CAIN Web Service

Comparative Perspectives on Housing Segregation
by Sandra J. Callaghan (2001)



[CAIN_Home]
[Key_Events] Key_Issues] [Conflict_Background]

Text and Research: Sandra J. Callaghan

Comparative Perspectives on Housing Segregation:
Northern Ireland and US Frostbelt Cities
Sandra J. Callaghan (2001)

This paper briefly explores comparative patterns of residential segregation in Northern Ireland to those of the United States Frostbelt region.

In 1989 Douglas S. Massey and Nancy A. Denton concluded that six US metropolitan areas: Baltimore; Chicago; Cleveland; Detroit; Milwaukee and Philadelphia, were the most "hypersegregated" metropolitan areas of the US. Massey and Denton based their findings on residential segregation patterns in sixty US metropolitan areas using 1980 US Census data. The 1980 rankings of residential segregation in US metropolitan areas were based upon comparative measurements of five different dimensions of segregation. The five dimensions and the corresponding measures used by Massey and Denton are defined in Table 1.1.

Table 1.1 – Definitions, Measures and Scoring for Five Dimensions of Residential Segregation

Segregation Dimension

Measures

Score

Evenness The most widely used measure of evenness; dissimilarity measures the proportion of members of one group that would have to change their area of residence to achieve perfect integration. 1 -> 1.0
0 = No Segregation
1 = Maximum Segregation
High = .600 +
Exposure Exposure measures the extent to which groups are exposed only to each other in their areas of residence. Commonly used exposure measures are Isolation and/or Correlation (Eta). 0 -> 1.0
0 = Interaction
1 = Isolation
High = .700 +
Clustering Clustering measures the extent to which census tracts inhabited by subordinate groups are contiguous. 0 -> 1.0+
>1 = Dominant/Subordinate Groups live closer to each other than to their own groups
1 = No differential clustering
1.0+ = Dominant /Subordinate groups live closer to one another than to members of the other group.
Centralization Centralization measures the extent to which a group is spatially located near the center of an urban area. -1.0 -> +1.0
-1 = Tendency to reside in outlying areas
+1 = Tendency to reside close to city center
0=Uniform distribution
High = .800 +
Concentration Concentration measures the amount of physical space occupied by subordinate groups so that as segregation increases subordinate groups are increasingly concentrated in smaller areas. -1.0 -> +1.0
-1.0 = Concentration of one group exceeds the other to the maximum possible extent
1 = Opposite concentration
0=Two groups equally concentrated
High = .700 +
Sources: Massey and Denton, 1989 and U.S. Census Bureau


All six of the "hypersegregated" metropolitan areas are located in an area known as the Frostbelt region. (The region is known as the Frostbelt because the winters are too long and cold which, of course, means the summers are far too short.)

1990 census data indicate five out of the six metropolitan areas earning Massey and Denton’s dubious distinction of being the most segregated metropolitan areas in the US in 1980 again exceeded all five measures in 1990. Chicago was the only metropolitan area not exceeding all five-dimension thresholds in 1990. Chicago’s centralization dropped from .872 in 1980 to .722 in 1990, but the other four dimensions, dissimilarity, isolation, clustering, and concentration exceeded the Massey and Denton’s suggested thresholds. Table 1.2 lists comparative 1990 residential segregation data for the Frostbelt’s fourteen largest metropolitan areas (excluding New York, New York). Map 1.1 illustrates 1990 comparative isolation indices for Frostbelt metropolitan areas listed in Table 1.2.

