Free Abstract Reasoning Test

What an abstract reasoning test actually measures

Abstract reasoning is the part of intelligence that has nothing to hold on to. You are shown a handful of shapes, told nothing about them, and expected to work out the rule that connects them and then apply it to a case you have never seen. Nothing you learned at school helps. There is no vocabulary to recall, no formula to remember and no general knowledge to fall back on, which is exactly why psychologists and employers keep coming back to this format. It gets close to what researchers call fluid reasoning: the raw ability to build a new rule on the spot rather than retrieve an old one.

That independence from background is the practical reason abstract tests dominate graduate recruitment. A verbal test quietly rewards anyone who reads a lot in the test language, and a numerical test rewards whoever last sat an exam full of percentages. A shape puzzle asks the same question of a candidate in Lisbon, Lagos or Leipzig, and the score means roughly the same thing in each case.

The rule families you will meet

Abstract puzzles look infinitely varied and are not. Almost every item on almost every test is built from a short list of transformations, and once you can name them the puzzles stop feeling like magic. The families are: change of quantity, where a count grows or shrinks in a fixed step; change of shading, where a fill cycles through solid, empty and half; rotation, where a figure turns by a constant angle at each step; reflection, where a figure flips across an axis and so becomes its own mirror image; movement, where an element travels around a frame one position at a time; and correspondence, where an inner element must agree with an outer one. Harder items simply stack two of these families on top of each other, so a figure might rotate while its dot count also increases.

The eight figure items on this page are drawn from those families deliberately. Some ask which of five figures breaks a rule that the other four obey — the classic odd one out. Others show three or four steps of a sequence and ask what comes next. In every case the wrong options are built by changing exactly one property of the correct figure, so there is never a second answer you could defend if pressed.

How the matrix questions work

The final twelve questions are three-by-three matrices with the bottom right cell missing. Read them twice: once along the rows and once down the columns. A rule that seems ambiguous in one direction is usually pinned down by the other. If the shapes multiply from left to right and the shading darkens from top to bottom, the missing cell has to satisfy both statements at once, and typically only one option does. When you cannot see any rule at all, invert the problem and eliminate: cross off every option that contradicts something you are sure of, and the survivors are few.

Reading your score

Your number of correct answers is mapped onto the standard scale, where 100 is average and 15 points is one standard deviation. Because 20 items is a short test, treat the result as a band of about ten points rather than a precise value.

Getting a score that reflects you

Timed reasoning tests punish two habits far more than they punish weak reasoning. The first is perfectionism: spending four minutes proving one item and then rushing six. Give yourself roughly forty seconds an item, and when an item resists, mark your best guess and move on. The second is reading the options first. Options are designed to look plausible, and glancing at them before you have formed your own answer plants a rule in your head that the puzzle never contained. Work out what the missing figure should look like, then go looking for it.

Beyond that, the ordinary advice holds and is not a cliché: take the test rested, in one sitting, on a screen large enough that you can see the whole matrix without scrolling. Fatigue costs more points on this format than on any verbal test, because every item demands a fresh start with no context to coast on.

How this differs from similar tests

Abstract reasoning, inductive reasoning and diagrammatic reasoning overlap heavily and the labels are used loosely across the industry. In practice, inductive tests lean on series completion, diagrammatic tests add process boxes and flow rules, and abstract tests centre on matrices and odd one out. A culture fair IQ test uses the same puzzle format but is scored against a general population norm rather than against a pool of applicants. If you score well here and poorly on a verbal test, that gap is genuinely informative: it usually says your reasoning is running ahead of your training in a particular language, not that one of the two scores is wrong.

What no online test can offer is a properly stratified norm sample or a supervised environment. Use this one to practise the format, to find out which rule family slows you down, and to see whether a real assessment is likely to go well. Do not use it as evidence in any decision that matters.