3. Fair tests

What Makes an Experiment a Fair Test?

A fair test is an investigation designed so that the effect of one factor can be studied without other factors interfering with the comparison.

In a simple controlled experiment:

one independent variable is deliberately changed

while:

other relevant variables are kept as constant as reasonably possible

and:

the dependent variable is measured consistently

A fair test allows scientists to make a meaningful comparison between different experimental conditions.

A useful idea is:

change one factor → control other relevant factors → measure the outcome → compare results

---

The Three Main Types of Variables

Understanding variables is essential for designing fair tests.

Independent Variable

The independent variable (IV) is the factor deliberately changed by the investigator.

For example:

Does water temperature affect how quickly sugar dissolves?

Independent variable:

water temperature

---

Dependent Variable

The dependent variable (DV) is the factor measured or observed.

In the sugar investigation:

Dependent variable:

time required for the sugar to dissolve

The dependent variable may change because the independent variable was changed.

---

Controlled Variables

Controlled variables are factors kept as constant as reasonably possible.

For the sugar investigation, these could include:

• Mass of sugar.
• Volume of water.
• Type of sugar.
• Container.
• Stirring method.
• Method used to determine when the sugar has dissolved.

These controls make the comparison more meaningful.

---

A Fair-Test Example

Suppose a student asks:

How does ramp height affect the speed of a toy car?

The student changes the ramp height and measures the car's speed.

Independent Variable

Ramp height.

Dependent Variable

Speed of the car.

Controlled Variables

• Same toy car.
• Same ramp.
• Same surface.
• Same travel distance.
• Same release point or release procedure.
• Same timing method.

If the student changes both the ramp height and the car, the comparison becomes less fair because either change could influence the result.

---

Why Controlled Variables Matter

Imagine testing whether temperature affects how quickly a tablet dissolves.

Trial A

Cold water + one tablet + 100 mL water.

Trial B

Hot water + two crushed tablets + 300 mL water.

This is not a fair comparison.

Several variables changed:

• Temperature.
• Number of tablets.
• Tablet particle size.
• Water volume.

If Trial B behaves differently, we cannot confidently determine which change was responsible.

---

Improving the Investigation

A better design would use:

Trial A

10°C water + one whole tablet + 100 mL water.

Trial B

30°C water + one whole tablet + 100 mL water.

Trial C

50°C water + one whole tablet + 100 mL water.

Now:

temperature changes

while the other important variables remain constant.

This provides a much fairer comparison.

---

Fair Does Not Mean Identical

A common misunderstanding is that everything must be kept the same.

If everything were identical, there would be no experiment.

One factor must deliberately change:

the independent variable

Therefore:

IV → changes

DV → measured

controlled variables → kept constant where relevant

---

Choosing an Appropriate Range

A good fair test usually investigates several useful values of the independent variable.

For example, instead of testing plant growth at only:

10°C and 30°C

a student might test:

10°C, 15°C, 20°C, 25°C and 30°C

This provides more information about the relationship between temperature and growth.

A suitable range should also be:

• Safe.
• Practical.
• Relevant to the research question.

---

Choosing Suitable Intervals

The values of the independent variable should allow useful comparisons.

Suppose a student investigates how ramp height affects travel time.

They could use:

5 cm, 10 cm, 15 cm, 20 cm, 25 cm

These regular intervals make it easier to identify patterns.

If the student used:

5 cm, 6 cm, 7 cm, 50 cm and 100 cm

the values might provide a less useful picture of the relationship, depending on the investigation.

---

Measuring the Dependent Variable

A fair test also requires the dependent variable to be measured consistently.

For example, if measuring plant growth, students should decide whether they will measure:

• Plant height.
• Change in height.
• Leaf number.
• Biomass.

They should then use the same method throughout the investigation.

Changing measurement methods halfway through an experiment reduces comparability.

---

Reliable Comparisons

A reliable comparison is one in which differences between groups are less likely to be caused by inconsistent procedures or random variation.

Ways to improve reliability include:

• Repeating measurements.
• Using multiple samples.
• Using consistent procedures.
• Using appropriate measuring equipment.
• Calculating averages where appropriate.

