← Back to the Introduction to Statistics outline The Course Maker
← Introduction to Statistics outline
Week 11 · Lecture outline

Week 11 — Lecture Outline · Confidence Intervals for Means

Introduction to Statistics · MATH 11 Fall 2026 · Prof. Rivera Fictional sample

Course: Introduction to Statistics (MATH 11) · Silver Oak University (fictional sample) · Prof. Rivera
Objectives covered: Objective 6 — Construct and interpret confidence intervals for means.
SLOs touched: A (reason quantitatively from data) · B (communicate results to a non-technical audience)
Meeting pattern: 2 sessions × 75 min = 150 min. Segment minutes below total ~150; scale to your own pattern.

Campus note: Veterans Day falls on Wednesday, Nov 11 this week — a campus holiday. Our Tue/Thu pattern (Tue Nov 10 / Thu Nov 12) is unaffected, but no makeup or office hours land on Wednesday. Mention it so students aren't surprised.

Week at a Glance
The week's big question "From one sample mean, how do we honestly state a range for the true population mean — and say exactly how confident we are without overclaiming?"
By the end of the week, students can… (1) explain why we use t instead of z when the population σ is unknown, and find the right degrees of freedom (df = n − 1); (2) construct a confidence interval for a mean as x̄ ± t*·(s/√n), using a supplied t* value; (3) compute and name the margin of error; (4) interpret a confidence interval correctly and catch the two classic misinterpretations.
Key vocabulary t-distribution, degrees of freedom (df), standard error (SE = s/√n), critical value t*, margin of error (ME), confidence level, confidence interval, point estimate, "95% confident," capture/long-run interpretation
Materials slides (Deck 11), the week's readings + video links, a spreadsheet (Google Sheets or Excel), one approved chatbot (Gemini / Claude / ChatGPT) for the AI-critique moment and the tutorial, and the embedded t* table below (we supply every critical value — no tables to look up)
Timing note 8 segments, ~150 min total. Session 1 = Segments 1–4 (~75). Session 2 = Segments 5–8 (~75).

The t* table we use all week (embedded — never look it up)

Rule for this course: every confidence-interval problem you see is engineered to use one of these exact t* values. The materials hand you t*; you never recall or look it up. (df = n − 1.)

Confidence df = 9 (n=10) df = 15 (n=16) df = 24 (n=25) df = ∞ (the z value)
90% 1.833 — — 1.645
95% 2.262 2.131 2.064 1.960

Read the last column as the punchline of Segment 2: as df grows, t* slides down toward the familiar z value (1.960 at 95%). Big samples → t looks like z.

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

Hook. Put a real headline on the board: "A new poll finds 54% of voters approve of the measure — margin of error ±3 percentage points."
- "You have read a sentence exactly like this a hundred times. Today you learn what that little ± 3 actually is, where it comes from, and — just as important — what it does not mean."
- "Last week we learned that a sample mean x̄ bounces around the true mean μ from sample to sample (the sampling distribution, the Central Limit Theorem). That bouncing is the problem. This week is the fix: instead of pretending our one x̄ is μ, we wrap an honest range around it and attach a confidence level."

The promise (write it on the board): "By the end of this week you can take a single sample — a mean, a standard deviation, a sample size — and produce a sentence like '±3' yourself: an interval for the true mean, plus an honest statement of how confident you are. And you'll never again say the two sentences that get this wrong."

Why it matters line (memory hook): "A statistic is a guess at one point. A confidence interval is an honest guess — it admits it might be off, and says by how much."

Segment 2 — Why t, Not z? The t-Distribution & Degrees of Freedom (22 min)

Plain language first.
- Last week, when we knew the population standard deviation σ, the sample mean's spread was σ/√n and we used the normal (z) model. In real life we almost never know σ — we only have the sample standard deviation s.
- Swapping s in for σ adds a second source of uncertainty: s is itself just an estimate, and in small samples it's a shaky one. To pay for that extra uncertainty, we switch from z to a slightly wider model: the t-distribution.
- The t-distribution looks like the normal bell but with fatter tails — it gives a little more room for being wrong, exactly because s might be off. The smaller the sample, the fatter the tails.

Degrees of freedom (the t-distribution's dial).
- The t-distribution isn't one curve; it's a family, indexed by degrees of freedom (df). For a one-sample mean, df = n − 1.
- Plain-language df: "the number of values free to vary once the mean is fixed." Don't over-explain it on day one — the operational rule is df = n − 1, and that's what picks the row in our t* table.

