On this page
Imagine two students comparing how quickly different antacid tablets neutralise acid. One uses warm acid and crushes her tablet; the other uses cold acid and drops his in whole. They get different results, but they can’t say which tablet is better, because they changed three things at once. Their test wasn’t fair.
The idea of a fair test is one of the foundations of experimental science. It’s simple to state and surprisingly easy to get wrong.
The definition
A fair test is an experiment in which only the independent variable is changed, while all other variables that could affect the result are kept the same (controlled). The dependent variable is measured to see the effect.
That way, any change in the dependent variable can be linked to the independent variable, and nothing else. See independent, dependent and control variables.
An unfair test, fixed
Question: Does increasing the concentration of hydrochloric acid increase the rate of reaction with marble chips?
Unfair version:
- Run 1: 0.5 mol/dm³ acid, 25 °C, large marble chips
- Run 2: 2.0 mol/dm³ acid, 40 °C, small marble chips
The second run is faster, but was it the concentration, the temperature or the smaller chips (bigger surface area)? It’s impossible to say. The variables are confounded.
Fair version:
- Change only the acid concentration: 0.5, 1.0, 1.5, 2.0 mol/dm³
- Keep the same: temperature (25 °C in a water bath), mass of marble (5.00 g), chip size (same batch, same size), volume of acid (50.0 cm³), no stirring
- Measure: mass lost as CO₂ escapes after 60 s (or volume of gas)
Now any difference in rate can be attributed to concentration.
More than just controlling variables
A fair test is essential, but a trustworthy experiment also needs:
Repeats
Doing each measurement several times (usually at least three) and calculating a mean:
- reduces the effect of random errors
- helps spot anomalies (results that don’t fit)
- shows how consistent the results are
A control experiment
A control experiment is a comparison run in which the factor being tested is absent. When testing whether manganese(IV) oxide catalyses the decomposition of hydrogen peroxide, the control is hydrogen peroxide without the catalyst under identical conditions. Without it, you can’t be sure the catalyst caused the effect.
An appropriate range and number of values
Test enough values of the independent variable (typically at least five) across a wide enough range to see a pattern, and include values close together where the change is most interesting.
Validity, repeatability, reproducibility, accuracy and precision
These words describe the quality of results, and they’re often confused.
| Term | Meaning | How to improve it |
|---|---|---|
| Valid | the experiment actually tests the question it claims to; it’s a fair test measuring the right thing | control variables; measure the correct dependent variable |
| Repeatable | the same person, using the same method and equipment, gets similar results | repeat measurements; careful technique |
| Reproducible | a different person, or different equipment or method, gets similar results | clear method; standard equipment |
| Accurate | close to the true value | reduce systematic errors; calibrate |
| Precise | repeated results are close to each other (small spread) | reduce random errors; better instruments |
An experiment can be precise but not accurate (consistent results that are all wrong, due to a systematic error), or accurate on average but not precise (widely scattered results whose mean is close to the true value). See random vs systematic errors.
Why fair tests are hard in chemistry
Some variables are easy to control; others are sneaky:
- Temperature: many reactions release or absorb heat, so the temperature changes during the reaction itself. Use a water bath and record the temperature.
- Surface area: “the same mass” of a solid can have very different surface areas if particle sizes differ.
- Concentration drift: solutions like sodium hydroxide absorb carbon dioxide from the air over time.
- Purity of reagents: different batches can differ slightly.
- Human judgement: judging colour changes or end points introduces random error.
- Catalysts you didn’t intend: traces of metal ions or dirt can catalyse reactions.
When a variable can’t be controlled perfectly, monitor it and discuss it in the evaluation.
Examples: fair or unfair?
1. A student compares the energy released by burning ethanol and propanol, using a spirit burner and a can of water. They use 100 cm³ of water for ethanol and 200 cm³ for propanol. Unfair. The mass of water heated is a control variable and must be the same (or the calculation must account for it).
2. A student tests the effect of temperature on the rate of reaction of magnesium ribbon with 1.0 mol/dm³ HCl, using 3 cm of ribbon each time, 25 cm³ of acid, and a water bath at 20, 30, 40 and 50 °C. Fair. Only temperature changes.
3. A student tests whether copper sulfate catalyses the reaction of zinc with sulfuric acid. They add copper sulfate to one test tube and nothing to another, but use zinc powder in the first and zinc granules in the second. Unfair. Surface area (powder vs granules) is also changed.
4. Two brands of antacid are tested by adding one tablet of each to 100 cm³ of 0.1 mol/dm³ HCl, measuring the pH after 5 minutes. Partly fair: same acid volume, concentration and time. But the tablets may have different masses, so comparing “per tablet” answers a different question from comparing “per gram”. A good investigation states which is being tested.
Fair tests beyond the lab
The same logic underpins much of modern science. Medical drug trials compare a treatment group with a control group given a placebo, randomly assigned so that other factors are balanced. Agricultural trials test fertilisers on plots with the same soil, water and crop. In every case, the aim is to change one thing, control the rest, and measure the effect.
Planning checklist for a fair test
Before starting any investigation, it helps to write down five things: the question you’re answering, the one variable you’ll change and the values you’ll use, what you’ll measure and with which instrument, every other factor that could affect the result and how you’ll keep each one constant, and how many repeats you’ll do. If you can fill in all five clearly, your test is very likely to be fair. If you can’t, the gaps show you what to fix before you pick up any equipment.
Key takeaways
- A fair test changes only the independent variable and controls all others.
- Confounded variables make it impossible to know what caused an effect.
- Repeats improve reliability; control experiments provide a baseline for comparison.
- Validity, repeatability, reproducibility, accuracy and precision describe different aspects of result quality.
- Monitor variables that can’t be fully controlled, and discuss them in your evaluation.
Advertisement