Developing and validating new theories in gifted education is often limited by two factors: imprecise measurement and small numbers of subjects. The author asserts that the meta-analysis procedure offers gifted educators and researchers an opportunity to overcome these obstacles. He reviews the technique, responds to critiques of meta-analysis, and presents current research employing this procedure.One of the strengths of the physical and biological sciences is that observations taken over hundreds of years from various parts of the world and by separate researchers can be combined to develop and test scientific theories. In the behavioral and many of the social sciences and in gifted education this has not generally been true. In the almost 100 years of psychological science, upon which a great deal of the practice of gifted education rests, it has been difficult to combine research findings meaningfully. There, of course, has always been the review of the related literature, but a true, quantitative integration of prior results has been difficult. The &dquo;box score&dquo; method (Light and Smith, 1971) of summing the results of a set of studies was the usual method. In the &dquo;box score&dquo; method all of the studies in the literature that related to a given theory were reviewed, and the number with statistically significant results were counted as were the number with non-significant results. The outcome that was predominant &dquo;won&dquo;, and the result was incorporated into the theory under scrutiny. However, it is now well known that this &dquo;box score&dquo; method of reviewing the literature can be badly misleading (Hedges, 1985). Measurement in the behavioral sciences is rather imprecise, and in gifted education the number of subjects is usually small. These two factors, imprecision and small numbers of subjects, combine to insure that results are obscure and that statistical significance is difficult to obtain. Type II errors of statistical inference abound. (Type II errors are those where the statistical null hypothesis of no difference among the means of the groups or no relationship between the variables, is not true but it cannot be rejected.) Thus we are likely to report results from our research in gifted education that say we could not show that a new method of instructing, guiding, or grouping gifted children was significantly different from a current standard method. Experiments in education are important to determine cause-and-effect relationships between educational treatments and educational outcomes (Asher, 1976). Experiments include true experiments, where subjects are randomly assigned to different experimental and control groups for differing educational treatments; quasi-experiments, where intact groups are pre and post tested to assess possible differential gains as the results of differing educational treatments, and time series studies, where one group is given a treatment, later it is withheld, then given again, etc. to study the treatment effects on a criterion. Education is...