← Back to the Introduction to Statistics outline The Course Maker
← Introduction to Statistics outline
Week 7 · Lecture outline

Week 7 — Lecture Outline · The Binomial Distribution

Introduction to Statistics Generic evergreen edition

Course: Introduction to Statistics (18-week generic edition)
Objectives covered: Objective 4 — Apply probability rules and random variables — this week, the binomial model: recognize the setting, compute exact probabilities, and give the mean and SD of a count of successes.
SLOs touched: A (reason quantitatively from data) · B (communicate results to a non-technical audience)
Meeting pattern: planned as 2 sessions × ~75 min = ~150 min. Segment minutes below total ~150; scale them to your own pattern.

Week at a Glance
The week's big question "When you try something n times and count the wins, what exactly are the chances — and what should you expect, give or take?"
By the end of the week, students can… (1) recognize a binomial setting with the B·I·N·S checklist (Binary outcome, Independent trials, Number of trials fixed, Same p) and name n and p; (2) compute P(X = k) with the binomial formula, showing the "ways" factor C(n, k); (3) use the complement for "at least one" — 1 − P(none); (4) state and interpret the mean np and SD √(np(1−p)); (5) compute with =BINOM.DIST (exact vs. cumulative) and describe the shape — including why big-n binomials start looking like a bell.
Key vocabulary binomial setting, trial, success/failure, n, p, binomial random variable, binomial coefficient C(n, k) ("ways"), probability distribution of a count, complement rule for "at least one," mean (expected value) np, standard deviation √(np(1−p)), cumulative probability, symmetric vs. skewed shape
Materials slides (Deck 7), the Week 7 chapter, the week's readings + video links, a spreadsheet (Google Sheets or Excel) for =BINOM.DIST, a Desmos-class stats tool for pictures, the student's chatbot for the AI-critique moment and the tutorial
Timing note 8 segments, ~150 min total. Session 1 = Segments 1–4 (~75). Session 2 = Segments 5–8 (~75).

Segment 1 — Hook & the Promise (8 min) · Session 1 opens

Hook. "Three questions, one skeleton. If you email 5 customers and each clicks with a 1-in-5 chance, how likely is it that exactly two click? If a 50-50 free-throw shooter takes 4 shots, how often does she make all 4? If you guess on every true/false question, how many do you get right?" Give them ten seconds. Then the reveal: these are all the same problem.

  • Fixed number of tries. Each try is a yes/no. Count the yeses. That pattern — repeated everywhere from marketing to seeds to test-taking — has one distribution that owns it: the binomial.
  • Last week we built distributions for random variables one row at a time, by hand. This week we meet the first famous random variable — one so common it comes with its own formula, its own shortcut for the mean, and its own spreadsheet function.

The promise (write it on the board): "By the end of this week you can spot a binomial setting in the wild, compute the exact chance of any number of successes, and say what count to expect — give or take how much."

Why it matters line (memory hook): "Same skeleton, a hundred costumes: fixed tries, yes/no, count the wins."

Segment 2 — The Binomial Setting: B·I·N·S (20 min)

Plain language first. Before any formula: the binomial distribution only applies when the situation passes a four-part checklist.

  • B — Binary. Each trial has exactly two outcomes — call one "success," the other "failure." Warning: "success" is just the thing we're counting; it can be a defect, a no-show, or a miss. Success ≠ good.
  • I — Independent. Trials don't affect each other. One click tells you nothing about the next.
  • N — Number fixed. The number of trials n is set before you start. Not "until the first win" — that's a different animal.
  • S — Same p. Every trial has the same probability of success, p.

Pass all four, and X = the number of successes in the n trials is a binomial random variable.

Memory hook (put it on a slide):

B·I·N·S — Binary, Independent, Number fixed, Same p. "If it fits in the BINS, it's binomial."

One fully worked example (the week's running example — set it up carefully).

An online shop sends a promotional email to 5 customers. Past campaigns show each recipient clicks the link with probability 0.2, independently of the others. Let X = the number who click.
- Binary? Click / no click. ✓
- Independent? One customer's click doesn't change another's. ✓ (stated)
- Number fixed? n = 5, decided before sending. ✓
- Same p? p = 0.2 for every recipient. ✓
So X is binomial with n = 5, p = 0.2. Possible values: 0, 1, 2, 3, 4, 5.

