Central Limit Theorem

What the Central Limit Theorem Says

The central limit theorem (CLT) is arguably the most important result in all of statistics. It says that when you add up (or average) a large number of independent random values, the distribution of that sum or average is approximately a normal distribution — the familiar bell curve — almost no matter what shape the original values came from. The individual pieces can be flat, skewed, lumpy, or weird; their sum still trends toward a bell. The simulator above demonstrates this with the simplest possible setup: rolling two six-sided dice and charting the distribution of the total.

This is a remarkable and slightly magical claim. It means the bell curve is not a special assumption you impose on data — it is what emerges naturally whenever many small, independent effects add together. That is why the normal distribution shows up everywhere: heights, measurement errors, test scores, and the averages behind almost every statistical confidence interval.

The 2d6 Intuition: From Flat to Triangular to Bell

Start with a single die. Each face — 1 through 6 — is equally likely, so its distribution is perfectly flat (uniform). There is nothing bell-shaped about one die.

Now add a second die and look at the sum. Suddenly the distribution is no longer flat — it is a triangle that peaks at 7. Why? Because there are six different ways to roll a 7 (1+6, 2+5, 3+4, 4+3, 5+2, 6+1) but only one way to roll a 2 (1+1) or a 12 (6+6). The middle totals have many more combinations than the extremes, so the middle stacks up tall and the ends stay short. That is exactly the triangular histogram you see in the simulator above when you roll 2d6 many times.

Keep going. Sum three dice and the triangle starts to round off at the corners. Sum five or ten dice and the shape becomes a smooth, symmetric bell. Each die you add is another flat, independent contribution — and the more of them you add, the more the combined distribution looks normal. Two dice already hint at the bell; more dice complete it. The underlying die never changes shape; the sum does, and that is the whole point of the theorem.

Why Sums and Averages Go Normal

The mechanism is the same combination-counting you can see with the dice. For a sum to land at an extreme value, every component has to cooperate — all dice must be high, or all must be low — and that is rare. For a sum to land in the middle, the components can balance each other in an enormous number of ways: a high here cancels a low there. Because there are vastly more ways to be average than to be extreme, probability piles up in the center and tapers symmetrically toward the tails. That pile-up is the bell curve. The theorem guarantees this happens for essentially any underlying distribution with a finite variance, as long as the pieces are independent and you add up enough of them.

Sample Size: How Many Is "Enough"?

The CLT is a statement about what happens as the number of terms grows, so a natural question is how many you need before the approximation is good. The honest answer is it depends on the underlying shape. If the original distribution is already symmetric and well-behaved (like a single die), even adding a handful of values gives a near-bell. If the original is heavily skewed, you need more terms to wash the skew out. A common rule of thumb in introductory statistics is that a sample size around 30 is "large enough" for the sampling distribution of the mean to look approximately normal — but that is a guideline, not a law. The dice demo lets you build the intuition for yourself: more rolls give a cleaner, smoother bell, and adding more dice per roll sharpens the shape faster than rolling the same two dice more times.

Why It Matters in the Real World

The central limit theorem is the engine under the hood of practical statistics. It is the reason polling works: even though any one person answers yes or no (a flat, two-valued variable), the average across a sample is approximately normal, which lets pollsters attach a margin of error. It is why manufacturing quality control, A/B testing, confidence intervals, and hypothesis tests can assume normality for sample means even when individual measurements are anything but normal. It also explains why so many natural quantities — where the outcome is the sum of countless small independent influences — end up bell-shaped on their own. Use the Export PNG button to capture the converging histogram for a lesson, or copy the embed code to drop the live demo onto a course page. Few abstractions in statistics become this obvious once you have watched a pile of dice rolls build a bell with your own eyes.