As discussed above, few robust associations between specific gene variants and complex behaviors have been identified. It may well be that very large sample sizes on the order of hundreds of thousands are required to provide sufficient statistical power. Even then, the gains may not be that great. A recent GWAS was able to identify three independent SNPs at genome-wide significance (p = 1 × 10−9) influencing educational attainment, a moderately heritable trait, in a discovery sample of 101,069 individuals and this finding was replicated in a sample of 25,490. However, the effect sizes for the SNPs were small (0.02%) and these three SNPs together accounted for only 2% of the variance in both educational attainment and cognitive function [80]. A GWAS in a community sample of Australian alcoholics also estimated that the effect size for SNPs was very small, each SNP contributing at most to 0.25% of the variance in AD risk [81].