Fairness and reliability are related, but they are not exactly the same idea.

---

Repeated Trials

Suppose a student measures the time for a toy car to travel down a ramp.

First result:

1.42 s

Should the student immediately accept this as the correct value?

Probably not.

The student could repeat the measurement:

Trial 1: 1.42 s

Trial 2: 1.38 s

Trial 3: 1.40 s

Average:

(1.42 + 1.38 + 1.40) / 3 = 1.40 s

Repeated trials reduce the influence of random variation.

---

Identifying Anomalies

An anomaly is a result that differs noticeably from the general pattern.

Suppose the measurements are:

1.42 s, 1.38 s, 2.91 s, 1.40 s, 1.41 s

The value:

2.91 s

appears unusual compared with the others.

The student should investigate possible reasons.

Perhaps:

• The timer was started late.
• The car hit something.
• The car was released incorrectly.
• The measurement was recorded incorrectly.

An anomalous result should not simply be deleted because it is inconvenient.

There should be a scientific reason for excluding or repeating it.

---

Accuracy

Accuracy describes how close a measurement is to the accepted or true value.

For example, suppose a balance should read:

100.0 g

but consistently reads:

104.0 g

The measurements may be consistent but inaccurate.

Checking and calibrating equipment can improve accuracy.

---

Precision

Precision refers to how closely repeated measurements agree with one another and, in measurement contexts, the resolution with which measurements are reported.

For example:

20.1 cm, 20.2 cm, 20.1 cm

are closely grouped.

Measurements such as:

18.4 cm, 22.7 cm, 20.0 cm

show much greater spread.

A good investigation aims for measurements that are both accurate and appropriately precise.

---

Validity

An investigation is valid when its design and measurements allow it to answer the intended research question appropriately.

Suppose the question is:

How does light intensity affect photosynthesis?

If the experiment actually changes both light intensity and temperature, it becomes difficult to determine whether light intensity caused the observed change.

This weakens the validity of the investigation.

---

Reliability and Validity Are Different

A procedure can produce very consistent results without answering the intended question properly.

For example, students might repeat a poorly designed experiment ten times.

The results could be highly consistent.

But if two important variables were changed simultaneously, the investigation would still have weak validity for isolating the effect of either one.

Therefore:

repetition improves reliability

but:

repetition cannot automatically fix a poor experimental design

---

What Is Bias?

Bias is a systematic influence that pushes results or interpretations in a particular direction.

Bias can enter an investigation through:

• Sample selection.
• Measurement methods.
• Expectations.
• Experimental procedures.
• Data collection.
• Data interpretation.

Scientists try to identify and reduce sources of bias.

---

Selection Bias

Selection bias occurs when the sample being studied is not selected in a way that fairly represents the population relevant to the question.

Suppose a student asks:

What is the average height of students at our school?

They measure only members of the senior basketball team.

The sample is unlikely to represent the entire school.

A better approach would involve selecting students from the relevant school population using an appropriate sampling method.

---

Measurement Bias

Measurement bias can occur when a measuring method consistently favors particular results.

For example, suppose a ruler has a damaged end and every measurement begins at the damaged edge.

All measurements might be systematically too large or too small.

This produces a systematic error.

---

Observer Bias

Expectations can sometimes influence observations.

Suppose a researcher knows which plants received a new treatment and expects those plants to look healthier.

If "healthier" is judged subjectively, expectations might influence the assessment.

A more objective measurement could be:

• Plant height.
• Dry mass.
• Number of leaves.

In some investigations, researchers can also use blinding so the person making measurements does not know which treatment each sample received.

---

Confirmation Bias

Confirmation bias is the tendency to focus more strongly on evidence that supports an existing expectation while overlooking contradictory evidence.

For example, a student predicts that fertilizer will increase plant growth.

If one fertilized plant grows poorly, the student should not simply ignore it.

Scientists should examine all relevant data, including results that do not support the original hypothesis.

---

Random Error

Not every difference is caused by bias.

Random errors produce unpredictable variation in measurements.

