The notion of measurement uncertainty

Measuring means comparing an unknown physical quantity with a quantity of the same nature taken as a reference, using an instrument and therefore giving a value to an observation.

The result of a measurement, is expressed through a numerical value, associated with a unit that recalls the nature of the reference and matched with additional information: its uncertainty.

The uncertainty of a measurement characterizes the dispersion of values that could reasonably be attributed to the result. It makes it possible to evaluate the errors that occur during measurement.

These errors are of two types:

Random error

This is the error that can appear when a large number of measurements are made, under the same conditions. It corresponds to the "normal" variability of a result.

Hence the notion of fidelity of a measurement result:
"This is the closeness of agreement between independent results obtained under certain conditions and the mean of the values found. It depends solely on the distribution of random errors and has no relation to the true value. It provides an indication of errors due to chance."

Systematic error

This is an error that always takes the same value during each measurement, it is a constant deviation from a true value.

Hence the notion of trueness of a measurement result:
"It is the closeness of agreement between the mean value obtained from a large series of results and an accepted reference value. It provides an indication of systematic errors (bias)."

Estimating uncertainties

Estimating measurement uncertainties can be carried out using several approaches.

Using our TITRIVINs, you estimate your uncertainties from the results obtained when setting up your internal quality control. This is then an intra-laboratory approach, as you use the internal data from your control charts.

In effect, these control charts made with TITRIVINs provide you with information on:

• random errors (fidelity of your results) in the form of an intra-laboratory reproducibility standard deviation,

• systematic errors (accuracy of your results) because the TITRIVIN has a robust reference value.

For a concentration level corresponding to the TITRIVIN analyzed, the expanded uncertainty will be estimated by:

With:

• an expansion factor k = 2

• σinternal repro: the standard deviation of the control chart with an internal quality control that corresponds to the uncertainty linked to random errors.

• accuracy: the standard uncertainty due to accuracy, estimated by

With:

• ref: the standard uncertainty on the reference value

• TargetCQI - ref: the mean deviation of the results obtained by the laboratory from the TITRIVIN reference value

Obviously, this approach is interesting for a laboratory because it's quick and easy to set up.
But we advise you to adopt an approach that combines the study of your intra- and inter-laboratory data. This method is far more relevant for laboratories that carry out both internal and external quality control, and has the advantage of being able to define an uncertainty over a measurement range.

.