Sec 7.
1 — Confidence Intervals for a Population Mean (σ
known)
Objectives
1. Construct + interpret confidence intervals for μ when σ is known
2. Find critical values
Statistical Process
1. Formulate questions
2. Collect data
3. Describe the data
4. Draw conclusions → confidence intervals is one main method
Point Estimate
● Best estimate of μ (population mean) = sample mean x̄
● Example: sample mean = 67.30 → this is point estimate
Confidence Interval
● Interval estimate used to estimate μ
● Written as: x̄ ± margin of error
Margin of Error
Margin of Error = (Critical Value) × (Standard Error)
Standard Error of x̄
{σ}/{sqrt{n}}
95% Confidence Case
● Critical value = 1.96 (because 95% of normal distribution is between -1.96 and +1.96)
● So 95% CI = x̄ ± 1.96(σ/√n)
Example
x̄ = 67.30
σ = 15
n = 100 → SE = 15/√100 = 1.5
Margin of Error = 1.96(1.5) = 2.94
→ CI: 64.36 < μ < 70.24
Interpretation: We are 95% confident the true μ is between 64.36 and 70.24.
Terminology
Term Meaning
Point estimate single number estimate (x̄)
Confidence interval interval estimate for μ
Confidence level % success rate of this method (ex
95%)
Margin of error critical value × standard error
Critical Values
● Each confidence level has its own critical z-value
● Common:
| Confidence | Critical z |
|------------|------------|
| 90% | 1.645 |
| 95% | 1.96 |
| 99% | 2.575 |
Rest Check from Table A.3
zα and zα/2 Notation
● zα = z score with α area to the right
● For CI → use zα/2 because CI splits area in both tails
Example: 92% CI → α = 0.08 → α/2 = 0.04 → critical value = 1.75
Technology Shortcuts
● TI-84 → Zinterval
● Excel
○ margin of error = [Link](α,σ,n)
○ critical value = [Link](prob)
Assumptions (σ known case)
● Simple random sample
● Population approx. normal OR sample size large (n > 30)
Section 7.2 — Confidence Intervals for the Population
Mean (σ unknown) — Detailed Notes
WHY this section exists
In real world statistics, we almost never know the population standard deviation σ.
So instead of using σ, we estimate it using the sample standard deviation s.
Because s is only an estimate ⇒ more uncertainty ⇒ we must account for that uncertainty.
This is the reason we switch from the standard normal z distribution to a t distribution.
t Distribution Characteristics
● symmetric, bell shaped like normal distribution
● centered at mean = 0
● more spread out / heavier tails than normal distribution
→ this compensates for using s instead of σ
● as sample size increases (n increases), t distribution gets closer to normal
Degrees of Freedom
df = n − 1
Degrees of freedom basically represent the amount of independent information in your sample.
Lower df = more spread out → larger critical values
Higher df (large n) → t gets closer to z
Confidence Interval Formula (σ unknown)
CI: xˉ±tα/2(s/sqrt n)
Where:
● x̄ = sample mean
● s = sample standard deviation
● n = sample size
● tα/2= critical t value with (n−1) degrees of freedom
Margin of Error (σ unknown)
E=tα/2(s/sqrt n)
Conditions / Assumptions
Use t-interval when:
1. σ is unknown (major reason)
2. Sample is random
3. Population approximately normal
(if n ≥ 30 → CLT makes it safe even if population is not perfectly normal)
Interpretation (VERY IMPORTANT FOR TESTS)
We say:
We are C% confident that the interval constructed contains the true population
mean μ.
We do NOT say there is a C% probability that μ is in this interval.
μ is a fixed value — the interval either contains it or doesn’t.
Confidence level refers to the reliability of the method, not the probability for a specific interval.
Technology Notes
TI-84: use TInterval
Excel: use [Link].2T(probability,df) for critical value or its CI functions
Effect of Confidence Level on Margin of Error
● ↑ confidence level → ↑ critical t value → ↑ margin of error (less precise interval)
● ↓ confidence level → ↓ critical t value → ↓ margin of error (more precise interval)