In the September 1995 issue of the Archives, Elkis et al1 conclude their meta-analysis of ventricular enlargement in patients with mood disorders with the statement that the statistical power of meta-analysis has allowed them to uncover significant evidence for a greater degree of ventricular enlargement in patients with schizophrenia compared with patients with mood disorders.
However, their emphasis on the small P value associated with the finding may be misplaced, as they neglect to discuss in detail the associated importance of their fail-safe N=10 value for their comparison of 11 studies, which were pooled to provide an effect size of 0.2.
It may be worth recalling the importance of the failsafe N statistic given that Elkis et al actually calculate this value but then fail to fully interpret it. The fail-safe N may be defined as the number of new, unpublished or unretrieved, nonsignificant "null-result" studies that, on average,