
use simulation to decide whether an observed result is consistent with a stated probability model, and use probability to make and evaluate fair decisions.
Instruction
The question. A friend flips a coin 100 times and gets 60 heads. Is the coin unfair? A fair coin does not give exactly 50 heads every time, so the question becomes: if the coin were fair, how often would chance alone produce a result as extreme as 60? If often, the data are consistent with fairness. If rarely, the data are evidence against it.
Simulation. We can answer by imitating the process many times. Using technology, simulate 100 flips of a fair coin and count heads; repeat 1,000 times. One such run of 1,000 simulated sets (with a fixed random seed) had a mean of 50.1 heads and a standard deviation of 4.9, and only 28 of the 1,000 sets had 60 or more heads, about 2.8%. The exact binomial probability is 0.0284, so the simulation is close. A result this extreme would happen by chance in fewer than 3 of 100 sets of fair flips, so the data give fairly strong evidence that the coin favors heads. Had the friend seen 55 heads, which a fair coin produces or exceeds about 18% of the time, we would say the result is consistent with a fair coin: not proof that it is fair, only no evidence that it is not.
The logic (MP.3). This argument has a fixed shape. State the model you are testing. Choose a statistic that measures what you observed. Simulate the model many times and see where your observed statistic falls in the simulated distribution. Then conclude consistent or not consistent, never “proved.” A second example: a game spinner is labeled as landing on blue 25% of the time, and in 200 spins it lands on blue 38 times, when 50 were expected. Under the claimed model, 38 or fewer blues happens with probability about 0.028. The claim is in doubt.
How rare is rare? There is no law of nature here. Many fields treat results that chance produces less than 5% of the time as statistically significant, but the cutoff is a convention and should depend on the stakes. You should report the proportion itself and let the reader see it.
Fair decisions (A2.DS.A.6). Probability also helps design fair processes. To choose one of three presenters, rolling a single die and assigning 1-2, 3-4 and 5-6 gives each a probability of 1/3. Asking “first to roll a sum of 7 with two dice wins” is fair between players who take turns only if you account for who goes first, because the first roller has an advantage. A decision is fair when each party has the probability the situation calls for, which you can check by computing it.
Analyzing decisions (A2.DS.A.6). Expected value weighs outcomes by their probabilities. A school fair game costs $1 to play: roll two dice and win $5 if the sum is 7 or 11. There are 36 equally likely outcomes and 8 give 7 or 11 (six ways to make 7, two to make 11), so P(win) = 8/36 ≈ 0.222. The expected net result for a player is 5(8/36) − 1 ≈ +$0.11 per play: the fair would lose money on average. Probability exposes the flaw before the fair opens. Expected value is not the only consideration in real decisions, which is why the lab asks you to weigh risk as well.
📖 Read for this lesson
The reading opens over the lesson; close it to come back to exactly where you were. The checks below draw on it.
Formative check
Work through these before moving on. They are not graded — they tell you whether the standard below has landed yet.
A fair coin gives 60 or more heads in 100 flips about 2.8% of the time. What is the best conclusion for a coin that gave 60 heads?
Simulation & tools
Simulating the spinner claim
Open JupyterLite and start a Python notebook. Type and run: import random, then results = [sum(random.random() < 0.25 for _ in range(200)) for _ in range(1000)], then print(sum(r <= 38 for r in results)). Record the count, then run the cell twice more. Question: in what proportion of simulated sets of 200 spins did the claimed 25% spinner land on blue 38 or fewer times, and is the observed 38 consistent with the claim? Compare with the exact value, about 0.028. No device: in groups, use a four-section paper spinner or a random-number table to run 20 simulated sets of 200 by sampling, pooling class results.
Open JupyterLite ↗ · JupyterLite, Project Jupyter · BSD-3-Clause
In the dice game, P(sum is 7 or 11) = /36, and the player’s expected net result per $1 play is about $.
Design a fair way to choose one of five club members using only a coin, and prove it is fair with probabilities. Then redesign the $1 dice game so that the fair expects to earn about $0.10 per play, and justify your design with an expected-value calculation.
How confident are you that you can use simulation to judge whether data are consistent with a model and use probability to make fair decisions?
Practice
Work these on paper or in your notebook, then open Check your answer. Aim for all of Fluency and Application; try at least one Challenge.
Printable version: this unit’s practice workbook (PDF)
Fluency
Build speed and accuracy with the core skill.
- X ~ Binomial(n = 19, p = 0.6). Find P(X = 8), P(X ≤ 8), and the mean and standard deviation of X.
Check your answer
Answer: P(X = 8) ≈ 0.0532; P(X ≤ 8) ≈ 0.0885; mean 11.40; SD ≈ 2.135P(X = k) = C(19, k) pk(1 − p)n−k. - X ~ Binomial(n = 17, p = 0.6). Find P(X = 12), P(X ≤ 12), and the mean and standard deviation of X.
