The Story of Statistics

Every tool you have used in this subject, averages, samples, significance, correlation, has a history. It was invented by particular people, for particular problems, at particular moments. Knowing that story does more than satisfy curiosity. It shows you that statistics was built to answer real questions about populations, risk, and chance, and that its methods still carry the fingerprints of the problems they were built to solve.

The history also carries a warning. Some of the founders of modern statistics used their new tools in the service of ideas we now reject, including eugenics. Seeing how brilliant methods and bad judgement lived side by side is a useful reminder that a technique is only as good as the questions and values guiding its use.

From counting the state to the word statistics

For most of history, gathering numbers meant counting for rulers: censuses, taxes, soldiers, harvests. The word statistics itself grew out of this. It is usually traced to eighteenth-century German scholars, with Gottfried Achenwall often credited for popularising the term Statistik, meaning knowledge about the state. Early statistics was descriptive bookkeeping about a country, not a method of inference.

Chance at the gaming table

A second root grew from gambling. In 1654 the French mathematicians Blaise Pascal and Pierre de Fermat exchanged letters about how to divide the stakes in an unfinished game of chance. That correspondence is widely regarded as the birth of probability theory. Others built on it: Christiaan Huygens wrote the first printed treatise on probability in 1657, and Jacob Bernoulli's law of large numbers, published posthumously in 1713, showed that observed frequencies settle toward their true probability as trials pile up. Counting the state and calculating the odds would eventually merge.

The average man and the rise of a discipline

In the nineteenth century the Belgian scholar Adolphe Quetelet took a bold step. He borrowed the error curve astronomers used for measurement mistakes, what we now call the normal distribution, and applied it to human beings, measuring traits such as height and chest size across populations. From this he proposed l'homme moyen, the average man, a statistical ideal around which real people scattered. His vision of a social physics helped turn statistics from bookkeeping into a science of variation and populations.

Graunt-style reasoning starts with a stable pattern in messy records. In London's christening records John Graunt noticed that boys consistently outnumbered girls, at a ratio of roughly 14 to 13. That works out to about 14 divided by 27, or 52 percent boys, not a fluke of one year but a regularity repeated year after year. From a heap of parish entries he had inferred something about human populations in general. Remarkably, the pattern holds: modern data still show roughly 105 boys born per 100 girls worldwide. Graunt's move, from tallies to a claim about the world, is the essence of statistical inference.

The average man also shows the danger of taking a statistic too literally. Quetelet's average person is a mathematical construct, not a real individual. A country might average 2.3 children per family, yet no household contains 2.3 children. Later thinkers warned that designing for the average can fit almost no one: a cockpit built for the average pilot matched hardly any actual pilot, because no one is average on every dimension at once. An average summarises a group; it does not describe a person.

Much of the machinery of modern statistics was forged in early-twentieth-century Britain. Karl Pearson (1857 to 1936) developed the correlation coefficient that still bears his name and the chi-squared test in 1900, founded the journal Biometrika, and built one of the first university statistics departments, at University College London. Ronald Fisher (1890 to 1962) followed with the design of experiments, randomisation, analysis of variance, and maximum likelihood; his books Statistical Methods for Research Workers (1925) and The Design of Experiments (1935) shaped how research is done to this day, including the popular 0.05 significance threshold.

Yet both men were leading advocates of eugenics, the belief that human populations should be improved through selective breeding, an idea later used to justify grave injustices. Francis Galton, Pearson's mentor and a pioneer of correlation and regression, coined the word eugenics in 1883. This is not a footnote to be hidden. It shows that powerful, genuinely useful methods can be developed by people whose broader beliefs were mistaken and harmful. The tools survived and were put to better uses; the ideology did not, and should not.

• Statistics began in the twentieth century. Its roots go back centuries, to Graunt in 1662 and the probabilists of the 1650s.
• The word statistics always meant number-crunching. It first meant knowledge about the state, closer to what we would call national accounts.
• The founders were neutral technicians. Several were deeply engaged in eugenics and other value-laden causes.
• The average describes a typical individual. An average summarises a group and may match no real person.

You are given shuffled milestones: the Pascal-Fermat letters, Graunt's Bills of Mortality, Quetelet's average man, Pearson's chi-squared test, and Fisher's design of experiments. Put them in order and, for each, say in one line what new idea it added to the toolkit.

Think Like a Maester: Statistics is not a timeless set of formulas but a human invention, shaped by the questions, and the blind spots, of the people who built it.

1. Name the two very different activities that gave statistics its early roots.
2. What did John Graunt do in 1662 that marks an origin of statistical thinking?
3. What did Quetelet mean by the average man, and what is one danger of the idea?
4. Name one contribution each from Karl Pearson and Ronald Fisher.
5. Why is it important to acknowledge the founders' involvement in eugenics?

Statistics grew from two roots: the state's need to count its people and the gambler's need to reckon with chance. John Graunt turned London's mortality records into inference in 1662; Pascal and Fermat opened probability theory in 1654; Quetelet made variation itself a subject of study with his average man. In the early twentieth century Pearson and Fisher assembled much of the modern toolkit, correlation, chi-squared, experimental design, significance testing, even as they championed eugenics. The story leaves us with both a richer sense of where our methods came from and a caution: technique and wisdom do not automatically travel together.

• What Is Statistics?
• Frequentist and Bayesian Thinking