An Intuitive, Interactive, Introduction to Biostatistics

Chapter-by-chapter summaries

Chapter 1: Introduction

Statistics is presented as a way to learn from evidence while quantifying uncertainty. The chapter connects a scientific question, a hypothesis, data collection, analysis, and communication. A population contains the broader cases of interest; a sample contains those actually observed. Parameters describe population quantities, while statistics summarize samples and can be used to estimate parameters. This framework explains why studying a sample can be informative without providing certainty. The examples also show that collecting data requires choices about what to measure and how to compare conditions. Evidence-based reasoning is not simply replacing opinion with a number: the number must come from a design and analysis appropriate to the question, and the remaining uncertainty must be communicated honestly.

Chapter 2: Data Summaries and Presentation

Data type determines which displays and summaries make sense. Categorical variables are described through counts and proportions, while quantitative variables support summaries of location and spread. The chapter compares graphical approaches and numerical measures, emphasizing that datasets with the same mean can have very different distributions. Means and medians respond differently to skew and outliers; percentiles and the interquartile range answer different questions from variance and standard deviation. The interactive activities let readers change values and see how summaries respond. This makes robustness and sensitivity visible rather than merely definitional. A good report combines appropriate numbers with a display and labels that preserve units and context. No single summary can replace understanding the pattern of the observations. For example, a highly skewed spending distribution can have a mean above the median because a small number of large values pull the arithmetic average upward. That is a feature to explain, not an automatic reason to discard observations. First check whether unusual values are plausible or reflect recording errors.

Chapter 3: Study Design

The chapter contrasts observational studies with experiments and develops randomized controlled clinical trials as an important experimental design. Random allocation helps make treatment groups comparable in expectation, while blinding limits influences arising from expectations or knowledge of assignment. Stratified randomization can help balance selected characteristics across groups. These procedures address different problems and should not be treated as synonyms. Observational evidence can reveal associations but requires careful consideration of confounding and selection. The chapter's examples demonstrate that a sophisticated analysis cannot fully repair a badly designed study. It is also important to distinguish random assignment, which supports comparisons between conditions, from random sampling, which concerns how a sample represents a population. Both the design and its implementation affect interpretation.

Chapter 4: Introduction to Probability and Simulation

Randomness means that an individual outcome cannot be predicted with certainty, even when a process has a describable long-run pattern. The chapter introduces sample spaces, events, and probability through examples such as rolling dice. Equally likely elementary outcomes can be counted, but apparently simple categories such as dice totals are not necessarily equally probable. Simulation repeats a specified random process to approximate its behavior. Short runs can differ noticeably from expected proportions, and longer runs tend to stabilize without guaranteeing an exact match. Interactive exercises connect observed frequencies with theoretical probabilities and model-based calculations. The essential distinction is between what happens on one trial and what a probability model says about repeated trials under the same conditions.

Chapter 5: Probability Distributions

A random variable assigns numerical values to outcomes, and a probability distribution describes how likely its possible values are. The chapter develops the connection among a distribution, its parameters, and data generated from it. The binomial model describes a count of successes in a fixed number of independent trials with a common success probability; changing the trial count or success probability changes its shape and center. The normal model describes a continuous distribution whose location and spread are controlled by its mean and standard deviation. Interactive simulations help distinguish a theoretical model from one realized sample. Expected value is a long-run average property, not a promise that the next observation will equal that value or even that it must be an attainable individual outcome.

Chapter 6: Sampling Distributions and the Central Limit Theorem

The chapter shifts attention from individual observations to statistics calculated from repeated samples. A sampling distribution describes how a statistic would vary if the same sampling process were repeated. A sample mean estimates a population mean, but different samples produce different estimates. Under suitable conditions, larger samples reduce the variability of the mean, and the central limit theorem explains why its sampling distribution can become approximately normal even when the underlying observations are not. The conditions matter; the theorem does not say that every small sample or every original dataset becomes normal. Simulations make the distinctions among a population distribution, one sample, and a distribution of sample means concrete. This prepares readers to understand standard errors and inference as descriptions of repeated-sampling uncertainty. Distinguish the standard deviation of individual observations from the standard error of an estimated mean. The first describes variation among observations; the second describes how much sample means vary under the sampling model. Increasing sample size can reduce the latter without making the underlying population less diverse.

Chapter 7: Introduction to Inference

Inference uses a sample and a model to make a statement about a population parameter. Confidence intervals express a range of estimates generated by a method with a specified long-run coverage property. The chapter's repeated-interval simulations show why a 95% procedure does not place 95% of individual observations in the interval and does not guarantee that every particular interval contains the parameter. Hypothesis testing compares data with a null model and distinguishes Type I from Type II errors. A p-value describes how unusual an observed or more extreme result would be under the null assumptions; it does not give the probability that a hypothesis is true. Confidence level, sample size, and variability affect precision, while statistical significance remains different from practical importance.

Supplementary synthesis: What to notice in an interactive simulation

Before running an application, identify the population or probability model, its fixed parameters, the sample size, and the statistic being displayed. Predict what should change when one setting changes, then compare the prediction with repeated runs. Increasing the number of simulations improves the approximation of a long-run pattern; increasing the sample size changes the sampling variability of the statistic. Those are separate operations. When studying confidence intervals, count how often the procedure covers the fixed parameter rather than imagining the parameter moving between runs. When studying a distribution, distinguish the theoretical shape from random irregularities in a finite sample. These questions connect the applications across chapters and prevent treating a simulation as a decorative animation.

Supplementary synthesis: Interpreting evidence without overstating it

A useful conclusion names the target population, what was measured, the comparison or association, and the uncertainty attached to the estimate. It also states what the design can support. Randomization does not make measurement flawless; a narrow interval does not remove selection bias; a small p-value does not establish a large or causal effect. Conversely, a result that is not statistically significant does not prove the absence of an important effect. These distinctions arise from combining the book's design, probability, and inference chapters. They are additional study synthesis rather than further original chapters.

Study instructions

Before an interactive exercise, predict what changing one input might do. Change one factor at a time and explain the resulting pattern in words. Record what is simulated, what is observed, and what assumptions the demonstration uses.

For a study example, identify the population, sample, measured variables, and design before considering an inference. Separate the size of an observed difference from uncertainty about that difference. Use the source explanations for the precise requirements of each statistical method.

Questions for reflection

• Which part of an example represents data and which represents a model?
• How might study design affect an interpretation?
• What can an interactive demonstration clarify that a definition alone does not?

The study suggestions and questions are independent Academic Master editorial material.

Sources and coverage

These original summaries draw on the source text, including chapter or section introductions and selected explanations, examples, exercises, and review material. They are condensed study guides, not a page-by-page account or a replacement for the complete resource. Additional synthesis sections are identified separately.

Official book source; Reviewed source edition.