Of course, the present investigation suffers from limitations. One is the number of electrodes used for recording P3 amplitude—one, or three in the case of the common factor approach. Although P3 amplitude is typically greatest at the site we used (Pz) and much of the relevant research establishing its status as an endophenotype has also used this site, several recent positive findings for individual SNPs or candidate genes, whether of P3 amplitude or event-related theta activity, have been for recordings over frontal brain regions. Our use of two different age cohorts, although allowing us to maximize sample size, may have obscured true effects that are expressed differently in late adolescence compared to adulthood. Even with the cohorts combined, the sample was small by current GWAS standards, if not when this project was first undertaken. We did not use bioinformatic methods that are designed to use additional information, such as knowledge about biological pathways and gene expression, to mine the p values produced by GWAS for patterns, or methods for selecting subsets of SNPs, although the approach we adopted is a reasonable starting point.
