Letter to Project RACE re: Multiple Response Sampling

March 31, 2000
U.S. Census Bureau
Letter to Project RACE
Regarding Multiple Response Sampling

Economics and Statistics Administration
U.S. Census Bureau
Washington, DC 20233-0001


March 31, 2000

Mr. James Landrith, Director
Ms. Susan Graham, President
Project Race, Inc.
P.O. Box 8208
Alexandria, VA 22306-8208

Dear Mr. Landrith and Ms. Graham:

Thank you for your cosigned letter regarding your concerns about the U.S. Census Bureau’s plans for collecting data on race and ethnicity during Census 2000.

The Census Bureau’s goal in Census 2000 is to take the most accurate census possible. Under current federal law, including the Voting Rights Act, the Community Reinvestment Act, and the Public Health Service Act, the Census Bureau must collect information on race and ethnicity. In addition, federal agencies are required to use census data on race and ethnicity as stipulated under the revised Office of Management and Budget Directive 15 entitled Standards for Maintaining, Collecting, and Presenting Federal Data on Race and Ethnicity (enclosed).

As you know, Census 2000 will accommodate the reporting of more than one race for the first time. The Census Bureau is implementing two projects that will enable Census Bureau staff to evaluate the quality of all race data collected in Census 2000, including the use of multiple-race categories. One is an experiment in which a small sample (10,000 households) will be sent a census questionnaire that is comparable to the questionnaire employed in the 1990 census. The purpose of this experiment is to determine the impact on census data of the whole package of questionnaire design and item changes that have been instituted since the 1990 census. The Census 2000 change to allow for the reporting of more than one race is only one difference from the 1990 census that will be analyzed. Other aspects include the effect of changes in format, instructions, and sequencing of questions. Similar experiments were incorporated in prior censuses. As the sample size for experiments are kept small, and randomly selected addresses are spread throughout the entire United States, there is little impact on the final census data.

Another evaluation of race data collected in Census 2000 involves a separate survey. In this survey a random sample of Census 2000 households (both with and without persons reporting more than one one race) will be contacted after they submit their questionnaire and will be asked some of the census questions, including the race question, a second time using the same wording as on the Census 2000 questionnaire. Additional questions will then be asked of persons reporting more than one race to identify how they would respond if asked to mark only one race category as in past censuses. From this evaluation, census staff will gain information that will aid in understanding more about the reporting of more than one race and its impact on statistical data uses. The results from this evaluation study will not be used to change the original race data as reported in Census 2000.

It is worth underscoring the Census Bureau’s commitment to protecting the confidentiality of all census data. No one except sworn Census Bureau employees can see individual census data. In fact, Census Bureau employees fave severe penalties, including fines and jail, if they in any way disclose confidential information about individual person or households. By law, the Census Bureau cannot share such information with the Internal Revenue Service, the Federal Bureau of Investigation, welfare and immigration officials – or any other government agency or court of law. The same law that keeps this information out of the hands of these agencies prevents the Census Bureau from selling or giving away any confidential data to anyone wanting it for any reason.

We hope this letter addresses your concerns. Thank you again for writing and for your interest in Census 2000.


Kenneth Prewitt


Leave a Reply

Your email address will not be published.