Accuracy versus precision is the distinction between being right and being consistent, and resolution is a third thing entirely. An instrument can be precise and wrong, accurate but coarse, or finely resolved and both wrong and erratic. Knowing which property is limiting a measurement tells you what to change.
Three properties that fail independently
Accuracy describes how close a reading is to the true value. Precision describes how closely repeated readings of the same quantity agree with one another. Resolution describes the smallest change in the quantity that the instrument can show at all. They are independent. A thermometer reading consistently two degrees high is precise and inaccurate. An instrument whose readings scatter widely around the correct value is accurate on average and imprecise. An instrument that displays in whole units cannot show a change smaller than one unit no matter how good its accuracy or precision. Diagnosing a measurement problem starts with working out which of the three is actually limiting you, because the corrective actions have nothing in common. Datasheets reinforce the confusion by listing all three together in a single specification block, which invites the assumption that a strong figure in one column implies strong figures in the others. It does not, and the three are frequently limited by entirely different parts of the instrument.
Why fine resolution feels like accuracy
Resolution is visible and immediate, so it dominates our impression of an instrument. A display that changes smoothly and shows small increments feels authoritative in a way that a coarse one does not, and that feeling is not evidence of anything. Resolution is a design choice about how finely the instrument divides its range. Accuracy is a separate claim, stated in the specification, about how far from the truth the reading may be. It is entirely normal for the accuracy specification to be many times larger than the smallest displayed increment, which means the last figures on the display are not carrying information about the quantity being measured. Reporting them implies a confidence the specification does not support.
When precision alone is exactly what you need
There is a large class of useful measurements where accuracy hardly matters. Comparing two components to find the odd one out, watching a value change while you adjust something, checking whether a reading is drifting, detecting that a temperature is rising, or finding the peak of a response all rely on precision and resolution rather than on absolute correctness. A consistent offset simply cancels out of the comparison. Recognizing that your question is comparative rather than absolute is liberating, because it means an inexpensive instrument can do the job well. It is also a trap in the other direction: results obtained that way must not later be quoted as absolute values, because nothing in the measurement established what the true value was.
Detecting which property is failing
The diagnostic procedure is simple. Take the same measurement repeatedly without changing anything and look at the spread; that characterizes precision. Then measure something whose value you know independently, such as a reference component or a calibration source, and compare; that characterizes accuracy at that point. Finally, make a deliberate small change to the quantity and see whether the display responds; that characterizes resolution in the situation you are actually in. A few minutes of this reveals more about an instrument than reading its datasheet, because it includes your leads, your connections and your environment. Repeat the accuracy check at more than one point, since offset and gain errors behave differently across the range.
Improving each one requires different action
Poor precision is addressed by averaging, by improving contacts, by shielding against interference, and by controlling how the measurement is set up each time. Poor accuracy is addressed by calibration, by correcting for known systematic effects such as lead resistance or loading, and by using a better reference. Insufficient resolution is addressed by changing range, by using an instrument with more resolution, or by amplifying the quantity before measuring it. Applying the wrong remedy is common and wasteful: averaging a thousand readings will not correct a calibration offset, and recalibrating will not quiet a noisy connection. Decide which property is limiting you first, then spend effort on that one alone.
What to check before you act
- Repeat a measurement unchanged to see the spread before blaming accuracy.
- Compare against a known reference to test accuracy at more than one point.
- Compare the accuracy specification against the smallest displayed increment.
- Decide whether your question is comparative or absolute before choosing an instrument.
Common questions
Can averaging make a measurement more accurate?
It improves precision by reducing random scatter, which can improve the accuracy of the average if the errors are random. It does nothing about a systematic offset.
Is a more expensive instrument always more accurate?
Not necessarily. Price often buys resolution, features and stability. Compare the published accuracy specifications over the range and conditions you will use.
The practical takeaway
Separate the three properties before acting. Repeat a reading to test precision, check against a known value to test accuracy, and compare the specification against the display’s finest increment to see how much of what you are reading is real. The remedy follows directly from which one is limiting you.
References and further reading
- CDC/NIOSH: electrical hazards and safe work practices
- NFPA: electrical safety resources and code background
- Department of Energy: comparing published performance figures when shopping
Numerical scenarios are illustrative unless identified otherwise. Follow the exact product instructions; component ratings and local installation requirements can differ.
