GIFSQ Tools · Module 15 — Metrology, Validation & Laboratory Accreditation · Tool 290 of 500
Plan a method verification study: how many replicates per level are needed to estimate the mean within your target margin, and the total number of analyses for the whole study.
Try sample values — fills the form with the worked example below.
–
Replicates per level
–
Total analyses
–
Analyses per level (with spares)
Equation: Replicates per level n = ceil[(t × SD ÷ margin)²]. Total analyses = n × levels.
Method: Sample-size planning for verification per ISO/IEC 17025 clause 7.2 (verification of a standard method) and Eurachem method-validation guidance: choose n so the confidence interval for the mean is no wider than the target margin, using the expected standard deviation from prior data.
Assumptions: the expected SD is a fair estimate for this matrix and level; the multiplier t ≈ 2 approximates 95% confidence for moderate n; replicates are independent.
Limitations: the formula uses a normal approximation — for very small n, use the exact t-value for n − 1 degrees of freedom and iterate; the plan covers precision only, not trueness, which needs spiked samples as well.
Rounding: replicates and totals are whole numbers (rounded up); intermediates unrounded.
This is a calculation aid, not a laboratory report — confirm against your accredited method before citing a value.
Method: Sample-size planning for verification per ISO/IEC 17025 clause 7.2 (verification of a standard method) and Eurachem method-validation guidance: choose n so the confidence interval for the mean is no wider than the target margin, using the expected standard deviation from prior data.
Assumptions: the expected SD is a fair estimate for this matrix and level; the multiplier t ≈ 2 approximates 95% confidence for moderate n; replicates are independent.
Limitations: the formula uses a normal approximation — for very small n, use the exact t-value for n − 1 degrees of freedom and iterate; the plan covers precision only, not trueness, which needs spiked samples as well.
Rounding: replicates and totals are whole numbers (rounded up); intermediates unrounded.
This is a calculation aid, not a laboratory report — confirm against your accredited method before citing a value.
Worked example: Expected SD 0.5 mg/kg, target margin 0.25 mg/kg, t = 2, 3 levels: n = ceil[(2 × 0.5 ÷ 0.25)²] = ceil[16] = 16 replicates per level, total = 48 analyses.
Enter values above to see results.
Add one spare replicate per level for re-runs — the plan above shows the minimum, not the bench-ready count. Tool 290 of 500 — GIFSQ
