Why Scientists Improve Experimental Procedures

An experimental procedure is a set of instructions describing how an investigation will be carried out.

A good procedure should allow scientists to collect evidence that is:

• Relevant to the research question.
• Reliable.
• As accurate and precise as reasonably possible.
• Consistent.
• Reproducible.
• Safe.

Scientists rarely assume that their first procedure is perfect. After completing an investigation, they examine the method and ask:

What could have affected the results, and how could the investigation be improved?

This process is called evaluation.

---

Identifying Weaknesses

A weakness is a feature of an experimental method that may reduce the quality of the evidence.

Common weaknesses include:

• Too few trials.
• Small sample size.
• Poor control of variables.
• Inappropriate measuring equipment.
• Inconsistent procedures.
• Human reaction time.
• Subjective measurements.
• Poorly defined endpoints.
• Large measurement uncertainty.
• Equipment that has not been calibrated.
• An inappropriate range of values.
• Missing control conditions.
• Unclear instructions.

Identifying a weakness is only the first step.

A strong evaluation also explains:

how the weakness affects the results

and:

how the procedure could realistically be improved.

---

The Weakness–Effect–Improvement Approach

A useful way to evaluate a procedure is:

Weakness → Effect on data → Specific improvement → Why it helps

For example:

Weakness: A student uses a handheld stopwatch to measure a very short travel time.

Effect: Human reaction time introduces variation into the measurements.

Improvement: Use electronic light gates.

Why: Light gates automatically detect when the object passes and reduce timing variation caused by human reaction time.

This is much stronger than simply writing:

"Use better equipment."

---

Avoid Vague Improvements

Weak scientific evaluations often contain statements such as:

• Be more careful.
• Do it better.
• Use better equipment.
• Take better measurements.
• Avoid mistakes.
• Repeat it more.
• Control everything.

These statements do not explain exactly what should change.

Instead, recommendations should be:

specific + realistic + scientifically justified

---

Example: Timing a Toy Car

Suppose students investigate:

How does ramp height affect the time taken for a toy car to travel 2.0 m?

They release the car by hand and use a handheld stopwatch.

Possible weaknesses include:

• Human reaction time.
• Inconsistent release of the car.
• Too few trials.
• Inaccurate measurement of ramp height.
• Changes in the surface.
• The car not travelling in a straight line.

---

Improving the Toy-Car Investigation

Weakness: Hand Timing

Human reaction time affects the start and stop of the stopwatch.

Improvement: Use electronic light gates or motion sensors.

Reason: Automatic timing reduces human reaction-time error.

---

Weakness: Hand Release

A student might accidentally push the car.

Improvement: Use a fixed release mechanism such as a gate or barrier that can be removed without pushing the car.

Reason: Each trial begins under more consistent conditions.

---

Weakness: One Trial

A single measurement may be affected by random variation.

Improvement: Repeat each ramp height at least three times and calculate a mean.

Reason: Repetition helps identify unusual results and reduces the influence of random variation on the final value.

---

Measurement Quality

Better procedures often begin with better measurements.

Scientists should consider:

• What quantity is being measured?
• What instrument is being used?
• What is the instrument's resolution?
• Is the instrument appropriate for the expected measurement?
• Has it been correctly zeroed or calibrated?
• Is the same technique used every time?

Choosing an appropriate instrument can significantly improve data quality.

---

Instrument Resolution

Resolution is the smallest change an instrument can display or distinguish.

Suppose two rulers are available.

Ruler A

Smallest division:

1 cm

Ruler B

Smallest division:

1 mm

For measuring the length of a small object, Ruler B allows finer measurements.

However, greater resolution does not automatically make a measurement accurate. The instrument must also be used correctly.

---

Choosing Suitable Equipment

Different measurements require different instruments.

The most suitable instrument depends on the required range and precision.

---

Measuring Volume

Suppose students need exactly 25.0 mL of solution.

Using a large beaker marked every 50 mL would provide poor measurement quality.

A more appropriate instrument might be:

• Measuring cylinder.
• Pipette.
• Burette.

The choice depends on the investigation and required precision.

---

Reading a Meniscus

When measuring the volume of many liquids in laboratory glassware, the surface curves.

This curved surface is called the:

meniscus

For water and many aqueous solutions, the measurement is generally read from the bottom of the meniscus, with the eye level with the liquid surface.

Viewing from above or below can introduce:

parallax error

---

Parallax Error

Parallax error occurs when a scale is viewed from an incorrect angle.

This can affect measurements from:

• Rulers.
• Measuring cylinders.
• Analogue meters.
• Thermometers with scales.

