Errors, Precision, and Accuracy Revision Notes | A Level AQA Physics

1. Why Measurement Quality Matters

Every measurement you take in physics carries some degree of uncertainty. The question is not whether your measurements are perfect — they never are — but whether you can characterise how imperfect they are and why.

At A Level, examiners expect you to go beyond recording numbers. You need to evaluate the quality of your data using precise language: terms like accuracy, precision, resolution, repeatability, and reproducibility each mean something specific, and using them correctly is what separates a Band 1 response from a Band 3 one.

This lesson builds the conceptual framework you need to analyse any experiment — whether it is your own required practical or an unfamiliar data set in a Paper 3 question.

2. Random and Systematic Errors

Picture a student measuring the period of a pendulum ten times. Some readings come out slightly high, some slightly low, and they scatter around a central value. That scatter is caused by random error — unpredictable, uncontrolled variation that affects each measurement independently.

Now imagine the student's stopwatch runs consistently slow, so every single reading is a little too large. All ten values are shifted in the same direction by the same amount. That is a systematic error — a consistent, repeatable offset that pushes every measurement away from the true value in one direction.

Because random errors scatter symmetrically, taking the mean of a large number of repeats causes the positive and negative deviations to cancel out, pulling the mean closer to the true value. This is why repeating measurements and averaging is a core experimental strategy.

Common sources of systematic error include a balance that reads 0.5 g when nothing is on it (a zero error), a thermometer calibrated incorrectly, or parallax error when a student always reads a scale from the same off-axis angle. The key diagnostic feature is that all readings are shifted the same way — the spread of data may look perfectly tight, yet every value is wrong by the same amount.

A useful mental model: think of a target at a shooting range. Random error is like shots scattered around the bullseye — imprecise but centred on the right place. Systematic error is like shots clustered tightly together but all hitting the wrong part of the target — precise but consistently off-centre.

3. Precision, Repeatability, and Reproducibility

With the two error types established, you can now build the vocabulary that describes the quality of a set of measurements. These terms are closely related but distinct — and examiners regularly set questions that require you to use them correctly.

Precision is entirely about the spread of repeated readings. A set of values that cluster tightly together is precise, regardless of whether they are close to the true value. This is why a systematically biased instrument can still produce precise results — the readings are consistent, even though they are consistently wrong.

Repeatability is essentially precision within a single experimental session. If you measure the resistance of a resistor five times in a row with the same multimeter and get very similar values each time, your measurement has high repeatability.

Reproducibility is the gold standard in science. When a result is reproducible, it means the finding is not an artefact of one particular instrument or one particular experimenter's technique. This is why peer review and independent replication are central to scientific practice.

The distinction between repeatability and reproducibility is one that students often conflate. Repeatability is about consistency within one person's work in one session; reproducibility is about consistency across different people, instruments, or laboratories. A result can be highly repeatable yet poorly reproducible if the method has a flaw that only shows up when a different instrument is used.

4. Resolution and Accuracy

Two more terms complete the framework: resolution and accuracy. These are often confused with each other and with precision, so it is worth being precise about what each one means.

Higher resolution does not automatically mean more accurate results. A digital thermometer displaying temperature to 0.001 °C has very fine resolution, but if it is poorly calibrated, every reading could still be systematically wrong. Resolution sets the limit of what you can distinguish, not the limit of how correct you are.

Accuracy and precision are independent properties. The target analogy from earlier captures this well: you can be precise without being accurate (tight cluster, wrong location), accurate without being precise (scattered readings that average to the true value), both, or neither. At A Level, you need to be able to describe which combination applies to a given data set and explain why.

5. Putting It Together: Analysing Experimental Data

Now that all five terms are defined, the real skill is applying them together when evaluating an experiment. Examiners frequently present a table of repeated measurements and ask you to comment on the quality of the data — this requires you to look at both the spread of readings and their relationship to a known true value.

Worked example: Evaluating a Set of Repeated Measurements

A student measures the acceleration due to gravity using a free-fall timer. The accepted value is $g = 9.81 \text{ m s}^{-2}$. Their five readings are: 9.74, 9.76, 9.75, 9.74, 9.76 $\text{m s}^{-2}$. Evaluate the precision and accuracy of these measurements.

1. $\bar{g} = \frac{9.74 + 9.76 + 9.75 + 9.74 + 9.76}{5} = \frac{48.75}{5} = 9.75 \text{ m s}^{-2}$ — The mean gives the best estimate of the true value from this data set. Always calculate the mean before commenting on accuracy.
2. Range $= 9.76 - 9.74 = 0.02 \text{ m s}^{-2}$ — The range is very small (0.02 m s⁻²), so the readings cluster tightly together. This indicates high precision — the measurement process is highly repeatable.
3. Deviation from true value $= 9.81 - 9.75 = 0.06 \text{ m s}^{-2}$ — The mean is 0.06 m s⁻² below the accepted value. All five readings are below 9.81 m s⁻², suggesting a systematic error — the timer or method consistently underestimates the true value.
4. All readings below true value by a consistent amount → systematic error — Because the deviation is in one direction and averaging did not bring the mean to 9.81 m s⁻², this is not random error. A likely cause is air resistance or friction in the apparatus, which consistently reduces the measured acceleration.
5. Precise (small range of 0.02 m s⁻²) but not accurate (mean deviates from true value by 0.06 m s⁻² due to systematic error) — This is the exam-standard conclusion. Use both terms explicitly and link each to the evidence from the data.

Answer: The measurements are precise (range = 0.02 m s⁻²) but not accurate (mean = 9.75 m s⁻², which is 0.06 m s⁻² below the accepted value of 9.81 m s⁻²). The consistent underestimate suggests a systematic error in the experimental method.

Notice how the worked example uses the data as evidence for every claim. Examiners award marks for statements that are explicitly supported by numbers from the question — vague comments like "the results are quite close" will not score.

6. Resolution in Practice: Choosing the Right Instrument

Resolution connects directly to the choice of instrument in an experiment. When you select a measuring device, you need to match its resolution to the scale of the changes you are trying to detect.

Consider measuring the extension of a spring under different loads. If the extensions range from 1 mm to 20 mm, a ruler with 1 mm resolution is borderline — you can just about detect the smallest extension, but your uncertainty is large relative to the measurement. A vernier calliper with 0.1 mm resolution would give you much more reliable data at the lower end of the range.

There is also a practical ceiling: using an instrument with finer resolution than the quantity warrants introduces false precision. If a quantity genuinely varies by ±0.5 cm due to random fluctuations, recording it to 0.001 cm implies a certainty that the data does not support. The resolution of your instrument should be appropriate to the precision achievable by the overall method.

7. Exam Technique and Common Mistakes

The vocabulary in this topic is precise, and examiners are strict about it. The most common way students lose marks is by using these terms interchangeably or by giving definitions that are too vague to earn credit.

For extended-response questions asking you to evaluate experimental data, structure your answer as: (1) state the precision using the range of readings, (2) state the accuracy by comparing the mean to the true value, (3) identify the type of error responsible for any offset or scatter, and (4) suggest a specific, actionable improvement. This four-part structure reliably covers the marking points.

Mastering the language of errors and measurement quality is not just about definitions — it is the foundation for every evaluation question you will encounter at A Level, from required practical write-ups to unseen data analysis in Paper 3. The five terms covered here (random error, systematic error, precision, accuracy, resolution, repeatability, reproducibility) form a complete toolkit for describing and improving any experimental measurement.