September 28, 2001
The Honorable Frank R. Wolf
Chairman, Subcommittee on Commerce, Justice,
State, the Judiciary and Related Agencies
Committee on Appropriations
U.S. House of Representatives
Washington, DC 20515-6017Dear Mr. Chairman:
House Report 106-1005 directed the U.S. Department of Commerce to submit to the Congress a written report on the issues associated with the inclusion of American citizens and their dependents living abroad in the decennial censuses for apportionment, redistricting, and other purposes for which census results are used. The enclosed document fulfills this request.
The U.S. Census Bureau is currently planning a conference in November 2001 with census stakeholders to obtain input into the policy and technical issues raised in this report. We will notify the Congress when the exact dates for this conference have been established.
If the Census Bureau determines that it is possible to proceed with the development of operations to enumerate American citizens and their dependents living abroad in the decennial census, methodologies will be tested in conjunction with the scheduled census tests in FY 2004, FY 2006, and the census dress rehearsal in FY 2008.
We look forward to your review of this document.
Sincerely,
(Signed)
William G. Barron, Jr.
Acting DirectorEnclosure
The Honorable Ernest F. Hollings
Chairman, Subcommittee on Commerce, Justice,
State, and the Judiciary
Committee on Appropriations
United States Senate
Washington, DC 20510-6027
Dear Mr. Chairman:
The Honorable Judd Gregg
Ranking Member
Subcommittee on Commerce, Justice,
State, and the Judiciary
Committee on Appropriations
United States Senate
Washington, DC 20510-6027
Dear Senator Gregg:
The Honorable Joseph I. Lieberman
Chairman, Committee on Governmental Affairs
United States Senate
Washington, DC 20510-6250
Dear Mr. Chairman:
The Honorable Fred Thompson
Ranking Member
Committee on Governmental Affairs
United States Senate
Washington, DC 20510-6250
Dear Senator Thompson:
The Honorable José, Serrano
Ranking Member
Subcommittee on Commerce, Justice,
State, the Judiciary and Related Agencies
Committee on Appropriations
U.S. House of Representatives
Washington, DC 20515-6027
Dear Representative Serrano:
The Honorable Dan Miller
Chairman, Subcommittee on the Census
Committee on Government Reform
U.S. House of Representatives
Washington, DC 20515-6143
Dear Mr. Chairman:
The Honorable Carolyn B. Maloney
Subcommittee on the Census
Committee on Government Reform
U.S. House of Representatives
Washington, DC 20515-6143
Dear Representative Maloney:
The Honorable William Lacy Clay
Ranking Member
Subcommittee on the Census
Committee on Government Reform
U.S. House of Representatives
Washington, DC 20515-6143
Dear Representative Clay:
The Honorable John D. Rockefeller, IV
United States Senate
Washington, DC 20510-4802
Dear Senator Rockefeller:
The Honorable James P. Moran
U.S. House of Representatives
Washington, DC 20515-4608
Dear Representative Moran:
The Honorable Benjamin A. Gilman
U.S. House of Representatives
Washington, DC 20515-3220
Dear Representative Gilman:
The Honorable Christopher B. Cannon
Subcommittee on the Census
Committee on Government Reform
U.S. House of Representatives
Washington, DC 20515-6143
Dear Representative Cannon:
U.S. Census Bureau
September 27, 2001
3.1 | What should the Census Bureau consider when assigning Americans living overseas to a home state or census block? |
3.2 | Should the Census Bureau strive for global coverage of Americans overseas? |
This report deals with the issues of including Americans overseas in future decennial censuses.
Policy issues: The U.S. Census Bureau considers prompt resolution of the policy issues a precondition to undertake research into the methodological issues. The direction of the methodological research could be dramatically different depending on the resolution of the policy issues.
