This project comprises three components: The first component considers two methods in the literature for combining data from a series of comparative trials. One method assumes a fixed effect while the other assigns a random effect to each trial at hand. The study evaluates the performance of each method under various heterogeneity assumptions and with several scales of measurement, including the risk difference and the odds ratio as measures of effect. The second component extends and adapts the meta-analytic methods appropriate for clinical trials to combining the evidence from a series of epidemiological studies (both case-control and cohort designs). Partial results from this section of the project are to appear in Journal of Clinical Epidemiology (in press). The third component develops and compares parametric and non-parametric tests to assess the assumption of homogeneity of effects in data from a series of trials. The study considers data from Gaussian, binomial, and poison sampling distributions. The tests under comparison include the C-alpha test of homogeneity, asymptotic maximum likelihood methods, non-parametric maximum likelihood methods, standard chi-squared tests of homogeneity, score and likelihood ratio tests.
