One observation corresponds to one person, with location randomized within their 2010 census block of residence. The files are broken up into CSVs by US metropolitan area. See accompanying crosswalk of CBSA codes to metropolitan area names for reference.