Did you know that a school project can contain a completely accurate graph and a conclusion that contradicts it? The mistake may hide in a single phrase such as ‘every day’ or ‘because of our design’. It can survive even when three pupils have checked the spelling and the sums.
The answer is a claim-to-graph check: underline each statement in the project conclusion, trace it to the exact plotted values, group, time window and unit, and ask whether the graph supports, contradicts or cannot establish it. Repair the claim before polishing the writing. This is a separate skill from checking that a table adds up correctly.
Why the last paragraph is often less reliable than the chart
In a group task, the person writing the conclusion may not be the person collecting data. A chart can be revised after an early summary was written. Students may also turn a visible pattern into an explanation of its cause without recognising that the chart recorded only a pattern. A good project makes the relationship between data and words inspectable, even by somebody outside the group.
A thoughtful conclusion answers the original question with the evidence actually collected. It should not merely sound persuasive. Read each important sentence slowly: what precise observation would make it true? Which observation might show that it was too strong? This habit teaches pupils to respect the reader, not just the marks.
Worked example: two reading corners and a mistaken claim
Imagine an entirely fictional school project counting visits to two reading corners. Corner A recorded 8 visits on Monday, 12 on Tuesday and 10 on Wednesday. Corner B recorded 12 on Monday, 9 on Tuesday and 13 on Wednesday. These are illustrative numbers, not real pupil results. The group draws a correct bar chart with six labelled bars.
Their first conclusion reads, ‘Corner B attracted more children every day, so its design made it more popular.’ Neither part is justified as written. On Tuesday, A had 12 visits while B had 9. Also, a visit is not necessarily a unique child; a student can enter twice. The graph contains no controlled test demonstrating that interior design caused the differences.
- Monday: A 8, B 12. B recorded four more visits.
- Tuesday: A 12, B 9. A recorded three more visits.
- Wednesday: A 10, B 13. B recorded three more visits.
- Total A: 8 + 12 + 10 = 30 recorded visits.
- Total B: 12 + 9 + 13 = 34 recorded visits, a difference of four over the three-day period.
An evidence-faithful revision is: ‘Across the three recorded days, Corner B received 34 visits and Corner A received 30. Corner B had more visits on Monday and Wednesday, while Corner A had more on Tuesday.’ Each sentence can be checked against an actual bar. It does not silently change ‘visits’ into ‘children’ or a three-day finding into an eternal rule.
Check the meaning of the unit before comparing numbers
The phrase ‘more children’ hides a measurement problem. The class counted visits, not distinct people. A reader might picture 34 different children, but the recorded dataset may include repeat visits. If pupils want to learn about distinct readers, they need an ethically appropriate method of counting individuals without unnecessarily collecting personal information. If they want to describe activity at the reading corners, visit counts can be perfectly useful—provided they are labelled honestly.
The same distinction matters in science and Mathematics: measuring height in centimetres is not the same as measuring growth in centimetres; recording the number of questions answered is not the same as recording how many questions a student understands; counting attempts is not necessarily counting people. Each unit answers its own question.
An evidence-check method every learner can practise
- Circle the claim word: every, most, doubled, improved, caused, increased or decreased.
- Point to the specific bars, points or table entries that the claim uses.
- Read the title, axes and units before interpreting the shapes.
- Compare the relevant values rather than trusting the visual impression.
- Mark the claim as supported, contradicted or not established.
- Rewrite the sentence using the same unit, group and observation window as the evidence.
This is not a call to weaken every sentence. ‘B had more visits on Monday’ is a confident statement supported by the fictional record. The problem is confidence that extends beyond the plotted data. Pupils need to learn the difference between justified certainty and unnecessary exaggeration.
Supported, contradicted, and not established are three different outcomes
A claim is supported when relevant observations make it reasonable within the stated scope. ‘B had 34 visits during the recorded period’ is supported by the totals. A claim is contradicted when a relevant observation shows it is false: ‘B led every day’ fails on Tuesday. A claim is not established when the evidence does not settle it: ‘Children like B more because its wall colour is blue’ would need additional evidence and an appropriate way to evaluate alternative causes.
This three-way distinction is more useful than a quick true-or-false mark. It teaches children that ‘we have not proved it’ is not equivalent to ‘we proved the opposite’. Parents can use these terms at the dinner table, and teachers can use them when checking a group’s work without making the process confrontational.
Why correct arithmetic does not guarantee a correct conclusion
If the class adds 30 + 34 and reports 64 visits, the arithmetic is correct. It still does not follow that 64 different children participated. Nor can those totals tell us whether the corners were open for equal hours. If one corner was unavailable for part of Tuesday, that needs to be documented before interpreting activity differences.
Numbers can be computed correctly while their real-world interpretation remains wrong. That is why the Core Aim guide to checking data-table totals solves an earlier task in the sequence, while the present lesson checks what a conclusion is allowed to say once the graph exists.
A second worked case: an upward endpoint is not a steady rise
Another fictional class graphs the number of weekly book loans: 20 in week 1, 24 in week 2, 21 in week 3 and 26 in week 4. The final point is higher than the first, and week 4 has the largest count. But the sentence ‘Loans rose every week’ is false because week 3 fell from 24 to 21. A better conclusion says, ‘Loans fluctuated, ending at 26 in week 4, above the initial 20.’
