Overview
ISO 11843-4:2003 - “Capability of detection - Part 4: Methodology for comparing the minimum detectable value with a given value” - defines a statistical procedure to assess whether a measurement method can detect a specified level of a net state variable (i.e., whether the minimum detectable value is less than or equal to a given value). Unlike Part 2, this part does not assume a linear calibration curve or fixed relationships between residual variability and the measured value, making it suitable for low-level and non-linear detection problems.
Key topics and requirements
- Purpose: Provide a criterion and experimental design to judge if the method’s minimum detectable value ≤ a given value (xg) instead of estimating the minimum detectable value directly.
- Assumptions: independent, normally distributed response measurements and that reference and test materials behave similarly in the measurement system.
- Experimental design: Use two reference states - the basic state (net state = 0, e.g., blank) and a sample with the given net value xg. Replication counts:
- N = number of replicates per reference material in capability assessment (preferably ≥ 5)
- J, K = replications used in routine application (often J = K = 1 in validation)
- Statistical criterion: The standard provides formulas for the critical response value and a power criterion (probability ≥ 1 − β) that the method will detect the specified xg given a type I error α. Simplified forms and confidence-limit-based confirmation procedures are included for practical use.
- Confirmation and reporting: The standard prescribes how to compute lower confidence limits for the relevant statistic from experimental data and specifies the information to report (reference material details, N, means and standard deviations, chosen α, β, J, K, and conclusion about capability).
Applications and users
- When to use: validation of measurement methods where linear calibration or homoscedasticity near zero are doubtful - common in trace analysis and low-level determinations.
- Typical sectors: environmental monitoring, clinical/biomedical assays, food safety, chemical analysis, materials testing, and any laboratory that must demonstrate a detection capability for a specified limit.
- Who benefits: method developers, laboratory validation teams, quality managers, accreditation bodies, regulators, and instrument manufacturers seeking to demonstrate or document detection capability without relying on linear calibration assumptions.
Related standards
Keywords: ISO 11843-4, capability of detection, minimum detectable value, detection limit, validation, measurement method, statistical criterion, critical value, low-level detection.