Planning for the 2010 Census and the ACS Survey

Looking beyond the 2000 census, this section covers planning for the 2010 census and the proposed American Community Survey.

ACS AS A CENSUS TESTING PLATFORM

The American Community Survey is a proposed national, continuous, mailout-mailback survey of 250,000 households per month, with field follow-up that makes use of techniques closely related to those used in the census. Therefore, rather than rely exclusively on the two or three large-scale census tests, which are always at least slightly limited in their generalizability by the specific locations selected, the Census Bureau could use the ACS as a platform for testing possible changes in the census. This work could serve as preliminary testing to the larger mid-decade tests for the census design.

A MATCH STUDY OF THE CENSUS SHORT FORM AND THE ACS

The decennial census makes use of one residence rule definition, the ACS uses a second, and a third approach is being tested in the alternative questionnaire study.12 As the Census Bureau is well aware (based on the

12

The difference between the census and the ACS residence rules stems from the fact that the ACS defines residence at a specific time while the decennial census defines residence as the predominant residence over a year.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 34
DESIGNING THE 2010 CENSUS: First Interim Report Planning for the 2010 Census and the ACS Survey Looking beyond the 2000 census, this section covers planning for the 2010 census and the proposed American Community Survey. ACS AS A CENSUS TESTING PLATFORM The American Community Survey is a proposed national, continuous, mailout-mailback survey of 250,000 households per month, with field follow-up that makes use of techniques closely related to those used in the census. Therefore, rather than rely exclusively on the two or three large-scale census tests, which are always at least slightly limited in their generalizability by the specific locations selected, the Census Bureau could use the ACS as a platform for testing possible changes in the census. This work could serve as preliminary testing to the larger mid-decade tests for the census design. A MATCH STUDY OF THE CENSUS SHORT FORM AND THE ACS The decennial census makes use of one residence rule definition, the ACS uses a second, and a third approach is being tested in the alternative questionnaire study.12 As the Census Bureau is well aware (based on the 12 The difference between the census and the ACS residence rules stems from the fact that the ACS defines residence at a specific time while the decennial census defines residence as the predominant residence over a year.

OCR for page 34
DESIGNING THE 2010 CENSUS: First Interim Report allocation of an experiment to this issue), confusion over residence rules is a source of possibly substantial error in the census. The panel suggests that the Census Bureau evaluate the methods used in an early 1990s study, the Living Situation Survey (Martin, 1999), in which it was shown that a large number of individuals, especially in Hispanic communities, were missed through use of a question similar to the current census question on a household's current residents. The Census Bureau needs to find the residence rule (within the set of rules satisfying legal and other restrictions) that results in the most accurate estimates. To learn more about this issue, the panel proposes an ACS-short-form match study in 2000 to examine this and other short-form measurement error issues.13 Another possible goal is to make use of some of the ideas in the Living Situation Survey in the 2010 version of the ACE survey. ADDRESS LIST IMPROVEMENT An accurate list of housing units will be needed for both the ACS and the 2010 census, making it a high-priority item. The accuracy of the address list may be the most important factor in determining the overall accuracy of a decennial census. The panel suggests some modifications to various aspects of both the MAF/TIGER software and the process for updating the master address file. First, MAF/TIGER needs to be modified so that it can accommodate inputs from geographic positioning system (GPS) technology to permit better identification of the location of rural addresses and boundaries in MAF/TIGER. Once this is accomplished, data collection augmented by GPS inputs should begin as soon as possible. Second, in the field check of addresses for MAF, every third address is directly visited, with indirect observation of the housing units on each side. It would be useful if the Census Bureau could make use of some form of evaluation to determine whether the information collected for housing units directly visited was better than that collected for neighboring housing units. Third, there needs to be a more specific plan on how to interact with localities to make the Community Address Updating System (CAUS) successful. For the 2000 census, the Census Bureau used information from 13 The ACS and the census long-form samples will not overlap, so a direct match study of ACS and census long forms is not possible.

OCR for page 34
DESIGNING THE 2010 CENSUS: First Interim Report the U.S. Postal Service as a source of address updates: the Address List Improvement Act of 1994, permitted for the first time, the transfer of data from the Postal Service to the Census Bureau for decennial census purposes. The Address List Improvement Act also established a program, called LUCA (Local Update of Census Addresses), which permits representatives of local governments (who have signed a confidentiality agreement) to review and provide feedback on the address list for their local areas. This program is based on the theory that local jurisdictions have unique access to administrative data and local knowledge. Ultimately, LUCA is supposed to be an integral part of CAUS, which is a larger program of continuous address list updating. Such updating is required by the ACS, which uses the MAF as a sampling frame, and will also be required for the 2010 decennial census. In order for CAUS to work, input from localities must be of high enough quality to be useful, and participation needs to be substantial and relatively uniform across the nation. There are several hurdles that the Census Bureau must overcome before any system of continuous address list updates can be put into place, including: establishing the strengths and weaknesses of the Postal Service Delivery Sequence File (DSF) updates; determining how other sources of information, including commercially available lists and input from local users, can be used to overcome these deficits; determining how to facilitate effective data transfer between Census Bureau staff in Washington, staff in the regional census offices, and representatives of localities; and gaining an in-depth understanding of what actually happened in the 2000 LUCA program (and its implications for the entire address list review process). Specifically, more needs to be known about the effectiveness of the DSF update in areas with city-style addresses as a tool for updating the MAF. Evaluations are needed to better determine MAF coverage (e.g., MAF's coverage of units in multiunit structures); determine the presence of newly constructed units in existing and new structures, including data on the lag between completion and reporting; and determine the accuracy of DSF updates for different types of areas. For the LUCA process in areas with city-style addresses, information is needed on what happened in 2000 by addressing two questions: How many areas participated in LUCA, with participation defined as the actual submission of address information? How were these participants distributed by census region, state, size of place, and other key demographic characteristics, such as population growth and level of undercount or overcount?

