Guides · Thinking Skills
Every Thinking Skills question type on the NSW Selective Test, with examples.
The Thinking Skills paper is 40 multiple-choice questions in 40 minutes, four options (A–D) each, no calculator. Underneath that simple shape, our question-by-question analysis of the Department's published papers shows the questions clustering into four distinct families — verbal argument, deductive reasoning, numerical reasoning, and spatial thinking — each testing a different kind of reasoning, in roughly fixed proportions from paper to paper. Below is every type in each family, with one fully worked example.
The four families, by share of the paper
Our taxonomy maps 21 distinct question types across these four families, and every practice paper we publish carries the same family mix as the Department's own papers. The approximate share each family takes of a 40-question paper is below.
| Family | Share of paper | What it tests | Example types |
|---|---|---|---|
| Verbal argument | ≈23% (9 of 40) | Reading a short argument and judging its logic — not its topic. | Strengthen/weaken an argument; identify a reasoning flaw |
| Deductive reasoning | ≈48% (19 of 40) | Formal logic — rules, constraints, orderings, ciphers. | Whose reasoning is correct; logic grids; arrangements; cryptograms; knights & liars |
| Numerical reasoning | ≈18% (7 of 40) | Word problems solved as reasoning puzzles, not routine arithmetic. | Systems of equations; minimum-cost combinations; table and chart reading |
| Spatial thinking | ≈13% (5 of 40) | Reasoning about shape, view, and pattern. | 3D views and pieces; folding and assembly; pattern and symmetry |
Family shares are rounded from the mean frequency across four official source papers — see the sourcing note below for exactly which papers.
Deductive reasoning alone is nearly half the paper, and within it one type — judging which of two children's inferences from a stated rule is valid — is the single most common question type in the whole test. If your child masters one thing before test day, that family, and that type within it, is the highest-leverage place to spend time.
Verbal argument (≈9 questions)
These are the paper's prose-heaviest questions and the only family with no diagrams at all. Each one presents a short, self-contained argument — a person's claim, a couple of sentences of reasoning — and asks your child to judge the argument itself, independent of whether they happen to agree with its conclusion. There are two types:
- Strengthen or weaken an argument. A short claim is given (about 60–140 words), and your child picks which of four additional facts would most strengthen it, or most weaken it, if true. The three wrong options are true-but-irrelevant, attack a side issue, or actually push in the opposite direction from what's asked.
- Identify the reasoning flaw. A character draws a conclusion that sounds reasonable on a first read but rests on a specific logical error — mistaking correlation for causation, treating "no one's been caught" as evidence no one's doing it, generalising from a small sample, and so on. Your child picks the sentence that names the actual gap, not just any true statement about the situation.
Worked example — weaken an argument
Gabriella says that zoos should be banned, because it is wrong to keep wild animals in captivity. She says animals should have the right to live in their natural habitats.
Which one of these statements, if true, most weakens the argument?
Answer: B.
Why: Gabriella's argument rests on the premise that captivity serves no purpose animals need. Option B directly undercuts that premise — it says captivity can be the only thing standing between a species and extinction, which is exactly the kind of purpose her argument assumes doesn't exist. Option A is true but doesn't touch her ethical claim; option C is true but supports her rather than weakening it; option D is true but describes conditions, not whether captivity itself is wrong.
Deductive reasoning (≈19 questions)
By far the biggest family, spanning ten distinct question types — formal logic applied to everyday scenarios rather than symbols. Almost all are pure logic puzzles with a single decidable answer; a few carry a small diagram (a symbol key, a seating schematic), but most are text-only. The ten types, briefly:
- Whose reasoning is correct (the single most common type on the whole paper). A boxed rule, then two children each draw a conclusion from it — your child judges whether each conclusion is valid (answer: "A only", "B only", "both", or "neither").
- Single-claim conditional operations. Given one or more rules, perform one logical step — restate the contrapositive, find what's impossible given a chain of rules, or work out which branch of a rule was taken.
- Constraint satisfaction. Classic logic-grid puzzles — assign people to positions, teams, or attributes under a set of rules, with exactly one valid solution.
- Line or circle arrangement. Place people or objects at fixed positions from adjacency and "who's opposite whom" clues.
- Deductive ordering. Rank people or things on one or two axes (age, speed, price) from partial comparative clues.
- Set syllogisms. Reason with "all / some / no" statements about overlapping groups — what must, or can't, be true of one member.
- Knights and liars. Truth-tellers and liars (or bounded numbers of false statements) — work out who's telling the truth.
- Cryptogram deduction. Infer a mapping between symbols, letters, or colours from partial examples, then apply it to a new case.
- Cycle reasoning. Reason over a repeating cycle with an offset — which season falls six months on, which letter a rotating count lands on.
