How to Validate Test Methods for Defensible Results

A test result is only as defensible as the method used to produce it. Knowing how to validate test methods is therefore fundamental for laboratories, manufacturers, fabricators, asset owners, and engineering teams making decisions about compliance, safety, material performance, or remaining service life. Validation establishes documented evidence that a method is fit for its intended purpose under defined conditions.

That distinction matters. A procedure may appear technically sound, produce plausible values, and still be unsuitable for the decision at hand. A corrosion-rate method used for trend monitoring, for example, may not have the accuracy or detection capability required to support a contractual acceptance decision. Effective validation connects the method, the sample matrix, the measurement range, and the required decision to objective performance data.

What Test Method Validation Actually Demonstrates

Method validation is a planned study that confirms a test method can consistently deliver results that meet predefined acceptance criteria. It applies to standardized methods, modified standards, laboratory-developed methods, and methods adapted for new materials, matrices, equipment, or reporting limits.

Validation should not be confused with method verification. Verification confirms that a laboratory can perform an established, validated standard method as intended within its own facility. Validation is more extensive. It is required when the laboratory develops a new method, makes a technically significant departure from a standard, combines procedures, changes the intended application, or uses a method outside the scope for which its performance has already been demonstrated.

For ISO/IEC 17025-aligned operations, the objective is not simply to generate a file of test data. The laboratory must demonstrate that the chosen method is appropriate, controlled, technically justified, and capable of supporting reliable reporting. The depth of validation should be proportionate to risk. A screening test used to prioritize samples does not require the same performance evidence as a method used to accept safety-critical weld material or quantify a regulated contaminant.

How to Validate Test Methods: Start With Intended Use

The strongest validation programs begin before samples enter the laboratory. Define what the method must achieve, who will use the result, and what decision will follow. This establishes the validation characteristics, sample population, acceptance criteria, and reporting requirements.

A useful method purpose statement identifies the material or matrix, measurand, measurement range, expected accuracy, reporting units, decision limits, and any applicable standard or client specification. For example, a positive material identification procedure for alloy sorting may need to distinguish between specific stainless steel grades in field conditions. A laboratory chemical analysis may instead need to quantify alloying elements at low concentrations with traceable calibration and stated measurement uncertainty.

This stage should also identify foreseeable sources of variation. These can include sample geometry, surface condition, coating residues, operator technique, instrument configuration, environmental conditions, reagent lot variation, specimen preparation, and data-processing rules. If a factor is likely to occur in routine use, it should be represented in the validation study rather than treated as an afterthought.

Select the Performance Characteristics That Matter

Not every validation requires every possible performance characteristic. The appropriate set depends on the method’s intended use and the consequence of an incorrect result. However, most laboratory and engineering test methods will require evidence across several core areas:

• Selectivity or specificity demonstrates that the method measures the intended property or analyte without unacceptable interference from other materials, components, or conditions.
• Accuracy or trueness evaluates the closeness of results to an accepted reference value, certified reference material, comparison method, or known addition.
• Precision measures agreement among repeated results. This should normally address repeatability and, where relevant, intermediate precision across analysts, days, equipment, or locations.
• Working range and linearity confirm that the method performs acceptably over the concentration, load, dimension, or response range that will be reported.
• Detection limit and quantitation limit establish the lowest level that can be detected or reliably quantified when the method is used for trace analysis or low-level measurement.
• Robustness assesses whether small, deliberate variations in controlled conditions materially affect the result.

For destructive mechanical testing, validation may focus on load-cell performance, extensometer calibration, specimen geometry, strain rate, temperature control, operator consistency, and comparison against reference materials or recognized methods. For coating inspection, relevant factors might include substrate profile, coating thickness range, instrument probe selection, surface curvature, and calibration check frequency. The performance criteria must reflect real operating conditions, not ideal laboratory conditions alone.

Develop a Written Validation Protocol

A validation protocol prevents a study from becoming a collection of unstructured trials. It should be approved before work begins and should explain exactly how performance will be assessed.