What should be the goal(s) of the Census Bureau in making the overseas counts? --The goal could simply be to provide a snapshot estimate for general information purposes. To go beyond that to assign that population to states for apportionment would go to a higher level of challenge for the Census Bureau. To develop estimates for redistricting or for funds allocation (particularly those taking into account intercensal population changes), would present very formidable challenges. Each of these uses has different implications as to the data that we would have to collect, the level of geography to which we would have to allocate this population, data quality requirements, data collection methodology, and costs. A use that requires collecting more data, at lower levels of geography and with higher levels of accuracy, would be more costly and difficult to develop and conduct.
What would be the implication of conducting a census count that would take a different form than the stateside enumeration because it could not use a master address file and would be voluntary in nature?--Any census conducted of Americans overseas would have to be voluntary in nature for methodological reasons. We use a master address file for the stateside enumeration to identify and follow up with nonrespondents. In this way, we assure that we get a complete count and, thus, compliance with the mandatory nature of the census for all U.S. residents. If we cannot develop a control list that would permit follow up of nonrespondents, we could include in the census only those Americans overseas who want to respond to the census or whom we could independently identify from administrative records. Such a census would likely yield incomplete and unreliable data of unknown coverage. This could distort the state populations, potentially affecting the apportionment of congressional seats. Should we include in the census population groups for which we cannot obtain complete coverage and cannot measure coverage? Would the States play any role in raising voluntary participation?
What are the implications of applying the concept of usual residence to Americans living overseas?--Counting Americans living overseas for purposes of apportionment, redistricting, and federal funds allocation requires that we ensure compliance with the concept of "usual residence." This concept is the principle underlying the stateside residence rules. "Usual residence" is the place where a person lives and sleeps most of the time. Many Americans living overseas do not satisfy this definition of stateside "usual residence," even if they could provide a home state of record, would be eligible to vote, or expressed an intent to return to the United States.
A change in the application of this concept to accommodate the counting of Americans overseas could have an impact on how this concept is applied stateside, for example, for people with multiple residences, people residing in group quarters, students, and people with no usual residence. A change in these rules could have dramatic effects on the allocation of population or money to the states.
Which components of the Americans overseas population should the Census Bureau count?--Before we can count American citizens overseas, the universe has to be well-defined and described. For example, would the universe include, singly or in combination:
The definition of this universe has implications for data quality, coverage, data requirements, and data collection methodology. The Census Bureau would have to develop procedures to identify and count only respondents who meet the conceptual criteria for inclusion.
Approach for reaching a decision on the policy issues: We are proposing the following process to decide these policy issues:
Conceptual issues: How we define key concepts would have an implication on methodological issues if we have to allocate overseas Americans to a stateside geographic area.
What should the Census Bureau consider when assigning Americans living overseas to a home state or census block?--To include Americans living overseas in the state apportionment population counts, we have to assign them to a home state. Several possible alternatives for assigning home state exist. Each can be interpreted differently and may yield different answers. A given alternative might give more weight to some states than to others and, thus, bias the congressional apportionment.
To include Americans overseas in the counts used for redistricting or funds allocation would require assigning them to a census block. This means that Americans overseas would have to report a specific current housing unit or other residential address in the United States. Many overseas Americans have not lived in the United States for a long time, and some may have never lived in the United States long enough to have a usual stateside residence.
Should the Census Bureau strive for global coverage of Americans overseas?--The Census Bureau would have to address many issues relating to the global coverage of an enumeration of Americans overseas. A key issue is whether including only selected countries or groups would introduce systematic bias that might result in favoring one state or location over another.
Methodological issues: The methodological issues include, at a minimum:
Estimating the size of the overseas population: At this time, the Census Bureau cannot estimate accurately the size of the universe of the overseas population or the specified components other than the federally affiliated groups. No acceptable tested methodology for providing an independent measure of the coverage of that population is available. We need to conduct extensive research and development work to see if we can develop an estimate that would meet our quality standards.
Determining data sources: The U.S. Government, private sector companies, churches, and other organizations maintain lists of their employees or constituents living abroad. We need to research the feasibility of using such lists. Also, we need to research the feasibility both of using records maintained by some of the countries in which Americans reside and unduplication of those records with the stateside enumeration. The accuracy of each source would no doubt vary enormously with different implications for individual States.
