Probability Detective Close this tab to return to the lesson.

For adults

Guide for parents and teachers

A short lesson (about 10 to 15 minutes) with replayable experiments for learners around age 12. Individual learners differ: some will want more time, co-play or the extension ideas below.

Learning objectives

  • Distinguish theoretical probability (from the rules) from observed frequency (from trials).
  • Describe why small samples can vary a lot.
  • Compare experiments with 10, 100 and 1,000 trials, knowing that larger samples often, but not always, land closer to the theory.
  • Explain fairness in a simple points game using expected points per round.

The lesson never says that randomness "owes" an outcome or that a long run guarantees an exact proportion.

Prerequisites

Simple fractions (1/2, 1/6, 6/36), percentages or decimals between 0 and 1, and reading a bar chart. Multiplying small whole numbers helps in Case 3.

Walkthrough

  1. Briefing (1 min): the "predict, test, explain" routine and the visual key (solid = observed, dashed = theory).
  2. Case 1, Coin Lab (4 to 5 min): demo of two 10-flip runs; predict; run 10, 100 and 1,000 flips of a fair or clearly labelled biased coin (0 to 1); spread test; streak check; two clue questions; fresh case: judge a mystery coin from evidence.
  3. Case 2, Dice (3 to 4 min): 36-outcome grid; predict the most common sum; chart plus equivalent table; two clue questions; fresh counting questions tested against 1,000 rolls.
  4. Case 3, Fair Game (3 to 4 min): two-coin warm-up (the "3 outcomes" trap); design a fair Owl-versus-Fox points game with exact arithmetic; final challenge on a new game.
  5. Report: tasks done, evidence from the learner's own runs, and a next-practice suggestion.

Fairness rule used

Each round, a side's expected points = (ways it scores out of 36 ÷ 36) × points. A game is called fair when both sides' expected points per round are equal. Equivalent whole-number test: Owl ways × Owl points = Fox ways × Fox points. Example: Owl scores 1 point on an even product (27 of 36) and Fox scores 3 points on an odd product (9 of 36): 27 × 1 = 9 × 3, so it is fair.

Simplification: equal expected points does not make every short match equally likely to be won. In that example, over 10 rounds Owl finishes ahead about 52.6% of the time. Discuss this with confident learners.

Adaptations and co-play

  • Support: hints, inline definitions, worked answers, "Larger text" and "Reduce motion" in Settings. Predictions can be skipped. Read the briefing aloud together.
  • Co-play: each person predicts before a run, then discuss whose prediction matched the theory (not the lucky result).
  • Extension: judge the mystery coin using only 100 flips at a time; find a fair design where both teams score 1 point (18 ways each, for example sums 3, 6, 7, 8 for Owl).
  • Keyboard, touch and screen-reader friendly: every chart has a data table and a text summary.

Privacy and safety

No accounts, names, ads, tracking, chat, uploads or network requests. Only task ticks and display settings are saved, in this browser's local storage, and can be cleared with "Reset all progress". Points are not money and have no prizes.

Model limitations

  • Randomness is generated by a seeded pseudo-random generator (Mulberry32) seeded from the browser's crypto source. It is suitable for simulation, not for security.
  • Coins and dice are idealised: no edge landings, wear or throwing technique.
  • The mystery-coin "usual range" is an exact binomial range holding at least 95% of fair-coin results. About 1 time in 20 a fair coin lands outside it, and the feedback says so.

Offline follow-up

With a partner, flip a real coin 20 times each and record the proportion of heads. Compare, then pool everyone's flips and see how the combined proportion compares to 0.5.

References

No claims are made about measured learning gains, diagnostic value or suitability for every child.