Did you know that a student can make a science report less accurate by adding more ‘possible errors’ to the last paragraph? A familiar school conclusion says the results were unreliable because someone ‘probably read the ruler wrongly’, ‘might have pushed too hard’ and ‘could have forgotten the stopwatch’. If nobody observed those mistakes, listing them as events makes the report sound less truthful, not more scientific.
An experimental limitation is a real constraint on what a study’s methods and evidence can establish. The immediate answer is to write what was observed, identify what the design cannot rule out, explain how that affects the conclusion, and suggest a realistic improvement. Separate documented problems from plausible risks and never claim an accident occurred merely because it could have occurred.
Start with three very different statements
- Observed: ‘Three paper aeroplane flights landed at different distances.’ This is supported by the recorded measurements.
- Documented method limit: ‘Only one paper aeroplane design was tested.’ This limits generalisation even if all measurements were accurate.
- Possible source of variation: ‘Differences in release technique might have affected distance, but we did not measure release technique.’ This is a hypothesis, not a confirmed mistake.
These distinctions help the reader understand both the value and boundaries of the result. A cautious sentence can be stronger than a confident list of imagined faults.
A worked example: three paper aeroplane flights
Consider an invented classroom exercise. One student folds a paper aeroplane according to the same pattern and launches it across a suitable open indoor space. A partner measures the distance from a fixed launch line to the first point where the plane touches the floor, to the nearest 0.1 metre. Three recorded distances are 12.4 m, 12.8 m and 11.9 m. These numbers are illustrative, not real pupil outcomes.
The mean distance is (12.4 + 12.8 + 11.9) ÷ 3 = 37.1 ÷ 3 = 12.366… metres, reported as about 12.4 m to one decimal place. The maximum recorded distance is 12.8 m, the minimum is 11.9 m, and their difference is 0.9 m. Check: 12.8 − 11.9 = 0.9.
What can we say? In this example, the three recorded flights had an average measured distance of about 12.4 m and a recorded spread of 0.9 m from shortest to longest. We cannot infer from three flights of one plane that this paper design performs better than a different one. Nor can we say how much of the spread came from the throw, air movement, measurement or the plane itself, because those contributions were not separately measured.
A responsible limitations paragraph could read: ‘Only three flights of one paper aeroplane were measured, so the result gives a limited picture of that launch setup and does not compare designs. The recorded flights ranged from 11.9 m to 12.8 m. Because launch angle and speed were not measured, their influence cannot be separated from other causes of variation. A next investigation could use a clearly specified release procedure, more recorded trials and a planned comparison between designs while keeping relevant conditions as consistent as practical.’
Notice what the paragraph does not say. It does not claim a pupil definitely pushed harder in the second flight. It does not announce that the metre measure was faulty. It does not turn the 0.9 m range into a formal measurement uncertainty interval. The range is simply the largest observed result minus the smallest; a defensible uncertainty analysis requires attention to method, instrument, variability and assumptions.
Why a limitation is not the same as a mistake
A limitation can exist even when everyone follows the instructions flawlessly. Measuring only one class, testing only one plant variety, observing for only one day, or using a limited range of temperatures all constrain the claims. By contrast, a mistake such as writing down the wrong number is an error in executing or recording the procedure. If a mistake occurred and is known, document it. If it is merely a possibility, label it as a potential source of error and explain how it could be investigated.
The US National Institute of Standards and Technology’s guidance on measurement uncertainty treats measurement uncertainty as a reasoned assessment using information about the measurement process, not a list of untested guesses. School students do not need to reproduce a metrology laboratory’s calculations, but they can learn the same basic honesty about the evidence.
How to diagnose weak limitations paragraphs
The paragraph lists everything that might happen
A report says, ‘Human error, wind, reading error, stopwatch error and equipment error.’ None is connected to a recorded condition or the result. Ask the learner: ‘Which part of the method makes each risk relevant? Did you observe it, or is it only possible? How would it change what you conclude?’ Remove vague items that cannot be explained.
The paragraph says the results prove nothing
A cautious learner writes, ‘There were only three flights, so the experiment failed.’ That is not necessary. The three measurements still describe those flights. State that valid narrower conclusion, then say what cannot be generalised. Honest limitations are boundaries, not a declaration that all observations were worthless.
