The evaluation is your critical reflection — the part where you honestly examine the weaknesses in your own method and suggest realistic fixes. This is where you prove you understand the scientific process, not just the calculations. The winning formula runs through the whole section: weakness → impact → improvement, every time, tied to your actual experiment.
📚 What you need to know
The evaluation assesses your data quality and how it affected your conclusion
Distinguish random errors (scatter, affect precision) from systematic errors (offset, affect accuracy)
Discuss weaknesses, limitations, and assumptions in your method
Every weakness needs a specific, realistic improvement a school lab could actually do
Close the loop each time: weakness → impact → improvement
Never blame “human error” — name the specific source instead
The two kinds of error
Almost every evaluation rests on telling these two apart. A random error scatters your readings unpredictably around the true value — it hurts precision. A systematic error shifts every reading the same way — it hurts accuracy. The picture below is the mental model to carry into any lab.
Random errors scatter points evenly around the true value; systematic errors push them all the same way.
Feature
Random error
Systematic error
Effect on data
scatter around the true value
consistent shift one way
Affects
precision
accuracy
Examples
reaction time, parallax, reading error
zero error, uncorrected heat loss
Reduce by
repeat trials & average; time many oscillations
recalibrate; correct the method
Here’s the tell that instantly sorts them: look at your scatter. If your repeats land on both sides of the expected value, that’s random — averaging will help. If they’re all too high or all too low by roughly the same amount, that’s systematic — and no amount of averaging will save you, because the error is baked into the method itself.
Systematic errors show up on the graph
One of the neatest skills in an evaluation is spotting a systematic error straight from your graph. Because a systematic error shifts every reading the same way, it offsets the whole line — the line of best fit no longer passes through the origin when theory says it should.
Same gradient, but the measured line is lifted off the origin — that y-intercept is the fingerprint of a systematic error.
Weaknesses, limitations, and assumptions
Beyond raw errors, a full evaluation also names three other things that shape how much your conclusion can be trusted.
Weaknesses — parts of your method that caused significant error. Example: timing a single swing, where reaction time is huge compared to the short period.
Limitations — boundaries on how far your conclusion applies. Example: only tested up to 1.00 m, so you can’t claim the relationship holds for very long pendulums.
Assumptions — simplifications that aren’t perfectly true. Example: treating the string as massless and air resistance as negligible.
A quick way to keep these straight: a weakness is something you did that added error, a limitation is a fence around where your conclusion is allowed to roam, and an assumption is a small “let’s pretend” you made to get the maths going. Examiners love seeing all three named specifically for your experiment — generic ones that could apply to any lab earn little.
Close the loop: weakness → impact → improvement
This is the single most important habit in the whole evaluation. For every weakness, don’t just name it — explain its impact on your result, then give a specific, realistic improvement. An improvement must be relevant (it fixes that exact weakness) and realistic (doable in a school lab).
Weakness what went wrong
leads to
Impact effect on result
fixed by
Improvement specific & realistic
WE 1
In a pendulum experiment for g, the length was measured to the bottom of the bob, not its centre. Write this up as weakness → impact → improvement.
Weakness (systematic)
Length L was measured to the bottom of the bob, not its centre of massImpact
Every L is consistently too short — since T ∝ √L, this shifts the whole dataset one way, biasing gImprovement
Measure to the bob’s geometric centre; find its diameter with vernier callipers and add half to the string length
A specific, realistic fixSee how it names a concrete cause, traces it to a consistent (systematic) shift, and gives a fix any school lab can do? That’s the full loop.
WE 2
The period was found by timing 20 oscillations with a hand stopwatch. Write the loop for the reaction-time issue.
Weakness (random)
Human reaction time when starting/stopping the watch
Impact
Adds scatter to the timings — visible as spread in repeats and points off the best-fit line
Improvement
Use a light gate at the bottom of the swing to record the period automatically, removing reaction time
Precision improvedTiming 20 swings already shrinks the error per swing; the light gate removes the human element entirely for an even tighter result.
💡 Top tips
Never write “human error” — name it: parallax, reaction time, a reading error.
Prioritise the one or two errors with the biggest impact on your result.
Close the loop every time: weakness → impact → improvement.
Refer to your own data, graph, and observations — not a generic experiment.
⚠ Common mistakes
Blaming vague “human error” instead of a specific source.
Suggesting unrealistic improvements (a bomb calorimeter in a school lab).
Listing weaknesses with no impact and no improvement.
Writing a generic evaluation that could fit any experiment.
Confusing a limitation (scope) with a weakness (source of error).
Quick recap:Random errors scatter (hurt precision); systematic errors shift one way (hurt accuracy) and offset the graph’s line. Name weaknesses, limitations, assumptions specific to your lab. For each, close the loop: weakness → impact → improvement, with a fix that’s relevant and realistic. Never write “human error.”
That completes the whole inquiry cycle — from collecting data all the way to critically evaluating your own method. These reflection skills are exactly what lift an Internal Assessment from a solid piece of work to a top-band one, and they carry straight into Paper 3 data questions too. From here, you’re ready to put the whole toolkit to work on the physics itself, starting with Motion, Forces & Energy.
Want an evaluation that hits the top band?
Book a free meeting and we’ll practise separating random from systematic errors, reading offsets off your graph, and closing the weakness–impact–improvement loop — the analysis that earns the highest IA marks.
