Matrix reasoning
A matrix item shows a grid of shapes, usually 3 by 3, with one cell missing. The shapes change across rows and columns according to one or more rules: a shape rotates, a fill darkens, an element is added. You choose the option that fits every rule at once.
Example in words: the top row shows one, two and three dots; the middle row two, three and four; the bottom row three, four and a blank. Each row adds one dot, so the missing cell has five. Matrices are the core of nonverbal tests such as Raven's Progressive Matrices and the Matrix Reasoning subtest of the Wechsler scales. Our page on matrix reasoning covers the rules in detail.
Number series and verbal analogies
Number series ask for the next number in a sequence. The rule can be a fixed step, a growing step, a multiplication, or two sequences interleaved. Example: 4, 9, 19, 39, ? Each number is double the previous one plus one, so the answer is 79. Letter series work the same way with the alphabet.
Verbal analogies give a pair of words with a relation and ask you to complete a second pair. Example: lock is to key as question is to … answer. The relation is "is opened or resolved by". Analogies test reasoning, but they also depend on vocabulary, which is why they are used less in tests meant to work across languages. Both types appear in employment tests such as the Wonderlic as well as in IQ batteries.
Spatial questions
Spatial items test how well you can picture and manipulate objects in your mind. The most common forms are:
- Mental rotation: which of four drawings is the target shape turned, rather than its mirror image?
- Paper folding: a sheet is folded twice and a hole punched through it; where are the holes when it is unfolded?
- Cube nets: which flat pattern folds into the cube shown?
- Block design: in clinical tests, arranging real red-and-white blocks to copy a pattern against the clock.
Example in words: picture a capital letter L lying on a table and turn it 180 degrees. The short arm now points left at the top, not right at the bottom. A mirror image would have the short arm pointing left at the bottom, and no rotation in the plane of the table can produce it.
Verbal and memory questions
Full clinical tests such as the WAIS-5 and Stanford-Binet 5 include tasks that depend on language and memory:
- Vocabulary: "What does candid mean?" (frank, open).
- Similarities: "How are a river and a road alike?" (both are routes for travel or transport).
- General knowledge: factual questions of increasing difficulty.
- Digit span: hear 7, 3, 9, 1 and repeat it backward: 1, 9, 3, 7. Sequences grow until you fail twice.
- Mental arithmetic: "Three pens cost 1.50. How much do seven cost?" (3.50), solved without paper.
These items measure crystallized knowledge and working memory; see fluid vs crystallized intelligence for the difference. Online tests use fewer of them because they depend on the language of the test and are harder to time and score without an examiner. Our WAIS-format test includes verbal and memory sections modeled on the format of the clinical test.
More worked examples, with the rule behind each
Short descriptions of each type are above; these are harder items of the kind that sit in the middle and at the end of a timed test. Try each before reading the answer.
A matrix with a hidden order. A 3 by 3 grid holds arrows: up, right, down on the first row, left, up, right on the second, down, left and a question mark on the third. Read the grid like a page, not row by row. Every arrow is the previous one turned a quarter turn clockwise, and the arrow before the question mark points left, so the missing one points up. The trap is looking at each row alone, where no pattern shows.
A matrix where shared symbols cancel. In each row the third cell keeps only the symbols that appear in exactly one of the first two cells; a symbol found in both disappears. Counting the symbols gives no clue, which is why this rule is among the harder ones.
Three number series, three rules. 5, 8, 14, 26, 50, ?: each number is double the last minus 2, so the next is 98. 2, 12, 4, 10, 6, 8, ?: two series are woven together, one rising by 2 and one falling by 2, so the next is 8. 3, 4, 8, 17, 33, ?: the gaps are 1, 4, 9 and 16, the square numbers, so the next gap is 25 and the answer is 58. When the gaps do not reveal a rule, ask whether the gaps form a series themselves.
Analogies and odd one out. Shepherd is to flock as conductor is to orchestra: say the first pair as a sentence (a shepherd leads a flock) and reuse it. Baton and score are tempting but name things a conductor uses, a different relation. Grain of sand is to beach as tree is to forest: a leaf is part of a tree, not the unit a forest is made of. In the odd-one-out set whisper, murmur, shout, mutter, all four are ways of speaking, but only shout is loud.
Direction and rotation. Walk 4 blocks north, 3 east, 2 south and 3 west: east and west cancel and north and south leave 2 blocks north. A clock hung upside down, turned 180 degrees, shows its hour hand at 2: half a turn moves every position six hours on, so it really points at 8. A mirror image would turn 2 into 10, and telling a rotation from a reflection is what most spatial items are about.
Memory with a twist. Read K 4 B 9 A 2 once, cover it and say the numbers low to high followed by the letters in order: 2 4 9 A B K. Holding six things is easy; sorting them in your head is the working-memory part.
How many questions, how long, and how hard
A clinical battery has 10 or more subtests and takes 60 to 90 minutes with a psychologist. A serious online test needs at least 20 to 25 items to estimate a band; ours uses 25 timed items and takes about 15 minutes. Most items are timed, either per item or per section, because speed is part of what is measured and untimed items invite outside help.
Items get harder as the test goes on. The first ones are solved by almost everyone; the last ones by only a few percent. A test where every item is easy cannot tell a 115 from a 135. That is also why sets of "hard IQ test questions" online are of limited use for practice: famous puzzles become easy once you have seen them, and a score earned on familiar items overstates ability.
Sources
Raven, J. C. Progressive Matrices: A Perceptual Test of Intelligence. H. K. Lewis, 1938.
Wechsler, D. WAIS-5 Administration and Scoring Manual. Pearson, 2024.
Condon, D. M. and Revelle, W. "The International Cognitive Ability Resource". Intelligence, 2014.
Frequently asked questions
What kind of questions are on an IQ test?
Matrix puzzles, number and letter series, verbal analogies, spatial items, vocabulary and similarities, and working-memory tasks such as digit span. Clinical tests also include processing-speed and mental-arithmetic tasks.
What questions are asked in IQ tests?
Questions that require reasoning rather than knowledge, such as finding the missing piece of a pattern or the next number in a series, plus vocabulary and memory tasks in full clinical tests. Difficulty rises as the test goes on.
What do IQ test questions look like?
Most online items look like puzzles: a 3 by 3 grid of shapes with one missing, a row of numbers with a gap, or a word pair to complete. Each has four to eight answer options and a time limit.
What are some questions on an IQ test?
Examples: what comes next in 4, 9, 19, 39 (79); lock is to key as question is to … (answer); repeat 7, 3, 9, 1 backward (1, 9, 3, 7); how are a river and a road alike (both are routes).
How many questions are on an IQ test?
It depends on the test. Clinical batteries contain well over 100 items across 10 or more subtests. Online tests usually have 20 to 40; ours has 25.
Are IQ tests timed?
Almost always. Clinical tests time many subtests individually, and online tests usually set a limit per item. An item left unanswered when time runs out counts as wrong.
Do IQ tests have math?
Some do. Number series need basic arithmetic, and the Wechsler scales include a mental-arithmetic subtest. Nothing beyond school-level arithmetic is required; the difficulty lies in spotting the rule.