Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Executive Summary THE PANEL ON RESEARCH ON FUTURE CENSUS METHODS has a broact charge to review the early planning process for the 2010 census. Its work includes observing the operation of the 2000 census, deriving lessons for 2010, anct acivising on effective eval- uations and tests. This is the panel's third report; we have previously issucct an interim report (National Research Council, 2000) offering suggestions on the Census Bureau's evaluation plan for 2000 anc! a letter report (National Research Council, 2001c) commenting on the bureau's proposed general structure for the 2010 census. The panel is expected to finish its work by the enc! of 2003, at which time it will issue a final report. EMERGING STRUCTURE OF THE 2010 CENSUS The Census Bureau's current plans for the 2010 census are predi- catec! on the completion of three major initiatives: 1. MAF/TIGER Enhancements Program. A specific set of improve- ments has been proposed to improve the Census Bureau's act- clress list (Master Aclclress File, or MAF) and geographic database (Topologically Integratecl Geographic Encocling and Referencing System, or TIGER).
2 PLANNING THE 2010 CENSUS 2. American Community Survey (ACS). The decennial census long form will be replacer! by a continuous survey, thus permitting a short-form-only census in 2010. The ACS covers the same social, economic, and demographic ciata as the census long form but will . . . . prov1c e estimates in a more time. y manner. 3. Early Integrated Planning. The Census Bureau hopes that early attention to planning will make census tests leacling up to 2010 more informative and useful. The Census Bureau's emerging 2010 census plan also includes clevel- opment of mobile computing crevices for use in nonresponse follow- up work and use of multiple response mocles (mail, Internet, and tele- phone). IMPORTANCE OF INTEGRATION Of the three legs of the Census Bureau's 2010 census strategy, the panel considers early integrated planning as perhaps the one most cru- cial to a successful 2010 census and, in some respects, the harclest to accomplish. In an earlier report, we recommenclect that: The Census Bureau should produce a"business plan" that provides an overall framework for development of the 2010 census. Such a plan should include: (1) a clear statement of objectives, (2) an approximate timeline for completion of tasks, (3) a cost-benefit analysis of the various components of the plan, and (4) a fuller expla- nation of how intra-Bureau efforts will be coordinated. In assessing the costs and benefits (both monetary and nonmonetary) of a reengineered 2010 census, attention should be given to potential effects of new processes on census coverage and differential undercount and their measurement. Clear articulation of such a plan, backoct by empirical evidence from evaluation studies anct census tests, anct with careful attention to both costs and benefits, woulc! help greatly in developing the design of the 2010 census.
EXECUTIVE SUMMARY 3 This interim report is neither intenclec! to be an exhaustive assess- ment of the Census Bureau's plans for the 2010 census nor a compre- hensive checklist of problem areas for which a solution would ensure a quality 2010 census. It concentrates primarily on two areas in which the Census Bureau has been particularly active anct in which informa- tion has been most available to the panel: modernization of the bureau's geographic resources anct efforts to moclel anct develop the technical in- frastructure of the census. This report also offers initial comment on the American Community Survey. In assessing these major initiatives, we have tried to suggest areas in which it is particularly important that the Census Bureau demonstrate how the initiatives support each other in an integrated fashion. The panel's discussions with the Census Bureau on several topics of interest, including strategies for coverage measurement, refining res- iclence rules, anc! new enumeration methods, are still developing. Thus we clefer cletailect discussion of these topics to the final report. REAL REENGINEERING: BUSINESS PROCESS AND TECHNICAL INFRASTRUCTURE The operations of the 2000 census were carrier! out, supported, or monitored by various software programs, the computer systems on which those software programs operated, anc! the telecommunication systems that connected the computer systems. The totality of these programs, and potentially other programs/systems carrying out many other functions, is referrec! to as the technical infrastructure of the census. The computer systems that supported the 2000 census were princi- pally implemented using customized in-house software programs. Col- lectively, the 2000 census technical infrastructure was functional but clevelopect at high cost anct high risk, without acloquate time for cle- velopment and testing to ensure that systems were capable of meeting functional neects. In orcler to develop an improver! technical infrastructure for the 2010 census, a necessary first step is to reengineer an appropriate logical infrastructure or business process moclel. In acictition to improving the bureau's computer systems, articulating a logical architecture for the 2010 census is important in orcler to develop a more complete
4 PLANNING THE 2010 CENSUS unclerstancling of how the 2010 census will operate anc! to provide a basis for comparing alternative design features. The panel is strongly impressed by a Census Bureau pilot study that clevelopec! a logical architecture for the 2000 census anc! subsequently aclaptect part of that architecture to reflect a limited set of 2010 cen- sus assumptions. The modeling language (IDEFO) anc! the software tool (System Architect) usect to support this work are both sounct. We therefore strongly encourage the Census Bureau to continue its infras- tructure reengineering activities and recommend that the bureau pro- ceecl as quickly as possible to develop alternative business process mocl- els for the 2010 census as a total system (Recommenciation TINS. Reengineering a comprehensive technical infrastructure supportive of the 2010 census business process moclel will require considerable re- sources anct close collaboration among all Census Bureau units. It will also require a strong coordinator, whom we refer to as the system ar- chitect for the decennial census, along with a cleclicatec! full-time staff (Recommenclation TINT. Finally, such an enterprise will not be suc- cessfully undertaken without strong commitment and ongoing support from the highest management levels at the Census Bureau. MODERNIZING GEOGRAPHIC RESOURCES Though the extent of inaccuracy in the TIGER database is un- known, evidence suggests that roacls, bounclaries, and other geographic features are misplaced with sufficient frequency that TIGER is in neec! of a comprehensive upclate. Hence, the panel supports the TIGER re- alignment portion of the MAF/TIGER Enhancements Program. That said, the plausibility of the Census Bureau's ambitious realignment timetable would be bolstered considerably through development of a cletailect work plan Including some notion of the orcler in which coun- ties will be initially upciatecI), specification of the clesirec! positional accuracy of the realigned TIGER file as well as the local anct tribal geographic files used to carry out realignment, and specification of the change-cletection program to acict upclates in later years. The panel also supports the Census Bureau's efforts to implement a moclern processing environment for its geographic resources using commercial off-the-shelf software.
