Abstract
Statistical inference is the process of using data to draw conclusions about unknown quantities. Statistical inference plays a large role both in designing clinical trials and in analyzing the resulting data. The two main schools of inference, frequentist and Bayesian, differ in how they estimate and quantify uncertainty in unknown quantities. Typically, Bayesian methods have clearer interpretation at the cost of specifying additional assumptions about the unknown quantities. This chapter reviews the philosophy behind these two frameworks including concepts such as p-values, Type I and Type II errors, confidence intervals, credible intervals, prior distributions, posterior distributions, and Bayes factors. Application of these ideas to various clinical trial designs including 3 + 3, Simon’s two-stage, interim safety and efficacy monitoring, basket, umbrella, and platform drug trials is discussed. Recent developments in computing power and statistical software now enable wide access to many novel trial designs with operating characteristics superior to classical methods.
Original language | English (US) |
---|---|
Title of host publication | Principles and Practice of Clinical Trials |
Publisher | Springer International Publishing |
Pages | 973-1002 |
Number of pages | 30 |
ISBN (Electronic) | 9783319526362 |
ISBN (Print) | 9783319526355 |
DOIs | |
State | Published - Jan 1 2022 |
Keywords
- Bayes factors
- Bayesian
- Confidence intervals
- Credible intervals
- Frequentist
- Hypothesis testing
- P-value
- Prior distribution
- Sequential designs
- Statistical inference
ASJC Scopus subject areas
- General Medicine
- General Mathematics