Sampling and the Central Limit Theorem

You can't measure everyone. So statistics does something audacious: it studies a small sample and draws conclusions about the whole population. One theorem makes that leap trustworthy.

Statistics vs. parameters

Keep two words straight:

• A parameter describes the whole population — the true mean μ, unknown.
• A statistic describes your sample — the sample mean x̄, which you can compute.

Every sample gives a slightly different x̄. So x̄ is itself random, with its own distribution — the sampling distribution of the mean.

The Central Limit Theorem

Here's the near-magical result:

However the population is shaped — even skewed — the sampling distribution of the mean is approximately normal for a large enough sample, centered on the true μ, with a spread that shrinks as the sample grows:

$\sigma_{\bar{x}} = \frac{\sigma}{\sqrt{n}}$

That √n is the whole game. To halve the uncertainty in your estimate you need four times the data — precision is expensive, but predictable.

Why this is the foundation

The Central Limit Theorem (CLT) is why the normal distribution rules statistics even when the raw data isn't normal. Because x̄ is reliably normal, we can attach exact probabilities to how far it might land from μ — and that's precisely what the next two lessons do: build confidence intervals and test hypotheses.

Why this matters

Polls, medical trials, and quality control all rest on this one idea: a well-drawn sample of a few thousand can pin down a nation of millions, and the CLT tells us exactly how much to trust it. Without it, inference would be guesswork; with it, it's arithmetic.