Another way to quantify the bias arising from the incomplete representation of relevant research is to calculate the "fail-safe N." The fail-safe N is the minimum number of additional studies with conflicting evidence thatwould be needed to overturn the conclusion reached in the review. Conflicting evidence is usually defined as a null result. If the meta-analysis has generated a statistically significant finding, the fail-safe N is the number of excluded studies averaging null results that would be needed to render that finding nonsignificant (Rosenthal 1979). The fail-safe N is directly related to the size of the effect and the number of studies (k) combined to estimate it in the meta-analysis. For example, if the results of fourteen studies were combined to yield a statistically significant mean effect size of r = .15, p = .018, it would require the addition of only nine further studies averaging a null effect to render this result statistically nonsignificant. If we could accept the possibility that there are at least nine "no effect" results buried in filing cabinets or published in obscure non-English journals, then we should be skeptical of the meta-analytic conclusion. However, if the fourteen studies returned a mean effect size of r = .30, then the fail-safe number would be a much higher seventy-eight studies. Thus, the fail-safe N describes the tolerance level of a result. The larger the N, the more tolerant the result will be of excluded null results.