Diagrammatic Reasoning Test: Examples, Operators and Preparation

Most guides file diagrammatic reasoning under abstract reasoning and move on. The two are not the same task. An abstract item shows a sequence and asks you to work out what is changing; a diagrammatic item hands you an input, a chain of labelled operators, and asks you to apply them. The rule is not hidden, it is printed on the page, so what is measured is execution rather than discovery. That makes the format closer to reading a short program than to spotting a pattern, and it explains why people who score well on abstract series stall here: they hunt for a concealed rule that was never concealed instead of running the one they were given, in the stated order. Order is the whole game, because operators usually do not commute. Rotate then invert and invert then rotate land on different panels, and the panel from the other order is sitting in the options.

Practising for a real assessment? The Spatial & Diagrammatic Reasoning Practice Pack runs these question types at full length — 5 timed tests, 150 questions, a worked explanation on every one.

The formats that sit under the diagrammatic reasoning label

Five mechanically different tasks share this label, and they fail in different ways, so one overall practice score tells you little.

Process or operator flow

An input panel enters a chain of labelled boxes, each carrying a defined transformation, and you choose the output. Operators rotate the panel a quarter or half turn, invert shading, add or delete an element, swap the contents of two named cells, or shift everything one position along. The definitions sit in a key above the item.

The mechanical skill is applying operators in sequence and, crucially, in the stated order, because most pairs do not commute. A rotation moves contents between cells; a swap defined in fixed panel coordinates always hits the same two cells, whatever is in them. Rotate then swap and you swap what the rotation has moved; swap then rotate and you rotate what the swap has moved. Both chains are legal, both look plausible, only one matches the arrows.

Operator deduction

You are shown an input and an output and asked which operator, or which chain, produced it. Some versions leave one box in a chain blank; others show several worked flows and ask which operator is faulty, meaning it behaves inconsistently across them.

This reverse direction is genuinely harder, because you must invert the transformation rather than apply it, and inverting is not always unique: a half turn is its own inverse, so is a colour inversion, and two chains can produce identical outputs from one input. Test a candidate operator forwards against every example shown, not just the one you guessed from.

Flowchart and decision-path items

A diamond holds a yes or no condition, such as whether the panel holds more than two shaded cells, and the answer routes the input down one branch. The trap is almost always the same: the condition must be evaluated against the intermediate state at that point in the flow, not the original input. Candidates carry the starting panel in their head, reach the diamond, test what they remember, and take the wrong branch. Every later step is then applied correctly to the wrong panel, which is why these produce confident wrong answers rather than blanks.

Rule-based sorting and network items

Symbols travel through a network with rules at the junctions: a filter passing only shaded symbols, a splitter sending circles left and the rest right, a merge combining two streams in a stated order. Unlike operator flow, the route is not fixed in advance. It depends on a property of the symbol, so you track a state and a position at once, and a symbol failing a filter may leave the problem entirely rather than be transformed.

Sequence panels with embedded operators

A row of panels progresses like an abstract series, but each change is expressed as a named operator, and you are asked for the next panel or for the operator filling a gap. These need the abstract skill of reading a progression and the diagrammatic skill of applying a rule exactly, which is why they sit at the harder end of a set.

Scoring and timing, and what they mean for strategy

Two scoring models are in use and they call for opposite behaviour on the last few items. Under raw scoring your mark is simply the number of correct answers, a blank and a wrong answer score the same, and leaving anything empty throws away a free chance. Under formula scoring a fraction of a mark is deducted for every wrong answer and nothing is deducted for an unanswered one, which is designed to cancel the expected gain from random guessing. If the marking deducts for errors and ignores blanks, answering an item you have not read properly costs you more than skipping it.

Timing models vary just as much. A fixed number of items against a fixed clock means every question is reachable and pacing is arithmetic. An indefinite bank against a clock means nobody finishes, so a chain you cannot resolve is worth abandoning early. Per-section timers stop you borrowing minutes from the flowchart block to rescue the operator-deduction block, so your weakest format must be managed inside its own budget. Per-question timers remove skipping and returning, which makes written intermediate states more valuable, not less, because you cannot afford to restart a chain. Adaptive versions choose the next item from your last answer, so early items carry unusual weight and a careless start can cap the difficulty, and therefore the score, you can reach.

Read the instructions for the scoring rule before you start. It decides what you do with the final minute.

A worked example: an operator chain

The key defines two operators. P rotates the whole panel a quarter turn clockwise, contents included, so a shape's own orientation turns with it. Q swaps the contents of the top-left and top-right cells, and those names refer to fixed positions, not to whatever was there at the start.

The input is a panel of four cells: a black circle top-left, an unshaded triangle pointing up bottom-left, the other two empty. The flow is input, then P, then Q.

Apply P first and write the result down. A quarter turn clockwise sends top-left to top-right, bottom-left to top-left, bottom-right to bottom-left, top-right to bottom-right. The circle is now top-right; the triangle is now top-left and, because contents turn too, points right. The bottom row is empty.

Now apply Q to that panel, not to the input. Swapping top-left and top-right gives a black circle top-left, a triangle pointing right top-right, an empty bottom row. That is the answer.

Each distractor comes from one specific misapplication. One shows a triangle pointing right top-left and a black circle bottom-right: Q then P, the right operators in the wrong order. One shows a triangle pointing down top-left with the circle bottom-right: P applied twice, then Q. One matches the answer except that the triangle still points up: the layout was rotated but not the contents, stopping the operator halfway. One shows a triangle pointing right top-left and a circle top-right: the output of P alone, from someone who stopped at the first box.

How to prepare for diagrammatic reasoning specifically

Write the intermediate state down. The dominant failure here is not misunderstanding an operator, it is losing a panel you were holding in your head while working out the next step. Invent a short notation you can produce quickly, for example TL, TR, BL, BR with a symbol and an orientation arrow after each, and use it every time. One line of scrap paper per operator box turns a memory task into a copying task.

Practise the reverse direction as its own exercise. Take a forward chain you have solved, hide the operators, and work from input and output back to the rule. Check your candidate against a second input, because operator deduction rewards testing forwards and punishes guessing from one case.

Test commutation deliberately. Pick two operators, apply them in both orders on one input, and see whether the outputs differ. A dozen repetitions build an instinct for which pairs are order-sensitive, and rotation combined with any position-fixed rule almost always is.

For flowcharts, re-read the current state immediately before each diamond, out loud if you can, so the condition is evaluated against the panel in front of you rather than the one you started with.

Finally, settle the scoring rule before you sit down, because it decides whether your last minute is spent filling blanks or leaving them.

Frequently asked questions

TestSolve is independent and not affiliated with any test provider or employer named on this page. All product names and trademarks belong to their respective owners.