How to use this book
R. Heller · Date: 2026-05-08 · Last changed: 2026-11-10
The book is organised around method topics, not courses. There are sixteen topic chapters, an unnumbered method chooser, five reporting-template chapters, and five appendices.
Three reading modes
As an author with data. Open Find your method. Answer the question “what are you trying to do?” until you land on a method. Read the method page top to bottom. Adapt and check the displayed R code against your data, package versions and assumptions. The Output & Results section explains what to inspect. Reporting templates are starting points to complete with your actual design and results, not descriptions of analyses you have not performed.
As a peer reviewer. Search for the method by name and jump straight to the For Reviewers section at the bottom of the page. It links to the shared checklist below and to that method’s assumptions and practical tips. Detailed method-specific reviewer notes remain editorial work; they are not yet supplied by the imported entries. Read Theory and Assumptions before judging an application.
As a learner.
Each topic chapter has labs under chapters/<topic>/labs/. Labs follow
Hypothesis → Visualise → Assumptions → Conduct → Conclude. Work
through them in order; the methods they exercise live in the same
chapter and are linked from the labs.
Conventions
- Examples are R, displayed without execution during the reference build. All 80 lab scripts pass syntax checks; the maintenance review executed the tidy-data and import/join labs. Other examples and scientific outputs still need checking in their intended data and package environment.
- Method pages organize background, prerequisites, theory, assumptions, code, interpretation and practical tips. Their reviewer sections provide navigation to the relevant checks rather than completed specialist appraisals.
- The Reporting Templates part contains draft Methods and Results wording. Replace every example-specific statement with the analysis actually performed.
Shared reviewer checklist
- Match the analysis to the question and design. Identify the estimand, outcome scale, sampling unit and dependence structure. Compare them with the selected method’s Theory and Assumptions sections.
- Ask for evidence about relevant assumptions. Locate the diagnostics and sensitivity analyses needed for this application. A successful package call alone does not establish that the assumptions are met.
- Reconcile the reported analysis. Check sample counts and denominators, exclusions, missing-data handling, transformations and uncertainty summaries against the stated model and supplied analysis code.
- Check reproducibility and interpretation. Look for data provenance, package versions and sufficient code to reproduce the reported calculation. Distinguish the model’s output from the scientific claim made from it.
- Check the Methods paragraph. It should identify the question, analysis unit, model or test, relevant settings, handling of missing observations and diagnostic checks. Record unresolved details rather than assuming defaults.
This is a general reading aid. Method-specific validity and reporting requirements need an appraisal of the actual study and relevant primary methodological sources.
What this book is not
It is not a linear introductory textbook. It does not derive every result from first principles. It does not cover SAS, Stata, SPSS, or Python. It does not host the interactive decision-tree wizard (Shiny apps remain on the live site).
Building locally
Rscript --vanilla -e '
install.packages(
c("bookdown", "rmarkdown", "knitr", "bslib",
"downlit", "xml2", "htmltools", "sessioninfo"),
repos = "https://cloud.r-project.org"
)'
python3 -m unittest discover -s tests
Rscript --vanilla -e 'bookdown::render_book("index.Rmd", output_format="bookdown::bs4_book")'
python3 scripts/build-download-manifest.py
python3 -m http.server 8000 --directory docsRun these commands from the repository root and open http://localhost:8000. HTML is the primary reading format. README.md also documents the verified EPUB build and its link-finalization step. PDF requires a separate TeX qualification.