EXECUTIVE SUMMARY 5 Though the panel is satisfied that the enhancements program may substantially improve TIGER, it is our view that the program falls se- riously short in enhancing the MAF. In part clue to as-yet unspecified plans for local partnerships and the Community Aclclress Upclating Sys- tem (CAUS), there is no comprehensive plan to improve the MAF by acicling new aciciresses, screening for duplicates, anc! generally ensuring that housing unit rosters are as complete anct accurate as possible. The Census Bureau's current strategy shows signs of repeating costly errors from the 2000 experience. The Census Bureau argues that the combination of three activities will result in an up-to-date aciciress list for the 2010 census: twice-yearly upclates basect on the U.S. Postal Service's Delivery Sequence File (DSF), CAUS, en c! implementation of a program akin to the Local Update of Census Aciciresses (LUCA) usect in the 2000 census. Specifically, the DSF upclates are intenclect to provide aciciress coverage in urban areas with city-style aciciresses, anct CAUS is intenclect to provide coverage in rural areas; these would be supplemented by local review. However, cluring construction of the 2000 census MAF, the Census Bureau argucct that a similar combination DSF updates couplet! with LUCA woulc! provide sufficient acictress updates. But the bureau later expressed doubts anct concluctec! a costly complete block canvass of aciciresses. Evaluation work on the contributions of DSF anct other sources to the 2000 MAF has not been completecl, so it has not been demonstrated that the DSF is better poisec! to be a backbone source of aciciresses in 2010 than it was in 2000. Meanwhile, CAUS raises concerns because of the not-yet-securec! funding for the ACS; since CAUS relies on a fully fielclect ACS, its budgetary viability is uncertain. It is also unclear how CAUS fielc! staff woulc! be targeted to particular geographic areas anc! to what extent CAUS acictress listing cluties might detract from their work in following up with ACS respondents. We are assuming that the Census Bureau hopes to avoict a complete block canvass prior to the 2010 census, given the cost of that operation anct the fact that it was treated as a last resort in 2000. However, it is difficult to see how a full block canvass can be averted without a clearer plan for CAUS anct without evidence that the combination of DSF anct LUCA reacting up to 2010 can overcome the last-minute doubts that arose in the late 1990s.
6 PLANNING THE 2010 CENSUS Accordingly, the panel recommends that the Census Bureau pro- vide clearer details on how the MAF/TIGER Enhancements Program and other geographic programs will add missing addresses, remove 1 1 · 1 1 1 1 1 1 ~ r ~ To ,~ 1 duplicate addresses, and generally correct the MAL t~ecommenda- tion MAFIA. In support thereof, the bureau should complete and improve evaluation work on address list development issues (Recom- mendation MAFIA. The panel further recommends that the Census Bureau would benefit from vesting responsibility for coordinating MAF improvement and research in one office, since this responsibil- ity is currently diffused over several divisions of the Census Bureau (Recommendation MAFIA. Finally, the panel suggests that the bureau promptly develop and describe plans for local geographic partnerships (Recommendation MAFIA. AMERICAN COMMUNITY SURVEY The American Community Survey (ACS) is a major household sur- vey anticipated to include 250,000 housing units each month. Relative to other national household surveys, the large sample size of the ACS should allow it to provide small-area information population charac- teristic profiles for counties, cities, and other local areas. However, in order to approximate the sample size of the census long form, ACS es- timates for smaller geographic and population groups would be based on 3- or 5-year moving averages. While the census long form can only provide these small area profiles in once-per-decade snapshots, the ACS collects information continuously throughout the decade. In the cen- sus context, the prime advantage of a full-fledged ACS is the resulting prospect of a short-form-only census. The most basic question the panel faces regarding the ACS is whether it is a satisfactory replacement for the census long form. We recognize that significant estimation and weighting challenges must be addressed and that more research is needed on the relative quality of ACS and long-form estimates, with particular focus on measurement error and error from nonresponse and imputation (Recommendation ACS-1~. The Census Bureau must do significant work in informing data users and stakeholders of the features and the problems of work- ing with moving average-based estimates (Recommendation ACS-3~. The Census Bureau must also complete evaluations comparing 2000
