these 416 cases are failed replications, they constitute the maximum number of biased reports that we could have not included in our database (Table S8). Under this extreme scenario, the 103 positive replications would not have been drawn from the 225 replication attempts gathered in our database, but from a larger set of 641 ( = 225+416) replication attempts in East Asians. This calculation allows us to estimate the lower-bound replicability rate at 16.1% (103 out of 641), which still departs from the null expectation of 5% (P = 10−16; Binomial test). In other words, the figure of 416 ungathered replication attempts is about an order of magnitude lower than the total number of unreported cases needed to explain all East Asian replications as type I errors (4,017, see previous paragraph) and, therefore, it is very unlikely that systematic reporting bias accounts for our results.