The key relationship (put it on a slide — this is the punchline):

As df grows, t* shrinks toward z. Look at the 95% column of our table: df 9 → 2.262, df 15 → 2.131, df 24 → 2.064, df ∞ → 1.960. The t value is always a little bigger than the z value (1.960), and it slides down toward 1.960 as the sample grows.
"t is z with a safety margin for small samples. Big sample, the margin almost vanishes."

One fully worked example (find df and pull t*):

A sample of n = 25 students; we want 95% confidence.
- df = n − 1 = 25 − 1 = 24.
- 95% confidence, df 24 → from our table, t* = 2.064. (We supplied it. No lookup, no guessing.)

Land the key idea: use t (with df = n − 1) whenever you're estimating a mean and you only have the sample's s. Reserve z for the rare case σ is genuinely known.

Segment 3 — Building a Confidence Interval for a Mean (25 min)

Plain language first — the anatomy of every interval.
- An interval has two parts: a center and a reach.
- Center = the point estimate, our single best guess: the sample mean x̄.
- Reach = the margin of error (ME) = how far we extend on each side.
- The whole formula, in words: "sample mean, give or take (a critical value) times (the standard error)."

The formula (notation after the idea):

CI for a mean: x̄ ± t* · (s / √n)
- s / √n is the standard error (SE) — the typical distance x̄ sits from μ (straight from last week's sampling distribution).
- t* · SE is the margin of error (ME).
- x̄ − ME is the lower endpoint; x̄ + ME is the upper endpoint.

One fully worked example — do every step out loud (numbers chosen so the arithmetic is clean):

A sample of n = 25 students reports a mean nightly sleep of x̄ = 50 (in some scaled "rest score"), with sample SD s = 10. Build a 95% confidence interval for the true mean.
1. Standard error: SE = s / √n = 10 / √25 = 10 / 5 = 2.
2. Degrees of freedom: df = n − 1 = 24. t* (95%, df 24) = 2.064 (supplied).
3. Margin of error: ME = t* · SE = 2.064 × 2 = 4.128 ≈ 4.13.
4. Interval: x̄ ± ME = 50 ± 4.13 → lower 50 − 4.13 = 45.87, upper 50 + 4.13 = 54.13.
5. Answer: the 95% confidence interval is (45.87, 54.13) — round to (45.9, 54.1) for reporting.

A second, even cleaner one (build it together, ~SE = 1):

n = 25, x̄ = 25, s = 5, 95% confidence. SE = 5/√25 = 5/5 = 1; t* = 2.064; ME = 2.064 × 1 = 2.06; interval 25 ± 2.06 = (22.94, 27.06).

The four-step recipe (give them this card):
1. SE = s ÷ √n. 2. t* = look it up in our supplied table (df = n−1, your confidence level). 3. ME = t* × SE. 4. Interval = x̄ − ME to x̄ + ME.

Segment 4 — The Margin of Error, and What Moves It (20 min) · Session 1 closes (~75)

Plain language: the margin of error is the "give or take" — the half-width of the interval, ME = t* · (s/√n). It is the single number a news story is reporting when it says "±3 points."

What makes the margin bigger or smaller (build intuition, not just formula):
- Bigger sample (n↑) → smaller ME. √n is in the denominator, so more data tightens the interval. (To halve the margin you must roughly quadruple n — because of the square root.)
- More spread (s↑) → bigger ME. Noisier data, less precision.
- Higher confidence (90% → 95% → 99%) → bigger t* → bigger ME. Being more sure costs width. You buy confidence with vagueness.

Worked contrast (same data, two confidence levels — show the trade-off):

n = 10, x̄ = 50, s = 10 (so SE = 10/√10 ≈ 3.162).
- At 90% (df 9, t* = 1.833): ME = 1.833 × 3.162 ≈ 5.80.
- At 95% (df 9, t* = 2.262): ME = 2.262 × 3.162 ≈ 7.15.
Same sample, wider net for more confidence. Land it: there is no free lunch — precision and confidence trade against each other.

Misconception + cure (the confidence-level one):
- ❌ "A 99% interval is better than a 95% one — more confidence is always better."
✅ Cure: higher confidence makes the interval wider (less useful). "I'm 100% confident the mean is between −∞ and +∞" is useless. The skill is choosing a confidence level that's honest and informative; 95% is the common compromise.

