A fail-safe N for effect size in meta-analysis.

摘要: Rosenthan's (1979) concept of fail-safe N has thus far been applied to probability levels exclusively. This note introduces a TV for effect size. Rosenthal's was an ingenious response the so-called file drawer problem in research integration, that it provided direct assessment threat posed by sampling bias literature search. The algorithm derived from Stouffer method (Mosteller & Bush, 1954), which overall Z-score is computed summing individual Z-scores and dividing square root number scores. A significant Z-score, Rosenthal reasoned, could be made nonsignificant adding some knowable hypothetical studies averaged null results. exact such needed bring p level up critical (e.g., .05) since become known as (Nfs). formula