Treat intelligence assessment as a layered writing problem: classify each statement as reporting, assumption, or judgement; attach calibrated likelihood and confidence language to judgements only; make key assumptions and drivers explicit; test a genuine alternative hypothesis before committing; and keep every judgement traceable to the reporting that supports it. Practise these moves on ordinary news paragraphs, then under time pressure.
Where the Reporting Stops and the Assessment Starts
Assessment material mixes observed events with inferences, sometimes inside the same sentence. The trainable skill is labelling every statement as fact, assessment, or assumption before you reason on it, because mislabelling at the start distorts every judgement built on top.
This distinction is harder than it looks because well-written reporting and well-written judgement use the same plain words. 'The vessel docked on 12 May' is reporting: it happened or it did not. 'The vessel is likely engaged in undeclared cargo transfers' is assessment: it rests on inference from a mismatch between manifests and observed behaviour, and it could turn out wrong even when every fact in it is right. Assumptions sit invisibly underneath both, such as the assumption that the reported observation itself is accurate.
Worked scenario: the stimulus gives three lines — a vessel docked; manifests listing grain; a second report of covered loading at night. The question asks what is happening. A weak answer asserts: 'The vessel is smuggling sanctioned goods.' A stronger answer writes: 'Assessment: the vessel is likely conducting an undeclared cargo transfer (moderate confidence), based on the mismatch between manifests and night-time loading; alternative: legitimate after-hours operations.' The weak version presents one inference as fact, cannot be reviewed for reasoning, and collapses if the observation report was wrong. The strong version shows the inferential steps a reader can follow and check.
Calibrating Likelihood and Confidence in Estimative Sentences
Estimative judgements need two separate dials: the probability language for the event and the confidence you hold in your own basis. Conflating them, or writing certainty when the scenario supports only tendency, is a core practicable skill.
Overclaiming converts a judgement into a promise; hedging too far strips the answer of usefulness. The disciplined middle ground pairs one calibrated phrase with one confidence term and states the drivers behind both. Treat the bands below as a self-training convention for drafting practice, not as official doctrine: what matters in an exam answer is that your chosen phrase is proportionate to the scenario's evidence and that you could defend it if challenged.
Apply the convention by rewriting categorical verbs into estimative ones. 'Group X will attack the convoy' becomes 'Group X is likely to target the convoy, based on its pattern of striking soft targets when escort levels drop.' Note the two moves: likelihood language sized to the evidence, and a stated driver that explains why the judgement is held. If you cannot name a driver, the judgement is a guess and the likelihood phrase is decoration — a useful self-test before you commit any estimative sentence to paper.
| Practice band | Phrase type | Common misuse | Self-check question |
|---|---|---|---|
| High likelihood | 'highly likely', 'very probable' | Writing 'will' or 'is' for an inferred future event | Does the scenario give multiple independent lines supporting this? |
| Moderate likelihood | 'likely', 'probable' | Reusing 'likely' for everything to sound cautious | Could the evidence also support a different explanation? If yes, lower or add drivers. |
| Genuinely open | 'even chance', 'unclear' | Claiming likelihood when evidence is thin | Have I said what would tip this either way? |
| Low likelihood | 'unlikely', 'probably not' | Burying a low-likelihood judgement in vague wording | Is the rejection based on evidence or on dislike of the alternative? |
Surfacing Assumptions and Key Drivers Hidden in the Stimulus
Every judgement rests on unstated conditions. Writing a short assumptions line and naming two or three key drivers turns invisible logic into checkable logic, and reveals where the scenario never actually supported you.
An assumption is a condition you have accepted without the scenario proving it — most commonly that a source reported accurately, that past behaviour predicts future behaviour, or that actors know what you assume they know. A driver is the factor your judgement most depends on: if the driver flips, the judgement flips with it. The two are different tools. Assumptions can be validated or invalidated; drivers tell a reader where the judgement is fragile.
Worked scenario: a stimulus describes rising local recruitment and intercepted communications referencing 'the visit', and asks about intent. A weak answer concludes the group plans an attack on the visit, resting entirely on the unexamined assumption that 'the visit' means the ministerial trip mentioned earlier. A stronger answer writes: 'Assessment: attack planning during the visit is plausible but not established (low-moderate confidence). This assessment assumes the communications refer to the ministerial visit rather than another event; key drivers are the meaning of the intercept and whether surveillance-type activity accompanies it.' The improved version is falsifiable and shows exactly what additional reporting would settle the question.
Building a Genuine Alternative Hypothesis Before Committing
A disciplined answer argues for a primary explanation only after constructing a real competing one. The competition forces you to find discriminators — evidence that would be true under one hypothesis but not the other.