A simple improvement is:

view the scale directly at eye level and perpendicular to the scale where appropriate.

---

Zeroing Equipment

Some instruments should be checked or zeroed before measurements begin.

For example, a balance should normally read:

0.00 g

before an object is measured.

If it reads:

2.00 g

when empty, every measurement could be systematically too high.

Repeated trials will not solve this problem.

The procedure should instead include:

check and zero the balance before collecting data.

---

Calibration

Calibration involves checking an instrument against known standards or reference values and adjusting or accounting for its response where appropriate.

Examples include:

• Checking temperature probes.
• Calibrating pH meters.
• Checking electronic balances.
• Calibrating sensors.

Calibration can help reduce systematic measurement problems.

---

Random Error

Random error produces unpredictable variation between measurements.

Possible causes include:

• Human reaction time.
• Small environmental fluctuations.
• Natural biological variation.
• Slight differences in release technique.
• Small fluctuations in instrument readings.

Random error cannot usually be eliminated completely.

However, its influence can often be reduced.

---

Reducing Random Error

Possible improvements include:

• Repeat measurements.
• Increase sample size.
• Calculate a mean.
• Use automated measurement equipment.
• Standardize procedures.
• Maintain stable environmental conditions.

The correct improvement depends on the source of variation.

---

Systematic Error

Systematic error causes measurements to be consistently shifted in a particular direction.

Examples include:

• A balance incorrectly reading +3 g.
• A thermometer consistently reading 2°C too high.
• A ruler with a damaged zero point.
• An incorrectly calibrated sensor.

Repeating the measurement does not normally eliminate systematic error.

The method itself must be corrected.

---

Improving Systematic Error

Possible improvements include:

• Zero equipment before use.
• Calibrate instruments.
• Replace damaged equipment.
• Check instruments against known standards.
• Correct the measurement technique.

Therefore:

random error → repetition may help

but:

systematic error → fix the method or equipment

---

Controlling Variables

Poor control of variables is a major weakness in experimental procedures.

Suppose students investigate:

How does temperature affect reaction rate?

They should normally control factors such as:

• Reactant concentration.
• Reactant volume.
• Surface area of solid reactants.
• Mixing method.

If concentration also changes, the investigation cannot clearly isolate the effect of temperature.

---

Improving Variable Control

Instead of saying:

"Control the variables better."

identify the specific variable.

For example:

Keep the hydrochloric acid concentration at 1.0 mol/L for every trial so that changes in reaction rate can be attributed more confidently to temperature rather than concentration.

This is a scientifically justified improvement.

---

Standardizing Procedures

A standardized procedure is performed in the same way during every trial.

Suppose students measure plant height.

Different students might measure:

• From the soil.
• From the bottom of the pot.
• To the highest leaf.
• To the top of the stem.

Their results would not be directly comparable.

The procedure should clearly define:

Measure vertically from the soil surface to the highest point of the main stem.

This improves consistency.

---

Defining an Endpoint

Some experiments require students to decide when a process is "finished."

Examples include:

• When a tablet has completely dissolved.
• When a color has disappeared.
• When a reaction stops producing gas.
• When a precipitate makes a mark impossible to see.

If the endpoint is subjective, different people may make different decisions.

This reduces reliability.

---

Improving Endpoints

Where possible, replace subjective judgments with objective measurements.

Instead of:

"Stop timing when the solution looks clear."

a procedure might use:

• A colorimeter.
• A light sensor.
• A gas sensor.
• A fixed threshold reading.

Automated or quantitative measurements can reduce observer variation.

---

Sample Size

Using too few samples can make an investigation strongly affected by natural variation.

Suppose students investigate fertilizer using:

one fertilized plant

and:

one control plant

The plants may naturally grow at different rates.

A better method would use several plants in each condition.

For example:

10 plants per condition

may provide stronger evidence than one plant per condition.

---

Why Larger Samples Help

Larger samples can:

• Better represent natural variation.
• Reduce the influence of unusual individuals.
• Allow averages to be calculated.
• Provide more reliable comparisons.

However, sample size should remain realistic given:

• Time.
• Equipment.
• Cost.
• Space.
• Ethical considerations.

"Use 10,000 plants" is not a realistic improvement for most school investigations.

---

Repeated Measurements

Suppose a student records:

Temperature = 23.4°C

once.

There is no information about measurement variation.

Instead, several readings or repeated trials may provide:

23.4°C, 23.6°C, 23.5°C

The student can examine the spread and calculate an appropriate mean.

Repeated measurements provide information about reliability that one measurement cannot.

---

Selecting an Appropriate Range

An experiment may have a poor range of independent-variable values.

