CLT for Sample Sums
What You'll Learn
You'll learn how the Central Limit Theorem applies not just to sample means, but also to the sum of random variables, and how to adjust the standardization accordingly.
The Core Idea
You've seen that the sample mean of many independent, identically distributed random variables becomes approximately normal as the sample size grows. But what about the sum itself?
If X₁, X₂, ..., Xₙ are independent random variables each with mean μ and variance σ², then their sum:
Sₙ = X₁ + X₂ + ... + Xₙ
also becomes approximately normal for large n. The difference is in the parameters:
- Mean of Sₙ: E[Sₙ] = nμ (not just μ)
- Variance of Sₙ: Var(Sₙ) = nσ² (not σ²/n)
- Standard Deviation of Sₙ: SD(Sₙ) = σ√n (not σ/√n)
The Standardization
To apply the CLT to a sum, standardize using:
Z = (Sₙ - nμ) / (σ√n)
This Z-score is approximately standard normal N(0,1) for large n.
The Connection
Remember: the sample mean is just X̄ₙ = Sₙ/n. When you standardize the mean, you get (X̄ₙ - μ)/(σ/√n). When you standardize the sum, you get (Sₙ - nμ)/(σ√n). These are mathematically equivalent—just multiplied by n!
When to Use Which
Use the sum formulation when you care about totals (total revenue, total errors, cumulative time). Use the mean formulation when you care about averages (average score, mean height).
Key Takeaway: The CLT applies to sample sums Sₙ with mean nμ and standard deviation σ√n, allowing you to find probabilities for totals using normal approximation.