Aim:Country-of-birth data contained in registers are often aggregated to create broad ancestry group categories. We examine how measures of residential segregation vary according to levels of aggregation.

Method:We use Swedish register data to calculate pairwise dissimilarity indices from 1990 to 2012 for ancestry groups defined at four nested levels of aggregation: (1) micro-groups containing 50 categories, (2) meso-groups containing 16 categories, (3) macro-groups containing six categories and (4) a broad Western/non-Western binary.

Results:We find variation in segregation levels between ancestry groups that is obscured by data aggregation.

Conclusions:This study demonstrates that the practice of aggregating country-of-birth statistics in register data can hinder the ability to identify highly segregated groups and therefore design effective policy to remedy both intergroup and intergenerational inequalities.