Key Topics

Types of error

There are two types of error: random errors and systematic errors. It is the presence of these errors that causes the uncertainty in measurement results. The aim of an uncertainty estimate is to combine the effect of all the sources of error in a measurement process to give a single value which indicates how sure the analyst is about the result reported. Note that in this context ‘error’ does not mean a mistake. Measurement uncertainty should reflect the variation in results when the method is being operated correctly (i.e. is operating under statistical control) so the effect of mistakes or ‘gross errors’ is not included in the uncertainty estimate.

Random errors cause results to vary in an unpredictable way from one measurement to the next. They are caused by chance variations in the measurement process that are outside of the analyst’s control. Random errors are present when any measurement is made but due to their unpredictable nature, they cannot be corrected for. The effect of random errors can be reduced, however, by making more measurements and reporting the mean. The magnitude of the random errors will determine the precision of the measurement results.

Systematic errors cause results to differ from the true value by the same amount (and in the same direction) each time a measurement is made. Systematic errors are not reduced by making repeat measurements as each measurement will be in error by the same amount. In some cases it may be possible to correct results to remove systematic errors. The magnitude (and direction) of systematic errors will determine the bias in the measurement results.