Eliciting and formalizing experts’ knowledge about effect sizes

Abstract

Experts’ knowledge is a valuable source of information for statistical inference, when data are not available, or predictions are needed. However, elicitation needs to be structured to avoid, or at least to minimize, possible biases. We propose a Shiny-App that aims to formalize experts’ judgments about effect-size for the comparison between two means (i.e., Cohen’s d). This is useful in psychology as many studies consider groups differences on specific dimensions using standardized measures. In the study we will evaluate the reliability of this procedure using as an example the difference between girls’ and boys’ average height at different ages. Future applications could be related to priors formalization in Bayesian analysis or power analysis considering experts’ judgement about plausible effect sizes.

Claudio Zandonella Callegher

Post-Doc Researcher

My research interests include the formalization of psychological thoeries, Bayesian methods in Behavioral Sciences, and everything related to programming in R!