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.
Modernizing Geographic Resources A BASIC TENET OF SURVEY RESEARCH is that the develop- ment of a sampling frame a listing of all units eligible for inclusion in the sample from which the sample is drawn is crucially important to the quality of the survey. Systematic biases or flaws in the frame may incluce serious errors of inference based on the survey results. Accorclingly, when considering a decennial census a survey of grant! scale it is difficult to overemphasize the importance of the underlying sampling frame. The quality of the acictress list to which questionnaires are mailed can lead to the omission or duplication of people or of entire housing units anct can hincler the goal of count- ing each resident once and only once within the precise geographic boundaries in which they belong. Hence, this panel stated in its first interim report that "the aciciress list may be the most important factor in determining the overall accuracy of a decennial census" (National Research Council, 2000:351. The "three-leggecl stool" strategy outlined by the Census Bureau in describing the early plans for the 2010 census includes attention to modernizing the Census Bureau's primary geographic resources: 39
40 PLANNING THE 2010 CENSUS · the Master Address File (MAF), the source of addresses not only for the decennial census, but also for the Census Bureau's numer- ous survey programs; and · the Topologically Integrated Geographic Encoding and Ref- erencing System (TIGER), a database describing the myriad geographic boundaries that partition the United States. The specific set of activities that the Census Bureau has described to achieve this modernization is known as the MAF/TIGER Enhance- ments Program (MTEP). In terms of its spirit and nominal goal, the MAF/TIGER Enhancements Program may be of paramount im- portance in terms of its potential impact on the quality of the 2010 census. In this chapter, we review the MAF/TIGER Enhancements Pro- gram. As we will describe in detail, we support completion of the En- nancements ~ ro "ram, which should provide some necessary improve- ments to the TIGER database. However, we are concerned that the 1 . TO Enhancements Package does little to enhance to improve the MAF. More generally, the Census Bureau's strategy for dealing with the MAF shows signs of repeating costly and chaotic processes from MAF con- struction in the 2000 census. OVERVIEW: CURRENT STATE OF MAF AND TIGER Before we discuss the specific enhancements program that has been initiated by the Census Bureau, it is useful to first briefly review the nature and status of the two geographic systems addressed by the package to get a sense of exactly what is in need of enhancement. The Master Address File Purpose and Scope The Census Bureau's Master Address File (MAF) is, in essence, pre- cisely what the name implies; it is the Census Bureau's complete inven- tory of known living quarters in the United States and its island areas. The MAF contains a mailing address for those living quarters, if one
MODERNIZING GEOGRAPHIC RESOURCES 41 exists. For housing units or living quarters without mail aciciresses, cle- scriptive aciciresses (e.g., "white house with brown shutters on left") may be coclect. The MAF also includes an intricate set of flags anct indicators that inclicate sources from which the aciciress was obtainer! anc! the time when it entered the MAF. In principle, the MAF is a constantly evolving en cl continually updated resource; the "snapshot" of the MAF that is extracted anct usect to conduct the decennial census is called the Decennial Master Aciciress File, or DMAF. Construction of the 2000 Census Master Address File The concept of a continuously maintained MAF is a relatively new one; in the 1990 anc! earlier censuses, aciciress lists were compiled from multiple sources prior to the census (most recently from commercial venclors) anc! were not retained after the census was complete. Follow- ing the 1990 census, the idea of maintaining the acictress list to sup- port not only the decennial census but also the Census Bureau's other survey programs took holcI. In part, writes Nash (2000:1), "a major impetus for this change was the unclercounts experienced in the 1990 anc! earlier decennial censuses, nearly a third of which was attributed to entirely missing housing units." An initial MAF was constructed using the city-style aciciresses1 on the Aciciress Control File (ACF) clevelopec! for the 1990 census (Hirschfelct, 2000~. To populate the MAF, the Census Bureau "clevisect a strategy of re- clunclancy using a variety of sources for aclclresses," thus "Eassuming] responsibility for developing a comprehensive, uncluplicatect file of act- clresses" (Nash, 2000:1~. Most prominent of the update sources were two that were endorsed by one of our predecessor Committee on Na- tional Statistics (CNSTAT) panels on the decennial census (National Research Council, 1995:5), which recommenclec! that the Census Bu- reau "clevelop cooperative arrangements with states en cl local govern- ments to develop an improver! master aciciress file" anc! that the U.S. 1A city-style address is one that can be specified by a numeric identifier (e.g., 305) in combination with a street name (e.g., Park Avenue), possibly with a specific subunit or apartment identifier. By comparison, non-city-style addresses are those that cannot be mapped to particular streets in this fashion, such as "Rural Route, Box 7" or a post office box.
42 PLANNING THE 2010 CENSUS Postal Service be given "an expanclec! role" in census aciciress list oper- ations. Both these recommendations were significant in that they re- quirect legislative authority in orcler to operate within the prohibition on release of confidential data coclifiec! in U.S. Cocle Title 13, the legal authority for census operations.2 Congress granted this authority in the Census Aciciress List Improvement Act of 1994 (Public Law 103- 4301. The Delivery Sequence File One provision of the Census Aciciress List Improvement Act authorized the Census Bureau to enter into a data sharing arrangement with the U.S. Postal Service, under which the Postal Service woulct regularly share its Delivery Sequence File (DSF) with the Census Bureau.3 The DSF is the Postal Service's master list of all delivery point acictresses servect by postal carriers.4 The name of the file derives from the Postal Service-specific data coclec! for each recorc! along with a stanclarctizect acictress anct ZIP cocle: namely, cocles that inclicate how the aciciress is server! by mail delivery (e.g., carrier route anct the sequential orcler in which the acictress is serviced on that route). The DSF recorc! for a particular aciciress also includes a cocle for cleliv- ery type that is meant to indicate whether the address is business or resiclential. 2In Baldridge v. Shapiro, 455 U.S. 345 (1982), the U.S. Supreme Court ruled that the Census Bureau's "address list ... is part of the raw census data intended by Congress to be protected" under the confidentiality provisions of Title 13. Accord- ingly, the court concluded that the bureau's address list is not subject to disclosure under the Freedom of Information Act or under the discovery process in civil court proceedings. Specifically, the legislation text indicates that "the Postal Service shall provide to the Secretary of Commerce for use by the Bureau of the Census such address informa- tion, address-related information, and point of postal delivery information, including postal delivery codes, as may be determined by the Secretary to be appropriate for any census or survey being conducted by the Bureau of the Census. The provision of such information under this subsection shall be in accordance with such mutually agreeable terms and conditions, including reimbursability, as the Postal Service and the Secretary of Commerce shall deem appropriate." 4The list does not include general delivery addresses. Additional information on the DSF and commercial programs under which private companies are able to match their own address lists against the DSF can be found on the Postal Ser- vice Web site at http://www.usps.com/ncsc/addressservices/addressqualityservices/ deliverysequence.htm.
