Mean and Variance of the Hypergeometric Distribution
What you'll learn: You'll derive and apply the formulas for the expected value and variance of a hypergeometric random variable.
Understanding the Formulas
Recall that the hypergeometric distribution models sampling without replacement from a finite population containing two types of items. If you draw n items from a population of N total items (with K "successes"), the number of successes X follows a hypergeometric distribution.
The mean (expected value) is:
E[X] = n · (K/N)
This makes intuitive sense! The proportion of successes in the population is K/N, and you're drawing n items. On average, you'd expect that same proportion among your draws.
Analogy: If 30% of marbles in a jar are red, and you grab 10 marbles, you'd expect about 3 red ones on average—whether you replace them or not.
The variance is:
Var(X) = n · (K/N) · (1 - K/N) · [(N - n)/(N - 1)]
Notice the term (N - n)/(N - 1) — this is the finite population correction factor. It accounts for sampling without replacement. When N is very large compared to n, this factor approaches 1, and the variance approaches the binomial variance n·p·(1-p).
Key Insight
The finite population correction is always ≤ 1, meaning the hypergeometric variance is smaller than the corresponding binomial variance. This makes sense: without replacement, there's less variability because each draw constrains future possibilities.
Key Takeaway: The hypergeometric mean is n·(K/N), identical in form to the binomial mean, while the variance includes a correction factor (N-n)/(N-1) that reduces variability due to sampling without replacement.