12.2 - Accuracy, Precision & Error
Accuracy and precision
| Term | What it means |
|---|---|
| Accurate | Close to the true value |
| Precise | Repeated readings are close to each other (a small spread) |
A set of results can be precise without being accurate. For example, weighing a 50.0 g mass on a balance not set to zero might give 52.0 g, 52.1 g, and 52.0 g. These readings are precise (they barely differ), but none is accurate (each is about 2 g too high).
Resolution
- Resolution is the smallest change an instrument can detect.
- A balance reading to 0.01 g has a higher resolution than one reading to 0.1 g.
- Use "higher resolution" for an instrument, not "more precise" - precise describes a set of readings that cluster closely, not an instrument.
Repeatable and reproducible
- Repeatable - You get similar results when you repeat the investigation yourself, with the same method and the same equipment.
- Reproducible - Someone else gets similar results, or you get them using a different method or different equipment.
- Reproducible results give more confidence than repeatable ones, because a similar result from a different set-up rules out a fault peculiar to one person or one piece of equipment.
Random and systematic error
- Random error - An unpredictable difference between readings, such as slight differences in judgement when reading a scale.
- Systematic error - A fault that shifts every reading by the same amount in the same direction, such as a balance not set to zero (a zero error).
How errors link to precision and accuracy
- Random error scatters readings, so it reduces precision. Taking repeat readings and calculating the mean reduces the effect of random error.
- Systematic error shifts every reading the same way, so it reduces accuracy. Taking a mean doesn't reduce systematic error; the method or equipment must be corrected.
- A set of results that is precise but not accurate indicates a systematic error.
TermDefinition
Accurate
Precise
Resolution
Repeatable
Reproducible
Random error
Systematic error
Zero error