Additional considerations need to be taken into account when designing a replication study. Firstly, the effect size found in original studies involving many variants is likely to be biased upwards as it is dependent on reaching statistical significance and being published. This was first described by Beavis 34 in the context of quantitative trait loci, and has since been described in other settings as the ‘Winner’s curse’ 35;36. A study designed to replicate a finding should therefore base sample size calculations on a smaller effect size than found in the original study. Secondly, a comparison has to be made between the origin of the population in which the replication study is conducted, and that of the original study. A true replication study will involve analysis of the same polymorphism in the same direction of effect in the same (ethnic) population measured on the same phenotype as the original study. If another ethnic population is considered, the study is in essence no longer a replication study, since causal pathways and the relative contribution of polymorphism to these pathways may differ between
