UCAT Decision Making Tutoring, Type by Type

Leading Tuition’s UCAT Decision Making tutoring is one-to-one preparation organised around the six question types the UCAT Consortium publishes for the subtest: syllogisms, logical puzzles, recognising assumptions, interpreting information, Venn diagrams, and probabilistic and statistical reasoning. Decision Making sets 35 questions in 37 minutes and returned a mean scaled score of 628 in 2025, against 661 for Quantitative Reasoning. We teach it type by type because the error differs by type.

Why We Teach Decision Making One Question Type at a Time

Decision Making is not a single skill with a single fix. The UCAT Consortium’s own Decision Making Question Tutorial lists six question types, and they ask for different things: syllogisms, logical puzzles, recognising assumptions, interpreting information and drawing conclusions, Venn diagrams, and probabilistic and statistical reasoning. The Consortium describes the subtest as testing whether a candidate can “apply logic to reach a decision or conclusion, evaluate arguments and analyse statistical information”, and notes that the questions may refer to text, charts, tables, graphs or diagrams and that all of them are standalone.

That last word is the one that shapes how we teach it. Because every question stands on its own, a student can be strong on four types and weak on two, and a mixed practice set will average those out into a score that hides the problem. Our first session is a diagnostic that sorts a student’s wrong answers into the six official buckets, and the order of everything after that is set by the result rather than by a fixed syllabus.

There is a second reason to work type by type, and it is structural. Decision Making is the only cognitive subtest that does not mark every question the same way. The Consortium’s scoring table reads: single answer questions are worth 1 mark, multiple statement questions are worth 2 marks, and 1 mark is awarded to partially correct responses on the multiple-statement questions. Verbal Reasoning and Quantitative Reasoning are flat — one mark a question, either way. In Decision Making, partial credit exists, and it exists on exactly the question types that look most intimidating on screen.

A student who reaches a five-statement drag-and-drop item, resolves three rows, cannot settle the fourth, and moves on with the boxes empty has given away marks they had already earned. That is not a reasoning failure. It is a rule nobody told them. We tell them in session one, and then we drill it until it holds under time.

The subtest is not the hardest of the three, and pretending otherwise would be a poor basis for a preparation plan. In 2025 the Consortium’s published test statistics reported a mean scaled score of 628 for Decision Making, between Verbal Reasoning on 602 and Quantitative Reasoning on 661, across 41,354 tests taken up to 26 September 2025. What makes it worth teaching properly is that most of the marks lost on it are lost to identifiable, repeatable habits rather than to raw ability.

What We Drill on Each of the Six Question Types

The table below is the working document behind a Decision Making programme. The middle column is the Consortium’s description of what the type asks. The two columns after it are ours: the error we see most often on that type in diagnostics, and the drill we use against it. If a student’s diagnostic shows no losses on a row, we skip that row.

Source: middle column summarised from the UCAT Consortium’s Decision Making Question Tutorial; the final two columns are Leading Tuition’s own teaching notes.

Syllogisms: Answering From the Premises, Not From What You Know

The official tutorial defines this type plainly: “In syllogisms, you are required to evaluate whether a series of conclusions arise from a given set of premises.” It adds two warnings that are worth reading together. The first is that questions sometimes use made-up words, and candidates should not be put off by them. The second is that quantifying words such as All, Some, None and Only “are not intended to trick you”.

That second line is the one students disbelieve, and disbelieving it is the error. A candidate reads a premise that is false in the world — and syllogism stems are often deliberately odd — decides the question must be a trap, and starts answering the trap instead of the question. The made-up words exist for exactly this reason: to remove anything the candidate could know independently, so that only the logic is left.

Our drill leans into that. We take a practice stem and substitute nonsense nouns for every real one, so there is no outside knowledge available to interfere, and the student has nothing to do but track the quantifier. Once they can work a set that way without slipping, we put the real nouns back and time it. The gap between those two scores is a direct measure of how much knowledge interference is costing, and it usually closes inside two sessions.

Syllogism items are frequently the multiple-statement, two-mark kind, which is why the partial-credit rule matters here before anywhere else. A student who settles most of the conclusions and leaves the rest blank has scored nothing on an item that was already paying.

Logical Puzzles: When the Grid Helps and When It Costs You Time

The Consortium describes logical puzzles as requiring the candidate to “use the information presented (in text, tables or other graphics) in order to arrive at a conclusion”, and gives three pieces of advice: look out for words such as MUST or MIGHT; remember that these are multiple-choice with only one correct answer, so eliminate options that cannot be true first; and note that candidates often draw a grid, beginning from the position of known facts.

Every preparation resource teaches the grid. Almost none teaches when not to draw one, and that is where the time actually goes. A puzzle with four people and three attributes rewards a grid. A puzzle with two variables and four answer options that are themselves statements is usually faster by elimination, and building a grid for it can consume most of the allowance for the question before any reasoning starts.

