Source: STAT 211 Engineering Statistics, Texas A&M University Difficulty: Intermediate | Prerequisites: Part 1 notes (descriptive statistics, probability, normal distribution, Z-scores)
Tags: sampling distribution, Central Limit Theorem, CLT, confidence interval, margin of error, t-distribution, standard error, point estimate, interval estimate, STAT 211, Texas A&M, FE exam
You now move from describing data you already have to making inferences about populations you cannot fully observe. The key bridge is the sampling distribution: if you understand how a sample statistic (like x̄) behaves across many hypothetical samples, you can quantify how confident you are in your estimate. Confidence intervals are the first major inferential tool, and they set up everything in hypothesis testing.
A sampling distribution describes the variability of a statistic across repeated samples. The Central Limit Theorem guarantees that the sampling distribution of x̄ is approximately normal for large n, regardless of the population shape. Confidence intervals use this fact to wrap a margin of error around a point estimate, giving you a plausible range for the true parameter.
Sampling distribution
The probability distribution of a statistic (e.g. x̄) computed from all possible samples of a given size n drawn from a population. Think of it as the answer to: "If I repeated this study thousands of times, what pattern would my sample means follow?"
Standard error (SE)
The standard deviation of a sampling distribution. For the sample mean: SE = σ / √n (when σ is known) or SE = s / √n (when σ is estimated). In simple terms, it measures how much your sample mean would jump around from sample to sample.
Central Limit Theorem (CLT)
Regardless of the population distribution, the sampling distribution of x̄ is approximately normal when the sample size n is sufficiently large (commonly n ≥ 30). This is the single most important theorem in the course. It is why normal-based methods work even when the underlying data are not normal.
Point estimate
A single number used to estimate a population parameter. For example, x̄ is the point estimate of μ.
Confidence interval (CI)
An interval of values, computed from sample data, that is likely to contain the true population parameter. A 95% CI means that if you repeated the sampling process many times, about 95% of the resulting intervals would capture the true parameter.
Confidence level
The probability (e.g. 90%, 95%, 99%) that the interval procedure will capture the true parameter across repeated sampling. It is not the probability that the true parameter lies in one specific interval you have already computed.
Margin of error (E)
Half the width of a confidence interval. E = (critical value) × (standard error). It quantifies the precision of your estimate.
Critical value
The Z or t value that corresponds to the tails of the distribution cut off by the confidence level. For a 95% CI using Z: z* = 1.96.
t-distribution
A bell-shaped distribution that is wider (heavier tails) than the standard normal. It is used in place of Z when the population standard deviation σ is unknown and must be estimated by s. The shape depends on degrees of freedom (df = n , 1).
Degrees of freedom (df)
For a one-sample t procedure: df = n , 1. As df increases, the t-distribution approaches the standard normal. Think of it as a correction for the extra uncertainty introduced by estimating σ.
If the population has mean μ and standard deviation σ, then the sampling distribution of x̄ has:
Mean: μ_x̄ = μ
Standard deviation (standard error): σ_x̄ = σ / √n
Increasing the sample size n reduces the standard error. Larger samples produce more precise estimates.
If the population itself is normal, the sampling distribution of x̄ is exactly normal for any n.
If the population is not normal, the CLT tells you the sampling distribution of x̄ is approximately normal once n is large enough (rule of thumb: n ≥ 30).
Use Z when the population standard deviation σ is known. This is rare in practice but common in textbook problems and on the FE exam.
Use t when σ is unknown and you substitute the sample standard deviation s. This is the realistic scenario for most engineering data.
As the sample size grows, the t-distribution converges to the standard normal, so for very large n the distinction becomes negligible.
Formula: x̄ ± z* · (σ / √n)
Choose z* based on the confidence level: 1.645 for 90%, 1.96 for 95%, 2.576 for 99%.
The interval is centred on x̄ and extends the margin of error in both directions.
Formula: x̄ ± t* · (s / √n), with df = n , 1
Look up t* from the t-table at the desired confidence level and degrees of freedom.
The interval is wider than the Z-interval because the t-distribution accounts for the extra uncertainty in estimating σ.
The sample proportion is p̂ = (number of successes) / n.
Formula: p̂ ± z* · √[p̂(1 , p̂) / n]
This uses the normal approximation, which requires np̂ ≥ 5 and n(1 , p̂) ≥ 5 (some texts use 10).
The standard error for a proportion is √[p̂(1 , p̂) / n], not σ / √n.
Correct: "We are 95% confident that the true population mean lies between [lower, upper]."
Incorrect: "There is a 95% probability that the true mean is in this interval." The true mean is fixed; it is either in the interval or it is not. The 95% refers to the long-run success rate of the method, not to any single interval.
