Reliability, usefulness and corroboration
This page is Question Three
- Every 91231 paper since 2022 has ended with a question about the sources themselves. Reliability of one source; usefulness of two; corroboration against others.
- It is worth a third of the paper, and it is the question most often left unattempted. That combination makes it the highest-value page in this topic.
- Three terms do the work, and they are not interchangeable: reliability, usefulness, corroboration.
Reliability — how much can this source be trusted about what it claims?
- Reliability is always reliability for a particular claim. No source is reliable in general. A handbill is unreliable about a candidate's merits and perfectly reliable about what her committee printed.
- Four questions establish it.
- Was the creator in a position to know? Present, informed, with access to the records?
- How close in time? Same week, or sixty years later?
- What interest did they have in the answer? Something to gain, defend, or sell?
- Does anything else confirm it? This is corroboration, below.
- Distance in time cuts both ways. A contemporary account has detail and no hindsight; a later account has hindsight and degraded detail. Say which you are getting.
- Bias is not a verdict, it is a direction. State which way the source leans, why its creator leaned that way, and what it remains good evidence for.
Usefulness — what can this source do for the historian's question?
- Usefulness is reliability plus relevance, measured against a specific question. The same source can be useless for one question and essential for the next.
- A source can be unreliable and highly useful. A propaganda poster tells you nothing true about the war and everything about what the government wanted believed.
- A source can be reliable and useless. An accurate table of rates income is impeccable and answers nothing about attitudes.
- The marking asks for strengths and weaknesses. The 2024 report names this exactly: Merit candidates identified both the strengths and weaknesses. Writing only limitations is half an answer.
- Say what it is useful for. Finish every usefulness judgement with the sentence "so it is useful for X and not for Y."
Corroboration — checking a source against other sources
- Corroboration is cross-checking, and 2024's Question Three required it outright: how reliable is Source H, using evidence from any TWO of Sources A–G.
- Three outcomes, and all three are worth marks.
- Agreement between independent sources strengthens both. Independent is the key word.
- Disagreement is not a failure — it locates the thing worth explaining, and explaining why two sources differ is a higher-level move than noting that they do.
- Silence — one source covers something no other mentions — leaves the claim unsupported, and saying so is an accurate finding.
- Two sources that share an origin do not corroborate each other. A newspaper report and a letter printed in the same newspaper passed through the same editor. Sources that agree because they came from the same place are one source wearing two hats.
- The strongest corroboration crosses source types: a contemporary record against a later history, a personal recollection against a table of figures.
Putting a judgement in writing
- Name the source and the claim. Source G is being used here as evidence about the 1911 campaign.
- State the position to know. Its speaker was present, but as a child, and speaking in 1978.
- State the interest. She has no stake in the outcome, but she is recalling an event that had since become celebrated.
- Corroborate. Source F confirms a woman stood in 1911; nothing confirms her account of the meeting.
- Land the judgement, split. So it is reliable for what the campaign felt like and how it was later remembered, and unreliable for the sequence of events.
- Five sentences. That structure earns Merit reliably and, with a raised question at the end, reaches Excellence.
Worked Example
Worked example
A Question Three asks: How reliable is Source G for a historian studying the 1911 Waipuna election campaign? Use evidence from any TWO of Sources C–F to support your answer.
Source G (invented) is an oral history recorded in 1978. The speaker, then aged about eight, recalls "the whole street" turning out for a meeting, and "everybody" saying Miss Renwick would win.
Answer:
Step 1 — establish the position to know, in my own words. The speaker was present in Waipuna in 1911, which no other source in the set can claim. But she was about eight years old, and she is speaking sixty-seven years later. So she has direct exposure and heavily degraded recall — a combination that makes her strong on impression and weak on fact.
Step 2 — identify the interest, which is subtler than it looks. She has nothing to gain from the account. But she is remembering an event that had become significant by 1978, and memories of events that later mattered tend to be reshaped to fit what came after. Her "everybody said she would win" is exactly the kind of detail hindsight manufactures.
Step 3 — corroborate, using two sources with different origins.
Against Source F, the table of election results: it confirms a woman did stand in 1911, so the core of her memory is sound. It also shows no woman was elected until 1927. If everybody expected Renwick to win, she lost anyway — so either the expectation was not general, or it was general and wrong. Source F does not contradict Source G, but it makes "everybody" hard to sustain.
Against Source E, the handbill: Renwick's committee campaigned on nine years of committee service and an argument from domesticity. A committee confident of victory does not usually spend its space justifying the candidate's right to stand. The handbill's defensiveness sits badly with a memory of assured triumph.
Step 4 — land the split judgement. Source G is reliable evidence that the 1911 campaign was a visible public event that made an impression on the town, corroborated by the fact that a campaign existed at all. It is unreliable evidence for what people expected, because the one contemporary source about the campaign's own strategy points the other way, and the result points the other way too.
Step 5 — raise the question. Was the speaker's family connected to the campaign? A child of a supporter would remember "the whole street" turning out because her street did. The booklet does not say, and that single missing fact would change how much weight the account can carry.