Writing · Oxbridge admissions
Reading a MAT question the way an examiner does
The test does not reward what school rewards. That is not an accident — it is the entire design, and it is why very good students from drilling-heavy systems come unstuck.
A parent said something to me last spring that has stayed with me. Her son had been getting full marks in school mathematics for four years, had done thousands of past paper questions, and had come out of a MAT mock with something in the forties. She wanted to know what was wrong with him.
Nothing was wrong with him. He had been trained, extremely effectively, for a different test.
What school mathematics selects for
A-level and most national curricula are built around a reasonable promise: here is a syllabus, here are the question types, master them and you will do well. The skill being developed is recognition followed by execution. You see the shape of a question, you identify which method it belongs to, you carry out the method carefully.
This is a genuinely useful skill and I do not want to be sniffy about it. Most of applied mathematics is exactly that. But it produces students who are fast and accurate on the recognised and nearly helpless on the unrecognised — and Oxford knows this, which is precisely why the MAT exists. If the test could be passed by recognition, it would tell them nothing they did not already have from school grades.
So the questions are built to be unrecognisable on purpose.
What the questions are actually doing
Take a typical multiple-choice item. It looks like it wants a value. A student trained on execution reads it, identifies the topic — say, a cubic and its turning points — and starts differentiating. Four minutes later they have an algebraic mess and about half of them will push on with it out of sunk cost.
What the question was actually asking is whether you noticed that the cubic is odd, or that the answer is forced by a symmetry, or that three of the five options can be eliminated by checking a single boundary value. The full computation is available and will get you there. It is just not what is being tested, and the time limit is set so that computation alone will not finish the paper.
This is the single most useful thing I can tell you about the MAT: if the calculation is getting ugly, you have almost certainly missed the observation. Ugly is a signal, not a challenge. Students who internalise that gain more marks than students who get faster at algebra.
The longer questions work differently. They are usually a single idea, developed over four or five parts, where each part is a hint about the next. Part (i) looks trivial and often is. It is not there for the mark. It is there because the thing you notice while doing it is what part (iv) requires. Students who rush part (i) to get to the real questions have thrown away the ladder they were being handed.
Why this hits some students harder
I work with students from Korea, China, Singapore, India and the Gulf, as well as the UK, and there is a pattern I would be doing them a disservice not to name.
Education systems that are exceptionally good at producing mathematical fluency through volume — and Korea and China are very good at it indeed — produce students whose default response to any question is to start computing. The reflex is deeply trained and it is usually correct. On the MAT it is a liability, and the students it hurts are frequently the hardest-working ones, which feels grotesquely unfair.
The fix is not more problems. It is a change in what you do in the first thirty seconds: read the whole question, ask what would make this easy, look for the symmetry or the substitution or the option you can eliminate, and only then pick up the pen. Thirty seconds of deliberate not-computing. It sounds trivially small and it takes months to become automatic, because you are working against four years of successful training.
The interview is a different test again
Then, in December, if the paper goes well, comes something that resembles nothing the student has ever sat.
An Oxford interview is not an examination. Nobody is checking whether you know the syllabus. You are given a problem you cannot do, and the tutor watches you not be able to do it. That is the format. They will give you a hint, and watch what you do with it. They will let you go down a wrong path for a while to see whether you notice. The problem is often chosen specifically because it is beyond you, because a problem you could solve tells them nothing.
What they are trying to find out is whether you would be good to teach for three years in a tutorial, one or two students to a tutor. That is the actual question behind the whole thing. Are you interesting to think with? Do you take a hint and run with it, or wait to be told the rest? When you are stuck, do you go quiet, or do you say what you have tried and why it failed?
Students from systems where interrupting a teacher is rude, or where visible uncertainty is shameful, are at a real disadvantage here, and it has nothing to do with their mathematics. I have worked with students who were plainly stronger mathematically than the candidate who got the offer and who sat in near-silence for twenty minutes because they had been taught that you speak when you have the answer.
This is trainable. It is not trainable by reading about it, and it is not trainable in three weeks in December. It takes months of sitting opposite someone who gives you problems you cannot do and refuses to rescue you — long enough that being visibly stuck stops feeling like failure and starts feeling like the ordinary condition of doing mathematics. Which is what it is, and what your tutors will assume you already understand.
Timing, which matters more than people think
The UCAS deadline is 15 October. The test follows shortly after, interviews in December.
By the time a student sits down in October, the strong candidates have been working since spring — not cramming, but slowly rebuilding how they approach an unfamiliar problem. That is not a process that compresses. You cannot drill a reflex out in six weeks, and the students who try tend to arrive at the test anxious and still computing.
If you are reading this in the spring of the year before, you have exactly the right amount of time. If you are reading it in September, you can still be helped, but the honest answer is that we will be doing damage limitation on the paper and putting most of the effort into the interview, where the returns are faster.
The thing I would tell every parent
Your child's school mathematics grade tells you very little about how they will do here. Not nothing — you need to be strong to be in the conversation at all — but the correlation is much weaker than everyone expects, and the students who surprise on the upside are usually the ones who liked the hard problem at the end of the sheet rather than the ones who finished the sheet fastest.
If you want to know how your child will do, do not look at their marks. Give them a problem they cannot do and watch what they do for the next ten minutes. You will learn more from that than from a year of reports, and it is roughly what Oxford is going to do in December.