Fig. 1.

Histograms for τ and h values observed in 50 metaanalyses of genetic associations from the candidate gene era (white boxes) (28), 10 prospective metaanalyses of genome-wide associations on type 2 diabetes and their replication efforts (black boxes) (29), and 31 metaanalyses of polymorphisms proposed to be associated with breast cancer through a two-stage genome-wide association study (gray boxes) (30). Data are shown separately for associations for which the summary effects have P < 0.05 (all candidate gene variants, all type 2 diabetes gene variants, and 11 of the breast cancer variants) (a) and for 20 proposed breast cancer gene variants for which the summary effects have P > 0.05 (i.e., not replicated even with conservative criteria) (b).