Examples include:

• Human reaction time.
• Small differences between biological organisms.
• Slight fluctuations in temperature.
• Small variations in equipment readings.

Repeating measurements can help reduce the influence of random error on the final result.

---

Systematic Error

A systematic error shifts measurements consistently in the same direction.

For example:

A balance reads:

+2.0 g

even when empty.

Every measurement may therefore be approximately 2.0 g too high unless the balance is correctly zeroed.

Repeating the measurement does not remove this systematic error.

The equipment or method must be corrected.

---

Random vs Systematic Error

Understanding this difference helps students improve experimental design.

---

Using a Control Group

Some fair tests require a control group.

Suppose the question is:

Does fertilizer increase plant growth?

Experimental Group

Plants receive fertilizer.

Control Group

Plants receive no fertilizer.

All other important conditions should be similar.

The control provides a baseline for comparison.

---

Control Group vs Controlled Variable

These are different concepts.

Control Group

A comparison group or condition.

Controlled Variable

A factor kept constant.

For the fertilizer investigation:

control group → plants receiving no fertilizer

while:

controlled variables → water, light, soil, temperature, plant species, pot size

Do not confuse the two.

---

Worked Example: Fertilizer and Plant Growth

Question:

How does fertilizer concentration affect bean plant growth?

Independent Variable

Fertilizer concentration.

Possible values:

0%, 1%, 2%, 3%, 4%

Dependent Variable

Change in plant height over 21 days.

Controlled Variables

• Bean variety.
• Starting size.
• Soil type and amount.
• Pot size.
• Water volume.
• Light exposure.
• Temperature.
• Growing time.

Control Condition

0% fertilizer

Reliability

Use several plants at each concentration.

Measurement

Measure plant height using the same method at regular intervals.

This is much stronger than using one plant for each concentration.

---

Worked Example: Ramp Investigation

Question:

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

Independent Variable

Ramp height.

Dependent Variable

Travel time.

Controlled Variables

• Same car.
• Same ramp.
• Same surface.
• Same travel distance.
• Same release procedure.
• Same timing method.

Reliability

Repeat each height at least three times.

Processing Data

Calculate the mean time for each ramp height.

The results can then be graphed.

---

Worked Example: Temperature and Dissolving

Question:

How does water temperature affect the time required for sugar to dissolve?

Independent Variable

Water temperature.

Dependent Variable

Dissolving time.

Controlled Variables

• Mass of sugar.
• Volume of water.
• Sugar particle size.
• Container.
• Stirring method.

Useful Temperatures

For example:

10°C, 20°C, 30°C, 40°C, 50°C

Reliability

Repeat each temperature several times.

---

Worked Example: Exercise and Heart Rate

Question:

How does exercise duration affect heart rate immediately after exercise?

Independent Variable

Exercise duration.

Dependent Variable

Heart rate immediately after exercise.

Variables Requiring Control

Depending on the design:

• Type of exercise.
• Exercise intensity.
• Measurement method.
• Timing of heart-rate measurement.
• Recovery conditions.

Biological investigations can be difficult because people naturally differ.

Using repeated measurements or appropriate groups can help account for variation.

---

Evaluating an Investigation

Consider this investigation:

A student wants to determine whether light affects plant growth.

Plant A

• 8 hours of light per day.
• 100 mL water per day.
• 25°C.

Plant B

• 2 hours of light per day.
• 30 mL water per day.
• 15°C.

Is this a fair test?

No.

Three important variables changed:

• Light.
• Water.
• Temperature.

If Plant A grows more, the student cannot determine which variable caused the difference.

---

Improving the Investigation

A better design could use:

Plant A

8 hours of light.

Plant B

2 hours of light.

Both receive:

• 100 mL water per day.
• Same temperature.
• Same soil.
• Same plant species.
• Same pot size.
• Same experimental duration.

Now the major deliberate difference is:

light exposure

The comparison is much fairer.

---

Sample Size

Using only one experimental subject can produce misleading results.

Suppose:

one fertilized plant grows 12 cm

and:

one control plant grows 8 cm

Can we confidently conclude that fertilizer caused the difference?

