M: Probability

Overview

Probability is AQA’s language for uncertainty. In specification M: Probability (3.14), candidates must move confidently between words, diagrams, fractions and formal notation. The topic rewards organised thinking: define the events, identify the relevant sample space, then choose a representation that makes the structure visible. A probability answer can be numerically correct but lose credit if the working does not show why a denominator, branch or overlap has been chosen.

This is a high-value synoptic topic. It connects directly to statistics, sampling, hypothesis testing and modelling, and it appears in both short skills questions and multi-step problem-solving questions. Typical AQA prompts include calculate, show that, hence, given that, at least one, exactly one, without replacement, independent and mutually exclusive. The command word tells you what evidence to show. In particular, a question containing given that changes the sample space, while without replacement changes later tree branches.

AQA papers are normally 120 minutes for 100 marks, so work at roughly 1.2 minutes per mark. The aim is not to memorise isolated rules. The aim is to make one defensible decision at a time, so that every line earns credit. Neat working is not decoration: it is evidence that each mathematical choice is valid during assessment.

Audio Revision

Use this approximately ten-minute tutor-led recap after a first study session, then pause before the quick-fire questions and answer aloud.

Key Concepts

1. Sample spaces, events and complements

A sample space is the complete set of possible outcomes. An event is a set of outcomes of interest. Probability must lie between 0 and 1 inclusive: 0 is impossible and 1 is certain. If outcomes are exhaustive and mutually exclusive, their probabilities add to 1.

The complement of event (A), written (A') or (A^c), means “not (A)”. Therefore (P(A')=1-P(A)). This is especially efficient for questions containing at least one. Rather than listing many successful cases, calculate the single unsuccessful case, “none”, then subtract from 1. For example, if a biased coin has (P(H)=0.7), then the probability of at least one head in two independent tosses is (1-P(\text{no heads})=1-(0.3)^2=0.91).

Examiner lens: Candidates gain method credit when the complementary event is explicitly identified. Writing only (1-0.09) is less secure than writing (1-P(\text{no heads})).

2. Union and intersection: translate the words first

The word orusually means a union, (A\cup B). The wordandmeans an intersection, (A\cap B). For any two events,

[
P(A\cup B)=P(A)+P(B)-P(A\cap B).
]

The subtraction is essential because the overlap has been counted twice. If the events cannot happen together, (P(A\cap B)=0), so the formula simplifies to (P(A\cup B)=P(A)+P(B)).

A frequent error is to add probabilities whenever the question contains the word “or”. Before adding, ask: can both events happen? If yes, find and subtract the overlap. If no, state that the events are mutually exclusive and use the simplified rule.

3. Mutually exclusive is not independent

These terms describe different relationships. Mutually exclusiveevents cannot happen together: (P(A\cap B)=0).Independent events can happen together, but knowing that one happened does not change the probability of the other. For independent events,

[
P(A\cap B)=P(A)P(B).
]

Except in a zero-probability edge case, two non-impossible mutually exclusive events are not independent. This is a standard AQA testing point. A clear comparison earns more credit than a vague statement such as “they are unrelated”.

Memory hook: Exclusive means exit together. Independent means influence does not enter. The first phrase reminds you that there is no overlap; the second reminds you that the probability is unchanged.

4. Venn diagrams: fill the overlap first

Use a Venn diagram when the data concern membership of two or more sets. Start with the centre, (A\cap B), because it belongs to both sets. Then fill the “A only” and “B only” regions by subtraction. Finally use the total of 1, or the total frequency, to find the outside region.

Suppose (P(M)=0.55), (P(P)=0.40) and (P(M\cap P)=0.20). Then maths only is (0.55-0.20=0.35), physics only is (0.40-0.20=0.20), and neither is (1-(0.35+0.20+0.20)=0.25). A completed diagram gives you a visual audit: every disjoint region must sum to 1.

5. Conditional probability: restrict, then divide

(P(A\mid B)) reads “the probability of Agiven B”. Once B has happened, outcomes outside B are no longer possible. The formula is

[
P(A\mid B)=\frac{P(A\cap B)}{P(B)},\quad P(B)>0.
]

The numerator is the part satisfying both A and B. The denominator is the whole restricted group B, because B is written afterthe vertical bar. If 15% of applicants study both maths and physics and 30% study physics, then (P(M\mid P)=0.15/0.30=0.5). The answer is not 0.15, because the question is no longer choosing from every applicant.

In frequency questions, use the same structure. If 18 of 24 tennis players are left-handed, (P(\text{left-handed}\mid\text{tennis})=18/24). Do not divide by the grand total if a condition has been supplied.

6. Two-way tables: totals are your error check

A two-way table is a compact alternative to a Venn diagram for two categorical variables. Enter every known inner cell and the row/column totals. Use subtraction to complete the table, then check that the row totals, column totals and grand total agree.

For conditional probability, select the row or column created by the condition and use its total as the denominator. For example, “probability of a student being a cyclist, given that they are in Year 13” means take the Year 13 cyclists over all Year 13 students. This representation is often the most reliable choice when the question gives counts rather than probabilities.

7. Tree diagrams: multiply along, add across

Use a tree diagram for consecutive stages, such as two selections, two tests or a sequence of matches. Branches leaving one node must total 1. Multiply along a complete route because every stage on that route must happen. Add across separate routes when any of them gives the required outcome.

If three red and two blue counters are selected twice without replacement, then

[
P(R\text{ then }B)=\frac35\times\frac24=\frac3{10}.
]

The probability of one of each colour requires both routes: red then blue or blue then red. Therefore,

[
P(\text{one of each})=\frac35\times\frac24+\frac25\times\frac34=\frac35.
]

The second denominator becomes 4 because one counter has gone. Candidates who retain a denominator of 5 have treated a dependent process as independent.

8. Replacement, dependence and modelling assumptions

With replacement, the composition returns to its starting state. The probabilities on the second branches are unchanged, so the selections are independent. Without replacement, the composition changes, so later probabilities must be recalculated.

Context questions also test modelling. “Selected at random” normally assumes each individual has an equal chance. A binomial-style situation may assume a constant probability and independent trials. Credit can be given for a sensible stated limitation, for example that a survey might be biased or that real customer choices may not be independent. Do not invent assumptions that the question has already specified; state only those needed to make the model usable.

Mathematical Relationships

For these probability identities, be prepared to recall and apply the relationship. A formula alone usually earns limited credit. Marks are awarded for substituting the correct event probabilities and interpreting the answer in context.

Data and Diagram Skills

Probability questions do not usually require a plotted graph, but they do require precise data handling. On a Venn diagram, write values into disjoint regions rather than repeatedly writing set totals. In a two-way table, show every row/column total. On a tree, label every branch and check each split sums to 1. Keep fractions during multi-step work where possible, then give a decimal only if asked. Probabilities have no units, but a frequency must be clearly distinguished from a probability: 18 students is not the same as (18/60).

Practical Applications

Probability models weather forecasts, insurance risk, medical screening, manufacturing quality control and A/B testing. The mathematical technique is the same, but a strong answer recognises when assumptions may be weak. For example, using a model for product faults assumes the fault rate is stable; using a survey assumes the sample represents the wider population. In real applications, conditional probability is particularly important because the relevant group is often already filtered, such as patients who tested positive or customers who clicked an advert.

There is no required practical for this A-Level Mathematics topic. The practical skill is instead the disciplined construction and checking of representations: Venn diagrams, two-way tables and tree diagrams.