What does an abstract reasoning test actually measure?
An abstract reasoning test measures how well you spot the rule that connects
a set of shapes, and how fast you can do it. It is a test of fluid
intelligence: the ability to reason with unfamiliar information rather than
recall anything you have learned. There is no maths, no vocabulary and no
subject knowledge involved, which is exactly why employers rely on it.
The skill transfers to real work. Someone who can look at unfamiliar data,
see the underlying pattern and act on it quickly is doing the same thing a
developer does with a new codebase or an analyst does with a dataset they
have never seen. Because the questions carry no cultural or educational
baggage, two candidates with very different backgrounds can be compared on
the same footing.
In practice you are shown shapes that change according to a hidden rule, and
you choose the option that fits. Almost every question is multiple choice
with four or five options, and you usually have a minute or less for each
one.
What types of question appear in an abstract reasoning test?
Most abstract reasoning questions fall into four recognisable formats, and
knowing them in advance saves you the time others spend working out what is
being asked.
- Shape sequences. A row of frames changes step by step and you pick the
frame that comes next. The rule is usually a combination of movement,
rotation and something appearing or disappearing on a fixed cycle.
- Matrices. A grid of shapes has one cell missing and you choose the
shape that completes it. The rule runs along the rows, down the columns,
or both at once.
- Odd one out. You are shown several shapes and select the one that
breaks the rule the others share, such as symmetry or the number of sides.
- Analogies. “Shape A is to shape B as shape C is to which option?” You
work out the change from A to B, then apply the same change to C.
The underlying logic is the same across all four. Once you can name the rule
types, the format the examiner has chosen matters far less.
How do you spot the rule in a shape sequence?
Work through one element at a time rather than staring at the whole frame.
A frame that looks chaotic is almost always two or three simple rules
running independently, and the mistake most people make is trying to read
them all at once.
A reliable method:
- Pick one element and track only it across every frame. Does the
circle rotate, move, change size, change shading, or appear on a cycle?
- Repeat for the next element. Treat each shape as its own separate
rule until you have accounted for everything that changes.
- Check the things that stay the same. A shape that never moves is still
information, and it rules out options that alter it.
- Apply every rule to predict the next frame, then find that frame in the
options. Predicting first stops you being talked into a wrong answer by
a plausible looking distractor.
Distractors are built from the common errors: an option that gets the
rotation right but the shading wrong, or the movement right but one step too
far. Predicting the answer before you look protects you from all of them.
How do you solve matrix reasoning questions?
Read a matrix along the rows and down the columns separately, because the
rule almost always works in one direction and sometimes in both. Matrix
questions, a three by three grid with the bottom-right cell missing, are one
of the most common formats in graduate testing and in the UCAT, so they are
worth drilling.
Look for how each shape changes as you move across a row: it may rotate a
fixed amount, gain or lose an element, or swap colour. Then check whether the
same or a different rule runs down the columns. When both directions carry a
rule, the missing cell has to satisfy both, which usually narrows the options
to one.
If a row rule and a column rule seem to disagree, you have misread one of
them. Go back to the corner cell that both a row and a column pass through
and check your rule against it, since that cell is the easiest place to catch
the error.
Almost every abstract reasoning question is built from a short list of
transformations, and learning them turns the test from guesswork into
recognition.
- Rotation. A shape turns by a fixed angle each step, usually 45 or 90
degrees, clockwise or anti-clockwise.
- Reflection. A shape flips to its mirror image. A reflection can never
be reproduced by rotation alone, which is often the key to an odd-one-out.
- Translation. An element moves a set distance or around the frame on a
fixed path.
- Alternation. Something switches on and off, or between two states, on a
regular cycle across the frames.
- Number. The count of sides, dots or shapes rises or falls by a constant
amount.
- Shading and size. Fill and scale change on their own cycle, independent
of position.
Real questions layer two or three of these at once. Once you can name them,
you stop trying to see the whole answer in one glance and instead check the
shapes off against a mental list.
How are abstract reasoning tests scored?
Most employers score you against other candidates, not against a fixed pass
mark. Your raw score, the number you get right, is converted into a
percentile by comparing it with a normative group of people who have taken
the same test. Being placed in the 70th percentile means you did better than
70% of that group.
This shapes how you should prepare. A raw score on its own tells you little,
because everything depends on the comparison group and the threshold that
employer has set. Competitive graduate schemes commonly sift at around the
70th to 80th percentile.
Two practical consequences follow. Speed matters as much as accuracy, because
most tests are timed so that few candidates finish, and unanswered questions
cost you position against people who attempted them. And since negative
marking is rare, a considered guess beats a blank when time is running out.
Which employers and test publishers use abstract reasoning tests?
Abstract reasoning tests are standard in graduate recruitment across
technology, engineering, consulting, finance and other data-driven sectors.
If a role rewards spotting patterns and learning quickly, expect one.
Software developers, engineers, analysts, consultants, project managers and
graduate scheme applicants are the most commonly assessed.
Most employers license their tests from a small number of publishers, and
the format varies noticeably between them, which is why practising the right
style matters more than practising more:
- SHL produce some of the most widely used tests in graduate recruitment.
- Saville Assessment, Cubiks, Cut-e/Aon and IBM Kenexa each
use their own timing and question conventions, including adaptive formats
where difficulty adjusts as you go.
- Talent Q, TalentLens, Criterion Partnership and Watson
Glaser appear for more specialised or senior roles.
Find out which publisher your employer uses before you prepare seriously. It
is often named in the invitation email, and it changes what good practice
looks like.
Is abstract reasoning the same as diagrammatic and inductive reasoning?
They overlap so heavily that publishers often use the names
interchangeably, but there are useful distinctions. All three are non-verbal
tests that ask you to find a rule in a set of figures and apply it.
- Abstract and inductive reasoning are the closest pair. Both centre on
shape sequences, matrices and odd-one-out questions, and many tests
labelled one could just as fairly be labelled the other.
- Diagrammatic reasoning leans towards process logic: following operators
or flowcharts that transform an input into an output, rather than pure
pattern spotting.
The practical takeaway is that practice transfers. Preparing on abstract
reasoning questions sharpens you for inductive and diagrammatic tests too, so
do not worry if the label on your invitation is not the one you expected.
What is the best way to prepare for an abstract reasoning test?
Practise under the real time limit from the start, because untimed practice
builds accuracy but not the speed the test actually measures. Around a minute
per question is the usual pressure, and most candidates only discover the
timing problem on test day if they have not rehearsed it.
A preparation sequence that works:
- Take one test slowly first, reading the worked solution for every
question, to learn the rule types rather than to score well.
- Then work under the clock, and expect your score to dip at first.
- Review the questions you got wrong, and the ones you guessed right. A
lucky guess is a gap you have not found yet.
- Practise the format your employer actually uses, once you know the
publisher.
Puzzle apps and books that involve pattern spotting are a useful supplement,
but nothing substitutes for timed questions in the real format.
