Crespi Catherine M. Power and Sample Size in R

Chapman and Hall/CRC, 2025. — 354 p. — (Chapman & Hall/CRC Biostatistics Series). — ISBN 978-0-429-48878-8.Power and Sample Size in R guides the reader through power and sample size calculations for a wide variety of study outcomes and designs and illustrates their implementation in R software. It is designed to be used as a learning tool for students as well as a resource for experienced statisticians and investigators.The book begins by explaining the process of power calculation step by step at an introductory level and then builds to increasingly complex and varied topics. For each type of study design, the information needed to perform a calculation and the factors that affect power are explained. Concepts are explained with statistical rigor but made accessible through intuition and examples. Practical advice for performing sample size and power calculations for real studies is given throughout.The book demonstrates calculations in R. It is integrated with the companion R package powertools and also draws on and summarizes the capabilities of other R packages. Only a basic proficiency in R is assumed.Topics include comparison of group means and proportions; ANOVA, including multiple comparisons; power for confidence intervals; multistage designs; linear, logistic and Poisson regression; crossover studies; multicenter, cluster randomized and stepped wedge designs; and time to event outcomes. Chapters are also devoted to designing noninferiority, superiority by a margin and equivalence studies and handling multiple primary endpoints.By emphasizing statistical thinking about the factors that influence power for different study designs and outcomes as well as providing R code, this book equips the reader with the knowledge and tools to perform their own calculations with confidence.Key Features:
Explains power and sample size calculation for a wide variety of study designs and outcomes
Suitable for both students and experienced researchers
Highlights key factors influencing power and provides practical tips for designing real studies
Includes extensive examples with R codePreamble
List of Figures
List of Tables
Preliminaries
R implementation
Probability distributions
Notation for common distributions
R functions for common distributions
Symmetric property of the normal distribution
Standardizing a normal distribution
Getting started: a first calculation
Steps in a sample size calculation
Hypothesis testing
A first calculation: one-sample z test
Effect size
Minimum detectable effect size
A general formula when the test statistic is normally distributed
R function for z tests
Sample size adjustments
Sensitivity analysis
Estimating power using simulation
Should I conduct a power analysis after my study is completed?
One or two means
One-sample t test
Two independent samples t test
Relative efficiency
Lognormal data
Paired t test
Remarks on R functions for t tests
Nonparametric tests of location
Hypotheses for different study objectives
Introduction
Test for nonequality
Test for superiority
Test for noninferiority
Test for superiority by a margin
Test for equivalence
Hypotheses when a lower mean corresponds to a better outcome
Remarks
Analysis of variance for comparing means
Introduction
One-way analysis of variance
Two-way analysis of variance
Analysis of covariance
Additional resources
Proportions: large sample methods
Preliminaries
One-sample proportion test
Test of two independent proportions
Test for two correlated proportions
Exact methods for proportions
One proportion: exact binomial test
Two-stage designs for single arm trials
Two proportions: Fisher exact tes
Two correlated proportions: exact test
Categorical variables
Chi-square goodness-of-fit test
Chi-square test of independence
Chi-square test for comparing two proportions
Ordinal categorical responses
Additional resources
Precision and confidence intervals
Introduction
Confidence intervals for means
Confidence intervals for proportions
Confidence intervals for relative risk
Confidence intervals for odds ratio
Additional resources
Correlation and linear regression
Pearson correlation coefficient
Simple linear regression
Multiple linear regression
Generalized linear regression
Power for generalized linear models
Logistic regression
Poisson regression
Additional resources
Crossover studies
Introduction
2 × 2 crossover design
(2 × 2)r crossover design
Efficiency of crossover designs
Additional resources
Multisite trials
Introduction
Multilevel data structure
Considerations for multisite trials
Model for continuous outcomes
Intraclass correlation coefficient
Power for test of average treatment effect
Power for test of heterogeneity of treatment effect
Binary outcomes
Additional resources
Cluster randomized trials: parallel designs
Introduction
Continuous outcomes
Binary outcomes
Additional resources for parallel cluster randomized trials
ndividually randomized group treatment trials
Other multilevel trial designs
Cluster randomized trials: longitudinal designs
Introduction
Modeling framework for continuous outcomes
Parallel cluster randomized trial with baseline measurement
Cluster randomized crossover designs
Stepped wedge designs
Time to event outcomes
Introduction
Concepts for time to event studies
Logrank test
Tests based on the Kaplan-Meier estimator
Distributions for survival, accrual and loss to follow up
Additional resources
Multiple primary endpoints
Introduction
Model
Co-primary endpoints
Alternative primary endpoints
Additional resources
Bibliography
Index