Measuring Residential Segregation

Abstract

The most common measures of segregation can only be used when block-level (or tract-level) statistics are available. The nineteenth-century manuscript census listings contain ordered listings of households, but (usually) no block informa tion. This article presents a method whereby the degree of residential segregation b y race can be measured from these listings, using the degree of departure from random ordering within the listings as the index. This index is based on the number of racially alike runs of adjacent households in the listings as a proportion of the number which would be expected, given random mterspersal of the two racial groups. The index is applied to datafor the city of Jacksonville, Florida, in 1870 and 1885.