A new paper published in eLife by Drs Robin Ince, Jim Kay, Angus Paton and Prof Philippe Schyns
Published: 3 March 2021
Bayesian inference of population prevalence
A new paper published in eLife by Drs Robin Ince, Angus Paton, Jim Kay and Prof Philippe Schyns: "Bayesian inference of population prevalence".
The paper develops a new statistical approach for psychology and neuroimaging which is based on evaluating effects within individual participants, viewing different participants as independent replications of the experiment and quantifying the population level replication probability (the prevalence) using Bayesian statistics.
It argues this may often provide a better match to researchers' questions (is this a reliable effect in the brains / behaviour of individuals?) than typical binary inference on the population mean (is the population mean of this effect different from zero, assuming a normal distribution?) and could improve robustness and replicability of results. See link here.
First published: 3 March 2021