Best (but oft forgotten) practices: sample size planning for powerful studies

https://doi.org/10.1093/ajcn/nqz058Get rights and content
Under an Elsevier user license
open archive

ABSTRACT

Given recent concerns regarding replicability and trustworthiness in several areas of science, it is vital to encourage researchers to conduct statistically rigorous studies. Achieving a high level of statistical power is one particularly important domain in which researchers can improve the quality and reproducibility of their studies. Although several factors influence statistical power, appropriate sample size planning is often under the control of the researcher and can result in powerful studies. However, the process of conducting sample size planning to achieve a specified level of desired statistical power is often complex and the literature can be difficult to navigate. This article aims to provide an approachable overview of statistical power and sample size planning, with emphasis on why statistical power is important for high-quality science. Thorough examples relevant to nutrition researchers are included to illustrate the process of sample size planning. Special consideration is also given to issues that may arise when conducting sample size planning in practice. The overarching goal is to provide nutrition researchers with the tools and expertise needed to conduct effective sample size planning for future studies.

Keywords:

sample size
statistical power
effect size
design
methodology

Abbreviations used:

AIPE
accuracy in parameter estimation
AJCN
American Journal of Clinical Nutrition
BUCSS
bias-uncertainty corrected sample size
cRCT
cluster randomized controlled trial
DE
design effect
FDR
false discovery rate
Ha
alternative hypothesis
H0
null hypothesis
HbA1c
glycated hemoglobin
ICC
intraclass correlation
MCP
multiple comparison procedure
QRP
questionable research practice
RCT
randomized controlled trial.

Cited by (0)