Introduction to the special section on metric in meta-analysis

A key issue for meta-analysts is how to represent study outcomes-in other words, what is the best metric for study outcomes? This introduction describes four articles in a special section on metric in meta-analysis. Do meta-analyses require standardized measures of study outcome? C. F. Bond, W. L. Wiitala, and F. D. Richard (2003) argue that standardization is often not needed and can obscure results. Another aspect of scaling outcomes is dichotomization. J. Sanchez-Meca, F. Marin-Martinez, and S. Chacon-Moscoso (2003) compare estimators appropriate when a meta-analysis includes treatment-control studies with dichotomized outcomes. Two articles deal with additional design complexities. R. Gillett (2003) describes mean-difference estimators, outlining indices comparable to effects from the t test. S. Olejnik and J. Algina (2003) provide a formula for generalized eta squared and omega squared statistics that are comparable across research designs.