Determining the "best" data collection strategy: The options for collecting data are self-enumeration, administrative records, or some combination of the two. We have to address many issues for each of these options. Whatever the methodology used, a key issue is the need to assure independent verifiability of data by the federal government.
The congressional conference report, H.R. 106-1005, requested a report from the Secretary of Commerce, due no later than September 30, 2001, on the:
"methodological, logistical, and other issues associated with the inclusion in future decennial censuses of American citizens and their dependents living abroad, for apportionment, redistricting, and other purposes for which decennial census results are used."
In response to this request, this report presents the issues the Census Bureau has to address to decide the feasibility of including Americans living abroad in future decennial censuses.
These detailed issues would serve as the foundation for developing a research and testing agenda for this program.
The Census Bureau considers prompt resolution of these issues a precondition to undertake research into the methodological issues of conducting a census of Americans living abroad. The direction of the methodological research could be dramatically different depending on the resolution of these issues. We can continue with some basic research activities while the process to decide these issues is underway. Still, we cannot conduct in-depth operational research and testing until we decide these issues. Examples of research that we can start include the identification of potential administrative records’ sources and research on barriers to obtain them.
The four major possible goals are:
Each of these uses has different implications as to the data that we would have to collect, the level of geography to which we would have to allocate this population, data quality requirements, data collection methodology, and costs. For example:
To assign that population to states for apportionment would go to a higher level of challenge for the Census Bureau. To develop estimates for redistricting or for funds allocation (particularly those taking into account intercensal population changes), would present very formidable challenges. A census use that requires collecting more data, at lower levels of geography, and with higher levels of accuracy would be more costly and difficult to develop and conduct.
Any census conducted of Americans overseas would have to be voluntary in nature for methodological reasons. At this time, we do not have a way of developing and maintaining a master address file or other comprehensive control list for Americans overseas, as we do stateside. We use such a control list for the stateside enumeration to:
In this way we assure compliance with the mandatory nature of the census for all U.S. residents. If we cannot develop a control list that would permit follow up of nonrespondents, we could include in the census only those Americans overseas who want to respond to the census or whom we could independently identify from administrative records. Such an exercise would likely yield incomplete and unreliable data. The result of an inaccurate count of overseas Americans could distort the state populations, potentially affecting the apportionment of congressional seats. Should we include in the census population groups for which we cannot obtain complete coverage and cannot measure coverage? Would the States play any role in raising voluntary participation?
Counting Americans living overseas for purposes of apportionment, redistricting, and federal funds allocation requires that we ensure compliance with the concept of "usual residence." This concept is the principle underlying the stateside residence rules. "Usual residence" is the place where a person lives and sleeps most of the time. Many Americans living overseas do not satisfy this definition of stateside "usual residence," even if they could provide a home state of record, would be eligible to vote, or expressed an intent to return to the United States.
A change in the application of this concept to accommodate the counting of Americans overseas could have an impact on how this concept is applied stateside, for example, for people with multiple residences, people residing in group quarters, students, and people with no usual residence. A change in these rules could have dramatic effects on the allocation of population or money to the states.
Before we can count Americans overseas, the universe has to be well-defined and described. For example, would a census of Americans overseas include, singly or in combination:
A related issue is determining which dependents living overseas we should include in the universe. Should we include, for example:
The definition of this universe has implications for data quality, coverage, data requirements, and data collection methodology. The Census Bureau would have to develop procedures to identify and include only respondents who meet the conceptual criteria for inclusion.
We are proposing the following process to decide these policy issues:
This section discusses the definition of three concepts that would have implications on the implementation methodology if use of the counts requires allocating overseas Americans to a stateside geographic area.
We would have to assign Americans living overseas to a home state to include them in the state apportionment population counts. Possible alternatives for assigning home state and issues associated with each are as follows:
Each method of reporting home state can be interpreted differently and may yield very different answers. The Census Bureau would need to deliberate carefully in deciding how to assign home state, as a given method might give more weight to some states than to others and, thus, bias the congressional apportionment.
