Binomial Parameters
A binomial distribution models the number of successes in a fixed number of independent trials, where each trial has the same probability of success. Two numbers — the mean and the variance — give us an instant summary of where the distribution is centred and how spread out it is, without listing every possible outcome. The mean (also called the expected value) of a binomial distribution tells you, on average, how many successes to expect. If you flip a fair coin 100 times, your intuition says 'about 50 heads' — and the formula confirms exactly that. The variance measures how much the actual count is likely to deviate from that average, capturing the randomness inherent in the process. Mastering these two parameters is essential because they appear in real-world applications ranging from quality control in manufacturing to predicting the spread of infectious diseases. Once you can calculate the mean and variance quickly, you can compare distributions, make predictions, and decide when observed data is surprisingly different from what chance alone would produce.
- Define the Binomial Setup
- Mean of a Single Bernoulli Trial
- Mean of the Full Binomial
- Variance of a Single Bernoulli Trial
- Variance of the Full Binomial
- Interpreting the Parameters
Define the Binomial Setup
A binomial random variable X counts the number of successes in n independent Bernoulli trials, each with probability of success p (and failure q = 1 − p). We write X ~ B(n, p). Each individual trial contributes either 1 (success) or 0 (failure), so X = X₁ + X₂ + ⋯ + Xₙ, where each Xᵢ is a Bernoulli(p) random variable.
Mean of a Single Bernoulli Trial
For one trial Xᵢ, E(Xᵢ) = 1 · p + 0 · (1 − p) = p. This is the expected value of a single Bernoulli random variable. The mean is simply the probability of success — nothing more.
Mean of the Full Binomial
By the linearity of expectation — a powerful rule that holds whether or not random variables are independent — E(X) = E(X₁ + X₂ + ⋯ + Xₙ) = E(X₁) + E(X₂) + ⋯ + E(Xₙ) = p + p + ⋯ + p (n times) = np. Therefore the mean of X ~ B(n, p) is μ = np.
Variance of a Single Bernoulli Trial
For one trial, Var(Xᵢ) = E(Xᵢ²) − [E(Xᵢ)]². Since Xᵢ only takes values 0 or 1, Xᵢ² = Xᵢ, so E(Xᵢ²) = p. Therefore Var(Xᵢ) = p − p² = p(1 − p) = pq.
Variance of the Full Binomial
Because the n trials are independent, variances add: Var(X) = Var(X₁) + Var(X₂) + ⋯ + Var(Xₙ) = pq + pq + ⋯ + pq (n times) = npq. Therefore the variance of X ~ B(n, p) is σ² = np(1 − p). The standard deviation is σ = √(np(1−p)), which gives the spread in the same units as X.
Interpreting the Parameters
Notice that variance is maximised when p = 0.5 (most uncertain) and shrinks toward zero as p approaches 0 or 1 (outcome becomes nearly certain). Also, even though μ = np grows linearly with n, σ = √(npq) grows only as √n — meaning large samples are proportionally more predictable, the mathematical foundation of why polling works.
The Key Equation
- \mu — Mean (Expected Value) (successes): The average number of successes expected over many repetitions of the n trials.
- \sigma^2 — Variance (successes squared): Measures the average squared deviation from the mean; indicates how spread out the distribution is.
- n — Number of Trials (trials): The fixed number of independent Bernoulli trials performed.
- p — Probability of Success (dimensionless): The probability that any single trial results in a success; must satisfy 0 ≤ p ≤ 1.
- 1-p — Probability of Failure (dimensionless): Often written as q; the probability that a single trial results in failure.
Worked Example
Problem. A factory produces light bulbs, and 8% of all bulbs are defective. A quality inspector randomly selects 150 bulbs. Let X be the number of defective bulbs found. (a) State the distribution of X, giving its parameters. (b) Calculate the mean number of defective bulbs. (c) Calculate the variance and standard deviation, giving your answers to 2 decimal places. (d) Would finding 25 defective bulbs be surprising? Justify your answer using the mean and standard deviation.
Answer. X ~ B(150, 0.08); mean = 12; variance = 11.04; standard deviation ≈ 3.32; finding 25 defectives is approximately 3.92 standard deviations above the mean and would be very surprising.
Origins · 1 minJakob Bernoulli and how this was discovered
Jakob Bernoulli was the first mathematician to systematically analyse what we now call Bernoulli trials — the binary success/failure experiments that are the building blocks of the binomial distribution. His landmark book Ars Conjectandi (The Art of Conjecturing), published posthumously in 1713, established the law of large numbers and laid the groundwork for probability as a rigorous mathematical discipline. The binomial distribution's individual trials (Bernoulli trials) and the Bernoulli distribution itself are named in his honour. Bernoulli was so fascinated by the logarithmic spiral that he requested it be engraved on his tombstone with the motto 'Eadem mutata resurgo' (Though changed, I rise the same). The stonemason accidentally carved an Archimedean spiral instead — a different curve entirely!
Related Concepts
- Binomial Distribution — The mean and variance are properties of the binomial distribution; understanding the full distribution is essential before working with its parameters.
- Bernoulli Trials — Each binomial experiment is built from independent Bernoulli trials; the mean of a Bernoulli trial (equal to p) directly underpins the binomial mean formula.
- Expected Value — The mean of a binomial distribution is its expected value; understanding E(X) and the linearity of expectation is the core tool used to derive μ = np.
- Variance and Standard Deviation — The binomial variance σ² = np(1−p) is a special case of the general concept of variance; knowing how variance behaves helps interpret the spread of any distribution.
- Normal Approximation to the Binomial — For large n, a binomial distribution is approximately normal with the same mean np and variance np(1−p); this connection shows why the parameters are so important.