Bradley Efron
To Think Like a Statistician
Is it a good idea to think like a statistician? A confident Yes here, admittedly by someone who earns his living doing so. We live in a random universe, where truth most often lies hidden beneath a fog of noise. Statistics is the science of peering through the fog and learning from experience, especially experience that arrives in sometimes contradictory pieces: Did the COVID vaccines save or cost lives? Is sugar good or bad for you? Are those paintings actually by Jean-Michel Basquiat?
This book is a tell-all of the tricks statisticians use to strip away the noise and see what's underneath (which is sometimes nothing at all). Definitely not a textbook, it's more of a guided cruise around the statistical seas, where dangers lurk for the unwary; no formulas, no theorems, no mind-numbing tables, but lots of examples from the media and my own statistical encounters (not all which were successful).
The detective shows I like to watch come in two varieties, the "arcs" that follow a single case all season, and "anthologies" where there's a new murder mystery each week. To Think Like a Statistician is more anthology than arc. Each chapter takes on a major statistical topic: correlation and causality; accuracy (Do the famous live longer than the rest of us?); prediction and the perverse law of waiting (the longer you wait, the longer you can expect to wait); comparisons (Is the new drug actually better than the old?); survival and risk; health news (Is anything good for you? Bad for you?); Bayes Rule and its dark side; and decisions and regrets (Should you open Monty Hall's third door?).
There is a little bit of "arc" in the book. Statisticians follow two competing philosophies, Bayesianism and frequentism, which lead to quite different methods of data analysis. Those differences show up in the algorithms that actually do the work described in my various chapters. This is one of those rare cases where philosophy directly affects practice, but a little philosophy goes a long way– advice I've tried to follow. Another arc traces the triumphs of a few famous statisticians, in particular the amazing Ronald Fisher, early 20th-century scientific polymath, who more than anyone else defined what it means to think like a statistician.
Ongoing thread. More from Bradley Efron to follow.
Curator: Bora Pajo
To Think Like a Statistician
Efron, B. (2026). To think like a statistician. Princeton University Press.
Available here →August 4, 2026
20260804-efron-to-think-like-a-statistician-in-a-nutshell
In a nutshell
The rest of this thread
Bradley Efron
Bradley Efron is Max Stein Professor emeritus of Statistics and of Biomedical Data Science at Stanford University. He is the inventor of the bootstrap method for assessing statistical accuracy and has published extensively on the theory and practice of data analysis. He has served as president of the American Statistical Association and the Institute of Mathematical Statistics. A MacArthur fellow, he is a member of the National Academy of Sciences and received the National Medal of Science in 2007. His remarkable contributions to his field were recognized again with the 2018 International Prize in Statistics, which noted his influence on "statistical thinking in every way, including new pathways in how we communicate and teach statistics to all."
Pull this thread
pulls againstpulls alongside
In noisy fields like health and politics the truth hides under variability and is recoverable only by averaging.
Against JP · The American Dream is the gateway drug, and meth cooking is a hyperbolic version of coming undone while pursuing it. On: The researcher's own first-person experience is admissible evidence in scholarly argument
Against DL · Spontaneous generation was a well-supported theory that happened to be false, and survived two millennia of hard testing. On: A workable boundary can be drawn between science and non-science
Against TP · Human heredity became a science inside insane asylums, and eugenic ambition was in it from the start. On: Instruments and measurement techniques create the objects they appear merely to record
Against HC · There is no scientific method, and invoking one as the guarantor of expertise now does more harm than good. On: Objectivity is a defensible working standard for inquiry and reporting
Alongside AR · Nihilism is true, because science cannot justify the morality nearly everyone accepts, and it is harmless. On: The researcher's own first-person experience is admissible evidence in scholarly argument
Alongside MM · The attack on the hockey stick was an industry-funded campaign built on fallacious statistics. On: A workable boundary can be drawn between science and non-science
Alongside DH · Atoms decay at unalterable rates and carry no memory, so they are better historians than any human record. On: Instruments and measurement techniques create the objects they appear merely to record
Alongside MK · Models of the mind are idealized like frictionless planes, and idealization does not buy antirealism. On: Objectivity is a defensible working standard for inquiry and reporting
Know something the rest of us don't?
Most threads on Rorotoko begin with our invitation. Any thread can begin with a proposal: you write it with the same prompts every thread answers, and we review it.
Propose a thread →