Analytic Multiplicity in Epidemiology

A Reproducible Workflow in R

Preface

Empirical data analysis involves a large number of defensible analytic choices: model form, covariate selection, transformations, and approaches to missing data. These decisions can meaningfully influence statistical results, yet they are rarely explored or reported systematically. This book presents a reproducible, R-based workflow for examining that analytic multiplicity using multiverse analyses and vibration-of-effects frameworks.

The workflow is implemented in the companion R package forkflow. Using R, analysts can enumerate plausible analytic specifications, run large sets of models programmatically, and summarise the resulting distribution of estimates rather than relying on a single analytic pathway. Examples come from applied epidemiology and public health, where alternative analytic choices can produce substantially different effect estimates even when each specification is methodologically defensible.

The useR! 2026 talk Managing Analytic Multiplicity in Epidemiology is an excerpt from this book, drawing chiefly on Chapters 1, 6, and 8.

How to read this book

The early chapters motivate the problem and locate the analytic choices. The middle chapters build the workflow one verb at a time: enumerate, fit, summarise, visualise, report. The case study runs the whole pipeline on a public dataset, and the closing chapter covers practical guidance and limits.

Setup

pak::pak("acolum/forkflow")
library(forkflow)