Alternative analysis in exam conditions does not require formal matrices. It requires that the alternative be genuinely different rather than a weak echo of your primary view, and that you identify at least one discriminator: an observable that would distinguish between them. Writing the alternative after you have fallen in love with your primary answer tends to produce a strawman; write it early, while you are still deciding.
Practical exercise: take any current-affairs paragraph and, in fifteen minutes, write a primary explanation, a genuinely different alternative, and three discriminators. Expected observations on first attempts: alternatives usually start as restatements ('maybe it is not smuggling, just unreported smuggling'), and discriminators start vague ('more information would help'). Rewrite until the alternative would lead a reasonable person to different predictions, and until each discriminator names something observable — a document, an action, a pattern of movement. Self-check rubric: two points if the alternative predicts differently, two if discriminators are observable, one if your primary answer explicitly explains more of the evidence than the alternative does. A rewrite loop of two or three passes is normal and is where the skill actually develops.
- Alternative is genuinely different: it predicts different outcomes, not just a reworded doubt.
- Discriminators are observable: each names evidence you could conceivably see or request.
- Primary still wins on the evidence: you can say why, in one sentence, you prefer it.
Mapping Problems to Community Roles and Coordinated Judgement
The National Intelligence Community comprises ten agencies with distinct responsibilities, functions and powers. Practise answering scenario questions by reasoning through function — what kind of problem it is — rather than reciting organisational charts.
Anchor your mapping to the functional categories the community's own careers material points to: protecting Australia and Australians, tracking criminals, supporting military operations, and analysing risks such as climate. When a scenario describes a domestic security investigation, foreign-focused collection, imagery or technical analysis, or whole-of-community coordination, frame your answer around that function and why it fits the problem, rather than asserting detailed agency powers you cannot support from the stimulus.
Coordination is the second behaviour worth practising. Community careers material describes roles including intelligence analyst, intelligence editor, and intelligence coordination, and highlights a culture of collegiality and contestability. In a scenario answer, this translates into recognising when a judgement touches another agency's domain and should be coordinated or contested rather than issued as a single-shop conclusion. Writing 'this assessment should be coordinated with the agency holding the relevant collection before it is finalised' demonstrates that you understand assessment as a community product, not a private opinion — a point the collaborative culture described publicly by the community reinforces.
Sourcing and Provenance: Keeping the Audit Trail Intact
Sourcing belongs on the reporting; confidence belongs on the judgement. A well-documented answer lets a reviewer trace every estimative sentence back to specific stimulus material without drowning each clause in hedges.
Conceptually, evaluate reporting along two axes: the reliability of the source and the credibility of the specific information. A normally reliable source can pass on wrong information; an untested source can be right. In exam answers, cite the scenario line that supports each factual claim — 'the night-loading report (second source)' — so a reviewer can check your chain. If a judgement rests on a single line of reporting, say so; that admission is a mark of discipline, not weakness.
The common drafting failure runs in both directions. Over-attribution hedges every clause ('it is possible that the report, which may be accurate, suggests...') until the answer says nothing. Under-attribution states inferences as though the stimulus proved them. The fix is structural: put sourcing phrases on reporting sentences, put likelihood and confidence on the judgement sentence, and keep them apart. Read your answer back asking one question per sentence — is this what happened, what I believe, or what I am taking for granted? If a sentence resists classification, rewrite it.
An Adaptable Preparation Sequence and Readiness Checks
Sequence your preparation from classification drills through estimation and alternatives, then to timed full scenarios. Treat readiness checks as learning milestones that show the craft is becoming automatic, not as predictions of any result.
A workable six-week sequence: weeks one and two, classification drills — take one news or scenario paragraph daily and label every sentence as reporting, assessment, or assumption, targeting near-zero misclassifications by the end of week two. Weeks three and four, estimation and alternatives — keep a short calibration diary where you record a judgement, its likelihood phrase, and later whether events supported it, and run the alternative-analysis drill from the earlier section twice weekly. Week five, community mapping and documentation: write answers that name functions, coordination points, and sourcing lines explicitly.
Week six, timed practice: complete full scenario sets, allocating roughly five minutes to draft a key judgement with drivers and a stated alternative before writing the body. Self-check rubric for any finished answer: one point each for clean fact/assessment/assumption separation, a proportionate likelihood phrase with named drivers, explicit assumptions, a genuine alternative with a discriminator, and intact sourcing. A total of four or five across repeated timed attempts is a reasonable milestone that your written craft is consolidating. If you stall, return to the classification drills — layering failures upstream usually explain downstream weaknesses in judgement quality.
- Classification check: a ten-sentence scenario labelled with no fact/assessment/assumption errors.
- Estimation check: every judgement carries one likelihood phrase, one confidence term, and named drivers.
- Alternatives check: each answer includes a different-predicting alternative and an observable discriminator.
- Documentation check: each factual claim traces to a specific line of stimulus material.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.