Course contentsShow
← Mathematics
Lesson 2982 of 3,12158. Joint and Limit BehaviorFree lesson

CLT for Non-Normal Populations

Apply the CLT to sample means from skewed, uniform, and other non-normal distributions.

CLT for Non-Normal Populations

What you'll learn: How the Central Limit Theorem transforms sample means from any distribution—skewed, uniform, or otherwise—into approximately normal distributions.

The Surprising Power of Averaging

By now you know the Central Limit Theorem works beautifully when sampling from normal populations. But here's the magic: the CLT doesn't care what shape your original distribution has. Whether you're sampling from a wildly right-skewed distribution, a uniform distribution, or anything else with finite variance, the sample mean will still follow an approximately normal distribution as sample size grows.

Why This Matters

Imagine measuring wait times at a hospital (exponential distribution—very right-skewed) or rolling dice (discrete uniform). Individual observations don't look normal at all. But when you take the average of many observations, something remarkable happens: those averages cluster into a bell curve.

The Mechanics

For a population with:

  • Mean μ (any value)
  • Standard deviation σ (finite)
  • Any shape (skewed, uniform, bimodal—doesn't matter!)

The sample mean X̄ from n independent observations has:

  • Mean: μ
  • Standard error: σ/√n
  • Distribution: approximately Normal(μ, σ²/n) when n is large enough

The key question: "How large is large enough?" It depends on how non-normal your population is. More skewness requires larger n (often n ≥ 30 is sufficient; severely skewed distributions might need n ≥ 50).

Practical Impact

This universality lets you make probability statements and confidence intervals about sample means even when you have no idea what the population distribution looks like—as long as you have enough data.

Key Takeaway: The CLT applies to sample means from any distribution with finite variance, making the normal distribution a universal tool for inference regardless of population shape.