Hypergeometric Distribution
How does sampling without replacement from a finite population introduce statistical dependence and define the Hypergeometric distribution?
How this lesson unfolds.
Begin with Why does the expected value of a Hypergeometric distribution equal n(K/N), exactly matching the Binomial mean despite trials being dependent? The sequence then connects The Canonical Four: Navigating the Hypergeometric Notation Table and builds toward The Canonical Four: Navigating the Hypergeometric Notation Table.
Why does the expected value of a Hypergeometric distribution equal n(K/N), exactly matching the Binomial mean despite trials being dependent?
Prior to observing earlier draws from an urn without replacement, every individual extraction carries an identical marginal success chance of K/N; because mathematical expectation adds linearly regardless of dependency, the expected total count sums unconditionally to n(K/N), such that for N=20, K=6, and n=5, E[X] = 5 × (6/20) = 1.5.
- $$E[X] = \sum_{i=1}^n E[I_i] = n \left(\frac{K}{N} ight), \quad P(X = 2) = \frac{\binom{6}{2}\binom{14}{3}}{\binom{20}{5}} = \frac{15 \times 364}{15{,}504} \approx 0.3522$$
- $$Because Linearity of Expectation holds regardless of dependence: E[X] = \sum_{i=1}^n E[I_i] = n \times (K/N), identical to sampling with replacement.$$
- Learning goalProve Hypergeometric expectation E[X] = n(K/N) using linearity over dependent indicator variables.
- Helpful foundationsLinearity of expectation · Indicator variables · Sampling without replacement · Foundations of the Theory of Probability (Kolmogorov, 1933) · An Essay towards solving a Problem in the Doctrine of Chances (Thomas Bayes, 1763)
- Common trapAssuming dependence alters expectation; expectation remains strictly linear regardless of indicator correlation.
- Next connectionMap narrative problem statements to parameters N, K, n, x.
More context & mechanics
In a finite population of size N containing K successes and N-K failures, we sample n items without replacement. Let X count the successes drawn. Define indicator variables Iᵢ = 1 if the i-th draw is a success, and 0 otherwise. By symmetry, without knowing prior outcomes, P(Iᵢ = 1) = K/N for every single draw i ∈ 1, 2, ṡ, n. Therefore, E[Iᵢ] = K/N. Invoking Linearity of Expectation: E[X] = E[ ∑_(i=1)^n Iᵢ ] = ∑_(i=1)^n E[Iᵢ] = n (K/N). The mean is totally immune to dependence!
In ecological wildlife tagging (Capture-Recapture / Lincoln-Petersen index), biologists capture K animals, tag them, release them back into population N, and draw a second sample of n animals. The expected number of tagged recaptures is n(K/N), enabling estimation of total population N = nK/X.
The capture-recapture method was developed by Pierre-Simon Laplace in 1786 to estimate the population of France, and independently formalized by C.G.J. Petersen in 1896 for marine fish populations.
The Immunity Rule: 'Expectation doesn't care about dependence'. Mean is simply: Sample size (n) × Population Success Ratio (K/N).
How do you systematically distinguish between population-level and sample-level parameters in the Hypergeometric framework?
Upper-case symbols specify the macro-environment: population size N and total target items K; lower-case symbols specify the micro-sample: items drawn n and observed successes x, bounded by max(0, n+K-N) ≤ x ≤ min(n, K).
- $$\max(0, n+K-N) \le x \le \min(n, K), \quad K \le N, \; n \le N$$
- $$Capital letters denote the population (N = total items, K = total successes); lowercase letters denote the sample (n = sample size, x = observed successes).$$
- Learning goalAssign parameters N, K, n, x from narrative problems and determine support bounds.
- Helpful foundationsFinite populations · Parameter identification · Support constraints · Foundations of the Theory of Probability (Andrey Kolmogorov, 1933) · An Essay towards solving a Problem in the Doctrine of Chances (Thomas Bayes, 1763)
- Common trapSwapping population successes K with sample size n; inverted roles invalidate combination counts.
- Next connectionExplore Finite Population Correction and full PMF derivations in depth.
More context & mechanics
The Hypergeometric distribution is governed by four interrelated integers:
- N (Population Size): Total number of items in the finite population.
- K (Population Successes): Total number of favorable items in the population (leaving N - K failures).
- n (Sample Size): Total number of items extracted without replacement.
- x (Sample Successes): The random variable outcome—how many of the n items are successes.
The support of x is bounded by physical reality: max(0, n - (N - K)) ≤ x ≤ min(n, K). You cannot draw more successes than exist in the population (x ≤ K) or in your sample (x ≤ n).
In quality assurance auditing, a batch of N=500 manufactured medical stents has K=15 defective units. An FDA compliance auditor draws a random sample of n=20 stents without replacement to check for x defective stents.
The systematic notation table was established in 20th-century acceptance sampling standards (such as MIL-STD-105E) used by the US Department of Defense during WWII.
The Case Rule: 'Capitals for Population (N, K), Lowercase for Sample (n, x)'. N contains K; n contains x.