|
Precision vs. Accuracy |
Precision and accuracy are not the same thing. Precision refers to the degree of specified detail which can be observed; while accuracy refers to the truthfulness, or correctness, of the specified data. Here's an example: The TV weather forecaster says that it will be between 40 and 60 degrees today. The actual reading turns out to be 53. Thus, the forecast was accurate, but not very precise. The forecaster provided a true statement but without enough detail for us to make plans. For tomorrow, the forecast is 52.47 degrees at 4 PM. It turns out to be 61 degrees. This forecast was very precise, but completely inaccurate.
So what good is one without the other? Not much! The above examples show that precision is useless without accuracy, but that accuracy with little precision does not tell much either.
In electronic test equipment, manufacturers attempt to balance the two factors. For example, a top quality 3½ digit DMM is accurate (truthful) enough for the smallest (or "least significant") digit to mean something. A careful review of the specifications on some units, however, can reveal that the stated precision, unsupported by the basic accuracy of the unit, is something of a marketing gimmick. Other factors, such as input impedance (which, if too low, can load the circuit under test), and the linearity of the analog to digital conversion process, can also undermine accuracy.
One must also be careful not to imply more accuracy than is warranted by the measurements when reporting calculated data. For example, suppose two measurements are made (perhaps voltage and current) to 1% accuracy or three digits of resolution. A result computed from these values (such as dividing to find resistance) can not be more accurate than the underlying data, and reporting that result to five or six digits of resolution (precision) creates a false impression. Don't be trapped into believing that because your pocket calculator displays eight decimal places of results (precision) that all eight places are accurate!
In general, then, it is good laboratory practice to measure quantities to the greatest degree of both accuracy and precision that the test equipment will permit. Report measured data and results computed from these measurements to consistent degrees of precision (resolution). And never, never believe that because your calculator shows you an answer it must be true!
