Mini Physics
Part of Mini EducationSwitch subject

Analysing and Evaluating A-Level Physics Experiments

Draw evidence-based practical conclusions, compare results with uncertainty, and write causal limitation–effect–improvement evaluations for A-Level Paper 4.

  • GCE A-Level H2 Physics 2027
On this page

Learning objectives

  • Analyse practical data, graphs, gradients and intercepts
  • Evaluate practical limitations and propose specific improvements

1. Conclusions must follow the evidence

A conclusion states what the processed data support, within the measured range and uncertainty. It should identify the trend or constant, quote relevant evidence and avoid claiming more than the experiment tested.

Examples:

  • “T² is proportional to L within the scatter because the best-fit graph is linear and its intercept is consistent with zero within uncertainty.”
  • “The component is non-ohmic over the measured range because V/I is not constant as current increases.”

2. Compare values using uncertainty

For a result x±Δ x, compare an accepted value with the interval rather than comparing rounded central values alone.

Percentage difference from a reference value x_ref is (|x-x_ref|)/|x_ref| × 100%.

This is not automatically the same as percentage uncertainty. State whether the reference value lies within the experimental uncertainty range, while recognising that agreement does not prove the method has no systematic bias.

3. Build a causal evaluation

Evidence-based experiment evaluationThree linked boxes show a named limitation causing a directional effect in the data and leading to a targeted improvement.1. Limitationtimer starts late becauseof reaction timename the mechanism2. Effectmeasured time is too short;calculated speed is too largestate the directionwhere possible3. Improvementuse a light gate triggeredby the moving objecttarget the mechanismGeneric “human error → repeat” does not establish this causal chain.
Scroll diagram horizontally to read all labels.
A useful evaluation names the mechanism, traces its effect into the data or final quantity, then proposes a feasible change that directly weakens that mechanism.

An evaluation becomes useful when each improvement directly addresses the named mechanism.

Weak statementStronger evaluation
“Human error affected timing.”“Starting the stopwatch after the marker passes makes the measured time too short; use a light gate triggered at the marker.”
“Heat was lost.”“Energy transfer to the cup and surroundings makes the calculated specific heat capacity too large; insulate the vessel and include the container heat capacity in the model.”
“Repeat for accuracy.”“Reaction-time scatter changes repeated periods; time more oscillations per trial and repeat the measurement to reduce fractional random uncertainty.”

4. Random and systematic effects

  • Random effects create unpredictable scatter. Repeats, a longer measurement interval and a wider data range can reduce their influence on a mean or gradient.
  • Systematic effects bias readings or the model in a consistent direction. Zero correction, calibration, a control experiment or a redesigned method is needed.
  • Model limitations occur when the governing approximation fails, such as large pendulum angles or component heating. Taking more readings under the same invalid conditions does not solve the problem.

5. Evaluating graphs

Check:

  • whether the proposed relationship matches the observed shape;
  • whether the intercept is physically meaningful or indicates an offset;
  • whether scatter is consistent across the range;
  • whether one point is anomalous and has been rechecked;
  • whether the fitted gradient is stable under acceptable steepest and shallowest lines when uncertainty analysis is required.

Do not describe every non-zero intercept as “zero error”. It may arise from an omitted physical term, a calibration offset, end correction or extrapolation outside the valid range.

5A. Common mistakes

  • Giving a generic improvement with no stated limitation.
  • Stating an effect without its direction when the direction can be inferred.
  • Claiming repeats improve accuracy when they address only random scatter.
  • Suggesting a higher-resolution instrument when the dominant limitation is heat loss or alignment.
  • Saying an accepted value “proves” the result; experimental evidence supports a conclusion within stated conditions.

6. Worked Examples

Modelled example 1

Evaluating an electrical heating experiment

Core

Problem

An experiment equates electrical energy VIt to mcΔ T to find c, but the vessel is not insulated. Explain the likely effect on calculated c and give a suitable improvement.
Study the worked solution

  1. Name the limitation

    Method

    Some input energy heats the vessel and transfers to the surroundings.

    Reason

    The uninsulated setup does not confine all electrical input to the sample.

    Working

    VIt = Qₛₐₘₚₗₑ + Qᵥₑₛₛₑₗ + Q_surroundings

  2. Expose the model error

    Method

    The model treats the full VIt as mcΔ T for the sample alone.

    Reason

    It omits the other receiving stores and environmental transfer.

    Working

    VIt > Qₛₐₘₚₗₑ = mcₜᵣᵤₑΔ T

  3. Infer the direction

    Method

    The calculated c = VIt/(mΔ T) is too large.

    Reason

    The numerator assigned to the sample exceeds the energy it actually gained.

    Working

    c_calculated > cₜᵣᵤₑ

  4. Improve the apparatus

    Method

    Insulate the vessel and use a lid.

    Reason

    These controls reduce energy transfer to the surroundings.

    Working

    Q_surroundings↓

  5. Improve the model

    Method

    Determine or include the vessel heat capacity and use cooling data if environmental transfer must be estimated.

    Reason

    This accounts for important energy paths that cannot be eliminated completely.

    Working

    VIt = (mc + Cᵥₑₛₛₑₗ)Δ T + Qₗₒₛₛ

Return to the A-Level Practical Hub, then take the Practical Quiz.