The paragraph promises the improvement will fix everything
A learner suggests ‘repeat 100 times to eliminate all error’. Repetition can provide more information about variation, but it does not automatically remove systematic bias or changes in technique. An improvement should target a named weakness and allow a better test, without claiming perfection.
A practical writing structure: evidence, boundary, effect, next check
- Evidence: name a real feature of the data or recorded method. Example: three flights of one design, measured to the nearest 0.1 m.
- Boundary: name a question that evidence cannot settle. Example: whether the design outperforms a competing design.
- Effect: explain why that limitation matters to the conclusion. Example: the result is descriptive of one setup, not a comparison.
- Next check: describe a feasible change that would address the boundary. Example: compare specified designs using consistent launch rules, repeated measurements and the same distance definition.
A teacher can annotate a draft with four small symbols for those four functions, then ask a learner to rewrite the paragraph in their own words. The goal is to make the relation between evidence and claim clear, not to force the same stock sentence into every practical report.
Guided practice, then independence
Guided task: give the paper-aeroplane record and four possible sentences: ‘Three flights were measured’; ‘Someone definitely threw harder’; ‘Only one design was tested’; ‘The mean was about 12.4 m.’ Ask pupils which are observations or method facts, which are unsupported, and which summarise the calculations. Make them justify each label.
Paired task: introduce a second fictional setup where one flight was interrupted by a person walking through the landing area, and the student recorded that event in the notebook. Partners now have a documented disruption. Ask how to label that trial, whether to repeat it under consistent conditions and how to preserve the original record. An interrupted trial should not be silently replaced by zero.
Independent task: change subjects. A student observes one potted seedling for seven days and records height daily. Ask for a truthful limitation. An appropriate answer is that one seedling over seven days cannot establish a growth pattern for all seedlings or isolate influences that were not controlled. An inappropriate answer is ‘the ruler was certainly read wrongly’ unless the record actually shows that.
Transfer task: ask learners to read a short history or survey conclusion and identify its sampling or evidence boundary. ‘We interviewed three people’ is a real method feature; ‘three people represent everyone’s experience’ is an unsupported leap. The skill applies wherever a conclusion may outrun its evidence.
How parents and teachers can recognise progress
- The learner distinguishes a recorded event from a hypothetical error without being reminded.
- The limitations paragraph includes at least one exact method feature or numerical result.
- Every proposed improvement responds to a named limitation rather than offering a generic ‘do it more carefully’.
- The pupil can state one conclusion that still follows from the data and one that does not.
- On a new scenario, the pupil avoids claiming uncertainty has been precisely quantified when no such analysis was done.
Parents can ask: ‘What did you actually notice?’, ‘What cannot be known from these measurements?’ and ‘What would you do differently to find out?’ A calm, specific conversation usually helps more than insisting the child add a fixed number of bullet points called ‘errors’.
Limits to the method itself
There is no universal template that covers every practical investigation. A descriptive observation, a controlled comparison and an advanced measurement study have different evidential needs. The right uncertainty language also depends on instrument properties, scientific conventions and what the investigator can reasonably verify. For curriculum-specific marking expectations, follow the actual assignment’s instructions and teacher guidance; this article offers a general reasoning method, not a substitute for a particular marking rubric.
For a related problem, read how to document a change to a science procedure. A change that actually occurred should be recorded precisely, unlike an imagined mistake. For another careful distinction, see why no visible change does not necessarily mean no effect.
Frequently asked questions
Must I mention ‘human error’ in every report?
No. Identify the specific human action that might matter, explain why, and say whether it was observed. Generic labels do not help the reader evaluate the study.
Is having only three measurements automatically wrong?
Not automatically. The appropriate number depends on the purpose, variability and constraints. But three measurements provide limited evidence for broad comparisons or general conclusions. State the scope accurately.
Can I mention something I suspect happened?
Yes, if you make the uncertainty explicit: ‘It is possible that…’ is different from ‘This happened because…’. Ideally identify a way to check the suspicion.
Is the range of my readings the instrument uncertainty?
No. The range describes the observed spread. Instrument resolution and other measurement-process information may be relevant to a formal uncertainty assessment; the range alone does not establish one.
The science habit worth keeping
A strong limitations paragraph does not apologise for being human. It tells a reader what the study found, why the conclusion has a boundary, and which next observation could move that boundary. The lesson is part of the same explanation-first, independently checked approach in the eduKate Education Hub and our Clementi learning reference. Describe reality as carefully as you can; where the evidence stops, let the writing stop too.