Suppose students investigate temperature and reaction rate using:

20°C and 21°C

The difference may be too small to reveal a useful pattern.

A more suitable range might be:

10°C, 20°C, 30°C, 40°C and 50°C

provided those temperatures are scientifically appropriate and safe.

A useful range should:

• Cover enough values to reveal a pattern.
• Include sensible intervals.
• Remain safe.
• Remain relevant to the research question.

---

More Data Points

Two values can show a difference, but several values can reveal a relationship.

For example:

10°C → 85 s

50°C → 21 s

shows a difference.

But:

10°C → 85 s

20°C → 63 s

30°C → 46 s

40°C → 32 s

50°C → 21 s

provides much more information about the pattern.

More appropriately spaced data points can improve the quality of the conclusion.

---

Worked Example: Pendulum

Suppose students investigate:

How does pendulum length affect period?

Weak method:

• Measure one swing.
• Use a handheld stopwatch.
• Test only two lengths.

Possible improvements:

Time Multiple Oscillations

Measure the time for 10 oscillations instead of one.

Then calculate:

Period = total time / 10

Why?

The total measured time is longer, so human reaction time represents a smaller percentage of the measurement.

Repeat Measurements

Repeat the measurement several times.

Test More Lengths

Use several appropriately spaced pendulum lengths.

Control Other Variables

Keep the release angle reasonably consistent.

---

Worked Example: Reaction Rate

Students investigate how temperature affects reaction rate.

They measure gas production using a balloon.

Weakness

Balloon size is difficult to measure accurately and may vary between balloons.

Improvement

Collect the gas using a gas syringe.

Why?

A gas syringe provides a direct quantitative measurement of gas volume.

This can improve precision and make comparisons easier.

---

Worked Example: Plant Growth

Students investigate the effect of fertilizer on plants.

They use:

• One plant with fertilizer.
• One plant without fertilizer.

Weakness

The sample size is too small to account well for natural variation.

Improvement

Use multiple plants in each condition.

Why?

A larger sample reduces the influence of unusual individual plants and provides a better estimate of typical growth.

---

Worked Example: Dissolving

Students investigate how temperature affects dissolving time.

Different students stir at different speeds.

Weakness

Stirring rate is not controlled.

Effect

Differences in dissolving time could be caused partly by stirring rather than temperature.

Improvement

Use the same stirring procedure for every trial, or use a mechanical stirrer set to the same speed.

Why?

This controls stirring and improves the validity of the comparison.

---

Worked Example: Measuring Temperature

A student removes a thermometer from a liquid before reading it.

Weakness

The thermometer may begin changing temperature after removal.

Improvement

Read the thermometer while it remains correctly positioned in the liquid, if appropriate for the instrument.

Why?

This reduces changes between the actual experimental temperature and the recorded measurement.

---

Worked Example: Free Fall

Students drop a ball and use a handheld stopwatch to measure a fall lasting approximately:

0.5 s

Human reaction time is a substantial fraction of such a short interval.

Improvement

Use:

• Light gates.
• Motion sensors.
• Video analysis with known frame rate.

These methods can provide more appropriate timing measurements.

---

Digital Measurement Techniques

Modern experiments can often use electronic sensors.

Examples include:

• Temperature probes.
• Light sensors.
• Force sensors.
• Motion sensors.
• pH probes.
• Pressure sensors.
• Gas sensors.

Sensors can:

• Take frequent measurements.
• Reduce some human reading errors.
• Record changes automatically.
• Produce large datasets.
• Measure rapidly changing quantities.

However, sensors still need appropriate calibration and correct use.

---

Video Analysis

Video can improve some motion experiments.

Students can record an object moving and analyze its position frame by frame.

If the frame rate is known, time intervals can be determined.

Video analysis can be useful for:

• Projectile motion.
• Falling objects.
• Collisions.
• Toy-car investigations.
• Sports movements.

It also creates a record that can be reanalyzed.

---

Avoiding Bias

Bias can also arise from the procedure.

Suppose a student expects fertilized plants to grow taller and unconsciously measures those plants differently.

Possible improvements include:

• Use objective measurement criteria.
• Label samples using codes.
• Use automated measurements where practical.
• Have measurements checked independently.

These strategies can reduce the influence of expectations.

---

Recording Data Immediately

Measurements should normally be recorded as they are collected.

Waiting until later can introduce:

• Memory errors.
• Missing data.
• Transcription mistakes.

A prepared data table can make data collection more consistent.

For example:

---

Recording Units

Every measured quantity should include appropriate units.

