The Dictionary of Human Geography

Comp. by: VPugazhenthi Stage : Revises1 ChapterID: 9781405132879_4_S Date:1/4/
09 Time:15:23:35 Filepath:H:/00_Blackwell/00_3B2/Gregory-9781405132879/appln/3B2/
revises/9781405132879_4_S.3d

segregation, measurement of Social scien-

tists have long tried to find ways of depicting

and measuring the degree of segregation

between groups in societies, especially cities.

In early discussions of this topic, such as the

classic studies by sociologists collectively

known as thechicago school, segregation

was portrayed in descriptive terms. Ernest

Burgess (1967), for example, included maps

that characterized areas of Chicago as the

‘Deutschland Ghetto’, ‘Little Sicily’ and

the ‘Black Belt’. But these terms were highly

generalized and none of the areas identified on

the maps held completely mono-ethnic popu-

lations. Over the past century, sociologists and

geographers have struggled to find more accur-

ate ways to visualize and analyse segregation.

The social geography of a specific group is

usually depicted inmapsthat are either based

on standard percentage figures or aslocation

quotients(LQs). These maps show areas

where a group is concentrated, providing a

visual demonstration of the degree of segrega-

tion involved. A group that is fully integrated

into the population at large would be evenly

distributed across the territory in question

(with all LQ figures close to 1.0), while a

completely segregated (ghettoized) group

would be concentrated in a single area, with

low LQ figures throughout the map except in

the area of concentration.

While maps such as these are undoubtedly

an improvement over the impressionistic gen-

eralizations made in early sociological studies,

they remain ambiguous. More precise statis-

tics designed to represent the degree of segre-

gation of groups were first introduced in the

late 1940s and two gained ascendancy after

a crucial paper published in the mid-1950s

(Duncan and Duncan, 1955). TheIndex of

Dissimilarityis a measure of the degree of resi-

dential isolation between two groups and is

calculated as:

IDxy¼

1

2

Xk

i¼ 1

[xiyi],

whereIDis the Index of Dissimilarity between

groupsxandy;xiandyiare the percentages of

the two groups that reside in a particular spa-

tial unit, such as a Census Tract, in a city

(note that the difference in these percentages

is taken as an absolute value and there are

no negative numbers used in the calculation

of theID); andkis the number of spatial units

that make up the whole. Arithmetically, the

index shows the proportion of groupxthat

would have to change its location in order to

match the distribution of groupy.AnIDvalue

of 0 means the two groups are precisely co-

located across the city or region in question,

while a value of 100 would mean that the two

do not overlap at all. TheIndex of Segregation

(IS) is calculated in much the same way, but

between an individual group and all other

groups in a society;ISvalues also range from

0 to 100.

Although theIDandIShave been the most

enduring measures of segregation, they share a

number of problems. First, there is no

straightforward way to decide when anIDor

IS value is significant. After considerable

debate, researchers have developed a consen-

sus that values under 25 indicate little or no

segregation, while those over 60 are inter-

preted as indicating a high level of segregation.

Second, both indices arescale-dependent:

they tend to be higher when a city or region

is divided into a larger number of administra-

tive units (Johnston, Forrest and Poulson,

2001). For this reason, comparisons of index

values over time or between cities are suspect,

unless the scale of spatial units is constant

(which is relatively rare, particularly in inter-

national comparisons; see Van Valley and

Roof, 1976). Third, index values for groups

with small populations tend to be higher than

those for groups that are large. Finally, indices

speak to the degree of isolation of a group but

do not describe particular spatial patterns.

For example, theISfor a group that is entirely

located in one inner-city neighbourhood,

which is the only group in that place, would

be 100. But theISwould also be 100 for a

group located in four different suburbs if it was

the only group in those areas. The reasons

for these segregated patterns could be entirely

different (the first group could be impover-

ished while the latter could be affluent), but

theISvalues would be the same.

A number of efforts have been made to pro-

vide better measures of segregation. These are

summarized by Massey and Denton (1986),

who discuss five dimensions of segregation

(and measures associated with each). In addi-

tion toisolation or dissimilarity, they describe:

concentration versus dispersion; centralization

versus dispersion;clustering among groups(e.g.

individual European-origin groups may be

isolated from one-another, but all European-

origin groups could be in the same general

area of a city that is distinct from the location

of Asians, Hispanics etc.); and the degree of

exposurebetween groups (also see Lieberson,

1981).

Gregory / The Dictionary of Human Geography 9781405132879_4_S Final Proof page 675 1.4.2009 3:23pm

SEGREGATION, MEASUREMENT OF