Counterexamples (name them as traps, one line each):
- "Count the emails you must send until the first click" — no fixed n (the count of trials is the random thing; that setting has its own name, but it's out of our scope — just recognize it's not binomial).
- "Draw 4 names from a box of 15, without replacement, and count part-timers" — not independent; each draw changes the mix, so p shifts trial to trial.
- "Record how long each seed takes to sprout" — not counting successes at all; that's a measurement, not a yes/no count.

Segment 3 — The Binomial Formula (25 min)

Build it, don't announce it. Ask for P(X = 2) in the email example and construct the answer in three moves.

Move 1 — one specific sequence. What's the chance the first two customers click and the last three don't? Independence lets us multiply:

0.2 × 0.2 × 0.8 × 0.8 × 0.8 = 0.2² × 0.8³ = 0.04 × 0.512 = 0.02048.

Move 2 — count the ways. But "exactly two click" doesn't care which two. Click-click-no-no-no is one way; click-no-click-no-no is another. The number of ways to choose which 2 of the 5 trials are the successes is the binomial coefficient:

C(5, 2) = 10 ways. (Say it as "5 choose 2." Every way has the same probability, 0.02048.)

Move 3 — assemble. Ten equally likely ways at 0.02048 each:

P(X = 2) = C(5, 2) × 0.2² × 0.8³ = 10 × 0.04 × 0.512 = 0.2048 — about a 20% chance.

The general formula (only now, after it's been built):

P(X = k) = C(n, k) × p^k × (1 − p)^(n − k) — (ways to place the successes) × (probability of the successes) × (probability of the failures).

The complement trick — "at least one" (work it immediately, same example):

P(at least one click) = 1 − P(no clicks) = 1 − 0.8⁵ = 1 − 0.32768 = 0.67232 — about a 67% chance.
Memory hook: "At least one = 1 − P(none)." One subtraction replaces five formula runs (k = 1, 2, 3, 4, 5).

The full distribution (show the table — students meet the mini-table they'll use all week):

The table below lists every value of X for the email example (n = 5, p = 0.2), with probabilities rounded to 4 decimal places.

k (clicks) 0 1 2 3 4 5
P(X = k) 0.3277 0.4096 0.2048 0.0512 0.0064 0.0003

Point at it: probabilities sum to 1 (a Week 6 rule); the most likely count is 1, not 0; and 5-for-5 is a 3-in-10,000 long shot.

Segment 4 — Misconceptions + Quick Interaction (22 min) · Session 1 closes (~75)

Name the misconceptions out loud, then cure each:

  • ❌ "P(exactly 2 of 5) = 0.2² × 0.8³ — done."
    ✅ Cure: that's the chance of one particular arrangement. Two clicks can arrive C(5,2) = 10 different ways, and you must count them all. The "ways" factor is the piece everyone forgets. (0.02048 vs. the correct 0.2048 — off by a factor of ten.)
  • ❌ "'At least 2' and 'exactly 2' are basically the same question."
    ✅ Cure: read the words like a lawyer. Exactly 2 is one bar of the histogram; at least 2 is a whole tail (k = 2, 3, 4, 5). For "at least one," the fast route is the complement: 1 − P(0).
  • ❌ "After three misses in a row, a make is due."
    ✅ Cure: the I in BINS. Independent trials don't remember. If p = 0.5, it's 0.5 on shot four whether the shooter is 3-for-3 or 0-for-3. "The trials don't remember." (This is Week 5's gambler's-fallacy cure wearing a new jersey — and it's this week's discussion.)
  • ❌ "Success means something good."
    ✅ Cure: "success" is only the outcome being counted. If you're counting no-shows, a no-show is a "success." Define it out loud in every problem, and check whether the stated percentage belongs to your success (p) or its opposite (1 − p).
  • ❌ "Binomial needs p = 0.5."
    ✅ Cure: p can be anything between 0 and 1 — 0.2 clicks, 0.9 germination. What must hold is that p is the same on every trial.

Interaction — Think-Pair-Share (binomial or not?, ~10 min):
Put 6 scenarios on a slide; students decide solo (30 sec), compare with a neighbor (1 min), class votes thumbs up (binomial) / down (not). Items: (1) count how many of 20 seedlings survive a frost, same survival chance, independent · (2) choose 5 of 30 lockers at random without replacement, count the rented ones · (3) count emails sent until the first bounce · (4) guess every one of 10 true/false questions, count correct · (5) shoot free throws for one minute, count makes · (6) each of 6 friends independently shows up to trivia night with probability 0.8, count who show.
(Answers: binomial · not — without replacement, dependent · not — no fixed n · binomial · not — n isn't fixed, time is · binomial.)
Debrief the sneaky one: #5 — "for one minute" fixes time, not the number of trials.

