Fig. 1
From: Combining controls can improve power in two-stage association studies
Diagram of methods a, b, and c. Method b differs by comparing \(C_{1}^{\prime }\) to pooled C0 and \(C_{0}^{\prime }\) at the replication stage to generate p-value Ps instead of Pr. Method c also pools controls at the discovery phase, comparing C1 to pooled C0 and \(C_{0}^{\prime }\) to generate p-values Pc instead of Pd. A ‘hit’ is declared in method a if Pd<α, Pr<β, Pm<γ, in method b if Pd<α, Ps<β∗, Pm<γ and in method c if Pc<α, Ps<β⊥, Pm<γ