Inquiry Question 3: How is the integrity of a scientific investigation judged?
Evaluate scientific investigations and findings in terms of reliability, validity, accuracy and precision of data
A focused answer to the HSC Investigating Science Module 5 dot point on reliability, validity, accuracy and precision. The four concepts every Investigating Science student must distinguish, with worked HSC past exam questions.
Reviewed by: AI editorial process; not yet individually human-reviewed
Have a quick question? Jump to the Q&A page
Jump to a section
What this dot point is asking
NESA wants you to distinguish reliability, validity, accuracy and precision, apply each to a given investigation, and identify which property of an investigation is in question when results disagree or fail to replicate. These four terms are the most heavily tested vocabulary in Investigating Science.
The answer
Four properties describe the quality of a scientific investigation and its data. They are independent: an investigation can be high on some and low on others.
Validity
The extent to which an investigation tests what it claims to test. A valid investigation:
- Uses appropriate variables and a clearly defined dependent variable.
- Holds controlled variables constant so the result is attributable to the independent variable.
- Includes a control group.
- Uses appropriate sample sizes and randomisation.
- Avoids confounders.
Threats to validity. Confounding variables, sampling bias, measurement instruments that do not measure what is claimed.
Reliability
The extent to which an investigation produces consistent results when repeated. A reliable investigation:
- Yields similar measurements when repeated by the same researcher.
- Yields similar measurements when repeated by different researchers (reproducibility).
- Has tightly clustered repeat measurements.
Threats to reliability. Random error, inconsistent technique, unstable equipment.
Accuracy
How close a measurement is to the true or accepted value. A measurement that consistently disagrees with the true value by a fixed amount is inaccurate, even if it is precise.
Threats to accuracy. Systematic error, calibration drift, observer bias.
Precision
How close repeated measurements are to each other, regardless of whether they hit the true value. Precision describes the spread of a measurement set.
Threats to precision. Random error, low-resolution instruments, careless technique.
The four together
Think of a dart board.
- High accuracy and high precision. All darts in the bullseye.
- High precision, low accuracy. All darts tightly clustered, but off-centre.
- High accuracy, low precision. Darts scattered around the bullseye on average.
- Low accuracy, low precision. Darts scattered randomly.
How to improve each
| Property | How to improve |
|---|---|
| Validity | Better experimental design; control more variables; add a control group |
| Reliability | More replicates; standardise procedure; use trained operators |
| Accuracy | Calibrate instruments; use a reference standard; remove systematic error |
| Precision | Use higher-resolution instruments; refine technique; reduce random variation |
Examples in context
Example 1. NSW EPA air-quality monitoring at Liverpool. The NSW Environment Protection Authority deploys reference-grade PM2.5 monitors at urban stations including Liverpool. To assess reliability, technicians collocate two identical TEOM analysers at the same site for one month and check whether hourly readings agree within 2 micrograms per cubic metre. Validity is checked against the European reference gravimetric method by running parallel filter samples for 24-hour periods. Accuracy is established by introducing certified PM2.5 reference aerosols through the instrument inlet and comparing readings to known concentrations. Precision is gauged from the standard deviation of repeated 1-hour averages during stable atmospheric conditions. Failing any one of these checks triggers calibration or replacement before the data enters the NSW Air Quality Index.
Example 2. Sydney Water reservoir thermometer drift. Sydney Water staff at Warragamba Dam noticed that one of three water-quality thermometers consistently read 1.2 degrees C higher than the other two during weekly cross-checks. The faulty unit was precise (within 0.1 degrees C across repeated readings) but inaccurate. Without the cross-check, the inaccurate readings would have entered the algal-bloom early-warning system, potentially triggering false alerts. The case shows why reliable, precise instruments are not enough; periodic calibration against a traceable reference standard is essential, and Sydney Water schedules quarterly recalibration of all temperature, pH and turbidity sensors against National Measurement Institute-traceable standards.
Try this
Q1. A school weather station records daily maximum temperature for one term. Repeated readings on the same calm day give 22.1, 22.2, 22.1, 22.2 degrees C, but the regional Bureau of Meteorology station 2 km away records 24.5 degrees C. Identify which of reliability, validity, accuracy and precision are satisfied, and which are not. [4 marks]
- Cue. Reliable: yes (consistent). Precise: yes (tight cluster). Accurate: no (offset of 2.3 degrees from BoM reference). Valid: questionable until siting is checked (shading, ground surface).
Q2. A clinical trial reports a mean blood pressure reduction of 3 mmHg with error bars of plus or minus 5 mmHg, with 40 participants. Discuss whether this result is reliable and whether it supports the trial's claim. [3 marks]
- Cue. Effect smaller than uncertainty; not statistically distinguishable from zero. Reliability of individual measurements may be fine, but the sample is underpowered to detect a 3 mmHg effect.
Q3. An NSW EPA report compares three air-quality monitoring methods at a single industrial site. (a) Define accuracy and precision. (b) Identify which property is improved by collocating duplicate monitors. (c) Identify which property is improved by calibrating against certified reference aerosols. [2+2+2 marks]
- Cue. (a) Accuracy: closeness to true value; precision: closeness of repeats. (b) Reliability and precision. (c) Accuracy.
Exam-style practice questions
Practice questions written in the style of NESA exam questions on this dot point, with worked answer explainers. The year tag is the paper they imitate, not the source.
2023 HSC4 marksDistinguish between reliability and validity in a scientific investigation. Use an example to illustrate each.Show worked answer →
A 4-mark answer needs clear definitions, the relationship between the two, and a worked example for each.
- Reliability
- The extent to which an investigation produces consistent results when repeated under the same conditions. A reliable investigation can be repeated by the same or different researchers and yield similar measurements.
- Validity
- The extent to which an investigation tests what it was intended to test. A valid investigation measures the dependent variable as a true response to the independent variable, with controlled variables held constant.
- Worked example for reliability
- A student measures the boiling point of water at sea level five times and gets 100.1, 100.0, 100.2, 100.1, 100.0 degrees Celsius. The investigation is reliable: the values are tightly clustered, so repeated measurements yield similar results.
- Worked example for validity
- A student wants to test whether sunlight affects plant growth. If different plant species are placed in sun and shade, the results are not valid because species variation is a confounder. The same species, soil and water schedule are required for the experiment to be valid.
- Relationship
- An investigation can be reliable without being valid (consistently wrong) but cannot be valid without being reliable. Markers reward both definitions and worked examples.
2022 HSC3 marksExplain the difference between accuracy and precision, with an example.Show worked answer →
A 3-mark answer needs both definitions and a target-style example.
- Accuracy
- How close a measurement is to the true or accepted value.
- Precision
- How close repeated measurements are to each other, regardless of the true value.
- Example
- A student measures the density of water four times:
- Set A: 1.00, 1.00, 1.01, 1.00 g/mL. Both accurate (true value is 1.00) and precise.
- Set B: 0.85, 0.86, 0.85, 0.86 g/mL. Precise (tightly clustered) but not accurate (systematic error).
- Set C: 0.90, 1.05, 0.95, 1.10 g/mL. On average accurate but not precise.
Relationship. Precision relates to random error; accuracy relates to systematic error (calibration, bias). A balance reading consistently 0.1 g too high is precise but not accurate. Improving precision requires more careful technique. Improving accuracy requires calibration.
Markers reward both definitions and an example that shows the two are independent properties of a measurement set.