Segment 5 — Mean & SD of a Binomial: np, Give or Take (20 min) · Session 2 opens

Hook back in: "Last session: the exact chance of any count. Today, the two-number summary — what to expect, give or take how much — without touching the formula."

Plain language first. If each of 100 tries succeeds 25% of the time, how many successes do you expect? A quarter of 100 — 25. That intuition is the theorem:
- Mean (expected value): μ = np.
- Standard deviation: σ = √(np(1 − p)). The give-or-take around np. (No derivation — Week 6 already taught what an SD of a random variable means; this is the binomial shortcut.)

One fully worked example (do every step out loud).

A packet of 100 prairie wildflower seeds; each germinates with probability 0.25, independently (a tough native species).
- Mean: μ = np = 100 × 0.25 = 25 seeds.
- Variance: np(1 − p) = 100 × 0.25 × 0.75 = 18.75.
- SD: σ = √18.75 ≈ 4.33 seeds.
- Say it in words: "Expect about 25 sprouts, give or take about 4." A tray with 24 or 29 sprouts is unremarkable; a tray with 5 means something's wrong with a lot more than luck.

Second quick example (the one they feel personally):

Guess on all 16 questions of a true/false quiz: n = 16, p = 0.5.
μ = 16 × 0.5 = 8; σ = √(16 × 0.5 × 0.5) = √4 = 2.
"Pure guessing: 8 right, give or take 2." Luck is not a study strategy.

Callback to Week 6 (close the loop — prove the shortcut agrees with the long way):

Week 6's method: E(X) = Σ k · P(k). Take a small binomial — n = 4, p = 0.5 (probabilities 0.0625, 0.25, 0.375, 0.25, 0.0625 for k = 0…4):
E(X) = 0(0.0625) + 1(0.25) + 2(0.375) + 3(0.25) + 4(0.0625) = 0 + 0.25 + 0.75 + 0.75 + 0.25 = 2 — and np = 4 × 0.5 = 2. ✓ Same answer, one multiplication instead of five. That's why formulas earn their place: they're the long way, pre-done.

Misconception + cure:
- ❌ "μ = 25 means 25 seeds will germinate."
✅ Cure: np is a long-run average, not a promise — "expected, not guaranteed." The SD is exactly the tool that says how far off a normal run can wander.

Segment 6 — Shape: From Lopsided to Bell (15 min)

Plain language: every binomial has a histogram — one bar per possible k. Three facts about its shape, shown not derived (sketch or use a Desmos-class tool live):

  • p = 0.5 → symmetric. Makes and misses trade places evenly (the n = 4, p = 0.5 histogram is a perfect little pyramid: 0.0625, 0.25, 0.375, 0.25, 0.0625).
  • p far from 0.5, small n → skewed. The email histogram (n = 5, p = 0.2) piles up near 0–1 and trails off right.
  • n grows → the bell emerges. Even at p = 0.2, by n = 100 the histogram is a smooth, symmetric-looking mound centered at np = 20. This is the week's cliffhanger: a completely different-looking machine — counts from yes/no trials — keeps producing the same bell silhouette.

The tease, said plainly: next week that silhouette gets a name — the normal distribution — and one curve will stand in for a thousand histograms. This week's job is only to see the bell coming.

Quick check (30 seconds, fingers): "n = 20. Which p gives the most symmetric histogram: 0.1, 0.5, or 0.9?" (0.5 — and 0.1 vs. 0.9 are mirror images of each other.)

Segment 7 — Technology Workflow + AI-Critique (25 min)

Technology workflow — =BINOM.DIST in a spreadsheet (exact steps, live):
1. Recreate the worked example: in any cell type =BINOM.DIST(2, 5, 0.2, FALSE) → 0.2048. Argument order: k, n, p, cumulative? — and FALSE = exactly k.
2. Change the last argument: =BINOM.DIST(1, 5, 0.2, TRUE) → 0.73728. TRUE = cumulative, P(X ≤ 1) — "at most 1," the first two bars added (0.32768 + 0.4096).
3. "At least one" by machine: =1 - BINOM.DIST(0, 5, 0.2, TRUE) → 0.67232 — matching Segment 3's complement by hand. (Same function in Google Sheets and Excel; Excel also keeps the older =BINOMDIST alive.)
4. Build the whole distribution in 30 seconds: k values 0–5 in column A, =BINOM.DIST(A2, 5, 0.2, FALSE) filled down in column B, then insert a column chart — the Segment 3 table, drawn.
5. In a Desmos-class stats tool, plot the same distribution and slide n up — watch the bell of Segment 6 emerge live.

