Evaluating Measurement Quality and Improving Accuracy

Hello again. In the previous lesson, you practised handling measurement values correctly: SI prefixes, cubic-unit conversions, scientific notation, and significant figures. Those skills make your numerical answers defensible. This lesson asks a prior question: are the measurements themselves good enough to trust?

For a density practical, poor mass or volume measurements can produce a poor density even if the calculation is perfect. You will learn to choose suitable instruments, distinguish accuracy from precision, spot anomalous results, and give specific improvements that earn evaluation marks.

---

1. Choose an instrument that fits the measurement

A measuring instrument is not simply “good” or “bad.” It is suitable only if its range and resolution fit the job.

• Range is the largest and smallest values an instrument can measure.
• Resolution is the smallest change that the instrument can show. It is often the size of the smallest scale division, or the final displayed digit on a digital instrument.

For example, suppose you need $5\mathrm{c}{\mathrm{m}}^3$ of water. A $100\mathrm{c}{\mathrm{m}}^3$ measuring cylinder can measure that volume, but its scale is relatively coarse. A $10\mathrm{c}{\mathrm{m}}^3$ measuring cylinder gives a more detailed reading and is therefore the better choice.

Likewise:

Measurement needed	Less suitable choice	More suitable choice	Why
$5\mathrm{c}{\mathrm{m}}^3$ liquid	$100\mathrm{c}{\mathrm{m}}^3$ measuring cylinder	$10\mathrm{c}{\mathrm{m}}^3$ measuring cylinder	Finer graduations for a small volume
$20\mathrm{cm}$ length	$20\mathrm{m}$ tape measure	$30\mathrm{cm}$ ruler	Scale is better matched to the length
$12.46\mathrm{g}$ mass	Balance reading nearest $1\mathrm{g}$	Digital balance reading nearest $0.01\mathrm{g}$	Much finer resolution
Diameter of a small sphere	Metre rule	Vernier calipers, if available	Designed for small diameters

Resolution sets a limit on what you can claim

A ruler with millimetre divisions lets you read length more finely than one with only centimetre divisions. You may usually estimate between scale marks on an analogue instrument, but you cannot honestly report a wildly precise value that the scale cannot support.

For an analogue scale, a common school-level uncertainty rule is:

$$\text{reading uncertainty}\approx \pm \frac{1}{2}\times \text{smallest scale division}$$

So a ruler marked every $1\mathrm{mm}$ has an estimated reading uncertainty of about:

$$\pm 0.5\mathrm{mm}$$

For a digital instrument, its final displayed place indicates its resolution. A balance that reads $23.46\mathrm{g}$ resolves $0.01\mathrm{g}$, whereas one that shows $23\mathrm{g}$ resolves only $1\mathrm{g}$.

A final warning: finer resolution can improve the precision of the reading, but it does not automatically guarantee that the result is accurate. A highly detailed balance that has not been zeroed can give very consistent but wrong measurements.

---

2. Accuracy and precision: two different questions

These terms are commonly confused because both describe data quality.

Accuracy asks:

Is the measurement close to the true or accepted value?

Precision asks:

Are repeated measurements close to one another?

Imagine the accepted density of a material is $2.70g/c{\mathrm{m}}^3$. Compare these sets of repeated student results:

Results in $g/c{\mathrm{m}}^3$	Accurate?	Precise?	Reason
2.69, 2.70, 2.71	Yes	Yes	Clustered near $2.70$
2.10, 2.11, 2.10	No	Yes	Clustered, but consistently too low
2.55, 2.71, 2.83	Roughly yes on average	No	Spread out, though centred near the accepted value
2.10, 2.56, 3.02	No	No	Spread out and not near the accepted value

The second row is a classic test question. The values are precise but inaccurate. This often points to a systematic error: something has shifted every result in the same direction.

You can judge precision just by comparing repeats. You can judge accuracy only if you have something trustworthy to compare against, such as an accepted value, a calibrated standard, or a correctly measured reference object.

The source of the problem matters

Two broad error patterns help you decide on an improvement.

Random errors make results vary unpredictably. They cause a wider spread of repeated readings. Examples include judging a liquid level slightly differently each time or changing eye position when reading a scale.

Systematic errors shift every measurement in the same direction by a similar amount. Examples include:

• using a balance that was not zeroed before weighing;
• measuring from the worn end of a ruler rather than its zero mark;
• using equipment with an uncorrected calibration error.

Repeating measurements helps deal with random error because averaging makes individual high and low variations less influential. Repeats do not remove a systematic error: averaging several wrongly zeroed-balance readings still gives a wrongly high mass.

---

3. Read scales properly: avoid parallax

Parallax error happens when you view a scale from an angle rather than directly in front of it. Your line of sight makes the object appear aligned with the wrong scale mark.

