U.S. Department of Commerce

Housing Patterns




APPENDIX A


MEASURING SEGREGATION USING ALTERNATIVE METHODS OF IDENTIFYING RACIAL/ETHNIC GROUP MEMBERS: "ALONE" VS. "ALONE OR IN COMBINATION"


Census 2000 was the first to allow respondents to identify themselves using multiple race categories. This change raises concerns of how to classify multi-race individuals when making demographic comparisons over time. One possibility is to include in the minority group of interest anyone designating themselves as a member of that racial group, e.g., people who self-identified as Black or African American alone or in combination with another group. An alternative is to include in a group's count only individuals identifying with that group alone. We have decided to use the first method ("alone or in combination") in this report for reasons described in chapter two. This means that the minority group definitions used in this report are not mutually exclusive. The purpose of this appendix is to show how segregation statistics differ across the two racial classification schemes. The reference group-- non-Hispanic Whites-- is always defined as those who report being White alone, and who are not of Hispanic origin. The count of Hispanics or Latinos is not affected by this issue since Hispanic ethnicity is determined by a separate census question, and Hispanics or Latinos can be of any race.

Table A-1 indicates that the differences across the methods are generally small across the 19 segregation indexes examined. The indexes in bold are the ones used throughout this report. As might be expected, segregation tends to be a little higher when using the "alone" classification scheme than the "alone or in combination" one (which includes multiracial individuals). Differences tend to be particularly small for both African Americans and Asians and Pacific Islanders; conclusions about the patterns of segregation for these groups would not change if the "alone" methodology were used instead of the "alone or in combination" one. The differences are modestly larger for American Indians and Alaska Natives. Whereas declines in segregation from 1980 to 2000 are registered across four of the five measures used in this report for this group when the "alone or in combination" scheme is used, this number falls to three when the "alone" category is used.

Table A-1. Comparison of Segregation Indexes for Racial Groups Defined Alone vs. Alone or in Combination: 1980 and 2000
  African American or Black American Indian and Alaska Native Asian and Pacific Islander (API)
  1980 2000 1980 2000 1980 2000
Index   Alone Alone or in combo   Alone Alone or in combo   API Alone Asian Alone NHOPI Alone API Alone or in combo

EVENNESS MEASURES
Dissimilarity 0.727 0.651 0.640 0.373 0.393 0.333 0.405 0.434 0.439 0.493 0.411
Gini 0.864 0.798 0.787 0.502 0.522 0.45 0.545 0.578 0.584 0.650 0.550
Entropy 0.546 0.447 0.434 0.125 0.144 0.111 0.151 0.180 0.183 0.177 0.165
Atkinson with b=.1 0.232 0.156 0.148 0.073 0.064 0.041 0.062 0.065 0.067 0.162 0.058
Atkinson with b=.5 0.701 0.587 0.57 0.252 0.266 0.198 0.261 0.286 0.291 0.392 0.258
Atkinson with b=.9 0.883 0.801 0.789 0.402 0.435 0.346 0.423 0.454 0.461 0.559 0.418

EXPOSURE MEASURES
Interaction 0.345 0.403 0.409 0.918 0.887 0.897 0.767 0.700 0.705 0.848 0.694
Isolation 0.655 0.597 0.591 0.082 0.113 0.103 0.233 0.300 0.295 0.152 0.306
Correlation ratio 0.571 0.478 0.468 0.068 0.089 0.071 0.120 0.168 0.170 0.099 0.158

CONCENTRATION MEASURES
Delta 0.834 0.796 0.793 0.695 0.692 0.676 0.741 0.753 0.756 0.747 0.743
Absolute concentration Index 0.888 0.883 0.881 0.863 0.871 0.882 0.869 0.892 0.899 0.894 0.876
Relative concentration 0.627 0.662 0.658 -1.423 -0.622 -0.261 0.483 0.614 0.622 0.346 0.588

CENTRALIZATION MEASURES
Absolute centralization 0.753 0.724 0.722 0.622 0.610 0.611 0.701 0.691 0.694 0.593 0.683
Relative centralization 0.314 0.294 0.290 0.003 0.051 0.067 0.194 0.214 0.218 0.097 0.202

CLUSTERING MEASURES
Absolute clustering 0.416 0.365 0.360 0.086 0.095 0.061 0.087 0.124 0.125 0.053 0.116
Spatial proximity 1.435 1.382 1.374 1.197 1.131 1.077 1.057 1.103 1.104 1.053 1.096
Relative clustering 2.139 1.259 1.192 6.605 2.731 1.206 0.766 0.555 0.584 0.712 0.454
Distance decay interaction 0.493 0.496 0.499 0.929 0.910 0.923 0.808 0.746 0.752 0.886 0.736
Distance decay isolation 0.507 0.504 0.501 0.069 0.090 0.077 0.189 0.254 0.248 0.114 0.264
Notes: Segregation scores represent weighted averages across all metropolitan areas. Indexes in bold are those highlighted in this report. NHOPI = Native Hawaiian and Other Pacific Islander.
Source: U.S. Census Bureau, Census 1980 and 2000 Summary File 1.

In short, the differences observed across racial classification methods are modest. They do not change our conclusions about trends in segregation over the 1980 to 2000 period. Alternative methods have a somewhat larger effect on American Indian and Alaska Native segregation scores than on the African American or Asian and Pacific Islander ones.



Go to Appendix B: Measures of Residential Segregation
Go to "Racial and Ethnic Residential Segregation in the United States: 1980-2000" in HTML format
Contact the Census Call Center Staff at 1-800-923-8282 (toll free) or visit ask.census.gov for further information on Housing Patterns Data.

Source: U.S. Census Bureau, Housing and Household Economic Statistics Division
Last Revised: October 31, 2011