To include Americans overseas in the counts used for redistricting or funds allocation would require assigning them to a census block--the lowest level of census geography. This task would be even more complicated than assigning the overseas residents to a home state, because it would require their being able to report a specific current housing unit or other residential address in the United States. Many overseas Americans have not lived in the United States for a long time, and some may have never lived in the United States.
The Census Bureau would have to address issues relating to the global coverage of an overseas census count.
In discussing the factors that counseled against counting private Americans living abroad for Census 2000, the Census Bureau emphasized that one of the most significant concerns was the possibility of systematic bias, which might result in favoring one state over another for purposes of congressional apportionment. In his June 9, 1999, testimony before the Subcommittee on the Census of the House Committee on Government Reform, then Census Bureau Director Kenneth Prewitt testified at some length on this point:
"Let us say that we did a particularly good job of counting in Mexico where many, many Americans live in retirement communities, and so forth. We did a very poor job in Canada, where there are also a very large number of overseas Americans . . . . Now, if that were to happen, we do better in some parts of the world than other parts of the world, the question we would have to ask ourselves: Are the people, the Americans who live in those different parts of the world, randomly distributed or proportionately distributed according to home of record? My guess is that the Americans living in Canada are much, much more likely to come from the northern tier of the states, and the Americans living in Mexico are much more likely to come from the southwestern tier of the States. So, if we did a very good job counting in Mexico and a very poor job counting in Canada, then we would have produced a distortion to the apportionment counts. It is a distortion to the magnitude that we could not estimate, nor could we correct for it."
In later responses to questions from Subcommittee members, Director Prewitt again stressed this point: "If we ended up undercounting, disproportionately, [Americans living in] Canada, and overcounting, disproportionately, Americans living in Mexico, then unless the people living in Canada and Mexico came equally from the same states, we would have penalized and rewarded states differently."
This section describes the methodological issues of estimating the size of this population and of counting Americans overseas.
Congress also requested that this report include "estimates of the number of Americans living abroad in the following categories: federal civilian employees, military personnel, employees of business enterprises, employees of nonprofit entities, and individuals not otherwise described." While data are available from Census 2000 on federal civilian employees and military personnel1, the Census Bureau does not know the number of private Americans living abroad under the other categories. No accurate estimate exists of the total number of Americans living abroad or of the other components of this population.
At this time, we cannot estimate accurately the size of the universe of the overseas population or the specified components other than the federally affiliated groups. A master address file or other comprehensive control list for this population does not exist, nor do we have an acceptable, tested methodology for providing an independent measure of the coverage of that population or its components. We need to conduct extensive research and development work to assess the feasibility of developing such a control list and an estimate that would meet our quality standards.
We can probably estimate the universe to a rough order of magnitude using existing administrative records. However, such an estimate would have significant limitations.
If we could get, merge, match, and unduplicate various of these administrative records files, then we could produce a better estimate of the size of this universe. Even so, to the extent that many Americans abroad may not be connected to any administrative records system, this estimate would not be accurate. Moreover, technical and resource issues then come to the forefront. We need to research and resolve issues such as standardization of file formats, availability of computer space and processing resources, and matching and unduplication methodology.
If we could resolve access, coverage, and technical issues, this unduplicated file could serve as a rough control file to gauge completeness of an enumeration of this population, to identify the addresses of Americans abroad, or to serve as a proxy for a direct enumeration. We will discuss these topics in more detail in the next sections of this report.
The resolution of the policy and conceptual issues we discussed earlier will drive the methods and procedures to count U.S. citizens abroad. One of the most significant of our concerns is how to eliminate the possibility of systematic biases that might result in favoring one state over another for purposes of congressional apportionment. We would have to develop and test methodologies that would not introduce any systematic bias into the overseas counts. This would not be an issue if we do not have to allocate the overseas counts back to a stateside geographic area.
