Uncertainty of a Different Kind….Estimation of Measurement Uncertainty in Testing Labs

Uncertainty of a Different Kind….Estimation of Measurement Uncertainty in Testing Labs

Uncertainty was one word used time and time again during the pandemic. Each of us have different approaches for trying to deal with expected change or apprehension about what lies ahead. I am certainly hopeful of brighter days ahead…This article, written by our Expert Tutor, Anne Marie Newell, talks us through estimating measurement uncertainty in testing laboratories.

Uncertainty suggests doubt, however within the testing laboratory uncertainty of measurement is actually a statement of confidence in the reported range in which the true measurement is expected to lie.

When analysts measure a chemical, physical or microbiological value in the lab they aim to obtain the true value of the measurement. However, there will always be some level of doubt around the measured value due to random and systematic variations that are present in the measuring process.  The measurement result will be just an estimate of the true value and the actual true value will remain unknown to us.

The term for the size of this level of doubt is known as the uncertainty.

Laboratory clients often say they  require results that are accurate, however accuracy is a qualitative term. What is accurate to one person may not apply to the next. Within the context of the laboratory, accuracy is characterised  by measurement uncertainty, which defines an interval around the measured value, where the true value  lies with some probability.  Uncertainty is a quantitative term with a ± figure

Testing laboratories accredited to the ISO17025 standard have always been required to estimate Measurement Uncertainty however new requirements in the 2017 update regarding Statements of Conformity (Pass / Fail decisions) have brought a renewed focus on uncertainty of measurement.

The statistical calculations used to calculate uncertainty appear daunting.  During the 2 day course on Estimation of Measurement Uncertainty for Testing Laboratories I start with the concept of measurement uncertainty and this sets the scene for the calculations used to estimate the level of precision (random variation) and bias (systematic) variation. For analysts that do not use some of these statistical calculations on a day to day basis I advise to start with what you do know and to build from there.

ISO 17025:2017 was updated with the objective of providing lab clients with greater reassurance in the results reported by the lab.  In my opinion this is an excellent standard and covers both the technical and quality management system requirements for testing laboratories.  There is a strong focus on ensuring the validity of the results reported. A  statement of uncertainty of measurement provides clients with the knowledge to enable them to make informed decisions. Lab reports typically refer to a 95% level of confidence and occasionally this is stated at 99.7% confidence.  Most would agree that to be given a set of test results along with the estimated range of uncertainty reported at a 95% confidence is very reassuring!

See more information or to book on our Introduction to Estimation of Measurement Uncertainty for Testing Laboratories