Bayesian inference of population prevalence

Area of Science:

• Neuroscience
• Psychology
• Neuroimaging

Background:

• Null hypothesis significance testing (NHST) of the population mean is prevalent in neuroscience, psychology, and neuroimaging.
• Current methods often lack population-level inference, especially in studies with small sample sizes.
• Replicability is a significant challenge in these scientific fields.

Purpose of the Study:

• To propose a novel Bayesian method for estimating population prevalence of effects.
• To offer advantages over traditional population mean NHST.
• To provide a quantitative population-level inference for studies with limited participant numbers.

Main Methods:

• Developed a Bayesian approach to estimate population prevalence based on individual participant NHST.
• Applied the method to address limitations in small-sample studies like psychophysics and precision imaging.

Main Results:

• The Bayesian prevalence method provides a quantitative population estimate with associated uncertainty.
• This approach moves beyond binary inferences common in NHST.
• The method is broadly applicable across neuroscience, psychology, and neuroimaging.

Conclusions:

• Bayesian prevalence offers a robust alternative to population mean NHST.
• The method enhances population-level inference, particularly for small-sample studies.
• Focusing on individual effects aids in addressing replicability issues in scientific research.