Evaluating experimental methods
Key idea: A useful evaluation links a specific limitation to its effect on the result and a change that reduces it. Repeats address random scatter, not a shared bias or an unsuitable model.
Offline pages
Saved on this device only. Connect at least once every 30 days to confirm Premium. Sign-out or a change of account removes saved copies.
Loading guide controls…
Guided lessons · coming soon
Guided lesson
Guided steps
On this page
You stopped at:
Continue“Human error” and “repeat for accuracy” say almost nothing about an experiment. A useful evaluation traces a specific limitation of the method into its effect on the result, then proposes a change that acts on that cause. Timing a moving object by hand shows how the chain works.
Build a causal evaluation
A useful evaluation names a limitation, traces its effect into an observation or calculated quantity, then proposes a feasible change that acts on that cause. Give the direction when the mechanism supports one; do not invent a sign when the information is insufficient.
For manual elapsed-time measurement, let δₛₜₐᵣₜ and δₛₜₒₚ be the delays after the intended start and stop events. Then
A late start alone does not determine the timing bias: what matters is which delay is larger. Equal delays cancel in the elapsed time, even though neither press occurs at its intended event.
Guided practice 1
Trace two delays into a speed estimate
Problem
Try this before viewing the solution
Show the steps Optional working
Hints
Hint 1
Use both timestamps
A late start reduces the elapsed interval; a late stop increases it.Hint 2
Trace the calculation
For the same measured distance, average speed is d/t. Compare your denominator with the true elapsed time.Show solution step by step
Combine the delays
Method
The indicated elapsed time is 4.90 s.Reason
The start delay exceeds the stop delay, shortening the measured interval.Working
Answer
t = 5.00 + 0.10-0.20 = 4.90 s.Trace the speed bias
Method
The calculated average speed is 4.08 m s⁻¹, above the true average 4.00 m s⁻¹.Reason
The fixed distance is divided by an underestimated elapsed time.Working
Answer
v = 20.0/4.90 = 4.08 m s⁻¹; vₜᵣᵤₑ = 20.0/5.00 = 4.00 m s⁻¹.Target the timing mechanism
Method
Use automatically triggered start and stop events at the relevant positions.Reason
This reduces manual reaction-time effects on the interval; check trigger positions and calibration as well.Working
Answer
Keep the same defined distance and corresponding timing events.
Quick check
Check your understanding 1 Explain your reasoning
Use the 20.0 m timing trial above. Explain the direction of the speed error and propose a change that acts on the timing delays.
Show answer
The 4.90 s interval is too short, so 20.0/4.90 = 4.08 m s⁻¹ is too high. Automatically trigger start and stop at the defined positions to reduce the manual delay difference. Check trigger alignment and timing calibration. Compare the short interval, division by that interval and an improvement that acts on its cause.
| Vague statement | A causal evaluation |
|---|---|
| “Human error affected timing.” | “If the start delay exceeds the stop delay, elapsed time is underestimated. Automatically trigger both events at the required positions to reduce reaction-time effects.” |
| “Heat was lost.” | “While the sample warms above ambient, assigning energy transferred to the vessel and surroundings to the sample overestimates its specific heat capacity. Reduce loss and include significant apparatus heat capacity.” |
| “Repeat for accuracy.” | “Comparable independent timings can reveal scatter and reduce its contribution to a mean. A consistent timing offset needs a separate check or changed technique.” |
An improvement need not remove every effect. Explain what it reduces, what remains and any new limitation it introduces. Automatic timing, for example, still needs correct trigger positions and suitable calibration.
Distinguish random, systematic and model effects
Random effects vary unpredictably across comparable observations. Independent repeats can reduce the random contribution to a mean. A longer interval can reduce fractional timing uncertainty when the absolute timing allowance is roughly fixed; a wider data range can improve a gradient estimate. Neither is a general cure for every uncertainty.
Systematic effects arise from a shared offset, calibration or method. Repeating the same biased procedure can give a very precise mean of the wrong value. Check zero and calibration or change the method according to the cause.
Model limitations arise when assumptions fail, such as a large pendulum angle, changing temperature in a component, or assigning all electrical input to one sample. More readings under the same unsuitable conditions do not repair that model.
Check your understanding 2
Reason backwards from a low speed Distance is measured correctly, but a repeated manual timing method consistently gives an average speed that is too low. Under the two-delay model, what must be true of the start and stop delays? Would averaging more such timings remove that bias? State a targeted change.
Show answer
The elapsed time is too long, so the stop delay exceeds the start delay. Repeats can expose random variation but do not remove that consistent difference. Automatically triggered start and stop events can reduce the reaction-time mechanism, provided their positions and timing calibration are appropriate.
Compare three points: a longer measured interval, stop delay larger than start delay, and a change that removes manual triggering rather than merely adding repeats.
Check your understanding 3
Choose a change for the mechanism An electrical-heating calculation assigns all supplied energy to a sample, although the vessel also warms and energy transfers to cooler surroundings. A student proposes a more finely graduated temperature sensor and more repeats. Explain what those changes can address and give a change that targets the energy-accounting problem.
Show answer
Finer temperature readings may reduce a reading contribution to uncertainty; comparable independent repeats may reveal or reduce random scatter. Neither makes all supplied energy enter the sample. Reduce environmental transfer with suitable insulation and a lid, account for significant apparatus heat capacity, and estimate remaining loss where needed. State the assumptions and limitations of any correction.
Check that you separate reading uncertainty from omitted energy, then name both reduced heat loss and vessel heat capacity.
Decide whether an improvement addresses the cause
A proposed change should alter the mechanism you identified. For a systematic timing difference, comparable repeats reveal its repeatability but do not remove the shared bias. For heat loss, increasing temperature-sensor resolution does not recover the missing energy. For a large-angle pendulum, collecting more large-angle timings does not make the small-angle model valid.
Choose a feasible action and explain its limits. “Use better equipment” needs a relevant feature: suitable resolution, calibrated response, controlled trigger or reduced energy transfer. A more elaborate apparatus can introduce new alignment, loading or calibration problems, so check what the change actually improves.
Common mistakes
- Writing “human error” or “reaction time” without a mechanism. Say which delay is larger and which way that shifts the result.
- Proposing repeats for a systematic error. Repeats reduce random scatter but leave a shared bias unchanged.
- Proposing “better equipment” without naming the feature that acts on the cause, such as resolution, calibration or a controlled trigger.
- Taking more readings to fix a model limitation. Readings under the same unsuitable conditions cannot make the model valid.
- Claiming an improvement removes every effect. State what it reduces, what remains and any new limitation it brings.
Spotted an error? Report a correction
Syllabus and review details
- GCE A-Level H2 Physics 2027 · 2027
Content Overview, PDF pages 9–10; Subject Content, PDF pages 11–30