Did you know that two trustworthy sources can report different numbers without either source lying? A student often encounters this when a textbook, public dashboard, survey or newspaper gives a figure that does not match another one. The tempting response is to choose the larger number, average them or quote whichever supports the assignment. None of those moves explains the disagreement.
Source reconciliation means testing why two apparently conflicting claims differ by checking what was measured, when, where, among whom and by which method. The immediate answer is to preserve both claims accurately, compare their definitions and evidence, find any compatible explanation, and write a conclusion that stays within what the sources actually support. When the conflict remains unresolved, say so rather than manufacture certainty.
The five checks to make before choosing a claim
- Exact wording: copy the source’s claim accurately, including units, labels and whether it says ‘about’, ‘at least’ or ‘estimated’.
- Population and place: identify which people, objects, schools, locations or sample the figure concerns.
- Time: separate a date of observation from a date of publication or later revision.
- Method: ask how information was gathered, who was excluded, and whether uncertainty or rounding matters.
- Original evidence: trace a quotation or statistic to its data table, study or original document rather than trusting a chain of summaries.
This is not an invitation to dismiss every answer as ‘just an opinion’. Some claims really are mistaken. The checks help learners distinguish an actual error from two correct descriptions of different things.
A worked example: two reports, two denominators
Imagine a fictional school garden project. An early report states, ’18 out of 30 pupils in the first survey group said they preferred watering the plants in the morning.’ A later report states, ’18 out of 36 respondents said they preferred morning watering.’ A pupil summarises them as ‘One source says 60%; the other says 50%; therefore the results are unreliable.’ That conclusion is too quick.
Check the calculations. For the earlier group, 18 ÷ 30 × 100 = 60%. For the later group, 18 ÷ 36 × 100 = 50%. Both percentages are arithmetically correct. If six additional people answered the same question and none chose morning watering, the numerator could stay 18 while the denominator rose from 30 to 36. The result would change from 60% to 50% without anyone changing their original response.
But we must not pretend we know how the later group was formed. The six-person explanation is only one plausible reconstruction, not a documented fact. The responsible student looks for the collection dates, response lists, question wording and eligibility rules. If those records are unavailable, the report can say: ‘The percentages use different numbers of respondents (30 and 36), so they are not directly interchangeable. The source documents do not establish why the samples differ.’
Those numbers, reports and pupils are invented instructional examples, not records from an actual eduKate class. The arithmetic illustrates a general rule: a percentage can change because the whole changes even when the counted number stays fixed.
A second example: when both temperature readings can be right
A classroom display says the air temperature was 28°C at 9:00 a.m., while another log says 32°C at 1:00 p.m.. A learner who copies only ‘temperature: 28°C’ versus ‘temperature: 32°C’ may see a contradiction. Once the times are restored, the two entries can both describe the same day at different hours. That does not prove the air warmed by exactly 4°C continuously between those times; it establishes only a 4°C difference between the two recorded readings, assuming instruments and locations are comparable.
A more serious mismatch arises if one sensor is in direct sunlight and another in a shaded shelter. Before describing an environmental trend, the report must check placement, calibration, time and conditions. A comparison without these details may be misleading even when both readings are accurately transcribed.
Diagnose the disagreement, not the student
Ask learners to point to the smallest piece of missing information that could decide whether the claims conflict. Some pupils need practice reading axis titles or units; others mistake publication year for observation year. Others can calculate correctly but treat two sample populations as if they were the same. Different errors need different feedback. ‘Find better sources’ is too vague when the actual gap is a changing denominator.
- If the learner drops units, have them rewrite both claims with units and measurement conditions.
- If they confuse populations, ask them to box the words ‘all pupils’, ‘respondents’ and ‘one class’.
- If they average incompatible values, ask what an average would mean and whether both groups have comparable weights.
- If they jump to ‘one source is fake’, ask what evidence would disprove each of the possible explanations.
- If they cannot decide, practise a qualified sentence: ‘The sources differ; the available information does not yet establish why.’
Build a source comparison note a reader can audit
A useful note has two source rows and four comparison fields: claim, measurement basis, date and original link. Add a final row for the unresolved question. Save the authors or organisations and page titles, not merely the names of websites. If a web page changes, an access date can be useful, while an archived version or official update history may clarify revisions.
For practice in bringing evidence together rather than stacking quotations, the Purdue Online Writing Lab’s guidance on synthesising sources explains how researchers connect information from several texts. Its source evaluation guidance also supports checking authorship, relevance and reliability.
From guided comparison to independent research
Guided: give two short garden survey snippets with 18/30 and 18/36 clearly printed. Students calculate both percentages, circle the two denominators and state why the headline figures differ. The teacher models how to resist inventing an explanation.
Paired: remove the explicit respondent counts. Provide a brief source note saying one report covers a morning session and another covers the full day. Partners draft two possible explanations and a question they would ask the original data owner. They must separate what is known from what remains hypothetical.
Independent: present an unfamiliar pair of statements: ‘The library recorded 240 borrowing transactions this month’ versus ‘The library served 190 different borrowers this month.’ The learner should explain that a transaction and a person are different counting units; one borrower can make more than one loan. They should avoid calculating a per-person borrowing rate without confirming that the 240 transactions and 190 borrowers cover the same period and population.
Transfer: invite the learner to apply the same method to two historical sources, science measurements or public data reports. A correct answer should identify each source’s basis and limits, not just decide which one seems more authoritative.
What parents and teachers can look for
- The learner quotes both original claims without altering their scope.
- A comparison chart includes units, dates and denominator or population where relevant.
- The pupil can say ‘not enough information’ and name precisely what is missing.
- A new, unrelated pair of sources is reconciled or correctly left open without teacher prompting.
- A final report carries an appropriately qualified conclusion, not a forced single figure.
Progress is observable when the student stops treating every difference as a battle for a winner and begins testing the conditions behind each statement. Parents can ask, ‘Are those two numbers counting the same thing?’ and ‘Which source would let you check?’ rather than deciding the right source for the child.
Important limits
Sometimes no source gives sufficient methods to settle a dispute. A student should then avoid presenting a speculative reconstruction as fact. Two apparently independent websites may reproduce the same original report and therefore supply only one evidence chain; our related guide on tracing multiple websites back to one original source helps with that problem.
Be especially cautious with health, safety, legal or admissions claims. Do not use a classroom reconciliation exercise to make a real-world decision when primary, authoritative and current information is needed. Similarly, the correct method for combining statistics depends on the study design; students should not pool two samples simply because they have numerical results.
Frequently asked questions
Should I always trust the newer report?
No. A newer publication can contain older observations. Compare observation dates and methods before deciding that one replaces the other.
Should I average two conflicting percentages?
Not automatically. Percentages may use different sample sizes, overlapping people, different questions or different periods. Resolve the bases first; an unweighted average may mislead.
Can two authors disagree without either being dishonest?
Yes. They may define the population differently, examine different evidence or apply different reasonable interpretations. A factual contradiction still needs checking, but difference alone does not establish deception.
A useful conclusion
Education asks learners to move from collecting facts to understanding what those facts describe. The strongest research paragraph does not hide conflicting evidence; it explains what could cause the difference and where the available knowledge stops. Continue with comparing surveys whose questions changed, the eduKate Education Hub and the diagnostic learning reference for more ways to turn uncertainty into careful, independent reasoning.