The protocol should state the method identification and revision, intended use, applicable standards, equipment, reference materials, sample selection approach, number of replicates, controlled variables, statistical methods, acceptance criteria, deviations process, and responsibilities. It should also specify what will happen if the method fails to meet a criterion. Repeating unfavorable results without a documented technical reason is not validation. It is a threat to data integrity.

Sample selection deserves particular attention. Validation samples must represent the method’s routine workload, including expected variation at the lower and upper ends of the working range. Where practical, include difficult samples: oxidized surfaces for material identification, heterogeneous concrete for chemical assessment, welded regions for metallography, or specimens with known interference risks for spectroscopic analysis.

Reference materials provide strong evidence of trueness, but they are not always available for every engineering application. In those cases, laboratories may use interlaboratory comparison, an independent reference method, spiked samples, designed experiments, or carefully characterized in-house reference materials. The technical rationale for the chosen approach should be clear and documented.

Execute the Study Under Controlled, Representative Conditions

Validation data should reflect normal operations while retaining enough control to identify why variation occurs. Confirm that instruments are calibrated and maintained, personnel are trained, software is version-controlled where applicable, and critical environmental conditions are monitored.

Precision studies should be designed to answer practical questions. Can different qualified analysts obtain comparable results? Does a change in test day affect the outcome? Does the method remain stable after routine recalibration? For methods used at multiple sites or in the field, reproducibility across locations or equipment may be more relevant than repeatability on one instrument.

Where a method involves a threshold or acceptance limit, examine performance around that limit. A method may perform well at mid-range values but show unacceptable bias or variability near the point where a component is accepted, rejected, repaired, or removed from service. This is where measurement uncertainty and decision rules become operationally significant.

Evaluate Results and Establish Measurement Uncertainty

Data evaluation should use suitable statistical methods, but statistics alone do not determine fitness for purpose. A small standard deviation can still be unacceptable if the method is biased. Conversely, a method with moderate variation may be suitable for a screening application where results are interpreted conservatively.

Compare each performance result with the pre-established acceptance criterion. Investigate outliers according to a documented procedure, considering equipment malfunction, transcription errors, sample anomalies, or procedural departures. Do not exclude data solely because it is inconvenient.

Measurement uncertainty should be evaluated where it is relevant to the validity of reported results. Its contributors may include calibration uncertainty, reference material uncertainty, repeatability, environmental effects, sample preparation, dimensional measurement, and operator influence. The uncertainty statement must be meaningful for the measurement range and decision being made. Where compliance statements are issued, laboratories should apply a defined decision rule that accounts for uncertainty and the risk of false acceptance or false rejection.

Document Approval, Training, and Ongoing Control

A completed validation report should present the method purpose, protocol, raw-data traceability, calculations, deviations, results, acceptance criteria, conclusion, limitations, and authorization for use. It should define the validated range, applicable matrices, reporting limits, and any restrictions. A result outside these boundaries should not be reported as though it carries the same validated confidence.

Method validation is not a one-time event. Ongoing quality control confirms that performance remains acceptable after implementation. Depending on the method, this may include control samples, duplicate testing, blanks, reference checks, proficiency testing, trend charts, equipment checks, and periodic technical review.

Revalidation, either full or partial, should be considered after changes such as a new instrument platform, significant software update, revised sample preparation, changed reagent source, expanded range, new material matrix, relocation, or a sustained quality-control failure. Change control is especially important when a method supports regulatory, contractual, or safety-critical decisions.

For complex materials investigations, nonroutine testing, or custom method development, independent technical review can identify gaps before results are relied upon. AECTL applies accredited laboratory discipline and multidisciplinary engineering expertise to help clients establish methods that are practical for the project while producing reliable, defensible evidence.

A well-validated method does more than satisfy a quality requirement. It gives project teams a defined basis for acting on the result, whether that means accepting material, investigating a failure mechanism, prioritizing repairs, or demonstrating compliance with confidence.