Statistics

Key people

• Florence NightingaleInvented the polar-area chart to prove hygiene saved soldiers' lives.Foundations
• Ronald FisherBuilt modern statistical inference — significance, experiments, and all.Inference
• David SpiegelhalterBritain's clearest public voice on risk, evidence and uncertainty.Today

Timeline

• 1662John Graunt reads London's death records — arguably the first data science.
• 1854John Snow maps cholera deaths and finds the Broad Street pump.
• 1925Fisher formalises significance testing in Statistical Methods.
• 2012'Data scientist' is called the sexiest job of the 21st century.

Read

• How to Lie with Statistics — Darrell HuffThe 1954 classic on the tricks numbers play. Still undefeated.
• The Art of Statistics — David SpiegelhalterHow to think clearly with data, from a master of the craft.

Watch

• The danger of mixing up causality and correlationIonica Smeets on the trap behind half the charts you see.TED-Ed · via YouTube
• The Joy of Stats — Hans Rosling200 countries, 200 years, 4 minutes — data made alive.Documentary · via YouTube

Listen

• Statistics Made SimpleBrad R. Fulton, PhDPodcast
• Department of StatisticsOxford UniversityPodcast
• Statistics with CrayonsPenn Dental MedicinePodcast
• Data Science Deep DiveINWT Statistics GmbHPodcast

Voices to follow

• Bjørn Lomborg@BjornLomborg · XDanish environmental economy author noted for his controversial views about climate change and resource scarcity
• Michael BagalmanYouTubeAmerican data science executive, author, and educator
• Carlos García Meixide@carmeiga · XSpanish mathematical statistician
• Dimitris Rizopoulos@drizopoulos · XResearcher (ORCID 0000-0001-9397-0900)

By the numbers

• p < 0.05the significance line much of science treats as 'real' — and why that's fragile.
• ~50%of published findings in some fields fail to replicate (the replication crisis).

Debates

• Is p < 0.05 a broken standard?A convenient line — or an engine of false discoveries?
• Can a statistic ever be neutral?Every choice of what to measure quietly encodes a value.

Glossary

• p-valueHow surprising the data would be if nothing were really going on.
• CorrelationTwo things moving together — which is not the same as causation.
• Sampling biasWhen who you happened to measure quietly decides your answer.

Careers

Where this shows up as work

Threads 13

Where this connects to other fields — and why it's worth knowing.

• Media Ethics & Representation Media

  If a newsroom only ever quotes rich men, its picture of the world comes out lopsided, no surprise. That's the same trap as a bad science survey: ask a skewed slice of people and you get a skewed answer. Whether it's whose voices make the news or whose faces train an AI, fair representation is really just the old problem of taking a fair sample.

• Democracy Politics

  Here's a surprising math fact: if each voter is even slightly more likely to be right than a coin flip, and they vote on their own, a big majority is almost always correct. The individual errors cancel out in the crowd, like averaging a survey. Democracy works not because each voter is a genius, but because averaging washes the mistakes away.

• Map Scale and Gerrymandering Geography

  Take the same map data and just redraw the boundaries of the regions, and you can flip a result from 'these two things rise together' to 'they move opposite.' It's a real trap called the Modifiable Areal Unit Problem. The scary part is that even honest researchers can accidentally 'prove' opposite conclusions from the very same numbers.

• Close Reading Literature

  Want to unmask an author hiding behind a fake name? Don't count their big fancy words, count the boring ones like 'the,' 'of,' and 'and.' Everyone sprinkles those in their own unconscious pattern, a fingerprint you can't fake because you don't even know you're doing it. Counting the dull words has exposed anonymous novelists for real.

• News, Journalism & Verification Media

  An editor deciding whether to publish a hot tip is basically a smoke detector. Too jumpy and it screams at burnt toast (printing a fake story); too dull and it misses a real fire (killing a true scoop). Every newsroom is really just choosing how sensitive to set that alarm, the same trade-off statisticians make.

• Decolonisation History

  Before colonial rulers, many identities in India were blurry and could shift over a lifetime. Then officials ran a census and forced everyone into fixed boxes for caste and tribe. Just counting people that rigid way hardened those boxes into real, permanent groups, some of which still vote as solid blocs today.

• NEP 2020 Education

  When India swaps memorize-and-recite exams for skill-based testing, there's a hidden cost: you can no longer compare this year's students to last year's. It's like changing your measuring tape mid-experiment; if scores rise, you can't tell if kids got smarter or the ruler just changed.

• Student Mental Health Education

  Grading 'on a curve' guarantees losers by pure math. If you rank students by percentile, exactly half must land 'below average' no matter how brilliant the whole class gets. The ranking manufactures failures out of arithmetic, even if everyone improved.

• Corruption and Governance Politics

  Made-up numbers leave a fingerprint. In real budgets and vote counts, the first digit is a 1 way more often than a 9, a pattern called Benford's Law. Plot the first digits, and faked figures that looked fine in a spreadsheet suddenly stick out.

• Business Case Studies & Corporate Collapses Business

  We love studying the famous companies that crashed and asking what killed them. But thousands of other firms made the exact same 'fatal' move and lived, and nobody wrote books about them. Studying only the dead is like judging a risky stunt by asking the people who survived it.

• Education and Social Mobility Sociology

  A scientist noticed that very tall parents usually have kids who are tall, but a bit shorter, drifting back toward average. That same math says huge advantages should fade over generations too. Yet inheritance and elite schools act like glue, holding the top families in place against that pull.

• Data Privacy Law Law

  Companies say your data is "anonymous" — no name attached. But just four time-and-place stamps (where you were, when) can pick you out of millions with about 95% accuracy. So the whole legal idea of "anonymized data" is basically a comforting fiction.

• Indian History History

  British colonial officials counted everyone and forced fluid caste and religious identities into fixed boxes on a form. That act of measuring made the categories feel rigid and permanent. And that hard counting helped make the violent split of Partition thinkable.