Poor:

Time = 14.2

Better:

Time = 14.2 s

Without units, the measurement is incomplete.

---

Recording Appropriate Precision

Measurements should reflect the resolution of the instrument.

If a ruler measures to the nearest millimetre, reporting:

12.638472 cm

would suggest a level of precision that the instrument did not provide.

Data should not imply greater measurement precision than was actually available.

---

Safety and Procedure Improvements

A procedure is not improved if it becomes unnecessarily dangerous.

Possible safety improvements include:

• Use lower concentrations where appropriate.
• Use smaller quantities.
• Wear appropriate eye protection.
• Use heatproof equipment.
• Secure unstable apparatus.
• Keep electrical equipment away from water.
• Use appropriate handling procedures.

Scientific quality and safety must be considered together.

---

Realistic Improvements

A recommendation should be practical.

For example:

Weak recommendation:

"Use a perfect timer with no error."

Such equipment does not exist.

Better recommendation:

"Use electronic light gates instead of hand timing to reduce variation caused by human reaction time."

Science aims to reduce and understand uncertainty, not pretend that uncertainty can be completely eliminated.

---

Matching Improvements to Problems

---

Improving Reliability

To improve reliability, scientists might:

• Repeat trials.
• Increase sample size.
• Standardize procedures.
• Use automated measurements.
• Reduce random variation.
• Calculate means where appropriate.
• Investigate anomalies.

---

Improving Validity

To improve validity, scientists might:

• Control important confounding variables.
• Change only the intended independent variable in a simple controlled experiment.
• Use appropriate control conditions.
• Measure the correct dependent variable.
• Use an appropriate experimental range.
• Reduce systematic errors.
• Ensure the procedure actually answers the research question.

---

Improving Accuracy

To improve accuracy, scientists might:

• Calibrate equipment.
• Zero instruments correctly.
• Use appropriate measuring instruments.
• Avoid parallax.
• Follow correct measurement procedures.
• Check instruments against known standards.

---

Improving Precision

To improve measurement precision where needed:

• Use instruments with suitable resolution.
• Use electronic sensors where appropriate.
• Measure longer intervals where this reduces percentage uncertainty.
• Standardize measurement techniques.

The most precise instrument is not always necessary. Equipment should be appropriate for the investigation.

---

Percentage Uncertainty

The importance of measurement uncertainty depends partly on the size of the measurement.

A useful approximation is:

percentage uncertainty = absolute uncertainty / measured value × 100%

Suppose timing uncertainty is approximately:

±0.2 s

For a measurement of:

1.0 s

percentage uncertainty is approximately:

0.2 / 1.0 × 100% = 20%

For a measurement of:

10.0 s

it is:

0.2 / 10.0 × 100% = 2%

This explains why measuring longer time intervals can sometimes improve measurement quality.

---

Justifying an Improvement

A strong scientific recommendation contains three parts.

Identify the Problem

The car is released by hand.

Recommend a Specific Improvement

Use a mechanical release mechanism.

Explain Why

This reduces variation in the initial push and gives each trial more consistent starting conditions.

Therefore:

problem → improvement → scientific reason

---

Example Evaluation Paragraph

A weak evaluation might say:

The experiment could be improved by being more careful and repeating it.

A stronger evaluation would say:

The travel time was measured using a handheld stopwatch, so human reaction time may have increased variation between trials. Electronic light gates could be placed at the start and finish points to measure the travel time automatically. This would reduce timing variation and improve the reliability and precision of the measurements.

The second response identifies:

• The weakness.
• Its effect.
• A realistic improvement.
• Why the improvement works.

---

Prioritizing Improvements

Not every weakness has the same importance.

Suppose an experiment contains:

• Slight uncertainty in measuring 100 mL of water.
• Only one trial.
• Two variables changed simultaneously.

The most serious issue is probably:

changing two variables simultaneously

because this directly affects whether the investigation can answer its research question.

Scientists should prioritize improvements that have the greatest effect on the quality and interpretation of the evidence.

---

Evaluating the Whole Procedure

When reviewing an experimental method, ask:

Variables

• Is the independent variable clearly changed?
• Is the dependent variable measured appropriately?
• Are important variables controlled?

Measurements

• Is the equipment suitable?
• Is its resolution appropriate?
• Is it correctly calibrated or zeroed?
• Are measurements taken consistently?

Reliability

• Are there repeated trials?
• Is the sample size sufficient?
• Are anomalies investigated?

Validity

• Does the method actually answer the research question?
• Are there important confounding variables?

Data Quality

• Are units recorded?
• Is precision appropriate?
• Is the range useful?

Safety

• Are hazards controlled appropriately?

