in materials testing, where results influence design, safety, and compliance. Interlaboratory studies (round-robin tests) are a central tool in estimating these metrics. Traditionally, ISO 5725-2 has provided the statistical foundation for such analyses, relying on classical ANOVA-based methods and fixed rules for outlier detection. With advances in computational statistics, including robust statistical methods and mixed linear models, the ISO 5725-2:2019 update now explicitly permits modern alternatives. This con-tribution explores how robust linear mixed models can enhance the analysis of precision and reproducibility data. The results demonstrate that robust methods provide greater stability against outliers and unbalanced designs, leading to more reliable estimates of variability components. The paper concludes with recom-mendations for practitioners and standards developers on adopting robust meth-odologies in precision studies.
Version
Published version
Access right on openHSU
Metadata only access