Although meta-analysis has an aura of objectivity about it, in practice it is riddled with judgment calls. How do we decide which studies to include in our analysis? How far do we go to deal with the file drawer problem? Do we exclude results not reported in the English language? Do we need to weight effect size estimates by the quality of the study? How do we gauge the quality of studies? How do we deal with the apples and oranges issue? Unfortunately, many meta-analyses are done mechanically, with little attention given to these issues. This leads to reviews that are undermined by both Type I and Type II errors.