
Measures of the degree of residential separation of subgroups within a wider population. The development of meaningful indices of segregation has been fundamental to the study of social stratification and residential differentiation in urban areas (see also factorial ecology; social area analysis).
A simple graphical method of showing segregation is the Lorenz curve. The figure shows curves for several ethnic groups in Great Britain in 1991. The x axis indicates the cumulative percentage of each ethnic group and the y axis the cumulative percentage of the total population over the districts (in this case wards) into which the country has been divided. A diagonal line indicates no segregation â€” i.e. the percentage of a group\'s population within each subarea is absolutely consistent with its percentage of the city population as a whole. Normally, the segregation line is a curve whose distance from the diagonal indicates the degree of segregation. The figure shows that at least 60 per cent of all of the ethnicminority groups are found in wards that contain only 30 per cent of the total population and that for the Indians, black Caribbeans, black Africans, Pakistanis and Bangladeshis about 70 per cent are found in wards which contain less than 10 per cent of the total population (Peach, 1996, p. 220). (The Lorenz curve is widely used in other contexts as a general measure of inequalities in distributions.)
{img src=show_image.php?name=bkhumgeofig31.gif }
indices of segregation
Most studies of segregation have used one of two simple indices to summarize differences between two spatial distributions. The indices vary from 0100 and indicate the percentage redistribution necessary before two groups are similarly distributed over a set of districts. First, the index of residential dissimilarity indicates the percentage difference between the distributions of two component groups of population:
{img src=show_image.php?name=bkhumgeofm13.gif }
where xi represents the percentage of the x population in the ith areal subunit, yi the percentage of the y population in the ith subunit, and the summation is given over all the k subunits making up the given territory, such as a city. Second, the index of residential segregation indicates the percentage difference between one group\'s distribution and that of the rest of the population:
{img src=show_image.php?name=bkhumgeofm14.gif }
where Idxn is the index of dissimilarity between group x and the total population y (including the subgroup), âˆ‘x represents the total number in group x in the city and âˆ‘ n represents the total population of the city. For example, in Britain in 1991, the index of dissimilarity for the Bangladeshi population compared to the white population was 73 per cent and the index of segregation for the Bangladeshis was 69, indicating their highly segregated distribution. A further simple measure is the location quotient, which shows the relative concentration of a population within any one subarea.
Variations on these basic measures, and the problems associated with scale, the size of subgroups and the nature of areal units have been much discussed. (See also ghetto; race; racism; urban geography.)Â (PEO)
Reference Peach, C. 1996: Does Britain have ghettos? Transactions, Institute of British Geographers NS 21: 21635.
Suggested Reading Jones, E. and Eyles, J. 1977: An introduction to social geography. Oxford and New York: Oxford University Press.Â Peach, C., ed., 1975: Urban social segregation. London and New York: Longman.Â Peach, C., Robinson, V and Smith, S., eds, 1981: Ethnic segregation in cities. London: Croom Helm; Athens, GA: University of Georgia Press.Â Plane, D.A. and Rogerson, D.A. 1994: The geographical analysis of population. New York and Chichester: John Wiley, ch. 10. 
