A statistical procedure addressing the file-drawer problem by computing the number of unretrieved studies, averaging an effect size of .00, that would have to exist in file drawers before the overall results of a meta-analysis would not be significant.