- Chain matching. Sequence tiles or words under an adjacency rule (each end must match the next start) and identify the missing pieces.
Worked example — whose reasoning is correct
Anyone who helped Mr Bamblett with the school garden last week will be given a ticket to the Country Fair.
Katherine: "I helped Mr Bamblett last Thursday, so I'll definitely get a ticket."
Kylie: "Fran is going to the Country Fair. She must have helped out too."
If the information in the box is true, whose reasoning is correct?
Answer: B.
Why: the rule only tells us that helping guarantees a ticket (helper → ticket) — it says nothing about how else a ticket might be earned. Katherine correctly applies the rule she was given: she helped, so she gets a ticket. Kylie runs the rule backwards — Fran having a ticket doesn't mean she helped, because Mr Bamblett's garden might not be the only way to get one. That's the exact error this question type is built to catch: reading a one-way rule as if it worked both ways.
Numerical reasoning (≈7 questions)
Arithmetic used as a reasoning tool rather than tested for its own sake — no formal algebra is assumed, but the scenarios are designed so guessing is slower than setting the problem up properly. Six types:
- Systems of equations, as a word problem. Two or more unknowns tied together by a scenario (prices and quantities, ages, coin counts) — solve for one of them.
- Minimum-cost combinations. Choose among discrete options (bus passes, bulk offers, pack sizes) to minimise cost or maximise what's achieved.
- Table analysis. Read a data table and compute a derived value — a rank, a per-unit rate, a column total.
- Chart reasoning. Match data to a pie or bar chart, or identify a missing slice or bar.
- Counting puzzles. Count elements under overlapping conditions — calendar events, overlapping groups, overlapping time windows.
- Small-graph reasoning. Reason over a simple network of routes or connections — a route of a given length, or the missing link that makes two points reachable.
Worked example — system of equations, as a word problem
A school canteen sells biscuits for $3 each and sausage rolls for $5 each. A customer buys 7 items altogether and pays $23.
How many biscuits did they buy?
Answer: C.
Why: let x be the number of biscuits and y the number of sausage rolls. Two facts are given — x + y = 7 (seven items total) and 3x + 5y = 23 (the total cost). Substituting y = 7 − x into the second equation gives 3x + 5(7 − x) = 23, which simplifies to 35 − 2x = 23, so x = 6. Six biscuits and one sausage roll: 6 × $3 + 1 × $5 = $23. Each wrong option corresponds to a plausible slip — for example, answering 7 assumes every item was a biscuit and ignores the higher price of a sausage roll.
Spatial thinking (≈5 questions)
The smallest family and the only one that's almost always built around a figure — a drawing of a 3D shape, a floor plan, a tile pattern. Three types:
- 3D views and pieces. Given a solid built from cubes (or a description of one), work out its top-down silhouette, its view from another angle, or which piece completes it.
- Assembly and layout. Folding, cutting, and fitting problems — which unfolded shape a folded-and-cut piece of paper produces, which piece completes a tiled square, how many tiles fit a floor and what the leftover fragment looks like.
- Pattern and symmetry. Infer a hidden or missing part of a repeating visual pattern, or apply a transformation rule (rotate, reflect, reorder) demonstrated on one example to a new case.
Most spatial questions need a diagram to work — but one recurring type doesn't, so we can show it here in full text.
Worked example — pattern and symmetry (digit rotation)
A date typed into a calculator in DDMMYYYY format reads as the same date when the calculator is turned upside down (so rotated 180°). This is possible only when every digit is one that looks like another valid digit when rotated: 0→0, 1→1, 8→8, 6→9.
Which one of the following dates would also read as the same date when rotated?
Answer: A.
Why: rotating the whole display 180° both reverses the digit order and flips each digit. For the date to read the same, every position's digit must be the rotation partner of the digit in the mirror-image position. 19/01/1061 reversed is 16011091, which flipped digit-by-digit (6→9, 0→0, 1→1) gives 19011061 — the original date. Every other option contains a digit (2, 3, 5, or 7) that has no valid rotation partner at all, so it's ruled out immediately.
Where this taxonomy comes from
This taxonomy is our own analysis. We classified all 160 questions across four distinct past and practice Thinking Skills papers — the Department's two 2026-entry practice tests, the 2023 paper, and the 2022 paper — question by question, into 21 named question types across these four families. Every practice paper we publish is built to that same map, in roughly the same proportions the source papers show, so it carries the same family mix as the real test. The four worked examples above are questions from our own papers, calibrated against those same source papers — not copied or reworded from anything published.
The format facts — 40 questions, 40 minutes, four options, no calculator — are the ones the NSW Selective Test guide verifies against the Department directly.