Chapter 5: Sample size superpowers

Introduction

The introduction explains why confidence intervals express uncertainty when a sample is used to estimate a population characteristic, and connects that idea to the central limit theorem.

• Confidence intervals and uncertainty: Explain why a confidence interval gives a range for a population characteristic instead of only a single estimate from a sample.
  Why it matters: A range makes the uncertainty in a sample-based estimate clearer than a single number does.
• Normal sample means connection: Describe how the approximate normal distribution of sample means motivates using a bell curve to capture ranges.
  Why it matters: This connection explains why the bell curve is relevant to ranges for sample-based estimates.

Key terms: sample, population, confidence interval, uncertainty, central limit theorem, normal distribution of sample means, bell curve

5.1 Applying the central limit theorem

Applies the central limit theorem to a sample-mean probability and shows how sample size changes the distribution of sample means.

• Distribution of sample means: Explain how the mean, standard deviation, and shape of the distribution of sample means relate to the population and sample size.
  Why it matters: These properties let you reason about averages of samples rather than individual observations.
• Body temperature probability example: Trace how the section uses the population temperature statistics and a sample of 31 adults to find the probability that their average is below 98°F.
  Why it matters: It shows which distribution to use when a probability question asks about a group's average.
• Sample size and variability: Explain how increasing sample size changes the distribution of sample means and why a single minimum sample-size rule is unreliable.
  Why it matters: It connects larger samples to more predictable averages without treating a rule of thumb as a guarantee.

Key terms: sample means, population mean, standard deviation of the sample means, sample size, population standard deviation, sample mean, probability distribution of the sample means, distribution of sample means, variability, tails

5.2 Confidence intervals with means

This section explains why a sample mean is an imperfect estimate of a population mean and introduces confidence intervals as a way to express that uncertainty.

• Sample mean as point estimate: Explain why a sizable sample can be used to estimate a population mean when surveying everyone is impractical, and identify the sample mean as a point estimate.
  Why it matters: Recognizing what a sample mean estimates makes its uncertainty easier to discuss.
• Sampling uncertainty and confidence intervals: Explain how the distribution of sample means motivates using a confidence interval to infer a range where the population mean might lie from one imperfect sample mean.
  Why it matters: A confidence interval expresses uncertainty that a single sample average cannot show on its own.

Key terms: sample, population, point estimate, population mean, distribution of sample means, sample mean, confidence interval (CI), bell curve

5.3 Critical Z-values

Shows how to find the critical Z-value that bounds a chosen symmetric center area of the standard normal distribution.

• Symmetric Center And Tail Areas: Explain how a desired center area leaves equal areas in the two tails of a standard normal distribution. For a center area of .95, each tail has area .025.
  Why it matters: The tail areas tell you which cumulative areas to use when finding the boundaries.
• Inverse Cumulative Area Lookup: Distinguish the CDF from the PPF and describe why the PPF is used to find a Z-boundary from a specified area.
  Why it matters: A target center area gives you probabilities first, so you need the inverse lookup to find its Z-boundaries.
• Critical Z Boundary Calculation: Trace how the two cumulative-area lookups give the boundaries of a .95 center area, and apply the expression PPF(1 − (1 − A)/2) to find the positive boundary for a center area A.
  Why it matters: The critical Z-value marks how far on either side of zero the chosen center area extends.

Key terms: standard normal distribution, center area, tails, cumulative distribution function (CDF), inverse CDF, percent point function (PPF), quantile function, critical Z-value, PPF, cumulative density

5.4 CIs with larger samples

This section shows how to construct and interpret confidence intervals for a population mean using a larger sample, and how confidence level and sample size affect interval width.

• Confidence intervals versus point estimates: Explain why reporting a confidence interval communicates a study finding more transparently than reporting only a sample mean.
  Why it matters: A range helps people make decisions without treating a sample estimate as exact.
• Larger-sample interval construction: Trace how a chosen confidence level becomes a confidence interval for a mean in the section’s larger-sample example.
  Why it matters: These steps connect the desired confidence level to a reportable range around the sample mean.
• Repeated-sampling confidence interpretation: Distinguish the repeated-sampling meaning of a 95% confidence interval from a probability claim about one calculated interval.
  Why it matters: This distinction prevents a common overstatement when presenting a particular study’s interval.
• Confidence and sample-size tradeoffs: Describe how increasing confidence or increasing sample size changes the width of a confidence interval.
  Why it matters: Interval width reflects both how certain you want to be and how much sample evidence you have.