We need to answer some or all of the following questions in selecting a methodology to count Americans abroad, depending on how the counts would be used. How can we:
Two basic possible methods of counting Americans overseas in a census are obtaining counts from administrative records or through self-enumeration. Using a combination of both methods is also possible.
Are there U.S.-based organizations or other countries that maintain administrative records of such quality (as to coverage and accuracy) and detail that we can use the records, themselves, to abstract the necessary information? We have to research and resolve many issues to answer this question. Besides the specific issues for each data source mentioned below, global issues include political, privacy, and legal barriers to obtaining the information the Census Bureau needs from each of these record sources. Other issues are duplication between the various sources, operational, logistical, content, and timing issues.
We can consult three types of administrative records’ sources; records from: official U.S. entities, private entities, and other countries.
This source includes records from:
We do not know if these files contain information required for uses beyond apportionment. We need to conduct research to make this assessment. Privacy concerns may be an issue.
However, inconsistency among the states, coverage, duplication, and availability of the SF-76 forms are key issues.
Even if we could obtain these forms before the states destroy them, other issues include timing of availability of the forms relative to Census Day and the legal deadlines for delivering apportionment and redistricting counts. Also, we must consider potential legal and privacy barriers of obtaining these forms from all states.
Private entities include trade and other professional organizations, social and religious organizations, universities and colleges with study abroad programs, and multinational corporations. Using administrative records maintained by private organizations could result in biasing apportionment in favor of states that have organizations within their borders that voluntarily submit such information to the Census Bureau for inclusion in the apportionment counts. To avoid this, we would have to identify the universe of these entities systematically and obtain their records.
We would have to conduct extensive research to answer these and other questions:
These include population registers, permit records, and other countries’ published census records. Intensive research would be required into both the kinds of official statistics published by counterpart statistical agencies and the kinds of routine administrative records maintained by other countries to track resident foreign populations.
A number of countries require foreigners to register with local or national authorities, obtain work permits, and even obtain permission to move within the country. Questions that research would answer include:
Another possibility is to consult the official published census records of other countries to see whether they include U.S. citizens as a separate group to get estimates of Americans living in those countries. The census publications for many countries, frequently going back a considerable time, are housed in the resource collection of the International Programs Center, Population Division, of the Census Bureau. But, earlier research shows that we cannot use this source to obtain census counts for apportionment, redistricting, or other census uses, as they do not contain the detailed information needed for these purposes.
A self-enumeration approach of the target population of Americans abroad presents a unique set of challenges. The Census Bureau would have to design and test a questionnaire that would allow us to identify and collect the required information for the agreed-upon census use from the target population only. This process would require the cooperation of the State Department and a variety of U.S. and foreign organizations, both public and private. Specific issues include:
The following are options to investigate and test, as appropriate:
To obtain responses from even a small segment of this population, the Census Bureau would need to launch an intensive effort to publicize the overseas census. The following are options to investigate:
Identifying and contacting all the possible organizations and media outlets would require a major effort from the Census Bureau. Undoubtedly, other outreach efforts, tailored to fit the needs of the diverse segments of the overseas population, would have to be conceived, investigated, and pursued to increase response.
The issues of developing a control list that the Census Bureau could use to track progress of the overseas enumeration and identify nonrespondents were already discussed. Even if we could develop such a control list, the legal and logistical issues of conducting a nonresponse follow-up operation would have to be considered.
If the Census Bureau is not able to conduct a follow up of nonrespondent households overseas, it could not estimate the number of Americans overseas missed and those counted more than once. Would results with an unknown coverage error be acceptable to stakeholders?
How do we ensure that the Americans overseas are not overcounted? How do we unduplicate within the overseas component and between the overseas and stateside counts? Any census would require means to identify and remove duplicates from the counts of U.S. citizens abroad. The Census Bureau has procedures to deal with this issue for the population resident in the United States, but developing such a system for people living outside the United States would present enormous operational problems. For example, there is no address list that will support an unduplication procedure.