Not necessarily.

The individual plants may naturally grow at different rates.

Using multiple plants in each condition provides stronger evidence.

---

Random Assignment

When experimental subjects naturally vary, random assignment can help reduce bias.

Suppose 40 similar plants are divided into:

• Fertilizer group.
• Control group.

If the researcher deliberately places the healthiest-looking plants in the fertilizer group, the experiment becomes biased.

Randomly assigning plants to groups helps reduce this problem.

---

Standardized Procedures

A standardized procedure means performing measurements and treatments consistently.

For example:

• Measure plants at the same time of day.
• Use the same measuring instrument.
• Use the same starting point.
• Use the same person or clearly defined method when appropriate.
• Follow the same procedure for every trial.

Standardization makes comparisons more reliable.

---

Designing a Fair Test

A useful procedure is:

Ask a Testable Question

For example:

How does temperature affect enzyme activity?

Identify the Independent Variable

Temperature.

Identify the Dependent Variable

Rate of enzyme activity.

Identify Controlled Variables

Keep relevant factors constant.

Select an Appropriate Range

Choose several suitable temperatures.

Decide How Measurements Will Be Made

Use a consistent and appropriate method.

Include a Control if Appropriate

Choose a meaningful baseline or comparison.

Repeat Trials

Collect enough data to assess variation.

Record All Results

Do not remove inconvenient results without justification.

Analyze the Data

Calculate averages where appropriate and look for patterns.

Evaluate the Method

Identify limitations and suggest realistic improvements.

---

Evaluating Scientific Validity

When evaluating an investigation, ask:

Does the experiment actually test the stated question?

Is only the intended independent variable deliberately changed?

Are important variables controlled?

Is the dependent variable measured appropriately?

Are measurements collected consistently?

Is the sample size adequate for the question?

Are trials repeated?

Are appropriate controls included where needed?

Could bias have influenced the results?

Are the conclusions supported by the data?

A scientifically strong investigation should survive this type of questioning.

---

Correlation Is Not Automatically Causation

Sometimes two variables change together.

This is called a correlation.

However, observing a correlation does not automatically prove that one variable caused the other.

For example, suppose researchers observe that two variables increase at the same time.

Possible explanations include:

• Variable A causes B.
• Variable B causes A.
• A third variable influences both.
• The relationship occurred by chance.

Well-controlled experiments can provide stronger evidence for causal relationships.

---

Fair Tests in Biology

Biological experiments can be challenging because organisms naturally vary.

Scientists may improve comparisons by:

• Increasing sample size.
• Randomly assigning organisms.
• Using organisms of similar age or size where appropriate.
• Maintaining consistent environmental conditions.
• Repeating experiments.

Natural variation cannot always be eliminated, but it can be managed.

---

Fair Tests in Chemistry

Chemistry experiments may require control of:

• Temperature.
• Concentration.
• Volume.
• Mass.
• Pressure.
• Surface area.
• Reaction time.

For example, when investigating concentration and reaction rate, temperature should normally remain constant unless temperature is the variable being investigated.

---

Fair Tests in Physics

Physics investigations may require control of:

• Mass.
• Distance.
• Angle.
• Surface.
• Starting position.
• Applied force.
• Equipment.
• Measurement method.

For example, when investigating ramp height and car speed, using a different car for every height could introduce another variable.

---

Fair Does Not Mean Perfect

Real experiments are rarely perfect.

Scientists may not be able to control every possible factor.

Instead, they try to:

• Control the most important variables.
• Measure carefully.
• Reduce bias.
• Repeat measurements.
• Quantify uncertainty where appropriate.
• Identify limitations.
• Improve future investigations.

Scientific validity comes from thoughtful experimental design and transparent evaluation, not from pretending uncertainty does not exist.

---

Common Mistakes

Changing More Than One Independent Variable

This makes cause and effect difficult to determine.

Confusing the Independent and Dependent Variables

Remember:

IV → deliberately changed

DV → measured

Confusing Control Groups and Controlled Variables

A control group provides a comparison.