EXECUTIVE SUMMARY 7 census long-form data with data from the ACS test sites, from the Census 2000 Supplementary Survey (a prototype ACS), and from the Supplementary Surveys concluctect in 2001 anct 2002. Making these data available (protecting confidentiality to the wicler research com- munity could help greatly in building the research base for the ACS (Recommenciation ACS-2~. We clo not see any looming flaw so large in magnitude that full ACS implementation should be set asicle. We therefore encourage full con- gressional funding of the ACS. It is important, though, that Congress recognize that functing of the ACS should be viewoct as a long-term commitment. Cuts in funding in subsequent years (anc! with them re- cluctions in sample size) will impair the overall quality of the survey, with first anc! most pronounced impact on the ability to produce esti- mates for small geographic areas and population groups. However, given that funding for the ACS is not yet securecl, the Census Bureau must begin contingency planning to be prepared in the event that full functing is not proviclect (Recommenclation ACS-4~. PLANS FOR TESTING The Census Bureau plans to conduct at least four major census tests prior to 2010 in orcler to try out new procedures anc! finalize program plans. In 2003, a national sample has been askoc! to participate in a test of possible response mocles (e.g., mail, Internet, anct telephone) anct of rewordings of the questions on race anc! Hispanic origin. This 2003 test is being aciministerect by mail anct cloes not involve an active fielct deployment of enumerators to conduct follow-up questioning. A 2004 Census Field Test will cover a wicler range of census operations, incluct- ing fielc! follow-up, in alreacly-cleterminec! sites in Georgia anc! New York. The 2004 fielct test should be the first major test of the Census Bureau's plans for using mobile computing crevices. In 2006, an as-yet unspecified test will focus on general anct reengineerect data capture, information, anc! management systems. Finally, a full-flecigec! ciress re- hearsal will be concluctect in 2008; the Census Bureau hopes that, by avoicting a late-clecacle crush in designing census plans as occurrect in 2000, the 2008 exercise will be a true rehearsal rather than a late ex- perimental test. These major test initiatives will be supplemented by
8 PLANNING THE 2010 CENSUS a variety of other small-scale feasibility studies anc! analytic work over the course of the clecacle. The Census Bureau shared plans for the 2003 census test with the panel at its September 2002 meeting en c! in subsequent discussions. However, the plans were shared with us too late for us to suggest anct for the Census Bureau to effect any meaningful change in the 2003 test plan. Plans for 2004 anct 2006 are still uncler development, anct the panel looks forwarc! to continucc! work with the Census Bureau on those plans. ASSESSING THE OVERALL STRATEGY A major conclusion of the panel is that discussion of the 2010 cen- sus design neects to be more fully informed by the evaluation of various tracle-offs the costs anct benefits of various reasonable approaches in orcler to make wise decisions. For example, there are costs anc! benefits associated with the following decisions: 1. How accurate will ACS information be relative to long-form in- formation? 2. How inaccurate is the TIGER database at present? What accu- racy will result from various approaches to its enhancement, anct at what cost per unit of enhanced accuracy? Of what magnitude are the cost reductions that may result from a geographically correct TIGER system, such as more accurate routing of nonre- sponse follow-up enumerators? 3. With respect to nonresponse follow-up en c! the use of various types of personal computing crevices, what benefits would offset their respective costs? 4. What cost reductions (monetary and nonmonetary) will result from greater use of the Internet and other high technology means of enumeration, and what are the costs of greater use of these enumeration mocles? These anct other fundamental questions neect to be acictressect so that clec~sion makers can make informer! selection among the various design options.
EXECUTIVE SUMMARY 9 To ciate, the plan for the 2010 census has been presented to the panel with little supporting analysis. In part, this is attributable to the Census Bureau's neect to clevote more time anct resources than expected to the intensive, specialized evaluation studies that surrounclec! the Ac- curacy anct Coverage Evaluation (ACE) anct the clebate over statistical adjustment of 2000 census results. Whatever the reason for the delays, we are concerned that the selection of design options may have been hampered by a failure to fully exploit the existing information that is available from various management information anct monitoring sys- tems used to support the 2000 census. The panel's recommendation in its letter report that contingency planning shoulc! be factorec! into census planning remains valicI. This is particularly true with regard to the still-uncertain budgetary prospects of the ACS. But the possibility exists that budget fortunes reacting to 2010 may not be as generous as in the 2000 cycle, and contingency plans must be aclaptect for various levels of budget support. The Census Bureau shoulc! complete the remaining items on its planned evaluation list as expeditiously as possible. That clone, the Census Bureau must subsequently take stock of what it has learned from the evaluation studies, flesh out the 2010 census plan with em- pirical support, and fill gaps in knowledge through further analysis of 2000 census operational data or through census tests. .
