Paper 2 tests things A-Level never explicitly teaches. Here’s the one that costs the most marks.
Take the statement:
\[x > 3 \implies x^2 > 9\]This is true. Being greater than $3$ is sufficient for $x^2 > 9$ — knowing it is enough to conclude it.
But it is not necessary. $x = -5$ has $x^2 = 25 > 9$ without $x > 3$. So the converse fails:
\[x^2 > 9 \not\Rightarrow x > 3\]The words are worth getting exactly right, because the exam uses them precisely:
A quick sanity check I find useful: sufficient conditions are small, necessary conditions are big. A sufficient condition is a subset of the thing it implies — enough, but possibly more than enough. A necessary condition is a superset — required, but possibly not enough on its own.
The classic TMUA question gives you a statement and four candidate conditions, and asks which is necessary but not sufficient. Students who know the definitions still get these wrong, for a predictable reason: they check one direction and stop.
So build the habit of always checking both, explicitly:
Two questions, every time. It takes ten seconds and it removes an entire category of error.
The other reliable mark-loser. What is the negation of:
For all real $x$, $f(x) > 0$.
It is not “for all real $x$, $f(x) \leq 0$.” That’s a much stronger claim.
The negation is:
There exists a real $x$ such that $f(x) \leq 0$.
The rule: negating flips the quantifier as well as the statement. $\forall$ becomes $\exists$, and $\exists$ becomes $\forall$. To disprove “all swans are white” you need one black swan, not a proof that every swan is black.
This shows up constantly on Paper 2, and it’s the single highest-value thing to drill because the fix is mechanical once you’ve seen it.
Paper 2 also gives you an argument that reaches a false conclusion and asks which line is wrong. Three culprits cover most cases:
Dividing by something that might be zero. The classic fake proof that $1 = 2$ divides by $a - b$ after assuming $a = b$.
Squaring both sides. $x = -2$ becomes $x^2 = 4$, which now also permits $x = 2$. Squaring introduces solutions; it doesn’t preserve equivalence.
Assuming the converse. The argument proves $P \implies Q$ and then uses $Q \implies P$. This one is deliberately hard to spot because the line reads perfectly naturally.
When you’re hunting, check every step that isn’t reversible. That’s where the flaw lives — almost without exception.
Less than you’d think. This is a small, closed set of ideas, and a few hours of deliberate work on them moves your Paper 2 score more than another twenty A-Level questions ever will.
Do real past papers, and after each one write down not just what you got wrong but which of these patterns it was. The list is short enough that you’ll start recognising them cold within a fortnight.
Free official TMUA past papers and worked solutions are linked on the TMUA subject page. Start with a full paper cold to get a baseline — it’s uncomfortable, and it’s the most useful hour of preparation you’ll spend.
Strong problem-solvers aren’t stuck less often. They’re just quicker to move through it, because they have a list of things to try. Here’s mine.
If the question is about $n$, try $n = 1$. Then $n = 2$. Then $n = 3$.
This feels like cheating. It isn’t — it’s the single most productive move in mathematics. Small cases show you the pattern, tell you what the answer should look like, and very often reveal why the general statement is true.
Suppose you’re asked to show that
\[1 + 2 + 3 + \cdots + n = \frac{n(n+1)}{2}\]and you can’t see where to begin. Write out $n=4$: $1+2+3+4 = 10$, and $\frac{4 \times 5}{2} = 10$. Now pair the terms from the outside in: $1+4 = 5$ and $2+3 = 5$. Two pairs of $5$. There’s your proof, and you found it by playing with a specific case rather than staring at the general one.
Even when it isn’t obviously a geometry question. Sketch the function. Draw the situation. Put the vectors on a diagram with the angle marked.
A surprising proportion of “I don’t know how to start” is really “I haven’t formed a picture of what’s going on”, and the picture usually suggests the method. In mechanics this is close to universal: almost every mechanics question that goes wrong went wrong because there was no force diagram.
Write down, in words, the thing you’re trying to find. Then write it as a symbol.
It sounds trivial. It is astonishingly effective, because a lot of stuckness comes from holding too much in your head at once. “I want the value of $k$ that makes the discriminant zero” is a sentence you can act on. “I want to do question 7” isn’t.
If the question tells you what to show — and in STEP, in proofs, and in “hence show that” questions, it usually does — start from the target and ask what would be enough to get there.
To show $x = 3$, it would be enough to show $x - 3 = 0$. It would be enough to show that $x$ is a root of some polynomial you can factorise. Now you have a smaller, more concrete goal than you started with.
This is the one I use most in exams, and it’s nearly always the answer to a “hence” question.
Exam questions are constructed. Every condition in the stem is there for a reason. If you’ve reached part (iii) and haven’t used the fact that the sequence is increasing, or that $n$ is even, or the result you proved in part (i) — that unused fact is your next step. Part (i) of a STEP question is almost never a warm-up. It’s the hint.
Notice that none of these require you to know more content. They’re all ways of changing your relationship to the question — from “I must recall the method for this” to “let me find out what’s going on here”.
That shift is most of what separates students who find maths frustrating from students who find it interesting. It’s also entirely learnable, which is the good news.
Try this: next time you’re stuck for more than two minutes, work down the list rather than re-reading the question a fourth time. Write down which one broke it open. After a few weeks you’ll have a personal ranking — and reaching for it will have become automatic.