The rule
percent error = |measured − true| ÷ |true| × 100
Why it works
Percent error compares the size of a mistake with the size of what was being measured. A 3-unit error means very different things when measuring 50 and when measuring 50,000, and dividing by the true value is what makes them comparable.
How to do it by hand
- Subtract the true value from the measured one
- Take the absolute value if you only want the size
- Divide by the true value
- Multiply by 100
What is worth knowing
The denominator is always the true value, never the measured one — that is what makes the figure meaningful, and swapping them is the usual mistake. Keeping the sign is often worth doing: a consistently positive error points at a systematic bias in the instrument, which is fixable, while errors scattered either side are random noise, which is not. In experimental work a percent error under 5 % is usually considered acceptable, though the threshold depends entirely on the field: analytical chemistry expects far better and field ecology far worse.