Chance and probability

Core idea

Probability is the numerical language of uncertainty. Every probability sits between 0 (impossible) and 1 (certain), and the rules for combining them are blissfully simple once you know which one to apply. For events that cannot both happen at once (mutually exclusive), P(A or B) = P(A) + P(B). For events whose outcomes do not influence each other (independent), P(A and B) = P(A) × P(B). Almost every probability puzzle reduces to identifying which rule applies and being honest about whether the trial outcomes are equally likely. Most everyday probability errors come from violating one of these two recognitions.

Why it matters

Probability underwrites the decision-making half of statistics. The framework for testing hypotheses, judging risk, pricing insurance, designing experiments, and reading medical-screening results all rest on the same handful of rules introduced here. The cost of poor probability intuition is high — the gambler's fallacy bankrupts lottery players, base-rate neglect causes misdiagnosis, and the misuse of expected value distorts insurance pricing. A solid grip on the basics is one of the highest-leverage pieces of mathematical literacy.

Mental model

Trials, outcomes, and events

Three terms that look interchangeable but are not. A trial is the experiment (rolling a die). An outcome is one possible result of one trial (rolling a 4). An event is an outcome or a combination of outcomes that share some property (rolling an even number combines three outcomes). Clarity about which is which is half the battle.

Copy sourceZoom

Choosing between add and multiply

Two rules cover almost every introductory probability question. The trick is to read the question for the connective word — "or" versus "and" — and then check whether the underlying events satisfy mutual exclusivity (for adding) or independence (for multiplying).

Copy sourceZoom

The tree diagram for sequences

For events that happen in sequence — first a coin toss, then a die roll — a tree diagram makes every path explicit. Probabilities along a path multiply; probabilities across paths add.

Copy sourceZoom

Practical application

Example

A small online retailer ships orders from two warehouses, A and B. Warehouse A handles 70 percent of orders; B handles 30 percent. Warehouse A makes a packing mistake on 2 percent of its orders; B makes a mistake on 5 percent. A customer's order arrives damaged — what is the probability it came from warehouse B?

Trial: a randomly received damaged order. Two paths can produce a damaged order: A-then-damaged, or B-then-damaged. The probabilities along the paths multiply:

• P(A and damaged) = 0.70 × 0.02 = 0.014
• P(B and damaged) = 0.30 × 0.05 = 0.015
• P(damaged) = 0.014 + 0.015 = 0.029

Given the order is damaged, the share from B is 0.015 / 0.029 ≈ 0.52. So even though A ships more than twice as many orders, slightly more than half of all damaged orders trace to B. The intuition fails because the larger error rate at B more than compensates for B's smaller order volume.

The general lesson: base rates matter, and so do conditional rates — the answer is almost never the rate that comes to mind first. A tree diagram with probabilities on every branch makes the calculation routine.

Related lessons