When taking measurements, there is often much confusion over the difference between being accurate and being precise. The two words are often used as synonyms for one another, yet the two actually have quite different meanings.
When talking about accuracy, we are referring to how true to life the measurement is – how close the measurement taken is to the actual value. Over a set of several measurements, the mean value of the results is used to determine accuracy. Results close to, but either side of the true measurement will produce a mean average close to the true value. Several near identical results that all lie to one side of the true measurement will produce a mean average that is further away from the true value and are therefore less accurate.
Precision is measured by taking several measurements of the same thing, with the same instrument and looking at the variance between them. The lower the standard deviation of a group of measurements (i.e. the closer together they are in value), the more precise it is. It is important to note that precision looks only at the distribution of results, and does not take into account how close the measurement is to its true value (i.e. its accuracy).
The shooting range example
It is possible for a measurement to be very accurate, but not precise, or very precise, but not accurate. The analogy of a series of shots on a target is frequently used as a way of explaining the difference between the two. If ten shots were taken, we would look at accuracy in terms of how close they were to the centre of the target, and precision, in terms of how close the shots were to one another.
1/ For the shots to be considered both accurate and precise, they would need to all be close to the centre of the target, and therefore to one another.
2/ If all the shots are clustered together but to one side of the target, the shots are precise, as they are all showing similar results. As the shots are not clustered around the centre of the target, they are not accurate.
3/ If all of the shots were relatively evenly distributed around the centre of the target, but not really that close to one another, we would say that the shots were accurate, as on average they were on target. As the shots are widely distributed, it is not very precise.
4/ If ten shots were taken and they were all scattered across one half of the target, they are neither accurate nor precise.
Tags: quality control , scanning technology
Post a comment: