Overview
ISO/TR 27877:2021 - "Statistical analysis for evaluating the precision of binary measurement methods and their results" is an ISO Technical Report that consolidates statistical approaches for assessing the precision of binary (two-class) measurement methods. It introduces five statistical methods and explains how to estimate precision measures using data from collaborative studies where each laboratory supplies repeated binary measurements. The report summarizes theory, provides worked examples (e.g., Listeria testing, h‑CLAT, histopathological classification), and compares the methods to guide selection and interpretation.
Key topics
- Scope and structure
- Five statistical methods grouped into two types:
- Type A: each laboratory repeatedly measures a single, nominally identical sample.
- Type B: extension with several sample levels.
- Methods introduced
- ISO 5725–based method (adaptation for binary outcomes)
- Accordance and concordance (within-lab vs between-lab agreement)
- ORDANOVA (ordinal dispersion-based approach; applicable to ordinal/binary)
- Confusion-matrix statistics: CM‑accuracy, sensitivity, specificity, CM‑precision, F‑measure
- Kappa coefficient (chance-corrected agreement)
- Estimation and interpretation
- How to estimate precision measures from collaborative study data.
- Worked cases and numerical examples demonstrating repeatability and reproducibility analogues for binary results.
- Comparative analysis
- Side-by-side comparison of mathematical expressions and numerical behaviours of the five methods.
Applications
ISO/TR 27877:2021 is practical for evaluating and documenting the precision of binary measurement methods in contexts such as:
- Clinical diagnostics (positive/negative test results)
- Food safety microbiology (e.g., Listeria detection)
- Toxicology and in vitro assays (e.g., h‑CLAT)
- Histopathology classification and other categorical lab outputs
- Method validation and inter-laboratory studies used for accreditation, regulatory submissions, and quality assurance
Who should use this standard
- Laboratory managers designing collaborative studies and method validations
- Statisticians analyzing binary outcome precision and agreement
- Test developers validating binary diagnostic/classification methods
- Regulatory bodies and accreditation organizations assessing method performance
- Researchers comparing statistical approaches for qualitative data
Related standards
- ISO 5725 (accuracy, repeatability and reproducibility of quantitative methods) - background principles
- ISO 3534‑1 (statistics - vocabulary and symbols)
ISO/TR 27877:2021 fills an important gap by providing guidance tailored to binary measurement precision, integrating confusion-matrix metrics, agreement statistics, and practical examples to support robust method evaluation and decision-making.