MODERNIZING GEOGRAPHIC RESOURCES 43 Because the census is concluctec! largely through mailed ques- tionnaires most of which are subsequently mailed back the U.S. Postal Service is a crucially important conduit in the census process. Moreover, the Postal Service is a constant presence in the fielcI, servic- ing existing anct emerging routes on a claily basis. For these reasons, securing access to the DSF was a major accomplishment. The DSF is an uncloubtecily vital source of acictress information, albeit an incomplete one for census purposes since the list of mail delivery aciciresses is only a subset of the complete list of housing units in the United States. Mail delivery listings may also be incomplete in distinguishing multiple housing units within the same structure. The Postal Service began sharing the DSF with the Census Bureau in the micI-1990s. Currently, as part of the Census Bureau's ongoing Geographic Base Support Program, new versions of the DSF are shared with the Census Bureau twice per year en c! updates or "refreshes" to the MAF are made at those times. Local Update of Census Addresses The Census Aciciress List Im- provement Act of 1994 also authorized the secretary of commerce anct the Census Bureau to provide officials who are designated as census liaisons by a local unit of general purpose government with access to census aciciress information for the purpose of verifying the accuracy of the acictress information of the bureau for cen- sus and survey purposes. The act obligated the Census Bureau to "responcl to each recommen- elation made by a census liaison concerning the accuracy of address in- formation, including the determination (anc! reasons therefor) of the bureau regarding each such recommenclation." Put another way, the act permitted the Census Bureau to share with a local or tribal govern- ment the address data it had on file for that locality, for their review and upclate. To preserve Title 13 confidentiality, limits were placed on the infor- mation to be proviclect; the information to be ctisclosect to any particular locality was limited to aciciress information anc! to the set of aciciresses for that area. Ultimately, the acictress information would only be shared
44 PLANNING THE 2010 CENSUS with local or tribal governments if they signet! an agreement to hole! the information as confidential anc! to dispose of it when finisher! with review. In August 1996, the Census Bureau initiated a program to ac- quire acictress list information from local governments. The Program for Aciciress List Supplementation (PALS) contacted local anc! tribal governments (along with regional planning agencies) anct solicited whatever lists of city-style aciciresses that they maintained for their jurisdictions. However, the Census Bureau quickly concluclect that the program was troubled; local aciciress lists were not necessarily in computer-reaclable format, anct then not formatted in such a way (in- clucling apartment and unit clesignators) as to match with the emerging coding system for MAF. More significantly, response by local govern- ments to an open-enclect query for local acictress lists ideally coclect to the appropriate census block was low. The program was officially ter- minatecl in September 1997 (U.S. Census Bureau, Geography Division, 1999~. The Census Bureau's next attempt at local geographic partnerships followoc! closer to the spirit of the Aciciress List Improvement Act by releasing parts of the Census Bureau's MAF for review rather than re- questing entire aciciress lists. The resulting program became known as the Local Upclate of Census Acictresses (LUCA), though it is also oc- casionally referred to as the Aciciress List Review Program. LUCA was concluctec! in two waves: · LUCA 98. In 1998, local anct tribal governments in areas with predominantly city-style addresses were given the opportunity to review the Census Bureau's address list. Census Bureau car- tographers used blue lines to distinguish city-style address areas from non-city-style areas on the maps that defined eligibility for LUCA. Hence, LUCA 98 was said to target localities lying "insicle the blue line." · LUCA 99. In 1999, attention turnec! to areas outside the "blue line," those with non-city-style acictresses.5 Local anct tribal gov- ernments were again offerec! the chance to review Census Bureau 5The "blue line" designating LUCA 98 and 99 areas was not constrained to follow borders of whole geographic locations, so many places and counties were eligible to participate in both waves of LUCA. In some localities, the blue line did not cleanly
MODERNIZING GEOGRAPHIC RESOURCES 45 materials, but this time the offer was to review block-level counts of housing units rather than actual aciciresses. To participate in LUCA, local anct tribal governments were required to identify liaisons to handle the aciciress list materials anc! to execute an oath of confidentiality. Materials were sent to the local anct tribal gov- ernments, which hac! a specifier! time perioc! to analyze them anc! sub- mit any proposed changes. These changes were then reviewocl by the Census Bureau, which often opted to reject part or all of the localities' suggested aciclitions or cleletions to the aciciress list. An appeals process was set up uncler the auspices of the Office of Management anct Budget (OMB), giving local and tribal governments a final opportunity if they founct grounds to quarrel with the Census Bureau's judgments. The Working Group on LUCA commissioned jointly by this panel anct the Panel to Review the 2000 Census has concluctect an extensive review of the LUCA process from the participant's (local government) perspective (Working Group on LUCA, 2001~. Block Canvass In the 1990 anct earlier censuses, when acictress lists were not maintained from census to census but rather assembled be- fore the decennial enumeration, a complete fielc! canvass of the city- style acictresses in clesignatect mailout/mailback areas was a stanclarct- but costly operation. The Census Bureau had hoped to avoid a com- plete block canvass before the 2000 census; in introducing the Acictress List Improvement Act of 1994, U.S. Representative Thomas Sawyer ex- pressecl hope that "collection and verification of address information in primarily electronic format" from the Postal Service anc! local govern- ments "will greatly recluce the amount of precensus fielct canvassing," activity that he inclicatec! hac! proven "expensive anc! often inaccurate."6 Rather than a complete block canvass, the Census Bureau planned to target specific areas with coverage gaps anc! focus fielc! canvass activities on those areas. In spring and summer 1997, as a continuous MAF began to take shape, optimism about the completeness of DSF updates gave way to distinguish between city-style and non-city-style areas, causing frustration for some LUCA participants (Working Group on LUCA, 2001~. The process for delineating city-style-address areas should be refined for future LUCA-type programs. Representative Sawyer's remarks can be found in the Congressional Record for the 103rd Congress, page H10618 (October 3, 1994~.
46 PLANNING THE 2010 CENSUS doubts, which were compounclec! by the failure of PALS to obtain acI- ciress information from local anc! tribal governments. Internal evalua- tions convinced the bureau that relying on DSF anct LUCA alone could leave gaps in MAF coverage; in particular, the bureau was concerned that "the DSF file missect too many acictresses for new construction anct was not updated at the same rate across all areas of the country" (Na- tional Research Council, 1999:39~. Accordingly, the Census Bureau opted to change course anc! con- cluct a full canvass of acictresses in mailout/mailback areas "in a manner similar to the traditional, blanket canvassing operations usec! in prior censuses." The bureau notect that the change would incur a large ex- pense but recognizing the bureau's concerns a previous CNSTAT panel "strongly enclorseLct] this change in plans" (National Research Council, 1999:25,39~. Plans for the complete block canvass overlapped with the emerg- ing plans for the LUCA program. The bureau originally planned for LUCA 98 to obtain feedback in early 1998, so that resulting changes to the MAF would be reacly for the block canvass in late 1999. However, delivery of MAF segments to most participating LUCA 98 localities was clelayoct. This led to a revisect plan that LUCA 98 changes would be compared to the MAF after block canvassing was complete. Further delays led to abandonment of a reconciliation operation in which ctis- crepancies between LUCA anc! block canvass observations woulc! have been reviewoct with localities; instead, localities received a list of ac- ceptec! en c! rejectee! aciciresses in LUCA's "final determination" phase and were given 30 clays to submit appeals to OMB's Census Aclclress List Appeals Office (Working Group on LUCA, 2001~. The TIGER Database Purpose and Scope The TIGER database is, effectively, a cartographic resource that cle- fines a complete digital map of the Unitec! States en c! its territories. It is intenclect to capture not only visible features the centerlines of streets, rivers, and railroacls, and the outlines of lakes, for instance but the myriac! political anc! administrative boundaries that may not corre- sponcl exactly with visible physical locales. Accorclingly, the TIGER
MODERNIZING GEOGRAPHIC RESOURCES database includes the political geography of 3,232 counties or county- level equivalents, over 30,000 county subdivisions or minor civil divi- 47 signs, and over 20,000 named places, among other political units. Of the many types of geography defined by the TIGER database, the most important are the boundaries of census blocks. Census blocks are the smallest unit of geography for which basic population data are tabulated in the census, and so it is these fine-resolution data at the block level that are aggregated to form political and other adminis- trative boundaries. TIGER's primary function in census operations is geoco~ing, the matching of a given address or location to the census block in which it lies. Once a location has been matched to the cor- rect census block, its location in higher-level geographic aggregates constructed from blocks is also known, and so census returns may be properly tabulated by geographic unit. In addition to the geocoding function, the Census Bureau has relied on TIGER for three other major uses (O'Grady and Godwin, 2000; U.S. Census Bureau, 20011: · geographic str?~ct?~re and relational analysis: how one geographic area relates to another, important for being able to aggregate small units like blocks into coherent higher-level aggregates; · geographic definitions: serving as a repository for the current def- initions of geography levels recognized by the Census Bureau; and · map production and dissemination: printing the maps used by cen- sus enumerators to carry out their assignments. . The full TIGER database maintained by the Census Bureau con- ta~ns point features along with linear features; in particular, points define the location of known housing units in areas without city-style addresses. However, most public exposure to the TIGER database comes via TIGER/Line files, a public excerpt of the TIGER database that contains only linear features such as roads, rails, and political boundaries (and, hence, not specific housing unit locations). The TIGER/Line files do contain complete street coverages with address ranges; it was the widespread availability of TIGER/Line files that facilitated the emergence and growth of the geographic information systems (GIS) industry.