So we teach the decision before the technique. The student reads the stem, counts the variables, looks at the shape of the answer options, and commits to grid or elimination within the first few seconds. Then we run the same puzzle both ways and time it, which settles the argument far more convincingly than being told. Most students discover they have been drawing grids out of habit on roughly half the puzzles that did not need one.

The MUST and MIGHT distinction gets its own short drill, because it changes the answer rather than the method. We give a set where the only difference between two versions of a question is that word, and mark them side by side.

Recognising Assumptions: Strongest Argument, Not Most Agreeable

The official guidance for this type is unusually specific. Candidates are asked to “evaluate the strength of arguments for and against a particular solution to a problem, and select the strongest argument”. Strong arguments, it says, are directly connected to the subject matter; weak arguments rely on assumption or opinion; there is only one correct answer, so statements that are assumptions rather than fact should be eliminated. And then the line that explains most of the marks lost here: “Remember that you may have to suspend your own beliefs or existing knowledge.”

This is the only Decision Making type where the candidate’s own opinion is the direct cause of the error. Given a question about, say, banning a product, a student with a settled view will reach for the option that argues their side well, rather than the option that bears most directly on the question asked. The two are often different options, and the test is built on that difference.

Our drill marks every option on two axes rather than ranking them on one. First, is the statement connected to the subject of the question, or is it about something adjacent? Second, does it assert a fact, or does it rest on something we would have to assume? An option can be persuasive and still fail both. Writing the two marks down beside each option, rather than holding a ranking in the head, is what makes the choice mechanical.

We then deliberately select stems where the strongest argument supports the position the student disagrees with, and we say so before they start. Knowing the trap is coming does not remove it, which is precisely the point worth demonstrating.

Interpreting Information: Reading the Chart the Question Gave You

Here the Consortium tells candidates they will be “presented with information in different formats (e.g. written passages, graphs, charts)” and must determine which conclusions follow, using drag-and-drop. Its two tips are to use only the information presented, without judging a conclusion by its plausibility, and to focus on what can be interpreted, since rounding numbers off may answer a question quickly.

Plausibility is the trap, and it is a well-designed one. Several of the distractor conclusions on a typical item are true statements about the world that the chart in front of the candidate simply does not establish. A student who knows the subject area is more likely to fall for them, not less.

The drill is a source-check pass. For every statement, before marking Yes or No, the student has to name out loud the specific row, axis label or sentence that settles it. If they cannot name one, the answer is No. That single rule converts an intuition into a search, and a search is something that can be trained and timed. It also produces a clean record of where a mark went wrong, because the student can see afterwards whether they misread the chart or never consulted it.

This type is also where the drag-and-drop mechanics cost time on test day, and where partial credit pays out most often. We rehearse the interface in the Consortium’s own practice tools rather than describing it, and we hold to the rule from session one: every row gets an answer, including the ones the student is unsure about.

Venn Diagrams: Three Question Shapes, One Habit

The official tutorial is explicit that this type is not one question shape but three. A candidate may be shown a Venn diagram and asked to select the best conclusion from a list of statements; may see a passage that has to be interpreted as a Venn diagram; or may be given a set of statements and asked which of four diagrams represents them. The tutorial closes with a plain instruction: refresh this area of maths if it is not familiar.

Students who prepare from a single worked example learn one of the three shapes and lose the other two. So the diagnostic set here deliberately contains all three, and a student who scores well on the read-the-diagram form and badly on the build-it-from-a-passage form gets a different plan from one whose problem is the reverse.

The error underneath most of the lost marks is the same in all three forms: a region total is confused with a whole-circle total. A passage saying that fifteen products contain wheat is describing the whole circle, not the part of it that sits outside every overlap, and a student who writes that figure into the outer segment will do faultless arithmetic on the wrong diagram. Our habit against it is inside-out filling. Start at the innermost overlap, write it in, then subtract outwards, so that every number placed is already net of what it contains.

The maths itself is short. One session on Venn regions and one on probability is usually the whole mathematical component of a Decision Making programme.

Probabilistic and Statistical Reasoning: Where At Least One Goes Wrong

The Consortium describes this type as presenting “a very short passage containing statistical information” from which the candidate has to indicate the uncertainty of particular outcomes, and advises revising probability, including multiple and conditional probabilities. Elsewhere in the same tutorial it says something students find harder to believe: advanced statistical knowledge, and knowledge of mathematical or logical reasoning terminology, is not required, because the questions rest on practical, everyday reasoning using statistical data.

Believing that matters, because the students who lose most here are usually the ones who have decided the subtest needs statistics they have not studied and have quietly written the type off. It does not. What it needs is a handful of rules applied without panic.

