Not all evidence deserves the same language. A statistic, an anecdote, a case study and a personal example may all be relevant, but they do different jobs.
The wider eduKateSG framework, Vocabulary for Junior College (JC1–JC2): Adult Thinking, Professional Writing, asks students to treat vocabulary as a tool for precision. Evidence vocabulary is where that precision becomes especially visible.
Statistics describe patterns
Statistical evidence can reveal frequency, change, distribution and association. Useful words include proportion, rate, median, trend, distribution, variation and correlation.
But a number without context can mislead. Students should ask: relative to what baseline, over what period, in which population, and with what measurement?
Examples illustrate
An example makes an abstract claim concrete. It may show that something can happen, reveal a mechanism or provide a counterexample. It does not automatically show how common the phenomenon is.
Anecdotes are vivid but narrow
Anecdotal evidence can reveal lived experience, edge cases and human consequences that aggregate data may hide. Its weakness is representativeness. One memorable story can dominate attention even when it is unusual.
Useful language includes illustrative but not representative, anecdotal, individual case, cannot be generalised from.
Case studies sit between story and pattern
A detailed case study can expose mechanism and context better than a broad statistic. Its strength is depth; its limitation is transfer. The question becomes whether the case is typical, exceptional or strategically chosen.
Worked comparison
Suppose an essay argues that remote work improves productivity. A worker’s personal story may illustrate flexibility. A company case study may reveal how work processes changed. A large dataset may indicate an average pattern. Each deserves different language and a different level of generalisation.
Vocabulary for evidence quality
- Representativeness: typical, unrepresentative, skewed
- Scale: individual, local, national, cross-national
- Consistency: convergent, mixed, contradictory
- Reliability: robust, preliminary, independently verified
- Transfer: generalisable, context-specific, limited in scope
The evidence hierarchy is question-dependent
There is no universal ranking in which statistics are always superior to stories. If the question concerns prevalence, broad data matter. If it concerns mechanism or experience, qualitative evidence may reveal something numbers conceal. Mature writing matches evidence type to the claim being made.
A JC evidence check
- What type of evidence is this?
- What can it legitimately show?
- What can it not show?
- How representative is it?
- What other evidence would strengthen the inference?
Students who can answer these questions stop treating every example as proof and every number as truth.
Evidence-Type Lab: Use the Right Evidence for the Right Question
Evidence quality cannot be judged without first asking what question the evidence is supposed to answer. A statistic may be excellent for estimating prevalence but poor for explaining lived experience. A case study may reveal mechanism in exquisite detail but tell us little about how common the pattern is.
The Evidence Matrix
- Anecdote: strongest for vivid experience and possibility; weak for prevalence.
- Case study: strongest for context and mechanism; limited for generalisation.
- Survey: useful for attitudes or self-reported behaviour; depends on sampling and question design.
- Administrative data: useful for large-scale recorded outcomes; limited by what institutions choose or are able to measure.
- Experiment: can strengthen causal inference under controlled conditions; may have limits in real-world transfer.
- Longitudinal evidence: tracks change over time; still requires care about confounding and attrition.
Quantitative and Qualitative Evidence Answer Different Questions
Quantitative evidence often asks how much, how often, how different. Qualitative evidence often asks how, why, what does this experience mean. Strong reasoning may combine both. A large dataset can reveal that a disparity exists; interviews or case studies can help identify mechanisms behind it.
Representativeness
A sample can be large and still be unrepresentative. Students should notice selection bias, self-selection, sampling frame and response bias. If a survey on workplace wellbeing reaches only people who choose to respond, the strongest opinions may be overrepresented.
Case Selection Matters
A dramatic success story can be chosen precisely because it is unusual. A failure can be equally unrepresentative. Ask whether the case is typical, deviant, extreme or strategically informative. An extreme case may still be useful if the aim is to show that an absolute claim has exceptions.
Triangulation
Confidence often improves when different evidence types converge. Survey data may show a pattern, interviews may reveal a mechanism, and administrative records may corroborate behaviour. This is triangulation: using different methods or sources to test whether the conclusion survives multiple perspectives.
Worked Example: Student Wellbeing
A single student’s account can illustrate intense examination stress. A school survey may estimate how widespread the experience is. Attendance or counselling records may provide behavioural indicators. None should automatically substitute for the others. Together, they can produce a more complete picture.
Evidence Selection Drill
- Write a claim about a GP topic.
- Identify whether the claim concerns prevalence, mechanism, causation, experience or comparison.
- Select the evidence type best matched to that job.
- Name its main limitation.
- Add a second evidence type that would compensate for that limitation.
Evidence-Type Checkpoint
- Am I using an anecdote to claim prevalence?
- Am I treating a national average as though it describes every subgroup?
- Is the case typical enough for the generalisation I want?
- Could another evidence type reveal a missing mechanism?
- Do independent methods point in the same direction?
Mature evidence use is not about collecting the largest quantity of material. It is about selecting evidence whose strengths match the question and whose weaknesses are visible to the writer.
For the complete JC vocabulary architecture, return to Vocabulary for Junior College (JC1–JC2): Adult Thinking, Professional Writing.