Hypergeometric vs. Binomial: When to Use Each
What you'll learn: You'll learn how to decide whether a problem calls for the binomial or hypergeometric distribution based on how sampling is done.
The Key Question: Replacement or Not?
Both the binomial and hypergeometric distributions count successes in a fixed number of trials, but they differ in one crucial way: how you sample.
Binomial Distribution
Use the binomial when:
- Trials are independent
- The probability of success stays constant across trials
- You're sampling with replacement (or from an infinite/very large population)
Example: Flipping a coin 10 times. Each flip has probability 0.5 of heads, regardless of previous outcomes.
Hypergeometric Distribution
Use the hypergeometric when:
- You're sampling without replacement from a finite population
- The probability of success changes from trial to trial
- You know the exact composition of the population
Example: Drawing 5 cards from a standard deck and counting aces. After drawing one ace, there are fewer aces remaining—the probability changes.
The Practical Rule of Thumb
If your sample size is small relative to the population (say, less than 5% of the population), the binomial provides an excellent approximation even when technically sampling without replacement. The probabilities don't change enough to matter.
For small populations or large sample-to-population ratios, use the hypergeometric for accuracy.
Key Takeaway: Use binomial for independent trials with constant success probability (sampling with replacement); use hypergeometric for dependent trials from finite populations (sampling without replacement).