--- name: ? status: compiling version: 0.0.0 maintainer: Neo dependencies: [patience] ---
drafting spec…
the universe did not have a file for this yet. writing one now. (first visit only: future readers will see this page instantly.)
--- name: ? status: compiling version: 0.0.0 maintainer: Neo dependencies: [patience] ---
the universe did not have a file for this yet. writing one now. (first visit only: future readers will see this page instantly.)
--- name: Statistics type: discipline status: running version: 4.1.2 released: 1749-01-01 maintainer: no single maintainer (disputed) dependencies: - mathematics - data - assumptions - human-willingness-to-be-uncertain license: Public Domain (results may vary) tags: - inference - probability - science - lying-carefully - uncertainty-management ---
The formal practice of making confident claims from incomplete information, while technically admitting you might be wrong.
You collect some observations. You cannot collect all observations. This gap between "some" and "all" is where statistics lives. The discipline provides a set of tools to navigate that gap without pretending it does not exist, which distinguishes it from opinion.
The core loop:
p < 0.05 is not a fact. It is a threshold. Someone picked 0.05. It was arbitrary.WARN_001: "statistically significant" != "important"
WARN_002: n=12 is not a large sample
ERR_403: cherry-picked confidence interval
ERR_500: model assumptions violated silently
FATAL: confusing "no evidence of effect" with "evidence of no effect"
Q: Can statistics prove anything? No. Statistics quantifies uncertainty. Proof is for mathematics and courtroom dramas.
Q: Is my poll accurate? Depends entirely on how you sampled. The margin of error is the least of your problems.
Q: What is a p-value, really? The probability of seeing results this extreme if the null hypothesis were true. Not the probability the null hypothesis is false. These are different. This difference has caused measurable harm.
Q: Should I use mean or median? Median. Almost always median. The mean is being held hostage by your outliers.
| Version | Note |
|---|---|
| 1.0 | Gauss, Laplace. Solid foundations. Few computers. |
| 2.0 | Fisher, Neyman, Pearson. Hypothesis testing wars begin. |
| 3.0 | Computers arrive. Everyone runs tests they do not understand. |
| 4.0 | Machine learning absorbs statistics, renames the parts, forgets the uncertainty. |
| 4.1.2 | Replication crisis acknowledged. Patch pending. |