Residential interracial exposure and isolation indices: Mean versus median indices, and the difference it makes

Interracial exposure and isolation (p*) indices have been widely used in studies of residential racial segregation. However, a recent pilot study raised serious issues about the use of these indices, because they are based on the mean statistic, which may yield misleading results in the case of skewed distributions, as is often the case with census tract racial compositions. An alternative median exposure index (p*-md) is proposed, and mean and median indices of white-to-Aftican-American and African-American-to-white exposure, as well as white and African-American isolation, are compared for the 50 largest metropolitan areas in the United States. The analysis shows that the mean and median measures produce different results, and that most of these differences are maximized in those areas that are most segregated and, especially for African-American-to-white exposure, in areas where the largest number of African Americans live. This creates significant problems in the interpretation and use of mean exposure and isolation indices, and in most cases, the median index yields a result more representative of the residential neighborhood situation of the majority of whites and African Americans. A particular problem with the mean exposure indices currently in common use is their tendency to overstate the degree of neighborhood-level interracial contact in U.S. metropolitan areas, and, in so doing, to understate the impact of racial housing discrimination.