Method Verification and Validation Studies

In modern metrology, the validity of a measurement result depends just as heavily on the mathematical methodology and automated software execution as it does on the physical hardware standards. Under ISO/IEC 17025:2017, laboratories cannot simply assume that a published test method, automated Python script, or complex Excel spreadsheet functions correctly. Every procedure and software tool deployed to collect, calculate, or report calibration data must be formally verified or validated before being placed into routine production service.

---

Method Verification vs. Method Validation per ISO/IEC 17025

A central concept tested on the ASQ CCT examination is the critical operational distinction between Method Verification and Method Validation governed by ISO/IEC 17025 Clause 7.2.

Method Verification (Clause 7.2.1.5)

Method verification is the provision of objective evidence that the laboratory can properly perform a recognized, published standard method before introducing it into service. Verification applies when the laboratory adopts:

• Consensus standards published by international or national standards organizations (e.g., ASTM E145 for ovens, ISO 6789 for hand torque tools, applicable electrical calibration methods (IEEE 488 is an instrument-interface standard)).
• Standard calibration procedures published in original equipment manufacturer (OEM) service and calibration manuals.
• Recognized metrology guides published by regional metrology organizations (such as EURAMET or SIM).

Verification Protocol: Demonstrate that the facility, personnel, equipment, and environment can achieve the method’s required performance. Reference artifacts, precision studies, uncertainty evaluation, and relevant comparisons can provide evidence. An interlaboratory comparison is not a mandatory component of every individual method verification.

Method Validation (Clause 7.2.2)

Method validation is the confirmation by examination and the provision of objective evidence that the particular requirements for a specific intended use are fulfilled. ISO/IEC 17025 requires validation to the extent necessary for intended use when the laboratory uses:

• Non-standard methods developed entirely in-house.
• Laboratory-designed custom test fixtures or automated procedures.
• Published standard methods that have been modified (e.g., changing test points, using alternative excitation voltages, or testing at non-standard temperatures).
• Standard methods applied outside their intended operational scope (e.g., using a procedure designed for laboratory ambient temperatures to calibrate sensors inside a cryogenic chamber).

---

Selecting validation characteristics for intended use

When validating a custom or modified calibration method, the Technical Manager selects relevant performance characteristics and acceptance criteria. The following are possible characteristics, not seven universal requirements for every calibration:

Trueness and Accuracy

Demonstrate that the method provides measurement results free from uncorrected systematic bias. Trueness is verified by comparing measurement results against certified reference materials (CRMs), primary intrinsic standards, or by participating in Proficiency Testing (PT) / Interlaboratory Comparisons (ILC).

Performance is evaluated using the Normalized Error Ratio ($E_n$):

$$E_n = \frac{|x_{\text{lab}} - x_{\text{ref}}|}{\sqrt{U_{\text{lab}}^2 + U_{\text{ref}}^2}}$$

Where $x_{\text{lab}}$ and $x_{\text{ref}}$ are the laboratory's measured value and the reference value, and $U_{\text{lab}}$ and $U_{\text{ref}}$ are their respective expanded measurement uncertainties ($k=2$). For a comparable, uncorrelated comparison, agreement is supported at the points tested if:

$$E_n \le 1.0$$

An absolute score above one prompts investigation of the results, uncertainty, correlation, and comparison conditions; it does not diagnose a unique cause.

Precision (Repeatability and Reproducibility)

• Repeatability ($s_r$): Closeness of agreement between independent test results obtained under identical conditions (same technician, same reference standard, same environmental setup, over a short time interval).
• Intermediate Precision: Variation under changed conditions within a laboratory, such as operators or days. Reproducibility conditions in VIM involve different locations and measuring systems. Choose replicate counts and study design for the intended performance claim; ten to thirty observations is not a universal minimum.

Linearity and Working Range

The LOD and LOQ factors below illustrate one blank-standard-deviation/slope model with its adopted criteria. They are not universal calibration requirements or definitions for every method. Verify the model, distribution, and intended detection or quantification performance. A nonlinear method can be valid if its model and uncertainty satisfy the intended use. Verify that the measurement response is directly proportional to the physical stimulus across the entire working span. The technician plots measured values against reference values and performs a least-squares linear regression:

$$y = m x + b$$

Set application-specific criteria. A large $R^2$ alone does not demonstrate acceptable residual errors or uncertainty and that the residuals exhibit a random distribution without systematic curvature.

Limit of Detection (LOD) and Limit of Quantitation (LOQ)

For sensitive instrumentation (e.g., optical power meters, picoammeters, micro-balances), the laboratory may need to establish applicable detection and quantitation limits:

• Limit of Detection (LOD): The lowest quantity that can be reliably distinguished from baseline electrical/mechanical noise:

$$\text{LOD}_{\text{example}} = 3.3 \cdot \frac{s_0}{S}$$

• Limit of Quantitation (LOQ): The lowest concentration or level that can be quantitatively measured with acceptable uncertainty:

$$\text{LOQ} = 10 \cdot \frac{s_0}{S}$$

where $s_0$ is the standard deviation of baseline blank readings, and $S$ is the slope (sensitivity) of the calibration curve.

Measurement Range and Dynamic Span

Establish the range over which the intended model, bias, precision, and uncertainty satisfy requirements. A result outside that range cannot be represented as validated routine work. Resolve any proposed extension through the appropriate verification or validation and agreed scope before making that claim.

Robustness and Ruggedness Testing

Evaluate whether performance remains adequate for intended use under small, deliberate changes. Robustness is not a promise of exactly unchanged readings. Choose safe perturbations within the applicable operating limits and examine resulting bias and uncertainty.

• Ambient temperature (e.g., $21^\circ\text{C}$ vs $25^\circ\text{C}$)
• Relative humidity (e.g., $30\%$ vs $65\%$)
• AC mains supply voltage (e.g., $110\text{ V}$ vs $125\text{ V}$)
• Test lead cable length (e.g., $1\text{ m}$ vs $2\text{ m}$) If these variations induce statistically significant shifts in results, the method must define strict operational tolerance limits.

Selectivity and Cross-Sensitivity

Verify that the method measures only the intended measurand without confounding interference from secondary physical influences. Examples include evaluating magnetic field interference on current clamps, thermal EMF interference on microvolt measurements, or barometric pressure sensitivity on quartz balances.