Table 1 showed that African Americans comprise just fewer than 12 percent of the population in the United States, while Afro-Caribbeans and Africans account for one-half of one percent and two-tenths of one percent, respectively. Thus, if these non-Hispanic black groups were distributed randomly (without regard to in-group preferences or discrimination), their isolation index values would be about 12, 0.5, and 0.2, respectively. Likewise, the exposure scores would match the population percentages of non-Hispanic blacks and whites in Table 1. Dissimilarity is a measure of evenness and thus captures how equally members of a given group are distributed across tracts compared to another group. A dissimilarity score less than 30 is generally thought to be indicative of low segregation, scores between 30 and 55 are moderate, and scores above 55 indicate high segregation.
Table 5 shows that exposure to whites is low and declining for each black group. Africans have the highest exposure to whites (in 2000, just under half of the people in the neighborhood where an average African person lived was white); Afro-Caribbeans now have the lowest exposure to whites (29.9 percent) among black groups. Conversely, dissimilarity scores indicate high, though slightly declining, segregation of all non-Hispanic black groups from whites. All dissimilarity scores from non-Hispanic whites are in excess of 60 percent.
The percent of African Americans in the neighborhood where an average African American person lives declined from 54.3 percent in 1990 to 49.4 percent in 2000. Because of their smaller size, other black groups have much lower isolation scores, though these were on an upward trajectory in the 1990s. While they live in neighborhoods where their own group tends to be a small minority, Afro-Caribbeans neighborhoods are on average close to 50% black. Africans, on the other hand, live in neighborhoods where blacks are outnumbered by whites. (though they did increase their exposure to blacks from 23.3% to 28.3%.)
Segregation of black groups from one another, as measured by the Index of Dissimilarity, is declining, but it is strikingly high. (Note that these average values are not symmetrical, because the average segregation of group X from group Y is weighted by the number of group X residents; segregation of group Y from group X is weighted by the number of group Y residents.) We caution, however, that because Africans and Afro-Caribbeans are found in very small numbers in many metropolitan areas, the national averages include many values for metro areas where the score is unreliable. It will be more revealing to assess dissimilarity scores among black groups in places like New York, Washington, D.C., and Atlanta where all three are found in larger numbers. We do this in Tables 6 and 7.