For a ruler, put the eye directly above the mark being read. For a measuring cylinder, put the eye level with the liquid surface and read the bottom of the curved meniscus for water-like liquids.

If a student looks from a different angle on each repeat, their readings may be scattered, producing random error. If they always use the same incorrect viewing angle, the error may become a consistent bias. In either situation, the practical improvement is specific:

Read the scale at eye level to avoid parallax error.

This is much stronger than writing “be more careful,” because it identifies both the error and the method for reducing it.

---

4. Repeats, means, and anomalous results

A result is anomalous when it does not fit the pattern of the other results collected under the same conditions. It is also called an outlier.

Consider five repeated measurements of a block’s volume:

$$12.4,12.5,18.9,12.6,12.5\mathrm{c}{\mathrm{m}}^3$$

The value $18.9\mathrm{c}{\mathrm{m}}^3$ is anomalous because it is far from the closely grouped values near $12.5\mathrm{c}{\mathrm{m}}^3$.

An anomaly might come from a genuine one-off mistake, such as:

• reading the scale from the wrong mark;
• recording $18.9$ instead of $12.9$;
• spilling water during a displacement measurement;
• failing to reset equipment for one trial.

However, do not simply delete any inconvenient value. A scientifically sound response is:

1. Identify the anomalous result.
2. Check for an identifiable mistake.
3. Repeat that measurement.
4. Exclude it from the mean only if there is good reason to treat it as an error, or the repeat confirms it does not represent the usual pattern.

For the volume results above, you would repeat the $18.9\mathrm{c}{\mathrm{m}}^3$ measurement. If the repeat is $12.5\mathrm{c}{\mathrm{m}}^3$, there is strong evidence that $18.9\mathrm{c}{\mathrm{m}}^3$ was a one-off error.

The mean of the four consistent readings is:

$$\frac{12.4+12.5+12.6+12.5}{4}=12.5\mathrm{c}{\mathrm{m}}^3$$

That mean is more reliable than using one reading alone.

Reliability, repeatability, and reproducibility

These terms may appear in a longer evaluation question.

• Reliable data: results are consistent enough that you can have confidence in the conclusion. Repeats and a mean usually improve reliability.
• Repeatable: the same person, equipment, and method produce similar results when repeated.
• Reproducible: a different person or different equipment can follow the method and obtain similar results.

For a short density test answer, “repeat measurements and calculate a mean” is usually an excellent improvement for reliability and random error. If asked how to test reproducibility, say that another student should repeat the investigation using the same written method.

---

5. Turning weaknesses into high-mark improvements

Evaluation questions reward a precise chain of reasoning:

Problem + specific improvement + why it helps

Use this structure rather than vague statements.

Weakness or error	Specific improvement	Why it helps
Volume is $5\mathrm{c}{\mathrm{m}}^3$, measured using a $100\mathrm{c}{\mathrm{m}}^3$ cylinder	Use a $10\mathrm{c}{\mathrm{m}}^3$ cylinder	Finer graduations reduce reading uncertainty
Balance was not checked before use	Tare or zero the balance before every mass measurement	Removes a systematic zero error
Measurements vary widely	Repeat each reading at least three times and calculate a mean	Reduces the effect of random error
Ruler read at an angle	View the scale at eye level	Reduces parallax error
One result is far from the others	Repeat the measurement and investigate the cause	Checks whether it is a genuine anomaly
Students use different techniques	Use a detailed, standardised method	Makes measurements more consistent and repeatable

Avoid these weak answers:

• “Use better equipment.”
• “Be more accurate.”
• “Do it again.”

Replace them with a named instrument or method and an explanation. For instance:

Use a digital balance that reads to $0.01\mathrm{g}$, rather than a balance reading to the nearest gram, to reduce the uncertainty in the mass measurement.

Or:

Repeat the volume measurement three times and calculate a mean, excluding a result only after repeating it and confirming it is anomalous; this reduces the effect of random error.

A rapid test-day checklist

When you see a table of repeated results or a practical method, ask:

1. Are the repeats close together? If yes, they are precise.
2. Is an accepted value given? If yes, compare to it to judge accuracy.
3. Is one value far from the rest or off the trend? It may be anomalous.
4. Is the instrument’s scale sensible for the size measured? If not, choose a smaller-range, finer-scale instrument.
5. Could the problem affect every reading similarly? Think systematic error: zero, calibration, or a consistent technique error.
6. Could the problem vary from trial to trial? Think random error: repeat and calculate a mean.
7. Can you state exactly how your change improves the data? Add that reason to secure the explanation mark.

---

You can now evaluate measurement quality rather than merely reading numbers. Choose equipment with an appropriate range and fine enough resolution; remember that accuracy means close to the true value, while precision means repeats close together; investigate anomalies rather than automatically deleting them; and match each error to a practical improvement.

Next, you will classify quantities as scalars or vectors and represent vectors with correctly labelled arrows.