48 PLANNING THE 2010 CENSUS The TIGER database is one part of a larger TIGER system, which includes the support structure of hardware anc! software necessary for maintaining the database. When TIGER was initially clevelopect, the database was compiled in a unique anc! home-grown language cle- finect by the Census Bureau; various software programs to upclate the database anc! to produce maps were similarly written to accommodate the custom, internal database language TIGER uses. As we will discuss, the proposer! MAF/TIGER enhancements make changes in both the database anct system senses, improving the content of the database as well as overhauling the support machinery around it. How the TIGER Database Began The TIGER database was clevelopect by the Census Bureau, with assistance from the U.S. Geological Survey (USGS), to support the 1990 census. "TIGER began life as a patchwork quilt of data sources" (O'Gracly anc! Goc~win, 2000:6), two of which were primary. One of these sources was the GBF/DIME files used by the Census Bureau to clo aciciress matching to street segments in the 1980 census.7 The GBF/DIME files foreshaclowoct TIGER in that they applied topological principles in piecing together points, lines, and polygons (Hirschfelcl, 2000~; they also began the move toward including more than streets anc! roacis in census maps, acicling features such as water, rail, anc! invis- ible boundaries. However, these files were limited in scope, covering the urban centers of 276 metropolitan areas "less than 2 percent of the land area but 60 percent of the people in the Unitecl States" (Car- baugh and Marx, 1990~. To complete the geographic coverage of the nation, the aciciress reference information in the GBF/DIME files was merged with computer-coclect versions of the water anct transportation features clefinec! by the USGS series of 1:100,000-scale topographic maps (Marx, 1 986~. As O'Gracly and Goclwin (2000:4) note, "accuracy was crucial" when TIGER was first assembled "but only in a relational sense." "The coordinate information presented in the TIGER/Line files is proviclec! for statistical analysis purposes only," wrote Carbaugh anc! Marx (1990~; "it is only a graphic representation of ground truth." 7GBF/DIME stands for Geographic Base File/Dual Independent Map Encoding.
MODERNIZING GEOGRAPHIC RESOURCES 49 Put another way, the priority in early TIGER was to achieve basic functionality for census purposes, which meant favoring relational accuracy Describing how geographic features relate to each other, such as whether census blocks are acljacent) over positional or locational accuracy (precise location of geographic features relative to a chosen stanciarcI). Hence, O'Gracly anc! Goc~win (2000:5-6) recall that the Census Bureau ctrew on properties of the USGS maps in publishing the following positional accuracy statement in the documentation for TIGER/Line files released in 1995: The positional accuracy varies with the source materials usecI, but at best meets the established National Map Accuracy stanclarcts (approximately +167 feet) where 1:100,000-scale maps from the USGS are the source. The Census Bureau cannot specify the accuracy of feature up- ciates aciclec! by its fielc! staff or of features clerivec! from the GBF/DIME-Files or other map sources. Thus, the level of positional accuracy in the 1995 TIGER/Line files is not suitable for high-precision measurement applications such as engineering problems, property transfers, or other uses that might require highly accurate measurements of the LEarth's] surface. In aciclition, the overall positional accuracy of early TIGER was lim- itect by shortcomings in the GBF/DIME files, which were also oriented towarc! relational accuracy. In particular, Census Bureau enumerators anct staff later founct that "hyctrographic features are not represented well" in TIGER database segments clerivec! from the GBF/DIME files (Rosenson, 2001:1~. Updates to TIGER Over the course of the 1990s, the TIGER database was updated using additional sources each with unique (ancl often unknown) levels of positional accuracy. Among those sources are the following programs that are likely to continue cluring anct after the MAF/TIGER Enhancements Program, although exactly how anct when the resulting information will be incorporated and how the programs might be restructured is as yet unspecifiecl:
50 PLANNING THE 2010 CENSUS · Boundary and Annexation Survey (BAS): an ongoing volun- tary survey of local anc! tribal governments, in which TIGER- generatect boundary maps are sent to governments for review anct update · MAT Geocoding Office Resolution (MAFGOR): a program in which city-style acictress records from the Postal Service Delivery Sequence File (DSF) of which more will be said later that can not be geococlec! in TIGER are referred to census regional offices for review · Targeted Map Update (TMU): a regular program in which cen- sus fielc! staff update aciciress ranges, acic! new streets, anc! update feature names in selected areas Digital Exchange One TIGER upclate mechanism of particular in- terest is the Digital Exchange (DEX) system, in which local en c! tribal geographic database files are used to update TIGER features. Builcling anc! improving upon DEX's capabilities will be a major part of TIGER realignment in the Enhancements Program. Developed in the late 1990s, the Census Bureau's DEX system is clescribect in greater detail by Rosenson (2001~. Given the unique struc- ture of TIGER anc! its interface software, DEX cloes not work directly with the local anct tribal geographic files but rather with a processed ex- tract thereof known as an "exchange file." In particular, DEX is strictly limited to working with roac! features anc! the attributes associated with them, including ZIP cocles. The exchange file clerivect from a local ge- ographic file is a street centerline database coclec! using TIGER's struc- ture. This exchange file is then matched to the TIGER file basect on spatial location as well as on attribute information (e.g., street name), beginning with matches on the intersection points between namect roact features in each file. After matching, one of the files is "rubber-sheetecl" meaning that its features are acljustec! to better match attributes in the other file, with neighboring attributes being adjusted simultaneously, as necessary. As Rosenson (2001) notes, this "rubber-sheeting" can be clone to either file but, at least in early DEX implementation, the process coulc! introduce topological errors such as lines that cross each other without a system-
MODERNIZING GEOGRAPHIC RESOURCES 51 clefinec! point marking their intersection. Thus, in orcler to preserve TIGER's topological structure, DEX manipulates the local "exchange file" to match certain TIGER features. Though some DEX capability hac! been clevelopec! anc! selectee! local geographic files were obtained prior to the 2000 census, active TIGER updating using DEX was cleferrec! cluring the actual conduct of the 2000 census. The Need to Modernize The development of TIGER is a milestone of which the Census Bureau should be extremely prouct. A home-grown database manage- ment system constructed to manipulate an enormously complex net- work of visible anct invisible boundaries, TIGER became an exemplar of what a geographic information system (GIS) can clot The example of TIGER ancl, significantly, the public availability of TIGER/Line files, a full anc! fine-scale public atlas of the Unitec! States toucher! off a commercial GIS revolution. Businesses anct organizations of all sizes are continuing to learn the power of spatial data analysis, anc! the work of TIGER to bring together anct make publicly available base geo- graphic layers helped make that possible. TIGER is also rightly a source of pricle because it successfully satisfied the operational clemancts of two decennial censuses. The coding system may be (in computer years) old anc! the structures arcane, but it is a rare in-house software product that can successfully cope with a production cycle of billions of printed maps anc! millions of aciciresses for geococling in the way TIGER clic! in the 1990 and 2000 censuses. But, as is the case with some revolutions, the first entrant ushers in tremendous change and then is unable to keep pace with the new world thus created so it is with TIGER. Though the text-basec! TIGER/Line files are parsable by commercial GIS applications, moclern database tools anc! the native TIGER database structure are not compatible. Hence, it has not been possible to directly update TIGER's street cov- erages using the GIS files upciatec! anc! maintained by local anc! tribal governments. The Census Bureau's unique role in delineating census blocks the base units that are aggregated to form most political districts anc! ongoing programs such as the Boundary anc! Annexa- tion Survey (BAS) give the Census Bureau advantages in defining the
52 PLANNING THE 2010 CENSUS invisible political boundaries that cross-cut the nation. But commercial GIS has macle it possible for external companies anc! local anc! tribal governments to extenct from a TIGER/Line base, realigning features when errors are fount! anc! making updates to street, rail, water, anc! other features to a clegree that Census Bureau resources have not permitted in the past. THE MAF/TIGER ENHANCEMENTS PACKAGE The bureau has set forth five objectives as essential steps in a com- prehensive MAF/TIGER modernization. They are spelled out with subtasks, as follows: 1. improve acictress/street location accuracy anct implement auto- matec! change detection; 2. implement a moclern processing environment; 3. expand and encourage geographic partnership options; 4. launch the Community Acictress Upclating System (CAUS), which has also been known as the American Community Survey Coverage Program; anct 5. implement periodic evaluation activities anct expand quality met- rics. Objective One: Address/Street Location Accuracy Objective One the actual realignment of TIGER geographic features is the centerpiece of the MTEP, enough so that it has ac- quirec! an acronym of its own. The contract to carry out Objective One also known as the MAF/TIGER Accuracy Improvement Project (MTAIP) was awarclec! to the Harris Corporation of Melbourne, Floricla, in June 2002. As described in documentation provided to the panel, the basic sub- tasks envisioned uncler Objective One are as follows:
MODERNIZING GEOGRAPHIC RESOURCES 53 1. correct (in TIGER) the location of every street anc! other map feature usec! by fielc! staff anc! governmental partners for orienta- tion, as well as the location of every boundary usect for tabulation of decennial census anc! household survey ciata; 2. correct (in the MAF) the location of every housing unit anct group quarters from which the decennial census anct the house- holc! surveys collect ciata; anc! . . 3. implement an effective change detection methodology to clocu- ment the location of every new street anct living quarters, along with the street name en c! aciciress for each. Means of Updating Accuracy As it has been explained to the panel, the basic idea of Objec- tive One is to perform a single, extensive upclate of TIGER for each county based on an external source with, presumably, more current anct accurately positioned feature information. These outside sources may include GIS files clevelopec! anc! maintained by local or tribal governments, commercial GIS files, or digital orthophotography/aerial photography. Once the TIGER data for a county is realigned, it is then poised for continual upclate through change detection for instance, aciclition of features through comparison of TIGER to newer aerial photographs of a region. Through this strategy extensive initial realignment, followoct by change detection the Census Bureau hopes to maintain TIGER so that its features are current to within 1 year. Embocliecl in this general framework is great flexibility for the Census Bureau anct its contractor to implement the TIGER upclate; at present, to the extent that plans have been shared with the panel, this great flexibility translates into very little specificity. The Request for Proposals (RFP) issucc! to solicit contractor bicis to perform Objective One indicates the Census Bureau's strong preference to use local or tribal government GIS files as the upclate source.8 However, the panel has not yet seen stanciarcis for the level of precision required of these local files, assessments of how many localities have GIS files that meet 8The REP and other documents related to Objective One, the MAF/TIGER Accu- racy Improvement project are archived at http://www.census.gov/geo/mod/maftiger. html.
