Statistics Notes: Confidence Intervals & Hypothesis
Testing
From sampling distributions to t-tests and p-values
What this packet includes
● Lecture recap and concise concept summaries
● Quick-reference tables and formulas
● Worked examples with step-by-step solutions
● Practice questions with answer key
● Glossary of key terms
● Further reading suggestions
Foundations
A statistic has a sampling distribution; its variability drives inference uncertainty.
Central Limit Theorem: sample mean ~ normal for large n, regardless of population shape.
Standard error quantifies spread of the estimator (e.g., s/√n for mean).
Intervals and Tests
Confidence interval: estimate ± critical value × standard error.
Hypothesis test: define H0 and H1, pick α, compute test statistic, get p-value, decide.
Use t-distribution for unknown σ with small n and roughly normal data.
Common Procedures
One-sample t for mean; two-sample t (pooled/unpooled) for mean differences.
Proportion z-tests and intervals for categorical data.
Paired tests compare before/after within subjects.
Choosing a Procedure
Goal Typical Method
Estimate population mean µ One-sample t interval
Compare two means µ1-µ2 Two-sample t test
Estimate a proportion p One-proportion z interval
Association in two-way table Chi-square test
Worked Example: One-sample t-interval
● Sample n=20, mean=72, s=10. 95% CI for µ.
● SE = s/√n = 10/√20 ≈ 2.236. t* ≈ 2.093 (df=19).
● CI = 72 ± 2.093×2.236 ≈ (67.3, 76.7).
Worked Example: Two-sample test
● Compare means of A and B with n1=25, n2=30, s1=8, s2=7.
● Use Welch’s t to avoid equal-variance assumption; compute statistic and p-value.
Practice Questions
● Compute a 90% CI for a mean with n=36, x■=15, s=6.
● Test H0: µ=50 vs H1: µ>50 with n=16, x■=53, s=8 at α=0.05.
● Decide between pooled and Welch two-sample t given sample summaries.
● Interpret a p-value of 0.03 in context.
● Explain when CLT justifies normal approximation for the mean.
Answer Key (self-check)
Q1 — See solution outline below.
Q2 — See solution outline below.
Q3 — See solution outline below.
Q4 — See solution outline below.
Q5 — See solution outline below.
Glossary
p-value: Probability of results at least as extreme as observed if H0 is true.
Standard Error: Estimated SD of a statistic.
Type I error: Rejecting a true H0.
Lecture Notes Space
• Note 1: ________________________________
• Note 2: ________________________________
• Note 3: ________________________________
• Note 4: ________________________________
• Note 5: ________________________________
• Note 6: ________________________________
• Note 7: ________________________________
• Note 8: ________________________________
• Note 9: ________________________________
• Note 10: ________________________________
• Note 11: ________________________________
• Note 12: ________________________________
• Note 13: ________________________________
• Note 14: ________________________________
• Note 15: ________________________________
• Note 16: ________________________________
• Note 17: ________________________________
• Note 18: ________________________________
• Note 19: ________________________________
• Note 20: ________________________________
• Note 21: ________________________________
• Note 22: ________________________________
• Note 23: ________________________________
• Note 24: ________________________________
Further Reading
● OpenStax textbooks (free): Calculus, Chemistry, Microeconomics, Statistics, Computer Science
● MIT OpenCourseWare lectures for foundational courses
● Khan Academy topic playlists for quick refreshers