Check your answer
Answer: P(X = 12) ≈ 0.1379; P(X ≤ 12) ≈ 0.8740; mean 10.20; SD ≈ 2.020P(X = k) = C(17, k) pk(1 − p)n−k. - X ~ Binomial(n = 6, p = 0.75). Find P(X = 5), P(X ≤ 5), and the mean and standard deviation of X.
Check your answer
Answer: P(X = 5) ≈ 0.3560; P(X ≤ 5) ≈ 0.8220; mean 4.50; SD ≈ 1.061P(X = k) = C(6, k) pk(1 − p)n−k. - A game pays −5 with probability 1/3; 35 with probability 5/12; 50 with probability 1/4. Find the expected value.
Check your answer
Answer: 305/12 ≈ 25.42Multiply each value by its probability and add. - A game pays 5 with probability 2/13; 15 with probability 6/13; 35 with probability 2/13; 45 with probability 3/13. Find the expected value.
Check your answer
Answer: 305/13 ≈ 23.46Multiply each value by its probability and add. - A game pays −10 with probability 2/7; 0 with probability 4/21; 25 with probability 5/21; 45 with probability 2/7. Find the expected value.
Check your answer
Answer: 335/21 ≈ 15.95Multiply each value by its probability and add.
Application
Use the skill in context. Show your reasoning.
- A fair coin is flipped 100 times. Find P(60 or more heads) and say whether 60 heads is evidence against fairness.
Check your answer
Answer: 0.0284; yes, fairly strong evidence (not proof) - A spinner is claimed to land on blue 25% of the time. In 200 spins it lands on blue 38 times. Find P(38 or fewer) under the claim and interpret.
Check your answer
Answer: 0.0276; the result is unusual under the claim, so the claim is in doubt - A $1 game pays $5 when two dice sum to 7 or 11. Find the player’s expected net result per play. Who gains in the long run?
Check your answer
Answer: 5(8/36) − 1 ≈ +$0.11; the player
Challenge
Stretch problems. Expect to think before you write.
- Design a fair way to choose one of five club members using only a coin, and show each member has probability 1/5.
Check your answer
Answer: Flip 3 times (8 equally likely outcomes); assign 5 outcomes to the members and reflip on the other 3. Each member has probability (1/8)/(5/8) = 1/5.
Review
Keep earlier skills sharp.
- Solve for x: 3x = 3.5. Give an exact answer using logarithms and a decimal to the nearest thousandth.
Check your answer
Answer: x = log 3.5 / log 3 ≈ 1.140Take the logarithm of both sides and use the power property. - Find the exact value of tan(330°).
Check your answer
Answer: −√3/3Use the reference angle and the sign of the function in that quadrant. - Two players take turns rolling two dice; the first to roll a sum of 7 wins. Find the probability the first player wins.
Check your answer
Answer: 6/11(1/6)/(1 − (5/6)2).
Two Simulation Records, the Exact Answers, and a Lottery Worth Arguing About
Simulation records generated for this course with Python’s random module, seed 2026, 1,000 repetitions each; Missouri Lottery Show Me Cash game structure (5 of 39, $1 per play) and prize odds as listed by LotteryUSA and Lottery.net (the Missouri Lottery’s own site blocks automated retrieval); odds recomputed from the rules; retrieved November 5, 2026 · SLS original, with public Missouri Lottery game rules
Why this reading. Your lab has to argue about whether data are consistent with a model, and the lesson quotes results from “one such run” of a simulation you have not seen. Two full simulated distributions are printed below, with the seed and the code, so you can reproduce them exactly — and with the exact probabilities beside them, so you can see how far a simulation of 1,000 sits from the truth. The claim to argue about is at the end, and it is a real one.
Record 1. One thousand sets of 100 fair coin flips
Generated with Python: rng = random.Random(2026), then [sum(rng.random() < 0.5 for _ in range(100)) for _ in range(1000)]. Anyone running those two lines gets this table.
| Heads in 100 flips | 32–35 | 36–39 | 40–43 | 44–47 | 48–51 | 52–55 | 56–59 | 60–63 | 64–67 | 68–71 |
|---|---|---|---|---|---|---|---|---|---|---|
| Sets | 2 | 15 | 74 | 214 | 297 | 249 | 114 | 28 | 6 | 1 |
Summary of the run: mean 50.2 heads, standard deviation 5.16. Sets with 60 or more heads: 35 of 1,000, or 3.5%. Sets with 55 or more: 207 of 1,000, or 20.7%.
Now compare with the exact answers. For a fair coin, the number of heads has a binomial distribution with mean 50 and standard deviation √(100 × 0.5 × 0.5) = 5.00. The exact probability of 60 or more heads is 0.0284; of 55 or more, 0.1841.
| This run (seed 2026) | The lesson’s run | Exact | |
|---|---|---|---|
| Mean heads | 50.2 | 50.1 | 50.00 |
| Standard deviation | 5.16 | 4.9 | 5.00 |
| P(60 or more) | 3.5% | 2.8% | 2.84% |
| P(55 or more) | 20.7% | — | 18.41% |
This is the most important row in the reading. Two honest simulations of the same model, each with 1,000 repetitions, report 3.5% and 2.8% for the same probability. Neither is wrong; both are estimates, and an estimate from 1,000 repetitions of something near 3% has a standard error of about √(0.03 × 0.97/1000) = 0.0054, so about half a percentage point. Both runs are within one standard error of the exact 2.84%. Report the proportion you got, say how many repetitions produced it, and compute the exact value when you can. A simulation is evidence about a model, and like all evidence it has a precision, which you are obliged to state.
