American Community Survey (ACS) Data Uncertainty and the Analysis of Segregation Dynamics

Abstract American Community Survey (ACS) data have become the workhorse for empirical analysis of segregation in U.S.A. during past decade. The increased frequency ACS offers over 10-year Census, which is main reason its popularity, comes with an level uncertainty published estimates due to reduced sampling ratio (1:40 households) relative Census (1:6 households). This paper introduces a new approach integrate into segregation. Our method relies on variance replicate 5-year and advances existing approaches by explicitly taking account covariance between when developing distributions indices. We illustrate our study comparative dynamics 29 metropolitan statistical areas California, using 2010–2014 2015–2019. methods yield different results than simulation technique described Napierala Denton (Demography 54(1):285–309, 2017). Taking estimate yields larger error margins those generated simulated number census tracts large minority percentage low, converse true small high.