The single most expensive error is the at-least-one outcome. Asked whether someone is more likely than not to win at least one of two independent games, a candidate adds the two separate probabilities and doubles a figure that was never additive. The correct route is the complement: one minus the probability of winning neither. Our drill is narrow on purpose. Every at-least-one question in the set gets answered by the complement, out loud, until the reach for addition stops happening.

The rest of the type is independence and conditional probability, taught in the same session, plus the habit of reading whether a passage gives a rate, a count or a proportion before any arithmetic starts. Half the errors we see here are not probability errors at all; they are reading errors about which of those three the passage supplied.

How a Decision Making Programme Runs With Us

Sessions are one-to-one, online or in person, and the shape is the same for every student even though the content is not.

It starts with a diagnostic: a mixed Decision Making set worked without a time limit, then worked again under time, with every wrong answer sorted into one of the six official types. Two students with the same score routinely produce different profiles, and the profile is what the programme is built from.

From there, one type at a time. Each type gets its rule established untimed first, then a set at a comfortable pace, then a set at test pace. The Consortium’s own general strategy for the subtest is the frame we teach inside: read the question carefully and only solve what you need to; practise the timing, since the average is approximately one minute per question; use the note-taking materials to write or draw information out; and flag harder questions for review rather than sitting on them. Those are test-day behaviours, and they need rehearsing under a clock rather than reading.

Two rules run across every session. Every row of a multiple-statement item gets an answer, because partial credit is real. And a question that has taken well past its share of the allowance gets flagged and left, because the standalone structure of the subtest means nothing later depends on it.

We work through the official materials rather than a product of our own. The UCAT Consortium states that it “does not work with any of these companies or endorse the use of their materials” when it comes to commercial preparation, and warns that commercial questions are “not necessarily of the standard you will encounter in the UCAT and this may distort your performance whilst practising”. We think that warning is fair, and we have no interest in arguing with the people who write the test. We do not sell a question bank. Sessions run against the Consortium’s question tutorials, its question banks and its four practice tests, which it describes as representative of the live test and publishes free at ucat.ac.uk. What a tutor adds is the diagnosis, the drill and the accountability, not the questions.

Students preparing across the whole test usually take Decision Making alongside the other subtests rather than in isolation; our one-to-one UCAT tutoring covers all four, and the wider UCAT preparation and UCAT tutoring pages set out the full test structure and the score benchmarks we work to. For applicants who want interview and personal statement support in the same programme, the medicine preparation hub is the place to start.

When We Tell Students Not to Prioritise Decision Making

Not every student who asks for Decision Making tutoring should have it, and saying so early saves an application.

The Consortium publishes a decile table each year, and the 2025 one is instructive. For Decision Making, the first decile sat at 520 and the ninth at 740. For Quantitative Reasoning, the first decile also sat at 520, but the ninth reached 820. The Quantitative Reasoning distribution stretches further at the top, which means the same amount of movement in raw performance converts into more scaled points there than it does in Decision Making. A candidate whose diagnostic puts them near the bottom decile in Quantitative Reasoning and mid-table in Decision Making has more to gain from an hour on the former.

The same argument runs the other way for a candidate whose Decision Making errors are concentrated in one or two types. Procedural errors — reading past a quantifier, leaving drag-and-drop rows blank, choosing the argument they agree with — are the cheapest marks on the whole test to recover, because the fix is a rule rather than a skill. That is a judgement we make from the diagnostic, not a promise made in advance.

What we will not do is quote you an expected score gain. We hold no audited outcome set for this subtest, and any figure we published would be marketing rather than evidence. Be careful with the comparability caveat you will see elsewhere, too: the Consortium's own 2025 statistics page says scores from 2025 are not directly comparable with those from before 2025, but its 2025 technical report is narrower than that, applying it to the total cognitive scaled score, which fell when Abstract Reasoning was removed, while saying of the subtests that “the test scores are consistent and comparable to previous cohorts”. So a Decision Making scaled score can be read against earlier years; a total cannot. The total cognitive mean in 2025 was 1891 on a scale running from 900 to 2700; where a particular student lands depends on far more than the hours they book.

Who Teaches Decision Making at Leading Tuition

Decision Making is taught by specialist tutors who have sat the UCAT themselves and scored in the upper deciles on it, and who have worked through the Consortium’s current materials rather than a version of the test that no longer exists. Teaching starts from the student’s diagnostic profile, so a student whose losses are concentrated in Venn diagrams and probability works on different material from one whose problem is argument evaluation.

On outcomes we publish what we can stand behind. 91% of our students achieve their desired grades. On Trustpilot, Leading Tuition is rated Excellent, 4.8 out of 5 from 58 reviews. Those are whole-service figures, not Decision Making figures, and we would rather say that than invent a subtest statistic.

Every programme begins with a free consultation: a conversation about where the student is in the cycle, what their practice scores look like by subtest, and whether Decision Making is actually the place their next ten hours should go.

Frequently Asked Questions About Decision Making Tutoring