[Wiley Series in Probability and Statistics] Statistical Meta-Analysis with Applications || Various Measures of Effect Size

✍ Scribed by Hartung, Joachim; Knapp, Guido; Sinha, Bimal K.

• Publisher: John Wiley & Sons, Inc.
• Year: 2008
• Tongue: English
• Weight: 574 KB
• Edition: 1
• Category: Article
• ISBN: 0470290897
• DOI: 10.1002/9780470386347.ch2

✦ Synopsis

Quite often the main objective in a study is to compare two treatments: experimental and control. When these treatments are applied to a set of experimental units, the outcomes can be of two types, qualitative and quantitative, leading to either proportions or means. Accordingly, effect sizes are also essentially of these two types: those based on differences of two means and those based on differences of two proportions. A third type of effect size, namely, correlation, arises when the objective in a study is to ascertain the nature and extent of the relationship between two variables.

2.1 EFFECT SIZE BASED ON MEANS

An effect size based on means is defined as follows. Denote the population means of the two groups (experimental and control) by p1 and pz and their variances by u! and oi, respectively. Then the effect size Q based on means is a standardized difference between and p2 and can be expressed as U