Null Hypotheses, Interval Estimation, and Bayesian Analysis

Abstract

Hypothesis testing using a frequentist approach is the mainstay of biostatistics and forms the foundation for assessing the significance of study results. This classical method has well-understood advantages as it determines whether data are statistically improbable and provides a threshold (ie, the P value) for delineating significance. Alternative statistical approaches have been proposed, including Bayesian analysis. This technique incorporates a prior probability as to what is already known clinically with the observed data. It is important for otolaryngologists to understand the advantages and disadvantages of these 2 approaches to conduct the most appropriate analyses.

Keywords: Bayesian analysis; frequentist approach; null hypothesis testing; prior probability; statistical significance.