Handbook of Biostatistics for Medical Research
Five chapters on the methodological surface of clinical and health-services research.
What this handbook is
This is the methodology reference I use in my own work and in client engagements. Each chapter is a structured guide (orientation, decision framework, worked examples drawn from the portfolio, further reading) covering one of the five surfaces a medical researcher, small company, or nonprofit needs to navigate when their work has to hold up under review.
How it works
The whole handbook is free. Each chapter is a structured methods reference on this site: orientation, a decision framework, worked examples drawn from the portfolio, and a curated reading list. Enough to navigate the decision for yourself.
If you’d rather take it in over time than read it all at once, The Confounder delivers the same methodological spine to your inbox, one piece at a time, alongside shorter dispatches on new research and methods. It is free, roughly every other week.
What The Confounder covers
Each issue is one of two things: a methods explainer drawn from these chapters, or a short dispatch on something timely.
- The explainers walk the spine: stating the question, defining populations and sample size, choosing a causal design, defending it with sensitivity analysis, and simulating when the formulas run out.
- The dispatches react to new papers, guidelines, and the methods debates worth following, and live on the site as dated posts.
- Reader questions shape what gets written.
Submitting a question
Methodology questions are welcome from any reader. Send yours to pauline@informedica.llc with “Confounder question” in the subject line. A few are selected for upcoming issues.
A few notes on the questions worth submitting:
- General methodology questions are the right fit: how to operationalize a study design, when to use which causal-inference approach, how to defend a sensitivity choice, what a reviewer-response should look like.
- Study-specific advice (your specific dataset, your specific protocol, your specific reviewer dispute) is more efficiently handled in a discovery call than in a Q&A column.
- Questions are answered anonymously by default. If you would prefer attribution (first name and role, e.g., “asked by a third-year resident”), say so in your email.
A standing note: The Confounder is general guidance, not consulting advice. For specific applications to your study, the discovery-call route is the right one.
Chapters
Chapter 1
From research question to study design
Research-question framing, PICO/PECO specification, target trial emulation, primary-vs-secondary endpoint logic, pre-registration. Move a clinical hypothesis to a defensible statistical study design.
Chapter 2
Defining populations and computing sample sizes
Inclusion and exclusion criteria, target-vs-study population, generalizability, power and sample-size calculation (closed-form and simulation-based), minimum-detectable-effect framing.
Chapter 3
Causal inference: a methods toolkit
Difference-in-differences, regression discontinuity, instrumental variables, synthetic control, propensity-score methods. When each applies, what each assumes, how to defend it.
Chapter 4
Sensitivity analysis and robustness as design defenses
Placebo design, falsification tests, E-values, Rosenbaum bounds, leave-one-out diagnostics, multiple imputation, missing-data sensitivity.
Chapter 5
Monte Carlo simulation for biostatistics
When to simulate, simulation-based sample size, bias quantification, type-I error rate calibration, parametric and non-parametric resampling.
Get the methods by email
The handbook is free to read here. The Confounder delivers it to your inbox over time, plus dispatches on new research and methods. Free, roughly every other week.