Home > Data Services > Census Support > Tools . 2010 . Topics . Experts

Search Data Services

Meta Search
search across all the following databases:

Data Catalog
Data and documentation

KnowledgeBase
Common questions and answers.

Resources
Entire collection of data resources.


Latest Data News

RSS Feed icon

Breaking News: Protect the 2020 Census

The Fastest Growing US Cities Are Mostly In the West

Another rule from NIH: Demographers might get to ignore this one

Research on Health Disparities: Incarceration Matters

Changing Race From One Census to the Next

Coding of Race in 2000

Data Documentation

In 1997, The Office of Management and Budget (OMB) recommended a significant change in the collection of race data for federal agencies. In the past, respondents self-identified with a single race category, which could be “other” if none of the listed races were appropriate. The new collection procedure allows for respondents to identify with one or more races. The six listed race categories are:

White
Black or African American
American Indian or Alaska Native
Asian
Native Hawaiian or Pacific Islander
Some Other Race

All possible combinations of six categories results in 63 possibilites: 6 single race choices, 15 combinations of two races, 20 combinations of three races, 15 combinations of four races, six combinations of five races, and one combination of all six races. race questions

Hispanic origin is asked in a separate question. This means that all of the race choices can be described as Hispanic or non-Hispanic. For instance, non-Hispanic white, Hispanic black, non-Hispanic Asian, American Indian, Other, etc. This results in a total of 126 possible race/Hispanic origin combinations.

OMB was well aware that changing the method of collecting race information would have serious consequences for summarizing racial distributions and examining trends over time. Thus, they provided an overview of issues to consider. It is likely to take some time for federal agencies, state and local governments, researchers and other data users to understand all the ramifications and consequences of moving from 5 or 6 race categories by Hispanic/non-Hispanic to 63 race combinations by Hispanic/non-Hispanic.

In the meantime, we are summarizing five ways to collapse the race information. The methods are (1) no collapsing; (2) maximums; (3) proportional combinations with minority preference; (4) straight proportional combinations; and (5) combinations based on forced single choice.

No collapsing

The no collapsing choice is what the Census Bureau uses. This method provides counts of each of the single race choices and then combines all other choices into 2+ races selected. Hispanic and non-Hispanic are either provided as a set of separate counts or the counts of the single race choices and 2+ races are tabulated by Hispanic/non-Hispanic origin.

An additional feature of this choice is to provide a listing of the largest multiple race choices for an area. For instance, any combination that is 2% (or some arbitrary percentage) of the total population would be reported under the 2+ races row or column.

Sample tabulation for Los Angeles county, California using this method.

SAS coding for determining largest multiple race choices.

Maximums

This choice maximizes the size of each of the six race choices. White would be defined as the sum of all combinations that include white. For instance, [(white only) + (white, black) + (white, american indian and native alaskan). . . . . . . + (white, black, american indican and native alaskan, asian, native hawaiian and pacific islander, other)].

Using maximums will always sum to more than the total population or if provided as percentages, will sum to more than 100 percent. This is because each combination is being counted two, three, four, five, or six times, depending on how many races are being combined.

If used as a denominator, maximums will minimize the rate for a population unless the numerator was also calculated this way (unlikely). Maximums should not be used in any statistic that is expecting mutually exclusive categories as any multi-race respondent will be represented in at least two categories.

Sample tabulation for Los Angeles county, California using this method.

SAS coding for maximum tabulation

FTP site index for recoded race data.

Proportional Combinations, with minority preference

This measure only counts a respondent one time so that population totals and percentages behave as expected.

If a respondent identifies with two or more races, he is split into two or more categories. If the race combination includes whites, the respondent is put into the other race choice(s), even if whites are the minority population in that area.

For instance, if a respondent selected white and asian as his race, he would be counted as asian. If a respondent chose white, black and asian as his race, the respondent would be evenly divided into .5 black and .5 asian and white would be dropped.

Sample tabulation for Los Angeles county, California using this method.

SAS coding for proportional combinations, with minority preference.

FTP site index for recoded race data.

Straight Proportional Combinations

This measure only counts a respondent one time so that population totals and percentages behave as expected.

Respondents are evenly divided into as many categories as they select for their race choice. If a respondent identifies with 2 races (e.g., black and asian), he is divided into .5 black and .5 asian.

Sample tabulation for Los Angeles county, California using this method.

SAS coding for straight proportional combinations

FTP site for recoded race data.

Proportional Combinations based on Forced Single Choice

This measure only counts a respondent one time so that population totals and percentages behave as expected.

Respondents are divided into as many categories as they select for their race choice. The algorithm that determines how a respondent is divided (.5 and .5, .6 and .4, .9 and .1, etc) is based on an NCHS survey . In the survey, respondents were asked to select their race with the choice of selecting multiple races. Any respondent who selected more than one race was also asked "Which of those groups would you say best describes your race?" The following table [link] summarizes the results, which provides the background for the algorithm used in calculating this method:

Sample tabulations for Los Angeles county, California using this method:

SAS program code for calculating proportional combination based on forced single choice.

FTP site for recoded race data.

Further information on race

United States Census 2000 Population With Bridged Race

Counting Multiracials in the 2000 Census: Implications for Asian Americans

Racial and Ethnic Classifications Used in Census 2000 and Beyond

Questions and Answers for Census 2000 Data on Race

Overview of Race and Hispanic Origin (Census 2000 Brief)

Office of Management and Budget (OMB) Provisional Guidelines on Data on Race and Ethnicity

Reporting Census 2000: A Guide for Journalists - Race

Draft Provisional Guidance on the Implementation of the 1997 Standards for Federal Data on Race and Ethnicity. (1999). Tabulation Working Group, Interagency Committee for the Review of Standards for Data on Race and Ethnicity. Washington, DC: Executive Office of the President.

Counting Multiracials in the 2000 Census: Implications for Asian American (2000). Albert Sanghyup Hahn. Asian American Policy Review, 9, pp. 56-75.

United States Census 2000 population with bridged race categories.(2003). Ingram DD, Parker JD, Schenker N, Weed JA, Hamilton B, Arias E, Madans JH. Vital Health Statistics 2(135). National Center for Health Statistics.

Identifying with Multiple Races: A Social Movement that Succeeded but Failed? (2001). Reynolds Farley. PSC Research Report 01-491. Ann Arbor, MI: University of Michigan, Population Studies Center.

Who Chooses to Choose Two? (2004). Sonya M. Tafoya, Hans Johnson, and Laura E. Hill. Washington, DC: Population Reference Bureau.