Meta-Analysis

This chapter provides a basic methodological guide to the conduct of meta-analyses, and it reviews the tools used for the purpose. It considers the specific features of empirical econometric analysis, and it cites as examples its application in policy/program evaluation. A salient feature of empirical studies in economics is that estimated effects mainly derive from econometric techniques based on multiple regressions on large samples, using different model specifications, datasets, identification strategies, and definitions of the variables of primary interest. A second salient feature is that, in economics, the analyst typically runs many regressions and the published results are severely selected from those of a larger set of econometric models. This may give rise to marked data dredging and publication bias. Accordingly, this chapter describes modern meta-analytic methods with which to derive generalized conclusions from heterogeneous results and to detect and correct for publication bias.