How AS & A Level Further Mathematics is marked
Further Mathematics is marked across fifteen papers (11, 12, 13, 21, 22, 23, 24, 31, 32, 33, 34, 41, 42, 43, 44), and in our bank the balance is 743 point-marked questions to 250 calculation questions overall — but this varies sharply by paper: papers 21 and 24 are almost entirely points, while the mechanics and statistics papers (32, 33, 41, 42, 43, 44) carry a much larger calculation share, in some cases outweighing points.
In this subject, 'points' questions are still frequently numerical — proving a result, deriving a formula, showing a given value — but the mark scheme awards discrete marks for specific correct steps or statements. 'Calculation' questions, concentrated in the mechanics and probability/statistics papers, are built around extended numerical methods (confidence intervals, hypothesis tests, generating functions) where marks are available for valid method even when the final answer goes wrong.
Because this is an advanced maths subject, a large share of marks lost are not conceptual but procedural: sign errors, choosing the wrong statistical test or critical value, mixing up a formula for a related-but-different quantity, or rounding too early in a multi-step calculation. The mark scheme is precise about which method is appropriate for which scenario (a t-value versus a z-value, a pooled variance versus separate variances), so accurately matching method to scenario is as important as the arithmetic itself.
The topic weighting in our bank is led by Matrices (38 questions), Inference and hypothesis testing (31) and Summation of series (28), with differential equations, mechanics topics (Hooke's law, projectiles, momentum, circular motion) and further pure topics (hyperbolic functions, complex numbers, polar coordinates) all carrying meaningful weight.
Where candidates actually lose marks
- Applying Vieta's formulas to a cubic, candidates make sign errors — for example giving p/2 instead of −p/2 by forgetting the coefficient of x is −p, not p — or divide through by the wrong leading coefficient. (try the question)
- Finding a confidence interval, candidates choose an incorrect z-value, use a pooled estimate for standard deviation instead of separate sample variances, or round prematurely during multi-step working, producing an inaccurate final interval; a very small number perform a hypothesis test instead of the confidence interval that was asked for. (try the question)
- On a chi-squared goodness-of-fit test, candidates fail to combine columns whose expected frequency is below 5, state the null hypothesis with insufficient detail (omitting reference to either the distribution or the data), use an incorrect tabular value, or conclude with no stated level of uncertainty. (try the question)
- Asked to justify a distinction numerically, candidates explain in words only, without providing the required numerical justification the question is actually asking for. (try the question)
- On a non-parametric (sign or rank-based) test, candidates use mu (the population mean) instead of m (the median) in the hypotheses, mishandle the test-statistic calculation, or — having correctly accepted the null hypothesis — draw the wrong practical conclusion from it. (try the question)
- Finding a cumulative distribution function, candidates omit the constant of integration, or incorrectly combine the tails of the distribution into a single blanket statement instead of stating each boundary case separately. (try the question)
Command words in AS & A Level Further Mathematics
| Command word | Times in our bank | What it demands |
|---|---|---|
| Find | 596 | The dominant command word in this subject: work out a specific value, expression or solution by a valid method, with working shown to support method marks. |
| Show | 231 | Requires you to derive a stated result rigorously from given information — using the target answer as a shortcut, or working to insufficient accuracy, forfeits the mark even if the final line matches. |
| Sketch | 44 | Wants a graph showing the correct key features (shape, intercepts, asymptotes) accurately placed, though not necessarily to a precise scale. |
| Test | 24 | Asks you to carry out a full statistical hypothesis test, including stating hypotheses with sufficient detail, computing the correct statistic, and reaching a conclusion with an appropriate level of uncertainty. |
| State | 21 | Wants a brief, exact answer with no derivation required. |
| Deduce | 17 | Asks you to use an already-established result to reach a further conclusion without repeating the original derivation. |
| Carry out | 16 | Requires you to perform a specified procedure or test in full, following through every step to a final result. |
| Calculate | 6 | Wants a specific numerical value obtained by computation, with method shown. |
How to revise with past papers
Because so many marks lost in this subject are procedural rather than conceptual, keep a running list of your own recurring slip types from past-paper practice — sign errors in Vieta's formulas, choosing a z-value where a t-value is needed, forgetting a constant of integration — and check for them specifically before submitting each new attempt.
For the statistics papers, practise matching method to scenario deliberately: before starting a hypothesis test or confidence interval, pause to confirm which test is appropriate (pooled or separate variance, z or t, parametric or non-parametric) rather than defaulting to the most familiar method, since choosing an inappropriate test means the working that follows cannot be credited.
For 'show that' questions specifically, always work to more decimal places or in exact fractions than the answer requires, since insufficient working precision can cost marks on these questions even when the final stated value is correct.
Practise real 9231 questions with AI marking →FAQs
Are marks available for a wrong final answer with correct method?
Yes — the calculation questions in this subject are built around extended methods where valid working typically earns credit even when the final answer contains an error, which is why showing full working consistently matters.
Why does choosing the right statistical test matter so much?
Choosing an inappropriate test (wrong distribution assumption, wrong variance treatment) means the method can't be credited even if the arithmetic that follows is correct, so confirming the right test before starting protects those marks.
Which topics carry the most weight in past papers?
In our bank, Matrices, Inference and hypothesis testing, and Summation of series are the most frequently examined topics, with mechanics and further pure topics also carrying substantial weight.
Do 'show that' questions need extra care?
Yes — working to insufficient precision (too few decimal places, or not using exact fractions) can lose marks on 'show that' questions even when the final line matches the target, because the working itself is the evidence being marked.