Amey-Thakur Probability fundamentals
Human intuition about probability is reliably wrong in specific ways: ignoring base rates, confusing P(A given B) with P(B given A), and misjudging rare events. A few fundamentals and an awareness of the traps prevent the errors that lead analysts and decision-makers badly astray.
Method
- Anchor on the base rate. The prior probability of an event dominates more than intuition allows. A test that is "95% accurate" for a disease that affects 1 in 1000 produces mostly false positives, because the base rate is tiny. Always ask "how common is this to begin with?" before updating on new evidence (this is the base-rate neglect that fools everyone; see mental-models).
- Keep conditional probabilities straight. P(A given B) is not P(B given A). P(positive test given disease) is high; P(disease given positive test) can be low, because it depends on the base rate. Bayes' rule is the correction: the posterior combines the evidence with the prior, and flipping the condition without it is a classic, costly error.
- Think in expected value for decisions under uncertainty. The value of an uncertain choice is the sum of each outcome's value times its probability. A small chance of a large loss can outweigh a large chance of a small gain; expected value makes the comparison explicit rather than trusting a gut that overweights the vivid outcome (see decision-matrix, risk-analysis).
- Distrust your sense of rare and extreme events. People overestimate vivid rare events (plane crashes) and underestimate mundane frequent ones, and misjudge compound probabilities (the chance of many things all going right). For a chain of independent steps, multiply the probabilities; the product is usually lower than intuition expects.
- Do not confuse independence and dependence. Independent events do not influence each other (the coin has no memory: the gambler's fallacy is expecting a "due" outcome); dependent events do. Treating correlated risks as independent (they all fail together in a crisis) or independent trials as linked are opposite, common errors with real consequences.
- Separate probability from outcome when judging decisions. A good decision (correct given what was known and the probabilities) can have a bad outcome, and vice versa; luck is not skill. Judge the reasoning and the odds it was based on, not only the result (resulting is the bias of grading decisions by outcomes; see decision-journals).
Boundaries
- These fundamentals guard against the common intuition traps; they are not the full field. Formal probability and statistics go deeper (see statistical-inference, pymc for Bayesian methods).
- Probabilities are only as good as the numbers behind them; a precise expected-value calculation on made-up probabilities is false precision. Be honest about where the numbers come from (see estimation-techniques).
- Genuine uncertainty (unknown probabilities, one-off events) resists exact calculation; there, reasoning about ranges, scenarios, and downside matters more than a single computed number (see risk-analysis, pre-mortem).