Racial Residential Segregation Measurement Project

U of M

    The Geographic Areas We Used

The Geographic Areas We Used


            To measure racial residential segregation, you need to make four decisions:


            Which one race or two are of interest to me?

            What is the largest geographic area of interest to me?  Is it a city, a county or a metropolis?

            What component parts of the larger geographic area are to be used in calculating measures of segregation?

            Which measure of segregation best satisfies my needs: the index of dissimilarity, an isolation measure or an exposure or interaction measure?


Geographic Units Used in the Census of 2000


Geographic Category

Number of Units in Category

Average Population




Entire Nation












Metropolitan Areas



Congressional Districts






Census Tracts



Block Groups











            Large Geographic Areas for Which Segregation Measures are Available


This website provides you will the ability to obtain racial residential segregation scores for: Entire states

                    Each of the 331 metropolitan areas

                    All central cities and other cities with 100,000 or more residents in 2000


Component Geographic Units Used in the Calculation of Segregation Measures


            For purposes of making sure that all housing units are enumerated in the decennial count, the Census Bureau designates census tracts.  These are contiguous geographic areas containing an average of 4200 residents.  They may be thought of as neighborhoods but they were delineated for purposes of census taking so they may or may not correspond to what residents think of as neighborhoods. 


            For enumeration purposes, the Census Bureau also delineates Block Groups made up of contiguous clusters of blocks.  These contained an average of about 1,300 residents in 2000.  Finally, the Census Bureau presented findings for blocks.  In metropolitan areas, these are typically city blocks bounded by four streets but they might be gore shaped areas bounded by three streets or might have a river or railroad as a boundary.  On average, blocks in 2000 contained just 34 residents.


How to Select a Component Geographic Area?


            The numerical value of segregation measures depends upon the geographic units used in their calculations.  A census tract might contain some blocks that are exclusively white while other blocks are exclusively black.  If this is a common occurrence in a metropolis, than a measure of segregation calculated from census tract data will be numerically larger than a measure of segregation calculated from block data.


            This website gives you your own choice of geographic units.  Traditionally, most segregation scores have been calculated using census tract data, partly because of the availability of such data and because it was simpler and faster to use fewer geographic units in the pre-computer days.  But census tracts are generally larger than the neighborhoods that most people have in mind when they think about residential segregation since most census tracts are home to more than 4,000 people.


            This implies that block groups are highly appropriate for the calculation of segregation measures.


            This website provides you with information about the number of census tracts, block groups and blocks that make up the larger geographic area for which segregation measures are calculated.


An Important Note of Caution


            The index of dissimilarity is statistically independent of the relative size of the two groups used in its calculation.  However, its value may be distorted if the number of geographic units used in its calculation exceeds the number of people in one of the two racial groups whose distributions are compared.  If the geographic distributions of blacks are whites are compared but the analysis is done with block data for a metropolis in which there are 15,000 blocks but only 3,000 blacks; there will necessarily be some block with no African American residents.  For that reason, indexes of dissimilarity must be interpreted with great caution if the size of one of the racial groups using in the calculation is less than five times the number of areal units used in the same calculation.  This implies that indexes of dissimilarity calculated from census tracts or block groups are often more valuable and accurate descriptions of racial residential segregation than indexes of dissimilarity calculated from block data.


            The isolation index and the exposure or interaction measures of segregation are not confounded in situations where one racial group is smaller than the number of areal units used in the calculation.


            Some counties in 2000 had very small populations so they included only one census tract or one block group.  In such counties, indexes of racial residential segregation could not be calculated since every person in the county lived in the same geographic unit.


Return to Home Page.