A controlled variable is kept constant.

Using Only One Trial

One result may be affected by random variation.

Using an Extremely Small Sample

Individual differences can strongly influence results.

Ignoring Anomalies

Unexpected results should be investigated rather than automatically deleted.

Thinking Repetition Fixes Systematic Error

Repeating a biased measurement usually reproduces the same bias.

Choosing Samples That Support the Hypothesis

Samples should be selected using an appropriate method rather than chosen to produce a desired outcome.

Changing the Measurement Method

The dependent variable should be measured consistently.

Claiming More Than the Evidence Shows

A conclusion should match the strength and limitations of the data.

---

Check Your Understanding

1. What is a fair test?

2. What is the independent variable?

3. What is the dependent variable?

4. What is a controlled variable?

5. Why should relevant variables be controlled?

6. Why does a fair test deliberately change one variable?

7. Identify the IV, DV, and three controlled variables in an investigation of ramp height and toy-car travel time.

8. Why are repeated trials useful?

9. What is an anomaly?

10. Should an anomalous result automatically be deleted? Explain.

11. What is the difference between accuracy and precision?

12. What is the difference between random and systematic error?

13. Why doesn't repeating an experiment necessarily remove systematic error?

14. What is experimental bias?

15. Give an example of selection bias.

16. How can observer bias influence results?

17. What is confirmation bias?

18. What is the difference between reliability and validity?

19. Why might a larger sample improve a biological investigation?

20. How can random assignment reduce bias?

21. Why are standardized procedures important?

22. A student changes both temperature and concentration while measuring reaction rate. Why is this not a fair test of temperature alone?

23. How could the investigation in Question 22 be improved?

24. Why does correlation not necessarily demonstrate causation?

25. Describe how you would design a fair test investigating how light exposure affects plant growth.

---

Key Terms

• Fair test – investigation designed to make meaningful comparisons by deliberately changing an independent variable while controlling other relevant factors.
• Independent variable – factor deliberately changed by the investigator.
• Dependent variable – factor measured or observed.
• Controlled variable – factor kept constant during an investigation.
• Control group – comparison group used to help interpret experimental results.
• Validity – extent to which an investigation appropriately answers the question it was designed to test.
• Reliability – extent to which measurements or findings are consistent when repeated under comparable conditions.
• Accuracy – closeness of a measurement to the accepted or true value.
• Precision – closeness of repeated measurements to one another and the resolution with which measurements are expressed.
• Bias – systematic influence that can push results or interpretations in a particular direction.
• Random error – unpredictable variation between measurements.
• Systematic error – consistent error that shifts measurements in a particular direction.
• Anomaly – result noticeably different from the general pattern.
• Sample size – number of experimental units or observations included.
• Random assignment – assigning experimental units to groups using chance.
• Standardized procedure – method performed consistently across trials or groups.
• Confounding variable – uncontrolled factor that could provide an alternative explanation for an observed result.
• Correlation – relationship in which variables change together.
• Causation – relationship in which a change in one factor produces a change in another.

---

Key Takeaways

• A fair test allows meaningful comparisons between experimental conditions.
• The independent variable is deliberately changed.
• The dependent variable is measured.
• Other relevant variables should be controlled.
• Fair testing does not mean keeping everything identical—the independent variable must change.
• Appropriate controls can provide useful baselines.
• Repeated trials help assess random variation and improve reliability.
• Larger sample sizes can reduce the influence of individual variation.
• Random assignment can reduce selection bias.
• Standardized procedures make comparisons more consistent.
• Random and systematic errors are different and require different solutions.
• Repetition does not automatically correct systematic error.
• Bias can enter an investigation through sampling, measurement, procedures, or interpretation.
• Scientists should examine evidence that both supports and contradicts their hypotheses.
• Reliability and validity are related but different.
• A reliable investigation is not necessarily valid if it does not properly test the intended question.
• A strong scientific conclusion should be supported by the data and acknowledge important limitations.
• A useful summary of fair testing is:

change one relevant factor → control other important factors → measure consistently → repeat → compare → evaluate