Record 2. One thousand sets of 200 spins at a claimed 25%
Same seed, [sum(rng.random() < 0.25 for _ in range(200)) for _ in range(1000)].
| Blues in 200 spins | 25–29 | 30–34 | 35–39 | 40–44 | 45–49 | 50–54 | 55–59 | 60–64 | 65–69 |
|---|---|---|---|---|---|---|---|---|---|
| Sets | 1 | 6 | 31 | 163 | 260 | 274 | 180 | 70 | 15 |
Mean 50.3, standard deviation 6.48, against a theoretical mean of 200 × 0.25 = 50 and standard deviation √(200 × 0.25 × 0.75) = 6.12. Sets landing on blue 38 or fewer times: 28 of 1,000, or 2.8%. The exact binomial probability is 0.0276. So the spinner’s observed 38 blues is about as extreme as the coin’s 60 heads, and the same conclusion follows: a result this unusual is evidence against the claimed model, not proof that the spinner is rigged.
The claim to argue about: a lottery is a fair decision procedure
The Missouri Lottery’s Show Me Cash game asks the player to choose five numbers from 1 to 39, for $1 a play, with a drawing every day. The jackpot requires all five. The number of possible tickets is
C(39, 5) = 39!/(5! 34!) = 575,757,
so the probability of the jackpot on one $1 play is 1 in 575,757, which is the figure published for the game. The lower prizes come from the same counting. Matching exactly four of the five drawn numbers means choosing 4 of the 5 winners and 1 of the 34 losers: C(5, 4) × C(34, 1) = 5 × 34 = 170 tickets, so about 1 in 3,387. Exactly three: C(5, 3) × C(34, 2) = 10 × 561 = 5,610 tickets, about 1 in 103. Exactly two: C(5, 2) × C(34, 3) = 10 × 5,984 = 59,840 tickets, about 1 in 10. You can verify all four from the game’s structure, with no data needed beyond the rules.
| Match | Prize | Winning tickets of 575,757 | Odds, 1 in |
|---|---|---|---|
| 5 | Jackpot (starts at $50,000) | 1 | 575,757 |
| 4 | $250 | 170 | 3,386.8 |
| 3 | $10 | 5,610 | 102.6 |
| 2 | $1 | 59,840 | 9.6 |
| Any prize | 65,621 | 8.77 |
Here is the claim. “A lottery is the fairest decision procedure we have. Every ticket has exactly the same probability, nobody can influence the draw, and the rules are published. If fairness means equal probability, a lottery is the definition of fair.”
Your lab should take that seriously before disagreeing with it, because the first three sentences are all true, and lotteries are used for exactly that reason in places where fairness is the point: randomized school admissions, jury pools, the allocation of scarce visas, and the random assignment in Lesson 4.1’s polio trial. Equal probability is a real and powerful kind of fairness.
Then push on it with the tools of this lesson.
- Equal probability is not equal expected value when the stakes differ. Two players buying one ticket each face identical odds. A Show Me Cash play costs $1; if one player spends $20 a week of a $400 budget and another spends $20 a week of a $4,000 one, the probabilities are the same and the decisions are not. Expected value alone cannot see that; A2.DS.A.6 asks you to weigh the payoff and the risk.
- A fair draw does not make a fair game. Compute the expected value of a $1 ticket as the lesson computes it for the dice game. The fixed prizes are worth (250 × 170 + 10 × 5,610 + 1 × 59,840)/575,757 = 158,440/575,757 ≈ $0.28 per play. A $50,000 jackpot, the starting amount, adds 50,000/575,757 ≈ $0.09, for a total of about $0.36 back on every $1 — an expected net result of about −$0.64, before taxes and before the chance of sharing a jackpot. Break-even would need a jackpot of about (1 − 0.2752) × 575,757 ≈ $417,000. The draw is unbiased; the transaction is not a coin flip, and the difference is what funds the public programs lottery revenue pays for.
- Where a lottery is the right answer. When a good is indivisible, demand exceeds supply, and no ranking criterion is agreed, equal probability is genuinely hard to beat — and you should say which of the three conditions is doing the work.
Write the argument with the numbers in it. “One in 575,757” and “one in three, by rolling a die and splitting 1–2, 3–4, 5–6” are both equal-probability procedures, and a good essay explains why nobody objects to the second.
Sources: LotteryUSA, Missouri Show Me Cash (how to play, prizes and odds) (lotteryusa.com); Lottery.net, Missouri Show Me Cash (lottery.net); Missouri Lottery, Show Me Cash (molottery.com); Python documentation, the random module (docs.python.org)