54 PLANNING THE 2010 CENSUS that stanciarcI, or specifications for the measurement of local GIS file quality (e.g., comparison with sample fielcI-collectec! GPS coorclinates). If local or tribal GIS files are not available, then alternative sources are to be usect. For example, the Census Bureau has concluctect exper- iments using subcontractors to perform updates based on digital or- thophotographs and other image sources. However, it is as yet unclear which mechanism the Census Bureau anct the Harris Corporation will favor in the absence of local files (or local files of insufficient quality) to perform the initial, global realignment. Since it is unclear what exact source will be usec! for the initial realignment, it is even less clear what source will be usect to upclate TIGER files in the change detection process, anc! with what frequency this will be clone. As we will discuss later, effective communication between the Census Bureau anc! state, local, anc! tribal governments must be established in orcler to accomplish Objective One realignment anc! change detection. Priorities Franz (2002) clescribect the following priority structure that the Census Bureau has iclentifiec! for carrying out Objective One realign- ment: 1. top priority: linear feature realignment across all areas; 2. establishing/correcting structure locations in areas outside the 2000 census mailout/mailback area; 3. establishing/correcting structure locations insicle the 2000 census mailout/mailback area; 4. establishing/correcting locations for residential structures over nonresidential structures, in carrying out the previous two steps. Schedl?~le Uncler plans clevelopect in 2002, the Census Bureau anct the Harris Corporation are supposed to realign 250 counties cluring fiscal year 2003. To meet the goal of completing Objective One by 2008, the timetable for realignment of remaining counties is as follows: 600 in
MODERNIZING GEOGRAPHIC RESOURCES 55 fiscal year 2004;9 700 in 2005; 700 in 2006; 600 in 2007; and 382 in 2008. In principle, change detection to make further alterations is supposec! to begin when counties are complete, so that 250 counties are slated for change detection in fiscal year 2004,850 in fiscal 2005, anc! so forth, until all counties are hancilect using change detection methods in 2009. A bar chart shared by the Census Bureau's Geography Division with the panel indicates the rough level of expected effort on each of the MTEP objectives for each fiscal year leacling to the 2010 census. The chart bears out the centerpiece nature of Objective One: cluring the peak years of activity (2004-2008), the estimated level of effort devoted to Objective One exceeds that given the other objectives combined. Objective Two: Modern Processing Environment Objective Two of the Enhancements Program targets TIGER in the systems sense, modernizing the structure of the database. The cur- rent home-grown TIGER system suffers from key limitations, promi- nent among them, the inability to directly link with commercial GIS packages (anc! hence local anc! tribal GIS files maintained using those packages) anc! the limitation that only one module (county) of TIGER may be "checkoct out" for upclating at any single time. Changes to the database structure also require that the suite of support software usec! to generate products from TIGER for instance, simply to print maps for fielc! enumeration must also be reauthorec! anc! tested. The Census Bureau's stated subtasks for Objective Two are as fol- lows: 1. make maximum possible use of commercial off-the-shelf (COTS) anct geographic information systems (GIS) tools to allow for rapid development of new applications; 2. customize the COTS/GIS tools to the minimum extent possible to avoid schedule and cost obstacles when the COTS/GIS ven- clors deploy new versions of their software. 9The figure of 600 counties is included in the detailed description of the Bush administration's budget request to Congress for fiscal year 2004.
56 PLANNING THE 2010 CENSUS Based on a chart provided to the panel, fiscal year 2003 is anticipated to be the peak year of Objective Two work, with some slight drop-off in fiscal 2004. Residual effort is expected in fiscal 2005 and 2006, with Objective Two not listed as an activity in 2007 or later years. Objective Three: Geographic Partnerships Objective Three acknowledges the crucial role of state, local, and tribal governments in maintaining geographic resources, not only for the TIGER realignment of Objective One but for continued update of the MAF, as in the LUCA program. Subtasks of Objective Three identified by the Census Bureau are as follows: 1. devise and deploy new strategies to communicate more effec- tively with governments to increase the level at which they participate in MAF/TIGER review and update activities; 2. devise and deploy new ways in which to integrate more effectively the address list review, street update, and boundary reporting ac- . . . . t1v1t1es to eat now exist as separate programs it, 3. establish new partnerships with other federal agencies and private-sector firms that have GIS and address files with infor- mation of value to an accurate and complete MAF/TIGER. Based on a chart provided to the panel, fiscal year 2004 is anticipated to be the peak year of Objective Three work; the level of effort expected on this objective in each of the years 2003 and 2005 through 2010 are shown to be roughly equivalent. Objective Four: Community Address Updating System Briefly known as the ACS Coverage Program, the Community Ad- dress Updating System (CAUS) is the address list update component of the proposed American Community Survey (ACS). The basic idea of the program is to make use of the continued field presence that would be necessary to conduct the ACS, allowing ACS enumerators the op- portunity to provide geographic updates. One hope is that the ACS enumerators might be particularly helpful in identifying geographic and
MODERNIZING GEOGRAPHIC RESOURCES 57 housing changes in rural areas, where local and tribal files might be less detailed (or unavailable). The Census Bureau has identified the following subtasks for Ob- . . ect~ve Four: 1. focus on predominately rural areas in which the U.S. Postal Ser- vice's Delivery Sequence File (DSF) does not effectively identify the existence or location of new housing units; 2. provide address list (and street) updates beyond what can be iden- tified through the current twice-yearly DSF "refresh" process to ensure a uniformly accurate sampling frame nationwide for the ACS and the other household surveys. Through contractors, the Census Bureau has developed prototype ~ 1 T · · 1 ~ r · T ~ ~ T ~ rut ~ 1 - Automated Listing and Clapping instrument tAL~ll' software, making use of a GPS receiver and a laptop computer. The ALMI system could permit ACS enumerators who encounter a new street that is undefined in TIGER to record a GPS trace as they drive along the street and to note location of houses along that street; these inputs could later be converted to TIGER. Based on a chart provided to the panel, the anticipated level of effort that the Census Bureau expects to expend on Objective Four is roughly equivalent during each of the fiscal years 2003-2010. Objective Five: Evaluation and Quality Metrics Finally, Objective Five raises the issue of assessing progress and quality; subtasks identified by the Census Bureau for this Objective include: 1. provide quality metrics information that will guide (target) areas in need of corrective action beyond the changes identified in the change detection and CAUS activities 2. document the progress being made to improve the accuracy and completeness of the street, address, and boundary information in MAF/TIGER; and