Table 1.2 – Residential Segregation in US Frostbelt Cities
Five Measures of Segregation (1990)

MSA / PMSA Name

Percent Minority Population

CENTRAL
-IZATION
Absolute Centralization Index (ACE)

CONCEN
-TRATION
Absolute Concentration Index (ACO)

EVENNESS
Dissimilarity Index (D)

CLUSTERING
Spatial Proximity Index (SP)

EXPOSURE
Isolation Index (xPx*)

EXPOSURE
Correlation Ratio or Eta Squared (V)

Baltimore, MD

25.9%

0.847

0.921

0.715

1.602

0.709

0.602

Boston, MA

6.6%

0.849

0.958

0.695

1.504

0.546

0.511

Buffalo, NY

10.3%

0.836

0.984

0.800

1.478

0.651

0.610

Chicago, IL

19.3%

0.722

0.929

0.842

1.858

0.814

0.760

Cincinnati, OH

12.5%

0.921

0.979

0.767

1.372

0.617

0.561

Cleveland, OH

17.3%

0.880

0.971

0.826

1.825

0.773

0.724

Columbus, OH

12.1%

0.875

0.966

0.683

1.331

0.539

0.473

Detroit, MI

22.1%

0.878

0.974

0.874

1.853

0.823

0.771

Indianapolis, IN

13.2%

0.866

0.970

0.751

1.453

0.604

0.542

Milwaukee WI

13.8%

0.889

0.976

0.826

1.761

0.725

0.678

Minneapolis, MN

3.5%

0.938

0.978

0.622

1.162

0.296

0.268

Philadelphia, PA

19.1%

0.822

0.918

0.769

1.688

0.720

0.649

Pittsburgh, PA

7.5%

0.831

0.967

0.708

1.281

0.518

0.479

St. Louis, MO

17.0%

0.913

0.966

0.782

1.637

0.707

0.646

US Frostbelt Median

13.5%

0.871

0.969

0.768

1.553

0.679

0.606

Source: US Census Bureau, Housing and Household Economic Statistics Division


Map 1.1 – US Frostbelt Isolation Indices, 1990


There are many causes of residential segregation. Individual choices to live in one area versus another is a certainly a primary cause of segregation. However, individual choices may be limited by constraints outside individual control. In the case of housing choices, constraints outside the control of individuals include available such things as the availability of affordable housing and discrimination (by mortgage lenders for buyers and by landlords for renters). One possible reason why Metropolitan Milwaukee ranks so highly on segregation is that Metropolitan Milwaukee had the highest racial loan gap for nine straight years (the difference between the rate of home mortgage loans approved for whites and those approved for blacks) (Milwaukee Journal Sentinel 1999 and Adversity).

Residential segregation also has many unintended consequences. Research has shown that segregation is linked to the economic fortunes and life changes of those living within highly segregated areas. One of the most negative affects of hypersegregation, as posited by Massey and Denton, is its relationship to concentrated poverty and related social dislocation problems (William Julius Wilson 1987) experienced by those living in such areas. Metropolitan Milwaukee also demonstrates unintended consequences of hypersegregation: Milwaukee had the highest ratio of black/white unemployment disparities from 1970 to 1996. In 1990, the area had the highest proportion of blacks living in higher poverty neighborhoods in the Frostbelt region (CED-UWM).

How do patterns of residential segregation in Northern Ireland compare to the same patterns in US Frostbelt metropolitan areas? Michael A. Poole and Paul Doherty (1996) compared residential segregation patterns in 39 Northern Ireland towns using 1981 census data. Poole and Doherty used two measures of Evenness (dissimilarity and replacement) and two measures of Exposure (isolation and Eta Squared) as well as a measure they called "Linear Dominance" to compare these 39 Northern Ireland towns (Appendix 1). Maps 1.2 and 1.3 represent Protestant and Catholic isolation indices as found by Poole and Doherty.

Map 1.2 – Protestant Isolation in Northern Ireland, 1981


Map 1.3 – Catholic Isolation in Northern Ireland, 1981


Maps 1.2 and 1.3 illustrate the rather symmetrical pattern of Northern Ireland’s residential isolation (exposure): towns in eastern Northern Ireland have higher Protestant isolation indices while those in western Northern Ireland have higher Catholic isolation indices. The numerical scale Poole and Doherty used for measuring segregation was a 1-100 rather scale than the 0.1-1 scale used by Massey and Denton. In the Northern Ireland case, therefore, an index above 70 indicates a high degree of isolation. Thirty of the towns studied by Poole and Doherty had isolation indices above 70 for either one of the groups (Catholics and Protestants). Table 1.3 lists the 30 towns for which either the Protestant or Catholic isolation index was above 70. (The table is sorted by the Catholic isolation index.)

