Good experiments are planned carefully
Experimental design is the plan behind an experiment. It decides what will be changed, what will be measured, what will be held constant, how many trials will be done and how the data will be analysed. A good design makes the results trustworthy.
Key features of strong design:
| Feature | Why it matters |
|---|---|
| One independent variable | So the cause of any change is clear |
| Controlled variables | So other factors don’t affect the result |
| A control group | Gives a baseline to compare against |
| Repeated trials | Reduces the effect of random variation |
| Random assignment | Prevents hidden bias in which subjects get which treatment |
| Blinding | Stops expectations from influencing results |
In a double-blind study, neither the subjects nor the researchers know who received the real treatment. This is standard in medical trials.
Every measurement has uncertainty
No measurement is perfect. The uncertainty is the range in which the true value probably lies. It is written with a ± sign.
A ruler marked in millimetres can usually be read to ±0.5 mm. If you measure a pencil as 152.0 mm, you’d report 152.0 ± 0.5 mm.
For repeated measurements, a simple estimate of uncertainty is half the range:
uncertainty ≈ (largest − smallest) ÷ 2
Five readings of 9.78, 9.82, 9.80, 9.85 and 9.75 m/s² give a mean of 9.80 and an uncertainty of about (9.85 − 9.75) ÷ 2 = ±0.05 m/s².
Types of error
| Type | What it is | Example | How to reduce it |
|---|---|---|---|
| Random error | Unpredictable scatter above and below the true value | Reaction time when using a stopwatch | Repeat trials and average |
| Systematic error | A consistent shift in one direction | A scale that reads 2 g too high | Calibrate equipment; check zero |
| Human mistake | A blunder, not a true error | Writing 15 instead of 51 | Check work; repeat the trial |
Averaging reduces random error, but it can’t fix systematic error. If every reading is too high by the same amount, the average will be too high too.
Precision and accuracy
- Precision is how close repeated measurements are to each other.
- Accuracy is how close they are to the true value.
Think of darts on a target:
| Pattern | Precise? | Accurate? |
|---|---|---|
| Tight cluster on the bullseye | Yes | Yes |
| Tight cluster off to one side | Yes | No (systematic error) |
| Scattered evenly around the bullseye | No | Roughly, on average |
| Scattered and off to one side | No | No |
Error bars
On graphs, uncertainty is often shown with error bars, small lines above and below each point. If the error bars of two results overlap a lot, the difference between them may not be meaningful.
Evaluating an experiment
When you evaluate, ask:
- Were there uncontrolled variables?
- Was the sample large enough?
- Were there signs of systematic error?
- How could the design be improved?
Common mistakes
- Thinking averaging fixes everything. It only reduces random error.
- Reporting too many decimal places, suggesting more certainty than you have.
- Calling any problem "human error" without saying exactly what went wrong.
Try it yourself
A student times a ball rolling down a ramp five times: 2.31 s, 2.28 s, 2.35 s, 2.29 s, 2.32 s. The stopwatch was later found to start 0.1 s late every time.
- Calculate the mean and the uncertainty.
- Is the stopwatch problem random or systematic error?
- How would it affect the mean?
- Suggest a better way to time the ball.
Quick recap
- Good experimental design uses controls, repetition, randomization and blinding.
- Every measurement has an uncertainty, written as ±.
- Random error scatters results; systematic error shifts them all one way.
- Precision is consistency; accuracy is closeness to the true value.
- Evaluate experiments by looking for sources of error and ways to improve.