58 PLANNING THE 2010 CENSUS 3. assure the availability of accurate anc! comprehensive metaciata that meet federal stanciarcis about the information in MAF/ TIGER. Basect on a chart proviclect to the panel, the anticipated level of effort that the Census Bureau expects to expend on Objective Five is roughly equivalent cluring each of the fiscal years 2003-2010. ASSESSMENT OF GEOGRAPHIC MODERNIZATION EFFORTS Locational Accuracy of TIGER Problems with the positional accuracy of TIGER have been ap- parent to the Census Bureau anct its users for some time; anecdotal experience of problems with TIGER representations clevelopec! from the experience of fielct enumerators cluring the 2000 census anct from feedback from local anc! tribal governments who participated in LUCA (Working Group on LUCA, 2001~. Quantitative evidence of TIGER discrepancies can be fount! in Liaclis (2000), the report of a Census Bureau experiment that collected GPS position readings for approx- imately 6,700 "anchor points" spread across selectee! census tracts in eight counties. Distances could then be computed between these "grounc! truth" coordinates anc! the longitucle/latitucle combination coclec! in TIGER. The results show evidence of strong local variation, even across tracts within the same county. The distance between TIGER representation anc! grounc! truth variec! with respect to the operation that introclucect the point into TIGER. Somewhat ironi- cally, more recent update programs which aclclecl features by digitally inserting them as freehand drawings accounted for the largest clevi- ations from grounc! truth, while pre-1990 sources (e.g., GBF/DIME files) anct programs involving direct use of local anct tribal geographic files (e.g., DEX) generally came closest to true locations. The Cen- sus Bureau's Geography Division also concluctect pilot experiments comparing for small geographic samples TIGER coordinates to a combination of GPS coordinates anct commercially available carto- graphic databases (U.S. Census Bureau, Geography Division, 2000) anc! to digital orthophotos giving an aerial view of grounc! features (O'Gracly, 2000~.
MODERNIZING GEOGRAPHIC RESOURCES 59 Though the extent of TIGER inaccuracy may be unknown, there is enough extant evidence available that the panel cleciclecily concurs re- garcting the basic nature of the problem. Roacts, boundaries, anct other geographic features are sufficiently misplacecl, and with enough regu- larity, that the TIGER database is in neect of a comprehensive upclate; moreover, raw TIGER/Line files cannot be fully trustee! for routine GIS anct non-GIS-relatect tasks. Accordingly, the major motivation behind Objective One of the enhancements package is well taken. Having concluclect that locational error in TIGER is sufficiently clear as to require correction, it follows naturally that accomplishing the basic task envisioned uncler Objective One is essential to the modernization of the census. GPS coordinates collected by MCDs are only useful to the extent that they can be accurately placed onto base maps with streets anct other key features. An accurately aligned TIGER, faithful to polygonal features such as municipal boundaries, can be passed along to localities and made available on the Internet, thereby allowing local anc! tribal entities the opportunity to report changes macle to both linear (e.g., roact anct railroad anct polygonal features (e.g., administrative borders collected by the Bounclary en cl Annexation Survey) in a more efficient anct accurate way. If localities can readily utilize an aligned TIGER for geococling their own address files, comparisons with (anct upclating of) the MAF can more closely resemble routine work. Hence, the panel supports Objective One of the enhancements package en cl it is heartened by the general steps taken to accomplish the objective. In particular, the panel views the acquisition of an out- sicle contractor as a sign of significant progress, rather than keeping TIGER updating as a purely in-house operation. As Census Bureau staff notect in an interview, it is incleect a "very major departure for us" to seek external help in retooling TIGER, but "we've come to the conclusion Ethat] we need to take advantage of Evenclors'] expertise anc! unclerstancling" (O'Hare anc! Caterinicchia, 2001~. In the panel's assessment, the Census Bureau deserves generally high gracles for Objective One anc! its determination to fix a major problem as well as the boldness of the approach. That said, concerns about the work remain, and the plausibility of the Census Bureau's am- bitious realignment timetable woulc! be bolstered considerably through attention to the following:
60 PLANNING THE 2010 CENSUS · a cletailec! work plan, including some notion of the orcler in which counties will be initially upciatecI; · specification of the clesirect positional accuracy stanclarct of the realigned TIGER, couplet! with specifications for the positional accuracy required of local anc! tribal GIS files; anc! · specification of plans for the post-realignment change-cletection program. In aciclition, a subtle point raised in our earlier discussion of the Census Bureau's Digital Exchange (DEX) program deserves fuller explication. Given two GIS files (a local file anct the TIGER clata), a "rubber-sheeting" process manipulates certain matched features in one file to conform to the other, shifting related features automatically. The Census Bureau's early DEX system altered the local file to follow known features in TIGER to avoict topological bugs that may result otherwise a justifiable choice, but one that intuitively runs counter to the basic purpose of updating the presumably misaligned TIGER based on presumably accurate local files. Aciclitional detail on how the Harris Corporation's alignment tools handle topological gaps and address the conflation between local anc! TIGER files coulc! strengthen confidence in the finished procluct. A point of some contention between the panel anc! the Census Bu- reau has been the orcler in which Objective One realignment will be performed. Asicle from indicating that jurisdictions involved in mict- clecacle census tests or dress rehearsals will be given priority, the Cen- sus Bureau has not given a clearer iclea of how it expects the flow of county-by-county processing to proceecl. The notion of ordering is unclerstanclably somewhat sensitive, since no locality would relish be- ing last in the queue. However, the ambitious timetable laid out ear- lier in this chapter is unrealistic at best without some sense of or- clering. The alternative effectively starting 3,232 inclepenclent upciat- ing efforts simultaneously and hoping that 850 fall into realignment by the enc! of 2004 cloes not inspire much confidence. There is no right answer to the question of ordering conceivable mechanisms include starting with urban counties or rural counties, starting with original GBF/DIME areas, sequencing by population, or sequencing by some assessment of how out of alignment TIGER is for an area. But provict-
MODERNIZING GEOGRAPHIC RESOURCES 61 ing some manner of structure to the task seems essential to measuring progress along the way to complete realignment anc! coulc! acic! plausi- bility to the hypothesized timetable. Modernizing the Toolset Similarly, we applaud the Census Bureau's efforts to adopt GPS technologies anc! a moclern processing environment using COTS procI- ucts to achieve Objectives One anct Two. This is already a long-term project anti, to our knowledge, database requirements have yet to be finalized. Clearly a considerable effort is neeclec! before the right combination of COTS products can be cleterminect, anct this neects to be carried out so that the system can be given a full-scale test prior to Census 2010. With respect to Objective Two, the conversion to a moclern database environment, the panel has two major points of concern. First, the difficulty of making the conversion work shoulc! not be unclerestimatect. In early discussions with the panel, the conversion was characterized as a fairly easy step: a new database structure woulc! be iclentifiect anct new support software would be written (anct tested, certified error free). Work on the TIGER database coulc! then be suspenclect for a period of a few clays, information ported over to the new structure, and the task would be clone. All experience with such upgrades particularly one that strives to raise the Census Bureau's Capability Maturity Moclel (CMM) score for software engineering at the same time that completely new systems are cleployoc! suggests that such a rosy scenario is misguiclecily optimistic. Second, the design of Objective Two anc! the new MAF/TIGER system should be coorctinatect with the broacler technical architecture program described in Chapter 2 and should follow similar techniques. As with MCDs, it is more important to moclel the work anct infor- mation flows that must be channeled in a moclernizec! MAF/TIGER environment anct to tailor the physical architecture accordingly, rather than lock into specific packages or programs too early. An architecture- basect approach coorctinatect with the rest of the Census Bureau may acict structure to the system of inputs anct outputs to MAF anct TIGER, including the information that is requested from state, local, anc! tribal governments.