Table 1.3 – Northern Ireland Towns with Isolation Indices above 70 (1981)

Town

EXPOSURE
Catholic Isolation Index

EXPOSURE
Protestant Isolation Index

Town

EXPOSURE
Catholic Isolation Index

EXPOSURE
Protestant Isolation Index

Strabane

89.7

20.6

Cookstown

48.5

65.9

Newry

88.7

32.6

Randalstown

44.2

67.9

Derry

86.8

66.8

Magherafelt

44.2

58.1

Downpatrick

86.8

34.2

Limavady

42.2

64.6

Lurgan

82.5

81.2

Ballynahinch

38.5

65.4

Warrenpoint

80.3

21.8

Antrim

33.5

73.2

Armagh

79.6

74.3

Ballymena

33.1

82.4

Omagh

75.7

63.2

Banbridge

31.4

74.1

Belfast

74.0

91.0

Larne

29.1

77.3

Dungannon

70.7

68.3

Portstewart

24.9

76.1

Portadown

67.6

90.5

Coleraine

24.3

79.8

Enniskillen

65.2

57.0

Ballymoney

23.7

82

Ballycastle

65.0

37.9

Portrush

23.5

81.3

Newcastle

63.2

43.2

Holywood

22.7

83.5

Kilkeel

53.6

69.2

Lisburn

20.9

86.8

Source: Poole and Doherty, 1996


Map 1.4 portrays the dissimilarity (evenness) comparisons of the 39 towns. Five of the 30 towns had dissimilarity indices above 60: Armagh, Belfast, Derry, Lurgan and Portadown. Table 1.4 lists the dissimilarity indices of the 39 towns (sorted by dissimilarity). Those towns that have dissimilarity indices above 60 have high degrees of dissimilarity.

Map 1.4 – Dissimilarity Indices of 39 Towns in Northern Ireland, 1981


Table 1.4 - Dissimilarity Indices of 39 Towns in Northern Ireland (1981)


Town

Dissimilarity Index

Town

Dissimilarity Index

Town

Dissimilarity Index

Belfast

76.0

Holywood

29.1

Ballyclare

19.8

Lurgan

72.9

Bangor

26.7

Coleraine

19.6

Armagh

72.0

Newtownards

26.3

Ballynahinch

19.1

Derry

70.9

Randalstown

24.4

Whitehead

16.7

Portadown

70.3

Antrim

23.8

Warrenpoint

16.4

Dungannon

56.9

Donaghadee

23.2

Ballycastle

13.9

Omagh

55.3

Larne

22.8

Comber

13.5

Newry

51.3

Limavady

22.3

Magherafelt

11.7

Downpatrick

46.6

Banbridge

22.2

Portstewart

8.6

Strabane

43.2

Carryduff

21.9

Kilkeel

42.9

Portrush

21.9

Enniskillen

42.8

Newcastle

21.1

Ballymena

38.2

Carrickfergus

20.0

Lisburn

32.4

Dromore

20.0

Cookstown

31.3

Ballymoney

19.9

Source: Poole and Doherty, 1996


Poole and Doherty used another evenness dimension measurement, which they called the "Linear Dominance Index" (LD). Poole and Doherty contend LD, while sharing the measurement of deviation from a benchmark with other measures of evenness such as dissimilarity and replacement is different from them. The way LD works is that if it is assumed two populations, A and B, are evenly distributed in a given geographic area the composition is 50/50. If they are not evenly distributed then one group is dominant within a given area.

Using LD as a measure of evenness Poole and Doherty found that 30 of Northern Ireland’s towns had LD measurements exceeding 50 and, because those towns were the majority of Northern Ireland’s urban settlements Northern Ireland’s larger towns were residentially segregated in 1981. The eleven towns with LD values below 50 represented only 9.1 percent of the total population in the 39 towns. Map 1.5 shows the LD indices of the 39 towns and Table 1.5 lists the indices (sorted by LD).