Using different methods for different overseas groups or for different countries might be possible. For example, we might use:
Administrative records to count the federally affiliated population abroad and possibly other segments of Americans overseas, if complete, accurate records for them are available that can be verified by the federal government.
Developing and controlling such a mix of methods would be time consuming, difficult to carry out, costly, and error prone.
A key requirement when attempting to count Americans living abroad is to assure independent verifiability by the federal government of the data required to decide if the person is eligible for inclusion in the census for the predefined census uses. The Census Bureau has serious concerns about the ability to validate responses of overseas Americans. For example:
Using passport numbers or SSNs for validation has the following limitations:
With respect to the use of administrative records, attempting to add persons from new administrative records sources that have not been carefully tested and reviewed would violate a tenet that the Census Bureau use verifiable methods for counting the population. Although the coverage error introduced by obtaining counts of persons living overseas from multiple, independently derived, nonverifiable administrative records is not quantifiable, one would expect it to be high relative to that of the resident population.
The research and development work required to address all the issues we mentioned is extensive. The table below presents broad research, testing, and consultation milestones the Census Bureau expects to accomplish over the next five years.
Milestones | Fiscal Year | |||
---|---|---|---|---|
02 | 03 | 04 | 05 | |
Consultation with stakeholders | ||||
Identify the stakeholders | X | |||
Hold a conference with stakeholders | X | |||
Plan for additional consultation activities | X | |||
Conduct additional consultation activities | X | X | X | X |
Research and development | ||||
Develop a detailed research, testing, and consultation plan | X | |||
Continue analysis of policy issues | X | |||
Document results of analysis of policy issues | X | |||
Continue analysis of conceptual issues | X | |||
Document results of analysis of conceptual issues | X | |||
Identify U.S.-based organizations that may provide administrative records | X | |||
Identify key contacts from those organizations | X | |||
Conduct research on issues relating to administrative records (access, privacy, accuracy, coverage, and content) from U.S.-based organizations | X | X | ||
Identify countries that maintain records of foreign residents | X | |||
Identify key contacts in those countries | X | |||
Conduct research on issues relating to records from countries that keep track of foreign residents | X | X | ||
Conduct research on issues relating to using a self-enumeration approach | X | X | ||
Conduct research on issues relating to validation, verification, and duplication | X | X | ||
Research published census reports from other countries | X | |||
Document results of methodological research | X | |||
Present research results and Census Bureau’s recommendations to Congress and other stakeholders | X | |||
Testing | ||||
Define goal, objectives, and strategy for testing possible enumeration methodologies in 2004 | X | |||
Conduct preparatory activities for 2004 test | X | X | ||
Carry out a test of possible methodologies in some countries | X | |||
Evaluate and document test results | X | |||
Present test results and Census Bureau’s recommendation to Congress and other stakeholders | X | |||
2010 Planning | ||||
Develop a detailed cost estimate of an overseas enumeration in the 2010 census, including additional tests possibly in 2006 and 2008 | X | |||
Define goal, objectives, and strategy for a test in 2006 | X | |||
Begin preparatory activities for a test in 2006 | X |
Planning for and taking a decennial census is a long, expensive, and time-intensive process. This would also be true for planning for a census of all Americans abroad. New methodologies and systems must undergo extensive research, testing, and analyses before being incorporated into a final comprehensive census plan. The planning process would involve coordination with multiple advisory/stakeholders groups, and reports to and oversight from Congress.
1 The Census 2000 apportionment counts included close to 580,000 federal employees and dependents (226,363 military personnel, 30,576 civilian employees, and 319,428 dependents of military and civilian employees).
[PDF] or denotes a file in Adobe’s Portable Document Format. To view the file, you will need the Adobe® Acrobat® Reader available free from Adobe. This symbol indicates a link to a non-government web site. Our linking to these sites does not constitute an endorsement of any products, services or the information found on them. Once you link to another site you are subject to the policies of the new site.