The rule of thumb to say out loud: formula by hand for small n (it shows you the machinery); technology for anything bigger. Nobody hand-computes C(100, 37).

AI-critique moment (students verify, not consume):

Paste this to your chatbot: "A shop emails 5 customers; each clicks with probability 0.2, independently. What's the probability exactly 2 click? Show your work."
Then check its work against your own 0.2048: chatbots frequently (a) drop the ways factor and report 0.02048, (b) quietly compute at most 2 (0.94208) instead of exactly 2, or (c) flub the powers of 0.8. You hold the vetted answer; make the machine match it. The tool drafts, you judge — and this week you can check it with three different weapons: the formula, the mini-table, and =BINOM.DIST.

Segment 8 — Callback, Tease & Hand-off (15 min) · Session 2 closes (~75)

Callback: "Week 5 gave us probability rules; Week 6 taught us to organize outcomes into a random variable with a mean and an SD. This week the two merged: the binomial is a random variable whose entire distribution comes from just two numbers, n and p. That's the whole trajectory of this course — from rules, to tables, to named machines."

Tease next week: "Keep the picture from Segment 6 in your head. Next week we stop drawing thousand-bar histograms and adopt the smooth curve they're all reaching for — the normal distribution, the most famous shape in statistics. Binomial counts are one of the many things it will model."

Hand-off (the week's work):
- Chapter 7 (the primary reading) — the binomial setting, formula, mean/SD, technology, shape — then Lecture Tutorial 7 (AI tutor, share-link submission), which drills all of it with the embedded mini-table.
- Data Lab 7 ("Two Hundred Rounds of Four") — simulate a 50-50 free-throw shooter and watch theory meet reality · Quiz 7 (end of week, closed to AI) · Discussion 7 ("The Hot-Hand Question") · Assignment 7 ("Expected, Not Guaranteed," AI-coached).

Instructor FAQ — Common Stumbles

Student says / does Quick cure
"Why multiply by C(n, k) at all?" p²(1−p)³ is the chance of one particular arrangement of wins and losses. "Exactly 2" is satisfied by C(5,2) = 10 different arrangements, each equally likely — so add it 10 times, i.e., multiply.
Uses p where 1 − p belongs (or vice versa). Make them define "success" in words first ("success = a click, p = 0.2"), then label both numbers before computing. Most sign-flips die at the labeling step.
"Is 'at least 2' the same as 'exactly 2'?" Exactly 2 = one bar. At least 2 = a tail of bars (2, 3, 4, 5). And "at least one" has the express lane: 1 − P(0).
"The mean is 25, so 25 will germinate." np is the long-run average — expected, not guaranteed. The SD (≈ 4.33) is the built-in "give or take."
"She's missed three straight — she's due." Independence: the trials don't remember. p stays p. (Send them to this week's discussion — that intuition is exactly what it interrogates.)
Counts a without-replacement draw as binomial. Each draw changes the remaining mix, so p shifts and trials aren't independent — the S and the I both fail. (Chance-picked with replacement, or independent trials, restore it.)
"Which do I use — formula, table, or =BINOM.DIST?" Small n by hand once (to own the machinery), the embedded table for the course's standard pairs, technology for everything bigger. All three must agree — that's the self-check.
Reads =BINOM.DIST(…, TRUE) as "exactly k." TRUE = cumulative, P(X ≤ k) — the running total of bars. FALSE = exactly k. Say "FALSE = one bar, TRUE = the bars so far."

Scope flag

This outline stays within Objective 4. Three touches are added context beyond the objective's strict floor: the name-drop that "trials until the first success" is a different (out-of-scope) distribution, the 16-equally-likely-sequences anatomy behind C(4, 2) = 6, and the Segment 6 shape preview of the normal model (Week 8 owns the actual bell). The np ≥ 10 / n(1−p) ≥ 10 "when is the bell trustworthy" condition is deliberately deferred to Week 8. Keep the previews — they make the bridge land — or cut them for a leaner session.