A novel application of the file drawer formula: rejected study bias in meta-analysis.

The file drawer formula in meta-analysis estimates the number of null difference effect sizes (ESs) from unpublished studies in researcher file drawers that are necessary to reduce the mean ES to a magnitude no longer considered practical or meaningful. Unlike the file drawer concept, this paper proposes a novel use of Orwin's 1983 formula that produces an estimate of the potential bias in the mean ES due to missing ESs from studies that were located and retrieved, but rejected because of insufficient data for ES calculation.