In a project of such a scale, the need for adequate data curation is paramount. Several measures were assessed to determine whether structures had values in the expected range for volume, hemispheric asymmetry, etc. In the end, the large number of contributing authors and contributing sites made it easier to identify sites that had outliers in their genotypic or brain measurement data, partly because normative data were becoming available from all the other sites. All relevant demographic factors were controlled for at each site, including population structure that might lead to ancestry differences masquerading as true genetic associations to brain traits.
