Meta-Analysis of Single-Case Research
Myles S. Faith David B. Allison Obesity Research Center St. Luke's/Roosevelt Hospital Center Columbia University College of Physicians and Surgeons
Bernard S. Gorman Hofstra University
Meta-analysis is a collection of methods designed to quantitatively summarize the results of separate studies. Unlike previous "box-score" techniques that simply tabulated the number of research studies passing some criterion (such as statistical significance or another a priori criterion of improvement), meta-analysis employs quantitative measures of the magnitude of effect of each study.
Meta-analyses generally have three basic goals. First, meta-analytic studies strive to provide a point estimate of the average effect size; that is, an overall, quantitative summary. The need for a point estimate not withstanding, it has been said that "A statistician is someone who can drown in a stream whose average depth is six inches," implying that an average can obscure a great deal of variability. Obviously, it is also desirable to establish boundaries around a point estimate. Therefore, as a second goal, meta-analyses strive to provide confidence intervals in which the "true" population effect size is likely to be found. Given confidence intervals, one could easily establish whether the average effect size is significantly different from a null value. Finally, if there is substantial variability around an average effect size, then meta-analytic techniques also permit the researcher to search for variables that moder-