Map 1.5 – Linear Dominance Indices of 39 Northern Ireland Towns (1981)


Table 1.5 – Linear Dominance Indices of 39 Northern Ireland Towns (1981)

Town

LINEAR DOMINANCE INDEX

Town

LINEAR DOMINANCE INDEX

Town

LINEAR DOMINANCE INDEX

Donaghadee

93.6

Armagh

70.3

Antrim

42.6

Comber

93.3

Downpatrick

69.9

Enniskillen

42.5

Ballyclare

92.4

Carryduff

65.6

Randalstown

39.2

Bangor

85.7

Holywood

64.9

Cookstown

32.6

Carrickfergus

85.0

Ballymoney

61.8

Limavady

29.1

Newtownards

84.7

Ballymena

61.0

Ballynahinch

28.0

Belfast

84.0

Portrush

60.7

Ballycastle

27.9

Portadown

81.6

Warrenpoint

59.7

Newcastle

26.6

Strabane

77.1

Omagh

58.3

Magherafelt

17.3

Whitehead

74.4

Coleraine

57.9

Derry

74.3

Dungannon

56.6

Newry

73.9

Larne

52.0

Dromore

73.3

Portstewart

51.8

Lurgan

72.9

Kilkeel

49.8

Lisburn

71.3

Banbridge

45.1

Source: Poole and Doherty 1996


When the differing dimensions of segregation measurements are combined and analyzed using the high thresholds for each of the measures (dissimilarity = 60+; isolation = 70+; and, LD = 50+) five of Northern Ireland’s towns showed highly segregated patterns in 1981 (Table 1.6). Table 1.6 is sorted alphabetically by Town.

Table 1.6 – Northern Ireland Towns Rating High on
Three Measures of Segregation Northern Ireland Towns

Town

EVENNESS
Dissimilarity

EXPOSURE
Catholic Isolation

EXPOSURE
Protestant Isolation

EVENNESS
Linear Dominance

Armagh

72.0

79.6

74.3

70.3

Belfast

76.0

74.0

91.0

84.0

Derry

70.9

86.8

66.8

74.3

Lurgan

72.9

82.5

81.2

72.9

Portadown

70.3

67.6

90.5

81.6

Source: Poole and Doherty 1996


Based upon the comparative analysis shown in Table 1.6 (above) towns in Northern Ireland certainly appeared to have been every bit as residentially segregated in 1981, as were large metropolitan areas in the US Frostbelt in 1990. Similarly, although not analyzed within the scope of this paper, patterns of residential segregation in both the US Frostbelt region and in Northern Ireland towns have endured for decades. Poole and Doherty concluded in the Northern Ireland case that:

The province is an irregular mosaic of radically different places—not just Catholic and Protestant, but with every conceivable level of mixing in between. All the pieces in the mosaic are tossed together into this single political unit, but the result is that nowhere is typical of Northern Ireland as a whole—not Belfast, not Ballymena, not Bushmills (262).

The same would certainly hold true for the US. No single place is typical of the whole US. In both Northern Ireland and the US however, patterns of residential segregation in certain areas are readily apparent to anyone traveling through them. Sadly, those patterns of residential segregation reflect the deep social divides within each society.



BIBLIOGRAPHY

Adversity.net, "Whites in Milwaukee rarely denied loans; theories on why abound" not dated, http://www.adversity.net/special/banking_housing_02.htm {external_link}

Center for Economic Development, University of Milwaukee-Wisconsin, The Economic State of Milwaukee: The City and the Region, {external_link} Levine, Marc V. with Callaghan, Sandra J., Milwaukee, Wisconsin, 1998

Massey, Douglas S. and Denton, Nancy A., "Hypersegregation in U.S. Metropolitan Areas: Black and Hispanic Segregation along Five Dimensions," Demography, Vol. 26, No. 3, August, 1989

