Reporting of Effect Size and Confidence Intervals: Review and Methods of Calculation

Abstract

Despite continuous criticism, significance tests remain the main procedure for statistical inference in psychology. In order to avoid some of the problems associated with them, it has been argued that authors should report effect sizes and confidence intervals as supplemental methods of data analysis. The present article discusses arguments in favor of reporting effect sizes and confidence intervals, and investigates how common such practices are by reviewing articles published in Bulgarian psychology journals. The results show that the majority of articles still don’t report effect sizes and that very few articles report confidence intervals. The article then outlines different measures of effect size and methods for their calculation. It also presents methods for calculating confidence intervals for effect sizes. While it is ultimately the authors’ responsibility to report them, it is argued that stronger editorial policies and actively teaching these concepts to psychology students can encourage psychologists to change the way they analyze their data.

Language: Bulgarian