Did you know that the words “shows,” “suggests,” and “proves” can change the truth of a school research paragraph even when all its numbers are correct? A pupil may copy the right percentage and still draw a conclusion that the study never tested.
Evidence strength describes how far available observations support a particular claim. The immediate answer is to name exactly who or what was studied, what was measured, what comparison was made, and what remains uncertain. Write the strongest conclusion the evidence actually supports—not the strongest conclusion that sounds impressive. “Suggests” is useful when evidence points towards an interpretation but important alternatives remain; “proves” is rarely justified by a single classroom study.
Diagnose the mismatch between data and claim
Give a learner this sentence: “Our survey proves that quizzes make all students learn better.” Before correcting the grammar, ask for the actual question, results, number of respondents and learning measure. If the only item asked which activity respondents liked, the study measured a preference, not an improvement in learning. The category of evidence is wrong even if every vote was counted accurately.
Underline the verb in a conclusion and ask a second question: what kind of study would be needed to justify that verb? A survey of enjoyment cannot on its own show causation; a single group comparison can be affected by group differences, changing conditions and uncertainty. Strong writing begins by protecting this distinction.
Worked example: twenty-four responses, one tempting overclaim
Invented example for practice, not a report of real students: A teacher asks twenty-four class members which of two revision activities they prefer: a short self-test or rereading notes. Fifteen choose the self-test and nine choose rereading. Assume all twenty-four answered once, and that these are the only options.
The calculation is 15 ÷ 24 × 100 = 62.5% selecting self-tests; 9 ÷ 24 × 100 = 37.5% selecting rereading. The percentages total 100%. That is all the arithmetic demonstrates. The survey did not measure test results, memory after one week or causes of students’ preferences.
Compare three sentences
- Too strong: “The survey proves that self-tests cause better marks for every student.” Neither marks nor cause were measured.
- Accurate: “Among the 24 pupils who answered this classroom question, 15 (62.5%) selected a short self-test as their preferred activity.” This states exactly what the responses show.
- Careful interpretation: “These answers suggest that short self-tests may be an acceptable revision option for many respondents; they do not establish which method produces greater learning.” This clearly separates observation and inference.
Suppose the survey excluded absent pupils, or the wording described quizzes as “fun and easy” but rereading as “boring.” Those details could affect interpretation. The Pew Research Center’s guidance on survey questions explains why wording, ordering and asking one concept at a time matter. A well-phrased claim cannot repair a leading survey, but it can avoid hiding the limitation.
A four-question evidence ladder
- Who or what? State the actual people, period, text, samples or measurements. Avoid broad words such as “everyone” when the evidence covers one class.
- What was observed? Quote or calculate the measured quantity, with its unit and denominator. Preferences, scores, attendance and recall are not interchangeable.
- What comparison exists? Was there a fair comparison, an earlier baseline, a control or merely a single observation?
- What cannot be ruled out? Note relevant alternative explanations, missing data, sample limitations or variation. Do not invent weaknesses just to sound scholarly.
Then choose a precise verb. Reports is suitable for describing what a source says; found can describe its measured result; is consistent with compares evidence with an explanation; suggests offers a guarded interpretation; demonstrates should be confined to what the design actually demonstrates. Replacing every verb with “may” is not careful reasoning. It can conceal what is known.
Why students overstate conclusions
Learners often feel pressure to finish an argument decisively. A bold final sentence sounds more certain than a specific, limited statement. Another trap is confusing the purpose of a source with the strength of its findings: a paper may investigate improvement without concluding that it has been established. A third trap comes from reading a summary or search snippet without checking the study’s method.
For source checking, pair this exercise with checking a search-result snippet against its original source. For reading passages, use questions, evidence and inference annotations. Both help the reader preserve where information came from.
A three-stage lesson: model, compare, transfer
Stage 1: adult models the claim repair
The adult reads the fictional survey result and narrates the check: “The denominator is 24 respondents, not the whole school. The question asked for preference, not improvement. Therefore I will report preference, give the numerator and denominator, and avoid a causal verb.” The learner highlights which phrase fixes each defect.
Stage 2: practise with a partly changed claim
Give the pupil another invented data sentence: “In five of six observation periods, the garden was warmer at noon than at 9 a.m.” Ask them to rewrite the overclaim “Gardens are always warmer at noon.” A suitable revision is “In five of the six periods recorded in this garden, noon was warmer than 9 a.m.; this small observation set does not establish an always-true rule.”
Stage 3: independent new context
Provide an unseen source about library book borrowing or a small science investigation. Without a model sentence, the learner writes three lines: one observation, one reasonable interpretation, and one limitation. They then point to the exact evidence behind each. A teacher can compare first-attempt and later independent explanations for fewer unsupported leaps.
How learners, parents and teachers can check progress
Look for observable changes: the learner states denominators, distinguishes opinion from performance, identifies at least one meaningful limitation, and rejects a stronger claim when supporting data are missing. A parent can ask “What did they actually measure?” A teacher can mark both factual accuracy and inferential restraint; a grammatically polished conclusion with the wrong causal claim needs conceptual feedback, not just editing.
Some evidence genuinely supports confident statements within narrow limits: for example, fifteen out of twenty-four recorded selections are 62.5% by arithmetic. Uncertainty belongs in the inference about why pupils chose an option or whether the result applies beyond those respondents, not in that straightforward division.
Limits and ethical use of evidence
A source may contain flaws that are not visible in a short abstract; full methods may matter. Some research uses sophisticated designs that justify stronger causal language than a classroom survey. Do not reduce all evidence to the same weak status. Equally, do not present invented classroom numbers as published research, or erase inconvenient observations after seeing the conclusion.
Frequently asked questions
Is “suggests” always safer than “shows”?
No. When a table unambiguously records 15 responses in a category, it is fair to say the table shows 15 responses. “Suggests” is more appropriate when moving from those observations to an interpretation not fully established.
Can a large sample prove the explanation?
A larger sample may reduce some sampling uncertainty, but it does not automatically remove biased wording, confounding or mistaken measurements. Design and question matter as well as size.
Does careful wording make writing less persuasive?
It makes the reasoning more trustworthy. A reader who understands precisely what has and has not been established is better equipped to use the result.
Continue learning
Try taking an existing paragraph and labelling every sentence as measurement, source report, interpretation or recommendation. Keep those categories separate, then connect them clearly. Explore the eduKate Education Hub and the Clementi Mathematics teaching reference for the same habit: diagnose the actual thinking step, explain the rule and check whether the learner can use it independently.