Statistics & Data

Watch

• How statistics can be misleading - Mark Liddell ↗TED-Ed
• Beyond the Numbers: A Data Analyst Journey | Anna Leach | TEDxPSU ↗TEDx Talks
• Why you should love statistics | Alan Smith ↗TED
• The NBA Data Scientist ↗Bloomberg Originals

Read

• How to Lie with Statistics ↗A timeless primer on the myriad ways data can mislead, essential for cultivating a healthy scepticism towards presented figures.Darrell Huff
• Naked Statistics: Stripping the Dread from the Data ↗Demystifies core statistical concepts with engaging real-world examples, making the subject accessible to the numerically wary.Charles Wheelan
• The Signal and the Noise: Why So Many Predictions Fail—but Some Don't ↗Explores the art and science of prediction, dissecting how data, probability, and human judgment intersect in forecasting everything from elections to epidemics.Nate Silver
• An Introduction to Statistical Learning: With Applications in R ↗A foundational textbook offering a rigorous yet accessible entry into modern statistical learning methods, indispensable for data practitioners.Gareth James, Daniela Witten, Trevor Hastie, Robert Tibshirani
• Why Most Published Research Findings Are False ↗A seminal critique revealing the systemic statistical and methodological flaws that undermine the reliability of much scientific literature.John P.A. Ioannidis

Voices to follow

• Nate Silver ↗Renowned for his pioneering work in predictive analytics, particularly in political forecasting and sports, demonstrating the power of data-driven probabilistic models.Statistician, founder of FiveThirtyEight
• Cathy O'Neil ↗A trenchant critic of algorithmic bias and the unchecked power of 'weapons of math destruction,' she illuminates the ethical pitfalls of data-driven decision-making.Data scientist, author, activist
• David Spiegelhalter ↗A leading voice in statistical literacy and risk communication, he excels at translating complex statistical concepts into understandable insights for the public and policymakers.Statistician, Emeritus Professor at the University of Cambridge
• Andrew Gelman ↗A prolific researcher and blogger, he offers incisive commentary on statistical methodology, Bayesian inference, and the proper application of data in social sciences.Professor of Statistics and Political Science, Columbia University
• Hilary Mason ↗A prominent practitioner and thought leader in applied machine learning, she bridges the gap between cutting-edge research and practical data product development.Data scientist, founder of Fast Forward Labs (acquired by Cloudera)

Glossary

• DataPieces of information or facts, like numbers, words, or measurements, that you collect about something. For example, if you count how many students in your class like pizza, the number you get is a piece of data.
• FrequencyHow often a particular value or item appears in a set of data. It's simply the count of how many times something happens. For example, if you count how many students in your class have blue eyes, and you find 7 students, then the frequency of blue eyes is 7.
• GraphA visual representation of data that uses shapes, lines, or bars to show relationships and patterns, making information easier to understand quickly. Common types include bar graphs, line graphs, and pie charts. For example, a bar graph can show the number of students who prefer different ice cream flavors.
• MeanThe average of a set of numbers, found by adding all the numbers together and then dividing by how many numbers there are. For example, if your test scores are 80, 90, and 70, the mean score is (80+90+70)/3 = 240/3 = 80.
• MedianThe middle value in a set of numbers when those numbers are arranged in order from smallest to largest. If there are two middle numbers, you find their average. For example, if your test scores are 70, 80, 90, the median is 80. If your scores are 70, 75, 80, 90, the median is (75+80)/2 = 77.5.
• ModeThe number or item that appears most often in a set of data. A data set can have one mode, more than one mode, or no mode at all. For example, if students' favorite colors are red, blue, green, red, yellow, red, then "red" is the mode because it appears most frequently.
• PopulationThe entire group of people, objects, or events that you are interested in studying. It's the complete set of everything you want to learn about. For example, if you want to know the average shoe size of all students in your school, then all the students in your school make up the population.
• RangeThe difference between the highest and lowest values in a set of data. It tells you how spread out the data is. For example, if the temperatures recorded in a week were 20, 22, 18, 25, 21, 19, 23 degrees Celsius, the highest is 25 and the lowest is 18, so the range is 25 - 18 = 7 degrees Celsius.
• SampleA smaller, representative group chosen from a larger population to study. We often use a sample because studying the entire population might be too difficult or time-consuming. For example, instead of asking every student in your school about their favorite sport (the population), you might ask 50 students from different grades (a sample) to get an idea.
• StatisticsThe study of collecting, organizing, analyzing, and presenting data to understand information and make decisions. It helps us make sense of large amounts of information. For example, using statistics, you can figure out the average height of students in your school or predict who might win an election.
• SurveyA method of collecting data by asking a group of people questions, usually through questionnaires or interviews. It's a way to gather opinions, facts, or preferences. For example, your school might conduct a survey to find out what new sports students would like to have.
• VariableA characteristic or feature that can be measured or observed and can change or have different values. For example, in a study about students, "age" is a variable because different students have different ages, and "favorite color" is also a variable because students have different favorite colors.

Threads 16

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 Political Science

  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.

• 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 Political Science

  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.

• Biomedical Engineering Technology

  Biomedical measurements need statistical analysis to distinguish reliable evidence from noise or variation.

  Sources: U.S. Bureau of Labor Statistics — Bioengineers and Biomedical Engineers ↗ · ABET engineering program criteria 2025–2026 ↗

• Industrial Engineering Technology

  Process measurements help distinguish random variation from evidence that an intervention changed performance.

  Sources: U.S. Bureau of Labor Statistics — Industrial Engineers ↗ · ABET engineering program criteria 2025–2026 ↗

• Disaster Risk Reduction Global Challenges

  Risk-reduction indicators can hide missing populations and unequal outcomes. Statistical reasoning helps distinguish observed associations from causal effects and evaluate whether prevention works.

  Sources: Global status of multi-hazard early warning systems 2024 ↗

• Food Security Global Challenges

  Different food-security indicators measure different conditions and populations. Statistical literacy prevents adding overlapping counts or mistaking a global trend for every household’s experience.

  Sources: Global hunger declines, but rises in Africa and western Asia: UN report ↗ · Acute food insecurity and malnutrition rose for sixth consecutive year in world’s most fragile regions ↗

• The Periodic Table & the Elements Science

  Reading periodic trends is pattern-finding in data — spotting the regularities that let you predict an element you have never measured.

  Sources: RSC — Periodic Table ↗