← Back to the Introduction to Statistics outline The Course Maker
← Introduction to Statistics outline
Week 6 · Lecture outline

Week 6 — Lecture Outline · Random Variables

Introduction to Statistics Generic evergreen edition

Course: Introduction to Statistics (18-week generic edition)
Objectives covered: Objective 4 — Use probability rules, conditional probability, and random variables to quantify uncertainty (this week: the random-variables half).
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 an outcome is uncertain but has numbers attached — a prize, a payout, a count — what is that uncertain number actually worth, and how wildly does it swing?"
By the end of the week, students can… (1) recognize a random variable and tell a discrete one from a continuous one; (2) check whether a table of values-and-probabilities is a legitimate probability distribution and read probabilities from it; (3) compute and interpret the expected value E(X) as a long-run average; (4) compute the variance and standard deviation of a discrete random variable; (5) predict what happens to the mean and SD under a linear transformation (a + bX); (6) explain why, for a continuous random variable, probability is area under a density curve and any single exact value has probability 0.
Key vocabulary random variable, discrete vs. continuous random variable, probability distribution, legitimate distribution (each probability in [0,1]; total exactly 1), expected value E(X) = μ_X, variance σ²_X, standard deviation σ_X, linear transformation (a + bX), density curve, uniform density
Materials slides (Deck 6), the Week 6 chapter, the week's readings + video links, a spreadsheet (Google Sheets or Excel), 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. "An arcade claw machine costs $1 a play. Most plays win nothing. Sometimes it drops a $1 keychain, occasionally a $5 plush, and once in a blue moon a $10 headset. Would you play? How would you even decide?" Take a quick show of hands — play / don't play / need more information. The "need more information" people are already doing statistics.

  • Last week we could answer "how likely is a win?" This week we answer the sharper question: "what is a play worth?" — because the outcomes aren't just events anymore, they're events with dollar amounts attached.
  • "By the end of this week, you'll compute exactly what one play of that machine is worth — and you'll be able to do the same for a warranty, a raffle ticket, and any other deal where chance carries a number."

The promise (write it on the board): "By the end of this week you can take any uncertain number — a payout, a count, a score — write down its possible values with their chances, and answer the two questions that matter: what's typical (the expected value) and how far off typical should I expect to be (the standard deviation)?"

Why it matters line (memory hook): "A random variable is just a number that hasn't happened yet — and we can still say what it's worth."

Segment 2 — Random Variables & Discrete Distributions (22 min)

Plain language first.
- A random variable is a numeric outcome of a chance process — a number whose value is decided by chance. Before the claw drops, "the cash value of my prize" is a random variable. After it drops, it's just data.
- We write random variables with capital letters (X, Y) and their possible values with small letters (x). "P(X = 5)" reads "the probability that X turns out to be 5."
- Two kinds, and the split echoes Week 1's variable types:
- Discrete — the possible values are separate, countable points (0, 1, 2, …). Counts of things. You can list them.
- Continuous — the possible values fill an entire interval; between any two values there are infinitely more. Measurements. You can't list them.
- Memory hook: "Discrete you count, continuous you measure" — the same instinct as NOIR in Week 1.
- The probability distribution of a discrete random variable is the complete table of its possible values with the probability of each. Two rules make a table legitimate:
1. every probability is between 0 and 1, and
2. the probabilities add up to exactly 1 (something must happen).

One fully worked example (build the table live).

The claw machine from the hook. X = the cash value of the prize from one $1 play. The arcade's own service data:

x (prize value) $0 $1 $5 $10
P(X = x) 0.70 0.20 0.08 0.02
  • Legitimacy check, out loud: every entry is between 0 and 1 ✓, and 0.70 + 0.20 + 0.08 + 0.02 = 1.00 ✓. This is a legitimate distribution.
  • Read from it: P(X = 0) = 0.70 — most plays win nothing. P(win something) = P(X ≥ 1) = 0.20 + 0.08 + 0.02 = 0.30. The complement rule from Week 5 still works: 1 − 0.70 = 0.30. Same machinery, new packaging.

Land the key idea: the distribution table is the entire personality of a discrete random variable — everything we compute this week comes from those two rows.