62 PLANNING THE 2010 CENSUS Quality Metrics Going further, we note that the Census Bureau has made significant steps toward establishing metrics to evaluate improvements in accuracy, as called for by Objective Five. Work with contractors has brought about an image-based rough assessment system that allows accuracy checks on incoming files, as well as progress on DEX evaluation of files on the basis of control points, and a soon-to-be installed system for quantifying and tracking MAF/TIGER errors over time. It is essential, in our view, that quality assessment through such metrics be an ongoing and well-timed process so that updating of the database achieves the apparent goal: information in MAF/TIGER maintained to a currency of 1 year or less at all times. WEAKNESS: ENHANCING THE MAF The MAF/TIGER Enhancements Program will make improve- ments to TIGER that are necessary, given known problems with TIGER accuracy. In other words, the MAF/TIGER Enhancements do show promise for enhancing TIGER. But, for the sake of census accuracy, a more important question is how the program will enhance the MAF that is, how it will add new addresses, screen for duplicates, and generally ensure that address rosters are as complete and accurate as possible. It is on this score that the Enhancements Program falls seriously short, in our view, due to lack of development in Objectives Three and Four. More generally, the Census Bureau's current strategy shows relative inattention to MAF improvement and shows signs of repeating costly errors from the 2000 experience. Alone, the magnitude of the Objective One task of realigning TIGER features and the monetary cost associated with it gives the Enhancements Program a TIGER-centric feel. But Objectives One, Two, and Five seem to speak to MAF largely as it inherits its quality from TIGER. Indeed, the Bush administration's fiscal year 2004 budget message to Congress described the geographic leg of the Census Bureau's 2010 strategy as a plan for "enhancing the Census Bureau's geographic database and associated address list." In line with our comments in opening this chapter, the MAF is too critical to the
MODERNIZING GEOGRAPHIC RESOURCES 63 quality of the census anc! other survey programs to be treated as an acicI-on or adjunct. Current Plans for MAP Updates for 2010 The Census Bureau argues that the combination of three activities- "the ongoing MAF/TIGER updating using the Delivery Sequence File, CAUS, anct enhancements incluclect in the proposed MAF/TIGER modernization initiative" "shoulc! result in an up-to-date aciciress list for the entire United States" (U.S. Census Bureau, 2003a:11~. More specifically, the update strategy is based on a rough urban/rural clichotomy: · The Postal Service's DSF is intenclec! to be the aciciress update source "in areas where DSF addresses can be assigned a physi- cal location, such as urban areas with city-style aclclresses" (U.S. Census Bureau, 2003a:9~. · "In rural areas with non-city-style acictresses, this EDSF upclate] process cannot be used," anc! so the Census Bureau intends to update this segment through CAUS. The bureau indicates that the areas for which DSF updates cannot be used "encompass the majority of the Nation's land area and about 15 percent of the population" (U.S. Census Bureau, 2003a:9~. These update sources would be supplemented in the MAF/TIGER En- hancements Program, which we interpret to mean a successor to the 2000 census LUCA program uncler Objective Three. The backbone of the Census Bureau's upclate strategy is the twice- yearly "refresh" that comes from the Postal Service's Delivery Sequence File. These regular updates are considered to be part of the bureau's Ge- ographic Support Base Program, not the MAF/TIGER Enhancements Program. As we notect earlier, the DSF is uncloubtecily an important source of aciciress information, anc! we clo not mean to imply in any way that its use is either wrong or inappropriate. However, reliance on the DSF as the principal source of address updates for (by the bureau's estimate) 85 percent of the household population raises concerns in at least three respects:
64 PLANNING THE 2010 CENSUS · Historical precedent in the 2000 MAF-b?~ilding process. As we in- clicatec! earlier, DSF updates were also viewoc! by the Census Bu- reau as a centerpiece acictress source after the 1994 passage of leg- islation that enabled sharing with the Postal Service. However, the bureau perceived problems with the level of DSF coverage in fast-growth en c! new construction areas anti, consequently, initi- atect a costly complete block canvass (National Research Council, 1999~. . · Effectiveness in 2000 unknown. As we will discuss in considerable detail later in this chapter, evaluation work that woulc! explain how the various sources that were merged to form the 2000 census MAF remains incomplete. Therefore, the effectiveness of DSF updates in providing valic! census aciciresses (as well as cluplicates) has not been empirically established. Worse, the Census Bureau's plannec! evaluation study on the DSF F.1, "Impact of the Delivery Sequence File Deliveries on the Master Aclclress File through Census 2000 Operations" (U.S. Census Bureau, 2002a) was inexplicably cancellecl in late 2002 (U.S. Census Bureau, 2003b).1° · Limitation of DSF to mail delivery population. Again, by clefini- tion, the DSF is intenclect to document mail delivery acictresses~ which is not equivalent to the complete list of housing units in the Unitecl States. The Census Bureau's planned activity to update addresses in rural areas is CAUS, which to briefly review is an associated program of the American Community Survey (ACS). Uncler CAUS, ACS field rep- resentatives would list acictresses (anct upclate streets, using traces from a 10The planned evaluation was intended to "assess the impact of each of the tDSF updates performed prior to 2000] through Census 2000 operations by profiling the number and characteristics of housing units added to and deleted from the MAP fol- lowing each delivery of the DSF" (U.S. Census Bureau, 2002a:C-56~. Of its cancel- lation, all that is said is that "this evaluation will not be conducted. In late 2002, the Census 2000 Evaluation Program was refined and priorities reassessed due to re- source constraints at the Census Bureau" (U.S. Census Bureau, 2003b:2~. The bureau's planned "synthesis report" on address list development evaluations may comment on DSF contributions, but that report has not been completed.
MODERNIZING GEOGRAPHIC RESOURCES 65 GPS receiver) through a laptop computer-basec! tool known as the Au- tomatec! Listing anc! Mapping Instrument (ALMI). However, general concerns raised by clepenclence on CAUS as an acictress upclate source include the following: · Linkage toACSfunding. Full funding for the ACS has not yet been securect; consequently, the budgetary viability of CAUS is not known. Full implementation of CAUS must also await full-up mobilization of ACS support staff (ancl, presumably, more elapsed time as establishing ACS operations takes prior- ity), which adds to the delay in the possible receipt of CAUS upclates. Finally, the number of possible CAUS fielct personnel is obviously linkoc! to the number of ACS enumerators; while it is hoped that budget commitments to ACS would not oscillate, CAUS effectiveness coulc! also be impaired if ACS funding is not stable over the years. · ACS workload management. It is unclear how much time en cl manpower ACS managers will commit to sicle work on acictress listing given the ambitious timetable of ACS data collection. · Unclear/unspecifiedt mechanism for targeting areas for update. The exact means by which CAUS representatives woulc! be cleployoc! to collect information in particular geographic areas are as yet un- specifiecI. One such means is for enumerators to list new streets or developments they find by happenstance in carrying out their regular ACS work, but that is surely an unreliable means of cover- ing the entire rural population. The draft ACS operations plan in- clicates that "AC S planners twilll use various methods for iclenti- fying where coverage is insufficient," including "work with com- munity officials to acquire information about new acictresses, new streets, ancl/or areas of significant growth" (U.S. Census Bureau, 2003a: 10~. But, again, the mechanics of this targeting are uncer- taln. The third element in the aciciress update strategy a LUCA-type program is a topic we will discuss in greater detail in a later section. But, for the purpose of the argument at hanct, the major concern re- garcling a new local aciciress review program is simply that no prototype plans have yet been clevelopecl.
