Session Number: 7b Session Title: Measurement of Segregation: New Directions and Results Session Organizers: a New Approach to Measuring Socio-spatial Economic Segregation

In this paper we propose two measures of income segregation, the rank-order information theory index (H R), and the rank-order variation ratio index (H R). We propose spatially-explicit versions of these indices as well. These indices have several appealing features that remedy flaws in existing measures. First, they are relatively easy to compute, since they require simply computing a series of pairwise segregation values using existing measures of segregation, fitting a polynomial regression line using WLS, and then computing a linear combination of the estimated parameters. Second, the measures are easily adapted to account for spatial proximity, following the approach of Reardon and O'Sullivan (2004). Third, the measures are largely insensitive to the set of income thresholds that define the categories in which income is reported, so long as the thresholds span most of the range of income percentiles. As a result, they do not require us to make assumptions about the shape of income distributions. Fourth, the measures are insensitive to rank-preserving changes in income, since the measures are based on the ranks of incomes rather than their numerical values. Finally, the indices can be interpreted in a variety of equivalent ways that illustrate their correspondence with standard notions of segregation.