Quick interaction — Think-Pair-Share (~6 min): Show two intervals for the same quantity: (48, 52) and (40, 60). "Which sample was probably bigger? Which reporter was probably more confident?" Surface that a narrower interval comes from a bigger n (or lower confidence or smaller s) — and that width and confidence level are different knobs.

Segment 5 — Interpreting a Confidence Interval CORRECTLY (25 min) · Session 2 opens

Hook back in: "Last session we built the interval. Today is the part students lose the most points on, and the part that actually matters in the real world: saying what it means — and refusing to say what it doesn't."

The one correct interpretation (make them memorize the template):

"We are 95% confident that the true population mean lies between 45.9 and 54.1."
What that sentence is really shorthand for: "This interval was produced by a method that captures the true mean 95% of the time in the long run. We can't see whether this particular interval is one of the lucky 95% or the unlucky 5% — but we trust the method."

The picture that makes it click (draw it):

Imagine taking 100 different samples and building 100 different 95% intervals. About 95 of them will contain the true μ; about 5 will miss it. The "95%" is a property of the procedure across many samples, not of any one interval. The true μ is a fixed number; it's the intervals that vary.

The TWO classic misinterpretations — name them, then cure each (this is the heart of the week):

  • ❌ Misinterpretation #1: "95% of the data values fall in the interval."
    ✅ Cure: No — a confidence interval is about the mean, not the spread of individual data points. Our interval (45.9, 54.1) is a range for the average, not a range that holds 95% of students. "A CI brackets the mean, not the people."

  • ❌ Misinterpretation #2: "There's a 95% chance the true mean is in THIS interval."
    ✅ Cure: Once the interval is computed, the true mean either is in it or isn't — there's no probability left to assign to this fixed interval. The 95% describes how often the method works before you draw the sample, across many possible samples — not the odds for the one interval in your hand. "The method is 95% reliable; this interval is already decided."

A third, sneakier slip to flag:
- ❌ "95% of all samples' means fall in this interval." / "We're 95% sure the sample mean is in here." ✅ Cure: the interval is about the unknown population mean μ, not about future sample means and certainly not the sample mean we already have (x̄ is the center — it's definitely "in there").

Memory hook (put it on a slide): "95% confident = the method catches the true mean 95% of the time. Not the data. Not a coin flip on this one interval."

Segment 6 — Conditions, and a Worked Interpretation Round (18 min)

Plain language — when are we allowed to do this? A t-interval for a mean is trustworthy when:
- Random: the data come from a random sample (or randomized experiment) — so the sample represents the population. (Callback to Week 1: method beats size.)
- Independent: observations don't influence each other (rough 10% rule: sample < 10% of the population when sampling without replacement).
- Normal/large enough: the population is roughly normal, or the sample is large enough (n ≥ 30 is the common rule of thumb) for the CLT to make x̄'s distribution near-normal. For small n, check the data have no strong skew or outliers.

Worked interpretation round (give them the interval, ask for the sentence):

Context: A random sample of n = 16 commute times has x̄ = 100 minutes/week, s = 8; the 95% interval is (95.7, 104.3). (SE = 8/√16 = 2; t* = 2.131, df 15; ME = 2.131×2 = 4.26.)
- Correct sentence: "We're 95% confident the true mean weekly commute for the population is between 95.7 and 104.3 minutes."
- Spot the error A: "95% of commuters spend between 95.7 and 104.3 minutes." → about individuals/data, not the mean. Wrong.
- Spot the error B: "There's a 95% probability the true mean is between 95.7 and 104.3." → assigns probability to a fixed interval. Wrong; it's the method that's 95% reliable.

Misconception + cure (z vs t, revisited):
- ❌ "With a small sample and unknown σ, just use 1.96 like last unit."
✅ Cure: 1.96 is the z value; with unknown σ you owe the extra width of t. For n = 16 at 95%, the right multiplier is 2.131, not 1.960. Using z here makes your interval too narrow and your confidence overstated.