66 PLANNING THE 2010 CENSUS We assume that the bureau hopes to avoic! a complete block canvass prior to the 2010 census, given the cost of that operation and that it was treated as a last resort in 2000. However, in the absence of evidence that the combination of DSF anc! LUCA leacling up to 2010 can overcome the last-minute doubts that arose in the late 1990s anct without a clearer plan for CAUS it is difficult to see how a full block canvass can be averted. The Census Bureau neecis to outline goals pertaining directly to MAF without particular regard to TIGER geography for example, in the development of quality metrics anct the identification of housing unit duplication. Overall milestones anc! tasks neec! to be specifically set for Objectives Three anct Four, with particular regard to the ways in which activities in these objectives may work to control housing unit duplication anct to more accurately identify anct account for multi- unit housing structures. It also neecis to expeditiously complete (anc! augment, as necessary) its evaluation work on the construction of the 2000 MAF, since that work is crucial to bolstering the case for the 2010 acictress upclate strategy. Recommendation MAF-1: In articulating the MAF/ TIGER Enhancements Program and defining its strategy for updating the MAF for the 2010 census, the Census Bureau should provide clearer details on how the MAF/ TIGER Enhancements Program and other geographic programs will add missing addresses, remove duplicate addresses, and generally correct the Master Address File, independent of benefits derived from being cross- referenced to an updated TIGER database. In particular, the Census Bureau should use data from the 2000 census process to test the adequacy of the U.S. Postal Service's Delivery Sequence Files, the Community Address Up- dating System, and as-yet unspecified local partnership programs as primary contributors of new addresses. Maintaining the MAF as a Housing Unit Inventory: Multi-Unit Structures and Duplication An important first step in enhancing the MAF is an examination of the definition, identification, anc! systematic cocling of housing units
MODERNIZING GEOGRAPHIC RESOURCES 67 (and, by extension, group quarters). For census purposes, the Master Address File's most fundamental purpose should be to serve as a com- plete register of housing units. The current MAF/TIGER Enhance- ments impart some benefit to MAF entries by virtue of their linkage to TIGER but do little to address two fundamental problems that hin- dered MAF's effectiveness as a housing unit roster in the 2000 census. The first of these are multi-unit structures physical buildings that contain more than one housing unit. Particularly problematic are small multi-unit structures, facilities, and homes with multiple residents but whose divisions into subhousing units is not obvious. A realigned TIGER database may offer a precise location for a structure an aerial photograph may confirm a structure's existence or point to the con- struction of a new one but that added precision is ultimately of little use if the address roster of subunits within the structure is unknown or inaccurate. Multi-unit structures pose problems conceptually (e.g., should a finished basement in a house that is sometimes offered for rent be counted as a unit?) and technically (e.g., do different data sources code an apartment as 3, 3A, or 3-A?), and deserve research during the intercensal decade. A second problem that hindered MAF in the 2000 census was hous- ing unit duplication. Duplication is an ever-present problem in a re- source like the MAF, an amalgam of various sources. It was evidence of housing unit duplication that prompted an unplanned, ad hoc process- effective but risky to filter potential duplicates during the actual con- duct of the census in 2000 (Nash, 2000~. It is possible that precise GPS coordinates may be useful in identifying some duplicates (e.g., struc- tures at a street intersection that may be recorded on one street in one source and on the cross street in another source), but broader, more systemic sources of duplication should also be a research and evalua- tion focus leading up to 2010. As we will discuss later, identification of MAF input sources that contributed duplicate addresses will provide vital evidence in remedying duplication problems. NEED FOR A MAF IMPROVEMENT COORDINATOR . . . . . ~ In (chapter 2, we advocated the creation ot a new position within the Census Bureau a system architect for the decennial census with the primary goal of integrating and coordinating effort on architecture re- modeling. It is our view that improving the MAF is likewise an area that
68 PLANNING THE 2010 CENSUS would benefit greatly from refocused staff effort. At least four major divisions within the Census Bureau (Geography, Field, Decennial Man- agement, and with the ACS Demographic Surveys) have a strong stake in the maintenance and use of the MAF. Given the legitimate (but sometimes competing) interests of the various divisions, it would be useful to vest responsibility for coordinating MAF improvement and research in one office with connections and the ability to work with all relevant divisions. Recommendation MAF-2: The Census Bureau should designate a resident expert to oversee the development and maintenance of the MAF as a housing unit inven- tory, with a focus on improving methods to designate, list, and update units. The bureau should give high priority to discussion and research, within the bureau and with experts outside the bureau, on the following: more effective means to define, list, and enumerate housing units and incorporate those changes into the housing unit inventory; · more effective ways to define, list, and enumerate group quarters arrangements; sources of address duplication and possible reme- dies; and · listing and enumeration in multi-unit structures. GEOGRAPHIC PARTNERSHIPS To its credit, the Census Bureau has recognized the importance of partnerships with local and tribal governments by designating their cre- ation and maintenance as Objective Three in the Enhancements Pro- gram. The Census Bureau's REP for the TIGER realignment of Ob- jective One makes this clear, noting that "the success of the Accuracy Improvement Project, and the continuous update of the information in MAF/TIGER, requires ongoing interaction between the Census Bu- reau and its federal, state, local, and tribal government geographic part- ners." To its cletriment, though, the Census Bureau has not provided
MODERNIZING GEOGRAPHIC RESOURCES 69 clear indication of how partnerships woulc! work. While the panel ac- knowleciges that the funcis available for expanding anc! encouraging ge- ographic partnership options have been limitecl, the cryptic descriptions of Objective Three that we have received to this juncture clo not make it clear how the Census Bureau intends to involve local anct tribal partners in these programs. A major stated role for local anct tribal geographic partners is to contribute to Objective One to share their current GIS files with the Census Bureau to support realignment. In this matter, anct in past ge- ographic interactions such as LUCA, the Census Bureau often has per- ceivecl "partnership" as a one-sided exchange: "partners" expend re- sources anc! turn information over to the bureau. The principal reward to a local or tribal government for entering into this kind of partnership is definitely not trivial: the prospect of a more accurate census count. The Census Bureau is not a funcI-granting organization anc! hence can not directly subsidize local or tribal governments to improve anct sub- mit their geographical resources. That saicl, the Census Bureau should explore means of building partnerships that are true exchanges of infor- mation: for instance, giving census fielc! anc! regional staff an increased role in interacting with local anct tribal governments anct collecting in- formation updates. At the very least, steps shoulc! be taken to lessen the burclen of partnership: conducting LUCA-like acictress list reviews in electronic form with submissions via the Internet anc! (as mentioned earlier) coordinating the various geographic data collection programs so that localities are not being askoct for similar information in ctiffer- ent formats by different clivisions of the Census Bureau. The Census Bureau neects to articulate a plan for communication with localities that takes advantage of existing structures, including the State Data Center Network, the Fecleral-State Cooperative Program for Population Estimates, State/Regional Councils of Governments, anc! other local governmental entities. The role of the Census Regional Of- fice Geographic Coordinators relative to these entities anc! to Census Bureau headquarters neects to be spelled out. The ability and willingness of different governments to join forces with the Census Bureau vary widely. It is inevitable that tensions will arise when local efforts are differentially expressed across different ar- eas of the nation, whether such effort be clevotec! to mapping, to acI- ctress listing, or to the nurturing of partnerships. Different areas should receive equal treatment in the spirit of fairness, yet local interest, fea-
70 PLANNING THE 2010 CENSUS sibility, anc! cost-effectiveness might well dictate otherwise. Moreover, geographic partnerships with local and tribal governments are useful to tap the knowledge anct expertise of those closest to the fielct, but those partnerships are not a panacea. Variation in geographic information sys- tems usage may impact the accuracy in local anct tribal government ge- ographic resources anc! coulc! in cases introduce error when mixed with census resources. In the interest of effectiveness, the successes anct failures of prior LUCA programs shoulc! be analyzer! in orcler to develop new commu- nity participation programs for 2010. Moreover, refinect evaluation of the 2000 address file by type of enumeration area, by dwelling type, by the contribution of geographic update programs like LUCA, and by re- gion of the country highlighting areas of the country where eliciting local anc! tribal information may be most beneficial is surely requirec! if the Census Bureau is going to maintain the MAF in a cost-effective manner in the years leacling to the 2010 census. The Census Bureau's future plans for LUCA and other partnerships programs should also include provision for evaluation of those very partnerships, not only to inform the effectiveness of local contributions from the census per- spective but also to provide feedback to participating local anc! tribal governments. Recommendation MAF-3: The Census Bureau and the Geography Division should move as expeditiously as possible to develop and describe plans for partnerships with state, local, and tribal governments in collecting address list and geographic information. Such plans should include a focus on adding incentive for localities to contribute data to the census effort, making it easier for localities and the bureau to exchange geographic in- formation. Plans for partnerships should clearly define benchmark standards for local data to be submitted to the bureau. THE KNOWLEDGE BASE FOR 2010: CURRENT AND FUTURE EVALUATION WORK A recurrent theme in our preceding remarks is that there is a strong need for empirical evaluation of the construction of the MAF for the
