Accuracy is influenced by both precision and bias
in the measurement process. To understand accuracy, one must understand
bias, systematic measurement error, random measurement error, measurement
error, and precision.
Manufacturer's accuracy specifications may be
misleading as they may not tell you how the measurement and test equipment
actually performs during calibration and use. We have
observed numerous instruments that do not meet the stated accuracy
If we only threw one dart and it hit the
target, this may be a single measurement bliss scenario. However,
multiple darts reveal random measurement error to be large (each additional
throw was about 160 mm off target in opposite directions). Did the manufacturer
use repeated measurements, or the first measurement when they wrote the
specification? If they used the mean of -160 mm, 0 mm and + 160
mm, the average would be zero.
|Accurate if we only look at the mean |
When we plot the precision of
the person throwing darts graphically, we can see the average would lie on our
target value. The standard deviation of these throws combined with
the bias (mean of the repeated throws) will give us the true accuracy.
|Precise throw with some bias.|
If we throw three darts and
miss the target by +50-60 mm, and the darts are about 10-15 mm apart, we can
say the precision is around 2.5 mm (standard deviation of 10, 12.5, 15 mm).
If the shooter can adjust their aim 60 mm to the left, with the same
throwing precision, they will hit the target with all three darts.
In the graph above the accuracy can be stated
by the lower, and upper tolerances. All measurements made fall into the
tolerance specifications. The manufacturer's accuracy specification would
be credible. In this scenario, it may be very easy to make
adjustments, which will lower the measurement error.
What is Accuracy?
- Accuracy is the function of bias and precision (estimate of systematic measurement error)
- The Qualitative term for the extent of the approximation of the measurement result to a "true value" (DIN55350)
- The closeness of agreement between a measured quantity value and the true quantity value of a measurand (VIM 2.13)
- Accuracy is a qualitative term as it is defined as a reference to a true value which is unknown
What is Bias?
- Measurement bias (bias): estimate of systematic measurement error (VIM 2.18)
- Bias is a Qualitative term
- Bias is how close something is to the true value
- Bias is often characterized by finding the mean of repeated measurements
What is Systematic Measurement Error?
Measurement Error: component of measurement error that in replicate
measurements remains constant or varies in a predictable manner (VIM
What is Measurement Error?
- Measurement Error: measured quantity value minus a reference quantity value (VIM 2.16)
What is Random Measurement Error?
Measurement Error: a component of measurement error that in replicate
measurements varies in an unpredictable manner (VIM 2.19)
- Note: Random Measurement Error equals Measurement Error minus Systemic Measurement Error
What is Precision?
- Precision is a measure of spread.
- Precision refers to the repeatability of the measurement
of agreement between indications or measured quantity values obtained
by replicate measurements on the same or similar objects. (VIM 2.15)
precision is used to define measurement repeatability, intermediate
measurement precision, and measurement reproducibility.
- Precision is often characterized by finding the standard deviation of multiple measurements
Regardless of the manufacturer's specification
sheet, it is important to have your equipment calibrated the way it is being
used. Best practice is to put in place a measurement assurance system
(practices put in place to monitor a testing or calibration process and to
ensure the calibration status of equipment, reference standards or reference
materials used in the measurement process), and have your equipment calibrated
at the proper intervals.
written by Henry Zumbrun
Please click here to subscribe to our newsletter.