Segment 3 — Expected Value: What a Chance Is Worth (23 min)

Plain language first. If you played the claw machine thousands of times, your winnings per play would settle down to one number — that's the expected value. It is a long-run average, computed by weighting each value by its probability:

E(X) = Σ x · P(x) — multiply each value by its probability, add them all up. Also written μ_X (it is the mean of the random variable).

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

The claw machine again. E(X) = 0(0.70) + 1(0.20) + 5(0.08) + 10(0.02)
= 0 + 0.20 + 0.40 + 0.20 = $0.80.
- Say it in words: "Over many, many plays, the machine pays out about 80 cents per play, on average."
- The play costs $1.00. So per play, the average player loses 1.00 − 0.80 = $0.20 — the arcade's edge. Play 100 times: spend $100, expect about $80 back.
- Now the punchline: $0.80 is not a possible prize. No single play ever pays 80 cents. The expected value is what your average approaches, not a value you should "expect" on any play.

Callback (Week 5): this is the long-run relative frequency idea wearing a new coat — in the Week 5 lab, the proportion of sixes wobbled early and settled near 1/6. An average settles the same way. Chance is wild in the short run, dependable in the long run.

Land the key idea (memory hook): "Expected value is what you'd average, not what you'd expect."

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

Name the misconceptions out loud, then cure each:

  • ❌ "The expected value is the most likely outcome."
    ✅ Cure: the claw machine's most likely outcome is $0 (probability 0.70); its expected value is $0.80 — a value it never produces. E(X) is a weighted average, not a prediction for one play.
  • ❌ "E(X) must be one of the possible values of X."
    ✅ Cure: averages routinely land between the values being averaged ("2.3 people per household"). If your computed E(X) happens to be impossible as a single outcome, that is not an error.
  • ❌ "Any table of values and probabilities is a distribution."
    ✅ Cure: run the two legitimacy rules first — each probability in [0,1], total exactly 1. If the total is 0.9 or 1.1, the table is broken before any computation.
  • ❌ "A random variable is the same thing as the data we collected."
    ✅ Cure: the random variable is the number before it happens (with chances attached); data are what you get after. The distribution plays the role the population played in Week 1 — the truth behind the numbers we see.

Interaction — Think-Pair-Share (rapid-fire, ~12 min):
Six quick prompts on a slide; solo answer (30 sec), compare with a neighbor (1 min), fingers vote. Suggested items:
1. X = the number of typos on a page you just typed — discrete or continuous?
2. X = the exact time you wait at a crosswalk — discrete or continuous?
3. A table lists probabilities 0.6, 0.3, 0.2 — legitimate?
4. A table lists probabilities 0.5, 0.25, 0.25 — legitimate?
5. X takes values 1 and 2, each with probability 0.5. E(X) = 1.5 — but 1.5 is impossible. Is the computation wrong?
6. Every value of X gets 10 added to it. Does the spread change?

(Answers: discrete · continuous · not legitimate — sums to 1.1 · legitimate · not wrong — E(X) is a long-run average, impossible values are fine · no — shifting every value moves the center, not the spread.)
Debrief items 5 and 6 — they preview Segments 5–6 and are the week's two favorite traps.

Segment 5 — Variance & SD of a Random Variable (25 min) · Session 2 opens

Hook back in: "Last session: E(X) tells you what's typical in the long run. But two deals can share an average and feel completely different — a steady $2 versus a coin flip between $0 and $4. The missing number is the spread."

Plain language first. In Week 3 the standard deviation measured how far data values typically sit from their mean. Same idea here, but the averaging is weighted by probability:

Var(X) = σ²_X = Σ (x − μ)² · P(x) — each squared distance from the mean, weighted by its probability.
SD(X) = σ_X = √Var(X) — back in the variable's own units.

One fully worked example (grind through every step — this is the week's longest computation).