Segment 7 — Technology Workflow + AI-Critique (20 min)

Technology workflow — build a confidence interval in a spreadsheet (exact steps):
1. Put your data in column A (say A2:A26 for 25 values). Compute the pieces:
- =AVERAGE(A2:A26) → x̄ (say it lands in C1).
- =STDEV.S(A2:A26) → the sample SD s (C2). (Use STDEV.S — the sample version — not STDEV.P.)
- =COUNT(A2:A26) → n (C3).
2. Standard error: =C2/SQRT(C3) → SE (C4).
3. Margin of error two ways:
- The honest "look it up in our table" way: ME = =2.064*C4 (drop in the supplied t* for df 24, 95%).
- The built-in way (interpret-the-output): =CONFIDENCE.T(0.05, C2, C3) returns the margin of error directly for a 95% interval. (0.05 = 1 − 0.95.) It should match t*·SE.
4. Endpoints: =C1 - ME and =C1 + ME.
- Sanity check: the two methods (supplied t* vs CONFIDENCE.T) should agree to a rounding wiggle. If they don't, you used STDEV.P or the wrong alpha.

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

Paste this to an approved chatbot: "A 95% confidence interval for the mean test score is (45.9, 54.1). Explain what this means."
Then audit the answer against today's two cures. Chatbots very often produce one of the banned sentences — "there's a 95% probability the true mean is in this interval" or "95% of scores fall in this range." Your job: catch it, and rewrite it as the correct long-run-method statement. The tool drafts; you judge. Also ask it to "construct a 95% CI from x̄=50, s=10, n=25" and check whether it correctly used t (2.064) rather than z (1.96) — models slip here too.

Segment 8 — Callback, Tease & Hand-off (12 min) · Session 2 closes (~75)

Callback: "Week 10 told us x̄ bounces around μ with spread σ/√n. This week we turned that bounce into an honest range — x̄ ± t*·(s/√n) — and learned to say what it means without overclaiming."

Tease next week: "We just did intervals for a mean (a quantity). Next week: confidence intervals for a proportion (a percentage — exactly the '54% approve, ±3 points' poll from Monday's hook), plus how pollsters choose a sample size to hit a target margin of error."

Hand-off (the week's graded work):
- Lecture Tutorial 11 (AI tutor, share-link submission) — t vs z, building a CI, margin of error, and the two interpretation traps.
- Quiz 11 (end of week) — 10 items on when to use t, constructing a CI, the margin of error, and correct interpretation.
- Discussion 11 — find a real reported margin of error or confidence interval (e.g., a poll's ±3%) and reason, with your chatbot, about what it actually means and the common misreadings.
- Assignment 11 — four problems: conditions/when t vs z; construct a CI; interpret correctly + catch a misinterpretation; explain "margin of error" / "95% confident" to a non-expert.

Instructor FAQ — Common Stumbles

Student says / does Quick cure
"When do I use t and when do I use z?" Estimating a mean and you only have the sample SD s (the usual case) → t, with df = n − 1. z is only for the rare case the population σ is truly known.
"What's the degrees of freedom again?" For a one-sample mean, df = n − 1. It's the row you read in our t* table. (n = 25 → df = 24.)
Uses 1.96 for a small sample with unknown σ. 1.96 is z. With unknown σ you owe the wider t value (e.g., 2.064 at df 24, 95%). Using z makes the interval too narrow and overstates confidence.
"95% of the data are in the interval." No — a CI brackets the mean, not individual values. It's a range for the average, not a range that holds 95% of people.
"There's a 95% chance the true mean is in this interval." The interval is already fixed; μ is either in it or not. The 95% describes the method over many samples (95 of 100 such intervals capture μ), not the odds for this one.
Forgets to divide s by √n. The standard error is s/√n, not s. The √n is what shrinks the margin as the sample grows.
Reports the interval as a single ± number with no center. Always state both: center (x̄) and margin (ME), or the two endpoints. "±4.13" alone is meaningless without the 50.
Thinks a 99% interval is "more accurate." Higher confidence → wider interval (bigger t*), i.e., less precise. More confidence is bought with more vagueness; 95% is the usual compromise.
In a spreadsheet, uses STDEV.P. Use STDEV.S (the sample SD) for s; STDEV.P is the population formula and gives the wrong SE here.

Scope flag

This outline stays within Objective 6 (confidence intervals for means). The conditions list in Segment 6 and the spreadsheet CONFIDENCE.T function are added rigor/context (helpful, not strictly demanded by the objective) — keep them for a complete treatment, or trim Segment 6's conditions to a single line for a leaner 60-minute version. Confidence intervals for proportions are deliberately held for Week 12.

~ Prof. Rivera's edition · Fall 2026 · built with thecoursemaker.com