This project consists of three components which continue from last year: The first component evaluates the practice of meta-analysis in the general medical literature to identify key areas where the methods of analysis are not valid and or the results are misleading. Based on the empirical evidence, this component develops standards to safeguard against the misuse of meta-analysis in practice. The second component compares parametric and non-parametric tests to assess the assumption of homogeneity of effects in data from a series of trials and develops more refined and more efficient tests to detect heterogeneity when it is not very large. This component considers data from Gaussian, binomial, and Poisson sampling distributions. Preliminary results from this component are submitted to Statistics in Medicine. The third component addresses issues that pertain to the use of meta-analysis in the design and monitoring of clinical trials and the application of meta-analysis to epidemiologic data. Preliminary results from this component appeared in Statistics in Medicine 15:1237-1248 (1996) and Chance 10:12-14, 1997.
