A quick, easy, and imperfect fix for using the Binomial Effect Size Display for correlations with continuous variables: divide by three

Read full text Summarise with VIRAT
Authors: Denis Lajoie
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The Binomial Effect Size Display (BESD; Rosenthal & Rubin, 1982) is a widely used heuristic for translating Pearson's r into a proportion of concordant pairs, yet it is known to systematically overestimate this proportion when applied to continuous variables. While Dunlap's (1994) CLr statistic correctly estimates the proportion of concordant pairs under normality, its reliance on an arcsine transformation limits its utility as a mental shorthand. This note proposes BESDc, a simple approximation of CLr obtained by dividing r by 3 rather than 2 before adding .5. Across the range of correlations typically observed in psychological research (r < .7), BESDc departs from CLr by less than .01, substantially outperforming the BESD. A practical decision tree is recommended: reporting the observed proportion of concordant pairs from raw data where possible, using CLr with a calculator when raw data are unavailable, and applying BESDc as a rapid mental approximation when neither option is convenient. BESDc will underestimate effect sizes for r > .7 and is not recommended for r > .9, where underestimation becomes larger than .05.

Fields of Study & Subjects

Meta-analysis and systematic reviews Psychology Research and Bibliometrics Psychometric Methodologies and Testing

Publication details

Journal / source
PsyArXiv (OSF Preprints)
Publication date
28 March 2026
Language
EN
Document type
Preprint