Mostly Harmless Statistics
Not rated yetRate this bookBuild introductory statistics across thirteen chapters, from data and probability to inference, ANOVA, regression and nonparametric methods. Original tasks emphasize questions, assumptions and interpretation.
Read on the publisher’s website. Opens in a new tab.
FreeAbout this book
Mostly Harmless Statistics by Rachel L. Webb develops introductory statistical reasoning from data collection and organization to probability and inference. Thirteen chapters connect graphical and numerical descriptions with discrete and continuous distributions, confidence intervals and hypothesis tests. Later chapters compare two populations, chi-square methods, analysis of variance, correlation, regression and nonparametric approaches.
The structure helps readers connect a method with the question and data it is intended to address. A computed result is not a complete conclusion without a population, design, assumptions and context. This guide adds original tasks using fictional examples to practice those distinctions. Its overviews reflect selected chapter openings rather than every procedure, software instruction or worked solution. The tasks emphasize explaining an analytical choice before calculation and separating association from a causal claim. Readers should use the full text for formulas and numerical practice, checking rendered notation where PDF extraction does not preserve mathematical expressions.
AI-assisted book summary
Reading guide
Define the data and study design before summaries. Build probability and distribution concepts before inference. Compare one-population, two-population and relationship questions before selecting a method.
Chapter-by-chapter guide
1: Introduction to Data
Data are introduced as observations within a statistical investigation. The question guides what should be collected and how conclusions can be drawn. Recording numbers alone does not establish a well-defined study.
Study task: Define a fictional study’s question, population and variables.
2: Organizing Data
Organizing data connects tables and graphs with recognizing patterns in a sample. A display should fit the variable and purpose. The way information is grouped can affect what becomes visible.
Study task: Choose a display for a fictional variable and explain the choice.
3: Descriptive Statistics
Descriptive statistics summarize center and other characteristics without automatically making a population inference. A single average can conceal variation. Interpret a summary through the distribution and context.
Study task: Compare two fictional datasets with similar centers and different patterns.
4: Probability
Probability provides language for uncertainty and events. The event and sample space must be defined before a probability statement is meaningful. Historical motivation is separate from a model’s assumptions.
Study task: Define a sample space and event for a simple fictional chance process.
5: Discrete Probability Distributions
Discrete distributions introduce numerical random variables with distinguishable possible values. A variable, an outcome and an event are different categories. State what is measured before choosing a distribution.
Study task: Specify a fictional discrete variable and its possible values.
6: Continuous Probability Distributions
Continuous distributions connect probabilities with intervals and an area representation. A measured quantity and its model should remain distinct. The model’s shape is an assumption to examine.
Study task: Compare a continuous measurement question with its proposed distribution model.
7: Confidence Intervals for One Population
One-population intervals connect sample information with uncertainty about a population quantity. An interval is not a guarantee about every observation. State the target parameter before interpreting the estimate.
Study task: Write a contextual interpretation of a fictional interval without claiming certainty.
8: Hypothesis Tests for One Population
One-population tests evaluate a claim using sample information under assumptions. A statistic and a parameter are not interchangeable. Evidence against a model does not establish every alternative explanation.
Study task: State a fictional null question and list the assumptions needed before testing.
9: Hypothesis Tests and Confidence Intervals for Two Populations
Two-population inference begins with dependent or matched observations. Pairing changes the analysis question and must be recognized. Similar sample sizes do not establish independence.
Study task: Compare paired and independent fictional study designs.
10: Chi-Square Tests
Chi-square learning introduces a distribution and later categorical-data questions. Identify the question and data structure before selecting a use. A reference distribution alone is not an analysis plan.
Study task: Describe a fictional categorical question and the counts required to investigate it.
11: Analysis of Variance
ANOVA compares group means within a defined design and model. An overall result does not identify every pairwise difference. Keep the global question distinct from follow-up comparisons.
Study task: Separate a fictional overall group question from a specific pairwise question.
12: Correlation and Regression
Correlation and regression examine quantitative relationships and prediction. An association does not establish causation. State the variables and design before interpreting a fitted relationship.
Study task: Rewrite a fictional association claim so it does not imply an unsupported cause.
13: Nonparametric Tests
Nonparametric methods provide alternative approaches for particular questions and data conditions. They still have assumptions and a scope. A label should not be treated as permission to ignore design.
Study task: Compare the questions and assumptions of two possible methods before choosing one.
How to study this book
For a fictional analysis, write the population, variables, design, question and assumptions first. Explain the result in context after calculation. Check notation and distinguish statistical evidence from practical importance or causation.
Reflection questions
- What population does the conclusion concern?
- Which assumptions support the chosen method?
- What does the result leave unresolved?
Sources and coverage
Source: Mostly Harmless Statistics — LibreTexts.
Original reader guide based on the verified contents and selected substantive chapter or section openings, introductions, reading guides and learning objectives in the linked source PDF. It is not a full review of every subsection, exercise, figure, derivation or linked media item. Study tasks are original additions, not exercises or solutions attributed to the source.
This page provides book information and study material. Full-text access is provided by the original publisher or authorised source; a listing does not grant permission to republish the book.
Read our Library Access & Copyright policyQuestion
Method choice begins with the study question and data.
Uncertainty
Inference depends on probability and design.
Interpretation
A numerical result needs context and limits.
Reader ratings & reviews
Share your thoughts
Your review helps other students choose the right resources.




