The Binomial and Poisson distributions
S2 Topic 1, spec point by spec point (The Binomial and Poisson distributions). Each spec point has the method, worked examples written the way the mark scheme wants them, how the marks go and the mistakes that cost the most marks.
The S2 paper (WST02) is 1 hour 30 minutes and 75 marks. It assumes S1 and the integration from P1 to P4. Hypothesis tests carry a large share of the marks: learn the full structure of a test.
How maths marks work
M marks are for a correct method, even with a slip in the numbers. A marks are for accurate answers and need the M mark before them. B marks stand alone (a value or a statement). dM means the method mark depends on an earlier M mark.
ft (follow through) lets a later mark be earned from your own earlier wrong value. cso (correct solution only) means every step must be right, which is common on "show that" answers. awrt means "answers which round to", so keep extra figures and round at the end. isw means extra working after a correct answer is ignored.
When a question says "use algebra", "show all stages of your working" or "solutions relying on calculator technology are not acceptable", an answer with no method scores nothing, even if it is right.
Each spec point below has the idea behind it first, then the method, worked examples written the way a mark scheme would want, how the marks go, and the mistakes that cost the most marks. The paper references, such as (WMA11 Jan 2022 Q4), are past questions from the IAL Prep question bank that use the point.
Spec1.1The binomial and Poisson distributions
Use the binomial and Poisson distributions to model real situations and comment on their suitability, finding probabilities by calculation or tables, and use the additive property of the Poisson distribution.
The binomial distribution, \(X \sim B(n, p)\)
Use it when:
- there is a fixed number \(n\) of trials;
- each trial has two outcomes (success or failure);
- the probability of success \(p\) is constant;
- the trials are independent.
\[P(X = r) = \binom{n}{r}p^r(1 - p)^{n - r}\]
Why: \(p^r(1 - p)^{n - r}\) is the probability of one particular order of \(r\) successes, and \(\binom{n}{r}\) counts the orders.
The Poisson distribution, \(X \sim \text{Po}(\lambda)\)
Use it for the number of events in a fixed interval (of time, length, area) when events happen:
- singly (one at a time);
- independently;
- randomly;
- at a constant average rate \(\lambda\) per interval.
\[P(X = r) = \frac{e^{-\lambda}\lambda^r}{r!}, \quad r = 0, 1, 2, \ldots\]
Additive property: if events per minute \(\sim \text{Po}(\lambda)\), then events per 5 minutes \(\sim \text{Po}(5\lambda)\). Independent Poisson variables add: \(\text{Po}(\lambda) + \text{Po}(\mu) = \text{Po}(\lambda + \mu)\).
Using tables (cumulative: \(P(X \leqslant x)\))
| You want | Use |
|---|---|
| \(P(X \leqslant 4)\) | read directly |
| \(P(X < 4)\) | \(P(X \leqslant 3)\) |
| \(P(X \geqslant 4)\) | \(1 - P(X \leqslant 3)\) |
| \(P(X > 4)\) | \(1 - P(X \leqslant 4)\) |
| \(P(3 \leqslant X \leqslant 7)\) | \(P(X \leqslant 7) - P(X \leqslant 2)\) |
| \(P(X = 4)\) | \(P(X \leqslant 4) - P(X \leqslant 3)\), or the formula |
binomial
Each rod has probability 0.05 of being too short. In a sample of 8, find \(P(\text{at least 3 too short})\).
\(X \sim B(8, 0.05)\). \(P(X \geqslant 3) = 1 - P(X \leqslant 2) = 0.0058\).
Poisson, then binomial
Customers arrive at 2 per 10 minutes on average. (a) Find \(P(\text{at most 3 in 10 minutes})\). (b) Find the probability that in 30 minutes between 4 and 7 inclusive arrive. (c) On 20 separate half-hours, find the probability that exactly 15 have between 4 and 7 arrivals.
(a) \(\text{Po}(2)\): \(P(X \leqslant 3) = 0.8571\).
(b) \(\text{Po}(6)\): \(P(X \leqslant 7) - P(X \leqslant 3) = 0.5928\).