A parking office tracks X = the number of vehicles a randomly chosen household in one neighborhood keeps registered:

x 0 1 2 3
P(X = x) 0.10 0.30 0.40 0.20

Step 1 — the mean: E(X) = 0(0.10) + 1(0.30) + 2(0.40) + 3(0.20) = 0 + 0.30 + 0.80 + 0.60 = 1.7 vehicles.
Step 2 — squared deviations, each weighted:
- (0 − 1.7)² = 2.89 → 2.89 × 0.10 = 0.289
- (1 − 1.7)² = 0.49 → 0.49 × 0.30 = 0.147
- (2 − 1.7)² = 0.09 → 0.09 × 0.40 = 0.036
- (3 − 1.7)² = 1.69 → 1.69 × 0.20 = 0.338
Step 3 — add: Var(X) = 0.289 + 0.147 + 0.036 + 0.338 = 0.81 vehicles².
Step 4 — square root: SD(X) = √0.81 = 0.9 vehicles.
Say it in words: "a randomly chosen household typically sits about 0.9 vehicles away from the mean of 1.7."

The trap to name (σ vs. σ²): variance comes out in squared units (vehicles²!) — nobody thinks in squared vehicles. Report the SD unless someone specifically asks for variance. If a computed "SD" looks suspiciously large, ask whether it's actually the variance still waiting for its square root.

Memory hook: "Variance is the math's scratch work; the SD is the answer you say out loud."

Segment 6 — Linear Transformations: Shifts & Stretches (18 min)

Plain language first. Real questions rarely stop at X. Fees get subtracted, rates get multiplied, units get converted. If Y = a + bX, you do not need to rebuild the distribution — two rules carry everything:

Mean: E(a + bX) = a + b·E(X) — the average goes through the same arithmetic.
SD: SD(a + bX) = |b|·SD(X) — only the multiplier touches the spread. Adding a constant slides every outcome by the same amount; the outcomes don't spread out. (Variance picks up b²: Var(a + bX) = b²·Var(X).)

Memory hook: "Adding shifts the center. Multiplying stretches both. A constant can't stretch anything."

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

A phone-repair kiosk's service records: X = the number of screen repairs in a day, with mean E(X) = 4 and SD(X) = 1.5. Each repair brings in $60, and the kiosk pays $50 a day for its mall stall. Daily profit: Y = 60X − 50.
- E(Y) = 60(4) − 50 = 240 − 50 = $190.
- SD(Y) = 60 × 1.5 = $90. The $50 stall fee does not appear — subtracting the same $50 every day shifts profits down but doesn't make them any more or less variable.
- Quick check on the trap: if the stall fee rises to $60, E(Y) drops $10 to $180 — and SD(Y) is still $90.

Misconception + cure:
- ❌ "Subtracting the $50 fee reduces the spread too."
✅ Cure: every possible day moves down by the same $50 — the gaps between good days and bad days are untouched. Only multiplication (the $60 rate) rescales the spread.

Segment 7 — Continuous Random Variables: Probability as Area (20 min)

Plain language first. A discrete variable's table lists each value with its chunk of probability. A continuous variable can't have a table — between any two values sit infinitely many more, so no single value can hold a chunk. Instead we draw a density curve and make one trade:

Probability = area under the density curve over an interval. The total area under any density curve is exactly 1. And a single exact value is an interval of width zero — so P(X = any exact value) = 0. Probability questions about continuous variables are always interval questions.

One fully worked example (the friendliest density there is).

Glance at a wall clock at a random moment. X = where the second hand points, measured in seconds from 0 up to 60. Every position is equally likely, so the density is a flat rectangle from 0 to 60.
- The rectangle's total area must be 1, and its base is 60 — so its height is 1/60 (≈ 0.0167). The height is not a probability — it's just the level that makes the area work.
- P(X ≤ 15) — the second hand is in the first quarter — = area = 15 × (1/60) = 0.25. Matches intuition: a quarter of the clock face.
- P(20 ≤ X ≤ 50) = 30 × (1/60) = 0.50.
- P(X = exactly 30) = a sliver of width 0 = 0. You can land near 30 easily; exactly 30.000000… never gets its own chunk of chance.

Misconception + cure:
- ❌ "P(X = a) is the height of the curve at a."
✅ Cure: height is not probability — area is. A tall thin curve and a short wide one can enclose the same area. (No width, no area, no probability.)
- Note the payoff of P(X = a) = 0: for continuous variables, P(X ≤ a) and P(X < a) are the same — the boundary point carries no probability.

Tease inside the segment: next week, one famous discrete variable gets a name (the binomial); the week after, the most famous density curve in the world (the normal). Both run on exactly this week's machinery.