OCR for page 34
DESIGNING THE 2010 CENSUS: First Interim Report Information is also needed about the effectiveness of the LUCA effort in areas without city-style addresses. Finally, after the above work has been completed, the Census Bureau needs a comprehensive plan to engage localities in CAUS, including: the process that will be used to link Census Bureau headquarters, the regional census offices, and local area representatives; methods for updating the address list; and interface with ACS staff for update and feedback purposes. TARGETED REPLACEMENT QUESTIONNAIRES It was clear from the dress rehearsal that a targeted replacement questionnaire—a second questionnaire that is mailed to housing units that fail to return the initial mailed census questionnaire—would be likely to raise the mail response rate by around 7 percent. Such a rise would significantly reduce census costs and greatly aid subsequent census operations, probably improving the ultimate accuracy of the data collected. Research in the early 1990s (Dillman et al., 1994) demonstrated the value of a targeted replacement questionnaire, and it was therefore included in decennial census plans through 1996. However, it was then discovered that commercial firms were unable to address the number of envelopes for housing units that would be expected not to return a questionnaire in the amount of time needed to keep on schedule. Given this problem and the number of duplicates that would probably have resulted from the use of a blanket replacement questionnaire (mailing of a replacement questionnaire to all housing units, including initial respondents), as demonstrated in the 1998 census dress rehearsal, the Census Bureau decided not to use a blanket replacement questionnaire in the 2000 census. Thus, the 2000 census will include neither blanket nor targeted replacement questionnaires. However, technology is quickly changing, and the feasibility of such an important improvement should be examined as soon as possible. The Census Bureau needs to find out how to make use of a targeted replacement questionnaire in 2010 (see National Research Council, 1999). LEGISLATIVE AND LEGAL INITIATIVES Several legislative and legal changes would be needed to implement some proposed features for the 2010 census. These changes include use of sampling for nonresponse follow-up; modifications to administrative records systems, such as inclusion of the address of residence on tax forms;

OCR for page 34
DESIGNING THE 2010 CENSUS: First Interim Report and, as discussed above, greater access to administrative records (e.g., food stamp records). These initiatives should receive attention as soon as possible. Also, the Census Bureau should inform policy makers that decisions on ACS have extremely important implications for the 2010 census and, therefore, that decisions about ACS should be made in light of that relationship. ACS has been correctly viewed as a substitute for the census long form, but there are many other relationships that should be considered, especially the continuous update of the MAF through ACS processes. AN ACS ADVISORY GROUP The development of the ACS raises a number of issues related to the quality of and planning for the 2010 census. There are also many other important technical issues raised by the introduction of the ACS into the federal statistical system. Formation of a technical working group could help to address many of these issues. LINKING RESPONSES FROM 1990 AND 2000 CENSUS QUESTIONS ON RACE AND ETHNICITY A sample of the Current Population Survey in the summer of 2000 will be used to develop a mapping between responses based on the question format and sequence relative to the race and ethnicity questions on the 1990 and the 2000 censuses. The panel is concerned about how the model that accomplishes this will be developed and evaluated. The statistical relationship between responses to the 1990- and 2000-style questions may exhibit strong local variation, which the sample may not be large enough to identify or which the model may not accommodate. Plans for model development and evaluation are crucial and may have important implications for the sample design. Moreover, the panel is concerned about how the results from the effort will be reported and disseminated. The Census Bureau could produce a simulated double-coded sample, as was done when the industry and occupation codes were changed from 1970 to 1980, but at present there appears to be no plan to do so.14 14 A simulated double-coded sample is a method in which two different coding systems are applied to a subset of the sample and the resulting joint distribution is used for (multiple) imputation of the missing codes for the remaining cases in the sample (for details, see Clogg et al., 1991).

OCR for page 34
DESIGNING THE 2010 CENSUS: First Interim Report As a final comment, we note that the Census Bureau has contracted to obtain advice on how to take advantage of new technologies. While this is a start on the need to incorporate the latest technology in census planning and operations, it would be helpful if familiarity with technology were integrated more fully into the staff. Also, it would be extremely useful if some sort of framework were established that made the Census Bureau aware, early on, of opportunities created by advances in technology, though the panel has no specific suggestions. Even more difficult, there is a need for the Census Bureau to be able to predict future technologies that would be useful in alternative census designs. To facilitate the use of the latest technologies in the planning for the 2010 census, testing and decisions of technology-intensive systems should be delayed as long as possible in the cycle for the 2010 census.