Next challenge the learner: can the line alone prove that a new display caused the increase? No. Other factors may have changed, and the design of the investigation matters. A descriptive trend is not the same as evidence that an intervention produced that trend. Both can be worth studying, but the report must not swap one for the other.
Teach graph reading and conclusion writing as one connected skill
Begin with a tiny graph and two contrasting sentences. Ask a pupil which bar decides the question. Then give a statement that is plausible but not actually tested, and explain why this differs from a contradiction. Only after the learner can classify those examples should they begin auditing a longer report. That progression prevents the assignment from becoming an unfocused exercise in proofreading.
The New South Wales Department of Education’s evaluating-scientific-data resource provides established guidance for assessing evidence and conclusions. The OECD’s science framework likewise discusses examining data, claims, evidence and limitations. Neither source suggests that one worksheet alone guarantees achievement; they support the importance of this kind of thinking.
What parents can ask after the project is finished
Ask a child to choose the strongest sentence in the conclusion and point to the bar that supports it. If the reply is simply ‘It looks higher’, ask for an exact value and the corresponding unit. If the pupil corrects ‘every’ to ‘two out of three days’, that is a genuine reasoning improvement. The revision matters more than whether the first draft sounded polished.
A parent need not mark an entire school report. Two well-chosen questions can reveal whether the child knows what the chart represents and why the conclusion is fair. When the child is uncertain, return to the table and count the observations together rather than rushing to provide the answer.
What teachers can check after group work
Invite someone who did not draw the graph to explain the conclusion and locate its evidence. This tests whether the group product is understandable and whether learning has travelled beyond one specialist member. Do not use the exercise to penalise the pupil who contributed illustrations or editing; rotate roles so each student gains experience with data and reasoning.
A useful individual exit question is, ‘Which sentence could the graph not justify, and how would you rewrite it?’ It is possible for a group to submit an excellent document while several children still cannot explain its central claim. Short independent checks help distinguish collaborative product quality from each learner’s understanding.
What if part of the graph has missing observations?
Suppose Wednesday was not measured, but a student joined Tuesday and Thursday with a smooth line and described an increasing Wednesday reading. The project has invented a value. Record missing measurements visibly, distinguish interpolation from actual observation, and avoid presenting an estimated value as though somebody collected it.
The earlier Core Aim article on missed observation times explains how to label data gaps. The issue here is later in the learning process: what the final written claim can justifiably infer from the completed display.
An independent transfer check and its answer
Give the learner an unfamiliar chart: a school garden counted 14 butterflies on Monday, 11 on Tuesday and 16 on Wednesday. The conclusion says, ‘Butterfly sightings increased every day, proving the garden is getting healthier.’ Ask the learner to classify both parts. ‘Increased every day’ is contradicted by the drop to 11. ‘Proving the garden is getting healthier’ is not established by three sighting counts alone; weather, observer effort and many other factors can affect what is seen.
A narrower supported description is ‘The recorded sightings were 14, 11 and 16; the highest count was on Wednesday.’ If the pupil can make this repair without seeing a model answer, and explain what additional observations would help test the health claim, the reasoning has started to transfer beyond the reading-corner example.
Limitations: not every chart can be checked by eye
Some data use unequal group sizes, logarithmic scales, uncertainty intervals or statistical models. A simple read-off check is the starting point, not a substitute for the relevant mathematics. The UK House of Commons Library’s guide to potentially confusing charts explains why axes and design choices can affect interpretation. Older learners may need to examine uncertainty and sampling design before reaching a conclusion.
The right response to an ambiguous figure is to check its source, consult the teacher and state the uncertainty. Do not secretly alter graph limits or omit awkward data so the sentence looks better. When reporting information about real classmates, use appropriate privacy safeguards and school-approved data handling.
Frequently asked questions
Can a graph be accurate but still support a misleading story?
Yes. Correct data can be paired with incorrect units, wrong time windows, overstated causal language or a conclusion that ignores part of the series. Accuracy of the plot and validity of the interpretation are separate checks.
Should children avoid words such as always and because?
No. Use them when the actual evidence supports that strength of claim. The habit to teach is matching the language to its burden of proof, not banning useful words.
What if a project conclusion is partly correct?
Preserve the supported part and repair only the unsupported wording. For example, ‘B had more total visits’ can remain while ‘B led every day’ must go.
Is a missing data point automatically a failure?
No. An honestly labelled missing observation can be responsible scientific reporting. It becomes a problem when the learner disguises it or draws unsupported conclusions from it.
How can a learner show progress?
Give a fresh graph after a delay and ask for a short, precise conclusion with units and limits. Look for fewer unsupported claims and more independent explanations, not merely neater presentation.
Continue through the eduKate Education library
Explore how to check whether a source supports a claim for an external-source version of this skill. The eduKate Education library connects good evidence use with independent learning, while the Clementi Mathematics teaching benchmark illustrates diagnosis, clear explanation, progressive practice and a later independent check.
The goal of the school project is not a paragraph that sounds certain. It is a conclusion a reader can trace back through the graph to the observations—and understand exactly what those observations do, and do not, establish.