(c) \(Y \sim B(20, 0.5928)\): \(P(Y = 15) = \binom{20}{15}(0.5928)^{15}(0.4072)^5 = 0.0681\).
the smallest n
Cars arrive as \(\text{Po}(6)\) per 10 minutes. Find the smallest \(n\) with \(P(X \geqslant n) < 0.05\).
Need \(P(X \leqslant n - 1) > 0.95\). Tables: \(P(X \leqslant 9) = 0.9161\), \(P(X \leqslant 10) = 0.9574\). So \(n - 1 = 10\), \(n = 11\).
- B1 for stating or using the right distribution with its parameters, M1 for the right cumulative expression (e.g. \(1 - P(X \leqslant 2)\)), A1 for awrt the value (WST02A Jan 2026 Q1b).
- Combinations: M1 for the new Poisson mean, M1 for the probability, M1 for the binomial built on it, A1 (WST02 May 2024 Q1).
- Conditions: give them in context ("calls arrive independently", "at a constant rate").
Often asked as: by far the most examined S2 point (167 questions). Probabilities from both distributions; change of interval; distribution of the number of intervals with a property; smallest or largest values; conditions in context.
\(P(X \geqslant 3) = 1 - P(X \leqslant 3)\).
Not scaling \(\lambda\) for a different interval.
Listing conditions with no context.
Spec1.2Mean and variance of the binomial and Poisson distributions
Know and use the mean and variance of the binomial and Poisson distributions.
The results
| Distribution | Mean | Variance |
|---|---|---|
| \(B(n, p)\) | \(np\) | \(np(1 - p)\) |
| \(\text{Po}(\lambda)\) | \(\lambda\) | \(\lambda\) |
The Poisson's mean equals its variance. So if data have a mean close to the variance, a Poisson model is plausible. If the variance is much bigger or smaller, doubt it.
For a binomial, the variance is always less than the mean.
supporting a Poisson model
Over 50 days, complaints give \(\sum x = 99\) and \(\sum x^2 = 294\). Do these support a Poisson model?
Mean \(= \frac{99}{50} = 1.98\). Variance \(= \frac{294}{50} - 1.98^2 = 1.96\).
The mean and variance are close, which supports the Poisson model.
binomial mean and standard deviation
700 adults, each with probability \(\frac{1}{200}\) of an allergy. Mean \(= 3.5\). Standard deviation \(= \sqrt{700 \times 0.005 \times 0.995} = 1.866\).
- B1 for the mean, B1 for the variance, B1 for the comment that they're similar (WST02 Oct 2025 Q2a).
- Binomial standard deviation: using the Poisson value instead scores M0 A0 (WST02 Jun 2017 Q4a).
- Using \(np\) to find \(n\) or \(p\), then deciding which value makes sense in context (WST02 Jun 2026 Q5).
Often asked as: find \(n\) and \(p\) from a given mean and variance; justify a Poisson model; compare a model's mean with data.
Giving the variance when the standard deviation is asked.
Spec1.3The Poisson approximation to the binomial
Use the Poisson distribution as an approximation to the binomial distribution.
When and why
If \(X \sim B(n, p)\) with \(n\) large and \(p\) small, then \(X \approx \text{Po}(np)\). A rule of thumb: \(n > 50\) and \(np < 5\). The binomial's variance \(np(1 - p)\) is then close to \(np\), matching the Poisson.
It makes calculations easier when tables don't cover \(n\) or \(p\).
approximation
Each rod is too short with probability 0.05. Use a suitable approximation for \(P(\text{more than 5 too short})\) in a sample of 60.
\(B(60, 0.05) \approx \text{Po}(3)\). \(P(Y > 5) = 1 - P(Y \leqslant 5) = 1 - 0.9161 = 0.0839\).
- B1 for \(\text{Po}(np)\), M1 for the cumulative expression, A1 (WST02A Jan 2026 Q1c).
- Reasons the Poisson is suitable: "\(n\) is large and \(p\) is small", or "\(p\) is not close to 0.5". "Mean approximately equals variance" is not accepted as a reason here (WST02A Oct 2025 Q3d).
Often asked as: approximate then compare with an exact value; choose between Poisson and Normal approximations.
Using the Poisson approximation when \(p\) is near 0.5 (use the Normal, 3.2).