Precision and Accuracy (SIN 701)

Precision and accuracy answer different questions. Accuracy asks: how close is a measurement to the true, correct value? Precision asks: how close are repeated measurements to each other, regardless of whether they're right? A scale that reads 2 grams too heavy every single time is perfectly precise and totally inaccurate, but you can fix it with a correction factor once you know the offset. A scale that wobbles randomly is useless no matter how you average it, because you can't predict which way it will be wrong next time. That's why a consistent instrument, even a wrong one, often beats an erratic one: consistent error is correctable, scatter is not.

The classic picture is a dartboard. Darts clustered tightly in one corner, far from the bullseye, show high precision and low accuracy. Darts scattered all over the board but centered on the bullseye on average show low precision and reasonable accuracy on average, though no single throw was trustworthy. In a data table, you check precision by looking only at the repeated trials themselves: find the range or spread, ignoring the true or accepted value entirely. You check accuracy by comparing the mean of those trials to the known or certified value.

This idea sits right behind experimental design in the science section: once you understand how to judge a single instrument's readings, you can evaluate whether a method, a setup, or a recalibration improves data quality or just hides error. A mean that matches the true value is not automatic proof of good measurement, which connects directly to spotting flawed study designs elsewhere in this unit.