Segment 8 — Technology Workflow + AI-Critique, Callback & Hand-off (12 min) · Session 2 closes (~75)

Technology workflow — E(X) and SD(X) in a spreadsheet (exact steps):
1. Put the values in A2:A5 (0, 1, 5, 10) and the probabilities in B2:B5 (0.70, 0.20, 0.08, 0.02) — the claw machine.
2. Expected value in one cell: =SUMPRODUCT(A2:A5,B2:B5) → 0.8. (SUMPRODUCT multiplies the columns pairwise and adds — exactly the E(X) formula.)
3. Sanity cell: =SUM(B2:B5) → must show exactly 1. If it doesn't, the distribution is broken; fix it before trusting anything.
4. Spread with a helper column: in C2, =(A2-0.8)^2*B2, fill down; Var = =SUM(C2:C5); SD = =SQRT(<that cell>). Google Sheets and Excel are identical here.
5. Preview of Data Lab 6: =RANDBETWEEN(1,100) will simulate draws from a distribution, and you'll watch the simulated average settle toward E(X). (Results are random — expect "close to," never "exactly.")

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

Paste this to your chatbot: "X takes the values 0, 1, 5, 10 with probabilities 0.70, 0.20, 0.08, 0.02. Find the expected value and interpret it."
Then check its work against SUMPRODUCT. Two classic chatbot stumbles: it averages the four values ignoring the probabilities — (0 + 1 + 5 + 10) ÷ 4 = 4, wildly wrong — or it computes 0.80 correctly and then interprets it as "you will most likely win about 80 cents," which is exactly this week's misconception. The tool drafts, you judge — same habit as every week.

Callback + tease:
- Callback: "Week 5 gave chances to events. This week attached numbers to those chances and asked what they're worth. That completes the probability toolkit."
- Tease next week: "One special random variable is so common it gets its own name, its own formula, and its own spreadsheet function: the number of successes in n tries — the binomial. Free-throws, click-throughs, guessing on a quiz — one distribution rules them all."

Hand-off (the week's work):
- Chapter 6 (the primary reading) if they haven't read it yet — then Lecture Tutorial 6 (AI tutor, share-link submission) — random variables, E(X), SD, transformations, density intro.
- Data Lab 6 (build and simulate a raffle in your spreadsheet) · Quiz 6 (end of week) · Discussion 6 (the extended-warranty question) · Assignment 6 (AI-coached).

Instructor FAQ — Common Stumbles

Student says / does Quick cure
"How can the expected value be 0.8 if you can never win 80 cents?" E(X) is the number your average over many plays homes in on — not a possible outcome. "Expected value is what you'd average, not what you'd expect."
Computes E(X) as the plain average of the values. That silently assumes all values are equally likely. Each value must be weighted by its probability — that's the whole formula. (Chatbots make this exact error; see Segment 8.)
"The probabilities sum to 0.98 — close enough?" Never. Legitimacy is exact: total = 1. A near-miss usually means a value is missing or a typo — find it.
Reports the variance when asked for the SD. Variance is in squared units (vehicles², dollars²). Take the square root and report the SD in real units. If a "SD" looks huge, ask if it's an un-rooted variance.
"Subtracting the fee makes profits less variable." Adding or subtracting a constant slides every outcome equally — spread untouched. Only the multiplier rescales SD (and variance gets b²).
"P(X = 30) should be 1/60 for the clock." Single exact values have width 0, so area 0, so probability 0 for continuous X. Only intervals carry probability; 1/60 is the density's height, not a probability.
Confuses P(X ≥ 1) with P(X = 1). "At least one" bundles several rows of the table (or use the Week 5 complement: 1 − P(X = 0)). Have them mark which rows the event includes before adding.
Writes probabilities for a continuous variable in a table. If you can't list all the values, there is no table — that's what makes it continuous. The density curve replaces the table; areas replace the entries.

Scope flag

This outline stays within Objective 4. The "house edge" gambling framing and the long-run-average reprise of Week 5's law-of-large-numbers demo are added context (kept because they make E(X) stick); the continuous-variable treatment is deliberately qualitative + uniform-density only — general density calculations, z-scores, and the normal model are Week 8's job, and the binomial shortcut formulas are Week 7's. Cut the second worked check in Segment 6 for a leaner session.