Milwaukee Journal Sentinel, "Racial lending gap here still too wide -- but there's hope," April 11, 1999, http://www.jsonline.com/news/editorials/0411loans.asp {external_link}

Poole, Michael A. and Doherty, Paul, Ethnic Residential Segregation in Northern Ireland, Centre for the Study of Conflict, University of Ulster, Coleraine, Northern Ireland, 1996

US Census Bureau, Housing and Household Economic Statistics Division, http://www.census.gov/hhes/www/housing/resseg/gettable.html {external_link}



APPENDIX 1
Measures of Residential Segregation in 39 Northern Ireland Towns (1981)

Town

EVENNESS
Dissimilarity Index

EVENNESS
Replacement Index

EXPOSURE
Catholic Isolation Index

EXPOSURE
Protestant Isolation Index

EXPOSURE
Eta Squared Index

LINEAR DOMINANCE INDEX

Antrim

23.8

19.5

33.5

73.2

6.8

42.6

Armagh

72.0

71.0

79.6

74.3

53.9

70.3

Ballycastle

13.9

12.8

65.0

37.9

2.9

27.9

Ballyclare

19.8

2.9

4.8

96.2

1.0

92.4

Ballymena

38.2

25.2

33.1

82.4

15.6

61.0

Ballymoney

19.9

12.3

23.7

82.0

5.7

61.8

Ballynahinch

19.1

17.6

38.5

65.4

3.9

28.0

Banbridge

22.2

17.7

31.4

74.1

5.5

45.1

Bangor

26.7

7.1

10.9

93.2

4.0

85.7

Belfast

76.0

59.0

74.0

91.0

65.0

84.0

Carrickfergus

20.0

5.5

9.1

92.6

1.8

85.0

Carryduff

21.9

12.5

20.2

83.4

3.6

65.6

Coleraine

19.6

13.0

24.3

79.8

4.1

57.9

Comber

13.5

1.7

3.8

96.7

0.4

93.3

Cookstown

31.3

30.0

48.5

65.9

14.4

32.6

Derry

70.9

57.6

86.8

66.8

53.6

74.3

Donaghadee

23.2

2.9

4.0

96.8

0.8

93.6

Downpatrick

46.6

26.0

86.8

34.2

20.9

69.9

Dromore

20.0

9.3

15.5

87.0

2.5

73.3

Dungannon

56.9

56.8

70.7

68.3

39

56.6

Enniskillen

42.8

42.3

65.2

57.0

22.1

42.5

Holywood

29.1

16.8

22.7

83.5

6.3

64.9

Kilkeel

42.9

41.2

53.6

69.2

22.9

49.8

Larne

22.8

16.7

29.1

77.3

6.5

52.0

Limavady

22.3

21.0

42.2

64.6

6.7

29.1

Lisburn

32.4

15.9

20.9

86.8

7.7

71.3

Lurgan

72.9

72.8

82.5

81.2

63.8

72.9

Magherafelt

11.7

11.5

44.2

58.1

2.3

17.3

Newcastle

21.1

20.1

63.2

43.2

6.4

26.6

Newry

51.3

25.2

88.7

32.6

21.3

73.9

Newtownards

26.3

7.4

10.2

92.6

2.8

84.7

Omagh

55.3

53.0

75.7

63.2

38.9

58.3

Portadown

70.3

49.3

67.6

90.5

58.1

81.6

Portrush

21.9

13.9

23.5

81.3

4.7

60.7

Portstewart

8.60

6.3

24.9

76.1

1.0

51.8

Randalstown

24.4

22.6

44.2

67.9

12.1

39.2

Strabane

43.2

17.5

89.7

20.6

10.3

77.1

Warrenpoint

16.4

10.5

80.3

21.8

2.1

59.7

Whitehead

16.7

7.5

14.4

87.4

1.8

74.4

Source: Poole and Doherty 1996


CAIN contains information and source material on the conflict and politics in Northern Ireland.
CAIN is based within the University of Ulster.


go to the top of this page go to the top of this page
Last modified :