What Statistics Is (and Isn't)

Every day the world throws piles of numbers at us — test scores, prices, wait times, click counts, temperatures. On their own, a thousand numbers are just noise. Statistics is the set of tools for turning that pile into something a human can actually understand: a typical value, a sense of the spread, a shape, a fair comparison, a trustworthy conclusion.

Put simply, statistics is the science of learning from data. It has two big jobs, and knowing which one you are doing is the first step to using it honestly.

Almost every interesting question is really about a population — the entire group you care about: every voter in a country, every battery a factory will ever make, every customer of a shop. Populations are usually too big, too expensive, or even impossible to measure completely.

So instead we collect a sample — a smaller group we actually observe — and use it to reason about the whole. A number that describes the population (like the true average height of all adults) is called a parameter; the matching number computed from your sample is a statistic. Much of inferential statistics is the art of using a sample statistic to estimate an unknown population parameter, while staying honest about how far off it might be.

Before you can summarize data, you have to know what type of data it is. A variable is any characteristic you record, and variables come in two main families:

• Categorical (also called qualitative) — labels or groups, like eye color, country, or a yes/no answer. You can count how many fall in each category, but adding or averaging the labels makes no sense.
• Numerical (quantitative) — actual quantities you can do arithmetic on. These split again: discrete values come in separate, countable steps (number of children, goals scored), while continuous values can fall anywhere on a scale (height, temperature, time).

The type decides everything downstream: which summary to compute, which chart to draw, which method to use. Averaging blood types is nonsense; counting how many people have each type is exactly right.

You cannot hold a thousand numbers in your head, but you can hold one. That is why the first move in statistics is almost always to summarize: replace the whole pile with a few well-chosen numbers that capture what matters. The most familiar summary is the mean — the ordinary average — which marks the balance point of the data:

A single summary is powerful, but it hides something important: the shape of the data. Two datasets can share the exact same average yet look completely different — one tightly bunched, one wildly spread, one lopsided, one with two separate peaks. The classic way to reveal that shape is a histogram: slice the range into bins and stack up how many values land in each.

The tool below lets you build one from scratch. Pick a kind of data, add values, and watch a shapeless list of numbers organize itself into a picture worth describing.