Index of Dissimilarity

The term index of dissimilarity refers to a standard measure of residential segregation, which gauges the extent to which two groups are evenly spread throughout neighborhoods in a given geographic area, usually a city or metropolitan area. It is interpreted as the percentage of either of the two groups that would have to move in order to achieve an even distribution across neighborhoods within the larger geographic unit of interest. An even distribution indicates that all of the neighborhoods would have the same distribution of the two groups as the larger, geographic unit of interest. Values of the index of dissimilarity range from 0 to 100, with values ranging from 0 to 30 indicating low levels of segregation or high levels of evenness, from 30 to 60 indicating moderate levels of segregation, and 60 or more indicating high levels of segregation or very low levels of evenness.

Results from Census 2000 reveal high levels of segregation for blacks and more moderate levels for Hispanics and Asians and Pacific Islanders, relative to non-Hispanic whites. More specifically, 64 percent of blacks, 51 percent of Hispanics, and 43 percent of Asians and Pacific Islanders would have to move to achieve an even distribution with whites, on average, in the metropolitan United States. Trends over time in index of dissimilarity scores reveal that blacks consistently are the most segregated group. In 1980, the dissimilarity scores for blacks, Hispanics, and Asians and Pacific Islanders were 73 percent, 50 percent, and 41 percent, respectively, relative to non-Hispanic whites. Although scores for blacks have declined over time, the level of segregation still remains within the high range, even nearly 40 years after the passage of the Fair Housing Act. Scores for Hispanics and Asians have remained relatively stable, in large part as a result of the trends in immigration of these groups.

Although the index of dissimilarity is a standard measure used to gauge levels of residential segregation, it is not free of limitations. One major limitation is that it can only examine the residential segregation of two groups at a time. Other measures, such as the multigroup information theory index, can measure the evenness of more than two groups. Another is that it is sensitive to the level of geography used as its building blocks. In general, census tract-level data are used as a proxy for the neighborhoods within which evenness is examined. However, an average census tract contains 4,000 residents. Census block groups are smaller areas of geography that are composed, on average, of 1,500 people. Studies have shown that when smaller areas of geography are used to calculate the index of dissimilarity, levels of segregation are actually higher. Other limitations with the index are that it cannot gauge how concentrated or centralized the populations of interest are. Nevertheless, because of its ease of interpretation and its detection of evenness, the index of dissimilarity remains a core measure used in the residential segregation literature.