Key terms: confidence interval, sample mean, population mean, margin of error, level of confidence, Z-value, standard normal distribution, 95% CI, true population mean, intervals

5.5 Smaller samples with t-distribution

This section explains when to use a t-distribution for a small-sample confidence interval and works through a candle burn-time example.

• Degrees of freedom and tails: Explain how degrees of freedom determine the t-distribution’s shape and why it approaches the normal distribution as sample size grows.
  Why it matters: The changing tails explain why sample size affects which distribution the section uses.
• Small-sample normality condition: Explain when this section substitutes a t-distribution for a normal distribution and what assumption that substitution requires.
  Why it matters: A small sample alone does not justify using a t-based interval under the assumptions given here.
• Candle burn-time interval: Apply the section’s t-distribution choice to the ten-candle example, stating its degrees of freedom and reported confidence interval.
  Why it matters: The example shows what changes—and what stays the same—when constructing a mean interval from a small sample.
• Confidence interval frequency meaning: Distinguish the repeated-sampling meaning of the candle example’s 99% confidence interval from a probability claim about its particular bounds.
  Why it matters: It keeps the advertised burn-time interval from being described with the wrong probability claim.

Key terms: t-distribution, degrees of freedom, fatter tails, normal distribution, population distribution, normal, central limit theorem, CI, sample mean, sample standard deviation

5.6 CLT and CI for proportions

The section explains when to measure an amount rather than a proportion and distinguishes a population proportion from a sample proportion.

• Means versus proportions: Explain why a measurable amount is often better analyzed with a confidence interval for a mean, while a clear yes-or-no characteristic can warrant a proportion.
  Why it matters: Choosing what to measure determines whether an analysis preserves useful information.
• Population and sample proportions: Distinguish a population proportion, denoted P, from a sample proportion, denoted p̂, using the video-game example.
  Why it matters: Keeping the population claim separate from the observed sample result prevents confusing what was measured with what is claimed about everyone.

Key terms: proportions, mean problems, CI approaches with means, categorical variable, binary outcome, population proportion, sample proportion, P, p̂, p-hat

5.7 CLT for proportions

The section explains how samples taken across venues form a distribution of sample proportions and what must change when using a normal approximation for proportions.

• Representative samples across venues: Explain why surveying only a venue such as Comic-Con can bias an estimate of video-game playing, and how polling across venues as unbiasedly as possible yields sample proportions whose distribution can support probability predictions.
  Why it matters: The distribution is useful for predictions only when the way samples are gathered does not systematically favor a particular group.
• Normal approximation for proportions: Distinguish applying the CLT to sample proportions from applying it to sample means: state that two conditions involving sample size n and population proportion p must hold for a reasonably accurate approximation, and that the bell curve uses standard error rather than sample standard deviation.
  Why it matters: Using a bell curve for proportions calls for an accuracy check and the appropriate measure of its spread.
• Limited population information: State that population data is rarely available in practice, so analysts are often limited to p-hat, and that the section says population conditions are assumed when proceeding to confidence intervals.
  Why it matters: It clarifies the gap between describing a population-based approximation and working with data actually available from a sample.

Key terms: sampling proportion, sampling distribution of sample proportions, random sample, CLT, sample proportions, sample means, standard error, sample standard deviation, n, population data

5.8 CIs for proportions

This section adapts confidence intervals for means to estimate population proportions and explains how to interpret a 95% interval.

• Proportion confidence interval calculation: Apply the proportion-specific standard error to form a confidence interval from a sample proportion, sample size, and critical Z-value. Describe how this differs from the standard error used for means.
  Why it matters: It lets you estimate a range for a population percentage from a sample percentage.
• Confidence interval frequency interpretation: Explain what the stated 95% confidence means for the example interval, distinguishing repeated-interval coverage from a probability assigned to this particular interval.
  Why it matters: It prevents a common misreading of what a reported confidence interval claims.

Key terms: CI for proportions, sample proportion, standard error, margin of error, critical Z-value, 95% confidence, true population proportion, such intervals

5.9 Summary