The significance of treatment effects across several independent studies

Standardized means can be generated from the t, F or chi 2 statistic used to test the significance of treatment effects in each of several independent studies. An unweighted means ANOVA, calculated from the standardized means, provides tests of significance for the treatment main effect across studies and for the differences in magnitudes of the treatment effects between studies. The protection against Type I errors (false positive conclusions) afforded by the standardized means ANOVA is evaluated in a series of Monte Carlo simulations involving both equal and unequal sample sizes and error variances. A detailed illustration of the method is provided. The combining of information from multiple independent trials in clinical psychopharmacology research is illustrated with regard to problems of establishing the general equivalence of two drugs, as well as in providing a comprehensive conclusion concerning efficacy across several studies that may not appear uniformly positive.