---

Common Mistakes

"Be More Careful"

This is too vague.

Explain exactly what should change.

"Repeat the Experiment to Make It Accurate"

Repetition primarily helps evaluate and reduce the influence of random variation. It does not automatically correct systematic error.

"Use Better Equipment"

Name the equipment and explain why it is better for the measurement.

"Control All Variables"

The independent variable must change. Relevant controlled variables should remain constant where appropriate.

"Remove the Anomaly"

Investigate the anomaly first. Do not delete data simply because it does not fit the expected pattern.

"Use More Decimal Places"

More decimal places do not automatically produce better data.

"Use a Bigger Sample" Without Explanation

Explain that a larger sample can reduce the influence of individual variation and provide a better estimate of typical behavior.

"Eliminate All Error"

No real experiment is completely free of uncertainty.

A better goal is to:

identify, reduce, measure, and account for uncertainty.

---

Check Your Understanding

1. What is meant by a weakness in an experimental procedure?

2. Why is "be more careful" a poor experimental improvement?

3. Explain the weakness–effect–improvement approach.

4. How can human reaction time affect stopwatch measurements?

5. How could light gates improve a toy-car investigation?

6. Why should a car be released using a consistent mechanism?

7. What is instrument resolution?

8. Why should equipment be zeroed before use?

9. What is calibration?

10. What is parallax error, and how can it be reduced?

11. Why might a measuring cylinder be more appropriate than a beaker for measuring liquid volume?

12. How can repeated trials improve data quality?

13. Why doesn't repetition normally eliminate systematic error?

14. How can increasing sample size improve a biological investigation?

15. Why should experimental procedures be standardized?

16. What is a subjective endpoint?

17. How could a subjective endpoint be made more objective?

18. Why should an experiment use an appropriate range of independent-variable values?

19. Why are several data points usually more informative than only two?

20. A pendulum period is very short. Explain why timing 10 oscillations may be better than timing one.

21. Students measure gas production using balloon size. Suggest and justify a better method.

22. Students investigate temperature and reaction rate but use different reactant concentrations. Identify the weakness and suggest an improvement.

23. A balance reads 2.0 g when empty. Explain how this should be addressed.

24. Explain the difference between reducing random error and correcting systematic error.

25. Write a complete justified improvement for an experiment in which students use a handheld stopwatch to measure a 0.4 s event.

---

Key Terms

• Experimental procedure – instructions describing how an investigation is performed.
• Evaluation – examination of the strengths, weaknesses, limitations, and quality of an investigation.
• Improvement – specific change intended to strengthen an experimental method.
• Random error – unpredictable variation between measurements.
• Systematic error – consistent error that shifts measurements in a particular direction.
• Resolution – smallest change an instrument can display or distinguish.
• Calibration – checking an instrument against known standards or reference values.
• Parallax error – measurement error caused by viewing a scale from an incorrect angle.
• Reliability – consistency of measurements or findings.
• Validity – extent to which an investigation appropriately addresses the intended question.
• Accuracy – closeness of a measurement to an accepted or true value.
• Precision – closeness of repeated measurements and the resolution with which measurements are expressed.
• Sample size – number of experimental units or observations included.
• Anomaly – result noticeably different from the general pattern.
• Endpoint – defined point at which a measurement or reaction is considered complete.
• Confounding variable – uncontrolled factor that provides an alternative explanation for a result.
• Percentage uncertainty – uncertainty expressed as a percentage of the measured value.

---

Key Takeaways

• Scientists evaluate procedures to identify weaknesses and improve the quality of evidence.
• A strong improvement is specific, realistic, and scientifically justified.
• A useful evaluation follows:

weakness → effect → improvement → reason

• Repeated trials help reduce the influence of random variation and assess reliability.
• Increasing sample size can reduce the influence of individual biological variation.
• Systematic errors usually require changes to equipment or technique rather than additional repetitions.
• Instruments should have an appropriate range and resolution.
• Equipment should be correctly zeroed and calibrated.
• Automated timing can reduce human reaction-time variation.
• Standardized procedures improve consistency.
• Objective measurements are generally preferable to subjective judgments.
• Important confounding variables should be controlled.
• Appropriate ranges and multiple data points make patterns easier to identify.
• Better measurement techniques can improve accuracy, precision, and reliability.
• Improvements should target the actual source of the problem.
• More expensive equipment is not automatically better; it must be appropriate for the measurement.
• Experimental improvements must remain safe and practical.
• Scientists cannot eliminate all uncertainty, but they can reduce and evaluate it.
• The strongest recommendations explain not only what should change, but also why that change improves the investigation.