The Central Limit Theorem

Picture rolling one die thousands of times. Every outcome from 1 to 6 is equally likely, so the distribution is flat. Now roll 30 dice and record the average each time. Most averages cluster near 3.5, with extreme averages rare, because getting all high or all low rolls together is unlikely. Averaging smooths out the randomness: high and low values tend to cancel, pulling most sample means toward the center. That canceling effect is the engine behind the Central Limit Theorem.

As sample size n increases, the sampling distribution of the sample mean looks more and more normal, even if the original population is strongly skewed. With small n, the sampling distribution still resembles the population's skew. As n grows, each sample contains a wider mix of values, so extreme means become rarer and the distribution tightens and symmetrizes around the true population mean. By around n = 30 for most moderately skewed populations, the approximation is usually good enough to use normal-based methods, though heavier skew needs larger n.

This result is what makes confidence intervals and hypothesis tests for means possible later in the course: those methods assume the sampling distribution of the mean is approximately normal, and the CLT is the justification. It connects back to sampling distributions in general, extending the idea that statistics vary from sample to sample, and shows why normal-based inference works even when raw data clearly isn't normal.