Endophenotypes have offered the promise of assisting scientists in identifying genes for psychiatric disorders, so much so that Miller and Rockstroh (2013) characterized the time since the publication of Gottesman and Gould’s influential paper in 2003 as the “decade of the endophenotype.” Against this backdrop, our failure to find genome-wide significant SNPs is disappointing. Null findings in individual GWA studies are commonplace, however. Assuming a two-tailed test at p < 5 × 10−8 and an effective sample size of 2,790, based on an intraclass correlation of .31 from a simple linear mixed model analysis with a random family-level intercept, we had 80% power to detect effects accounting for 1.4% of the variance in our phenotypes (Gauderman & Morrison, 2006). This is larger than the typical effect size in GWAS findings for quantitative traits (Visscher, Brown, McCarthy, & Yang, 2012). Although requiring one to argue that failing to reject the null hypothesis constitutes positive evidence, our findings support a polygenic model of inheritance, in which complex traits reflect the additive influence of many SNPs, each with very small effect. If the