To achieve a desired margin of error E with a given confidence level:
For a mean (σ known): n = (z* · σ / E)²
For a proportion: n = (z* / E)² · p̂(1 , p̂). When p̂ is unknown, use 0.5 for the most conservative estimate.
Always round n up to the next whole number.
Standard error of the mean (σ known): SE = σ / √n
Standard error of the mean (σ unknown): SE = s / √n
CI for μ (Z): x̄ ± z* · (σ / √n)
CI for μ (t): x̄ ± t* · (s / √n), df = n , 1
CI for p: p̂ ± z* · √[p̂(1 , p̂) / n]
Sample size for a desired margin of error (mean): n = (z* · σ / E)²
Sample size for a desired margin of error (proportion): n = (z* / E)² · p̂(1 , p̂)
Common critical Z-values: 90% → 1.645, 95% → 1.96, 99% → 2.576
Confidence intervals are how engineers report measurement uncertainty. When a materials lab reports the tensile strength of a steel alloy as "520 ± 12 MPa at 95% confidence," they are giving a confidence interval. Regulatory agencies and quality standards (ISO, ASTM) require interval estimates, not just single numbers, precisely because a point estimate on its own says nothing about precision.
Students frequently say "there is a 95% chance the true mean is in this interval." That is not what a confidence interval means. The true mean is a fixed number. The 95% describes the reliability of the procedure across many samples.
Confusing standard deviation (s) with standard error (s / √n) is one of the most common exam mistakes. Standard deviation describes data spread; standard error describes the precision of a sample statistic.
Using the Z-interval when σ is unknown is wrong. If you are plugging in s for σ, you must use t.
Students sometimes think a wider interval is "worse." It is less precise, yes, but it reflects either a smaller sample, more variability, or a higher confidence level. The interval is not wrong; it is appropriately cautious.
⚠️ The CLT is tested conceptually: expect questions about when it applies and what it guarantees.
⚠️ Know which CI formula to use. The decision tree is: σ known → Z; σ unknown → t; proportion → Z with the proportion standard error.
⚠️ Interpretation questions are worth easy marks but frequently lost. Memorise the correct phrasing.
⚠️ Sample size determination problems almost always appear on at least one exam. Remember to round up.
⚠️ The t-table: you need to be comfortable reading it under exam conditions. Practise finding t* for various df and confidence levels.
True or False: The Central Limit Theorem requires the population to be normally distributed.
Fill in the blank: As the sample size increases, the standard error ___.
True or False: A 99% confidence interval is narrower than a 95% confidence interval for the same data.
Fill in the blank: The degrees of freedom for a one-sample t-interval are ___.
True or False: When computing a CI for a proportion, you use the t-distribution.
Answers: 1. False (it works regardless of population shape, given large enough n). 2. Decreases. 3. False (it is wider). 4. n , 1. 5. False (you use the Z-distribution with the proportion standard error).
Q: A sample of 36 measurements has x̄ = 50 and σ = 12. Construct a 95% confidence interval for μ.
A: SE = 12 / √36 = 2. CI = 50 ± 1.96(2) = 50 ± 3.92 = (46.08, 53.92).
Q: Why do we use the t-distribution instead of the Z-distribution when σ is unknown?
A: Because substituting s for σ introduces additional uncertainty. The t-distribution has heavier tails to account for this, producing wider intervals that are appropriately conservative.
Q: A survey finds that 120 out of 400 respondents prefer Brand A. Construct a 95% CI for the population proportion.
A: p̂ = 120/400 = 0.30. SE = √[0.30(0.70)/400] = √(0.000525) ≈ 0.0229. CI = 0.30 ± 1.96(0.0229) = 0.30 ± 0.045 = (0.255, 0.345).
Q: You want a margin of error of 2 units for a 95% CI, and σ = 10. What sample size do you need?
A: n = (1.96 · 10 / 2)² = (9.8)² = 96.04. Round up to n = 97.
Q: What happens to the width of a confidence interval if you increase the confidence level from 95% to 99%, holding everything else constant?
A: The interval becomes wider, because the critical value increases from 1.96 to 2.576.
Sampling distributions are the theoretical backbone of hypothesis testing (Weeks 10, 11). The test statistic in a hypothesis test is computed the same way as the CI boundary, just rearranged. The confidence interval for a proportion connects directly to testing two proportions in Week 13. Regression confidence intervals (Week 15) follow the same t-based logic with a different standard error formula.
sampling distribution, standard error, Central Limit Theorem, CLT, confidence interval, CI, margin of error, critical value, Z-interval, t-interval, t-distribution, degrees of freedom, point estimate, interval estimate, proportion confidence interval, sample size determination, precision, reliability of estimation, STAT 211, engineering statistics, FE exam, Texas A&M