MODERNIZING GEOGRAPHIC RESOURCES 71 2000 census. The foundation of the Census Bureau's evaluation studies along these lines is known as the MAF Extract. Related to the Decen- nial Master Address File the "snapshot" of the MAF that was used to generate census mailing labels and to monitor mail response the MAF Extract includes "flags" that indicate which of several sources contributed the address to the MAF. The MAF Extract also contains selected outcome measures, such as whether the address record was actually used in the 2000 census and whether it was tagged as a potential duplicate during the ad hoc duplicate screening program of early to mid-2000 (Nash, 20001. The MAF Extract has certain liabilities, chief among them that the system of flags used to indicate the source of an address does not constitute a true history of the address on the MAF. Other than rough temporal ordering of the input sources themselves, it is usually impos- sible to determine which source first contributed the address. Still, the extract is critical to answering key questions about the MAF-building process, and the panel continues to urge that the data resource be tapped for as much information as possible. Analyses of the extract should be conducted with respect to the type of enumeration area the address belonged to for the 2000 census (e.g., mailout/mailback or update/leave), as well as by geographic re- gion of the nation. The main objective of analysis of the MAF Extract is not to highlight how different areas of the country may have fared under various programs in place at the time. Areas of the country do differ, but knowledge of how they respond and interact with census activities is essential knowledge for the planning of future census pro- grams (see Question 6 below, for instance). Some key questions to address through Census 2000 evaluations are the following: 1. Why were addresses included in the MAF but not in the 2000 census? This question provides perspective for the others on this list and is a good starting place. 2. How useful were the DSF updates in the identification of new units, especially in high-growth areas of the nation? The goal is to examine how much of the newest housing was picked up in a timely fashion by the U.S. Postal Service. An-
72 PLANNING THE 2010 CENSUS swers provide valuable clues about the effort the Census Bureau shoulc! put into other avenues (e.g., new construction program) as sources of information on new housing. 3. How effective were LUCA inputs relative to what was already known (or was promptly seen) in a DSF update? Of those con- tributions that can be cleterminect as "unique," how many gov- ernments were represented anc! what kind of housing clo these acictresses represent? While LUCA must be concluctec! in 2010, the resources the Cen- sus Bureau chooses to expend on it can vary dramatically. Also, the answer to this question can inform strategies for the LUCA program in 2010. 4. What were the original sources of acictress records that were cleletec! as duplicates in the act hoc duplicate identification anc! removal process concluctec! in 2000? Duplication that is tied to acictress listing anomalies can be recti- fiec! once the problems with duplicate aciciresses have been iclenti- fiect. Identifying the original contributing source of affected act- clresses is a prime means for doing that. 5. What were the original sources of aciciresses that were flagged as potential duplicates but later reinstated? This acictresses the hypothesis that some acictresses, originally consiclerec! as potential duplicates, were put back into the census in error. The Census Bureau already has an estimate of this number. By identifying the original sources of the aciciresses, the bureau will have valuable clues about what proclucect this problem and how to avoid it in the future. 6. What were the original sources of aciciresses for housing units where an interview was not obtained in nonresponse follow-up (NRFU) ? One hypothesis regarding the shortfall of long-form data in the 2000 census has to do with NRFU enumerators encoun- tering high levels of resistance from respondents who were being enumerated for the first time ever (some were there in 1990 but escaped cletection). Where did the addresses of these
MODERNIZING GEOGRAPHIC RESOURCES 73 tough-to-enumerate units fall? (Of course, this is not the only hypothesis that coulc! explain problematic long-form ciata; it is almost certainly not the most likely hypothesis, either. But it is an intriguing question that shoulc! be aciciressable using internal Census Bureau data on the MAF.) 7. What were the original sources of acictresses for housing units when the housing unit was cleclarec! to be nonexistent or coulc! not be found in NRFU? NRFU enumerators hac! the option of entering cocles for "can- not locate," "cluplicate," and "nonresiclential," among others, as reasons for listing a unit as "nonexistent." Were these potential duplicates aciclec! back in? Were erroneous aciciresses brought in from LUCA that were not cletectect by the Census Bureau? Or were these aciciresses disproportionately from some other original source? 8. For cases where a unit was cleterminect not to exist in coverage improvement follow-up (CIFU; the final follow-up stage cluring the actual fielcting of the census), what was the original source of the aciciress? How many aciciresses were erroneously kept in the census and then cleletecl when the bureau went out to check in CIFU? The 1990 ACE, the initial 1997 DSF update, anc! block canvassing account for a very large percentage of all acictresses in the 2000 census in mailout/mailback areas. In absolute terms, these sources will clominate any original sources in a volume analysis. Nonetheless, normalizations are possible so that the Census Bureau can more properly quantify the real contributions of various inputs to those acictresses that were consict- erecl correct in the 2000 census. Most especially the effect, and perhaps differential effect, of LUCA programs neecis immediate attention. In this last assessment, account should be macle of the sometimes faulty nature of the LUCA program in the 1990s. timing anc! man accuracy problems among them. 1 ~ Completed Evaluations As part of the evaluation process for the 2000 census, an evaluation "topic report" on aciciress list development was scheclulec! for release
74 PLANNING THE 2010 CENSUS in Tune 2003.~ This report will synthesize the results of individual evaluation reports related to the general topic. It is expected that the detailed individual reports will be released in conjunction with the topic report. Consistent with its previous recommendations, the panel urges the Census Bureau to continue to make evaluations related to address list development a priority and to actively incorporate them into 2010 planning, including the census field test in 2004. The panel has received access to a small number of individual topic evaluations that are not yet publicly available. These include: Assess- ment of Field Verification (Tenebaum, 2002~; The Address Listing Oper- ation and its Impact on the Master Address File (Ruhnke, 2002~; Block Canvassing Operation (Burcham, 2002~; Evaluation of the Local Update of Census Addresses 99 (LUCA 99) (Owens, 2002~; and List/Enumerate (Zajac, 2002~. Though they are not yet publicly available, we do wish to offer some comment on them to help guide future evaluation work. Field Verification In field verification, enumerators visited the locations of units with returned questionnaires lacking an assigned census ID number, to ver- ify existence. These responses came from the Be Counted Program, from Telephone Questionnaire Assistance, and other alternative re- sponse modes. Some 885,000 cases were subject to this verification step. About half of them were coded as valid; about a third of them were coded as deletes; the remainder as duplicates. Of particular interest: more than half of the addresses that had been deleted in two or more previous operations were coded as valid addresses. Tenebaum (2002:11) suggests "that the Bureau may need to conduct additional research into the source of the double deletes with a mail return to . . . . . . . . . try to determine why they were deleted In two or more previous operations." We would like to see this research, with an emphasis on address histories and especially original sources, and with further detail on geographic locations and multi-unit dwellings. Some of the geographic detail is here by regional census office and by type of local Spas of the end of July 2003, the report had not yet been made publicly available. Lithe Be Counted program allowed respondents who felt that they had been missed in the mailout of census forms to pick up a census form from public offices and submit it.
MODERNIZING GEOGRAPHIC RESOURCES 75 census office; some data are also available on the multi-unit aciciresses involvec! in fielc! verification. Address Listing Operation The Aciciress Listing Operation (Ruhnke, 2002113 appears to have been quite successful, if not efficient. It is noteworthy that among some 22 million acictresses aciclect to the MAF by this operation, 99 percent of them were deliverable to the DMAF anc! 43 percent of them matched to acictresses iclentifiect as residential on or before the September 1998 DSF. However, the performance of Aciciress Listing in handling multi- unit structures is hindered by a flaw in the definition of the MAF vari- able containing the number of separate housing units at a basic street acictress (BSA). Specifically, all non-city-style acictresses which consti- tute at least 14 percent of the cases aciclec! by aciciress listing are auto- matically consiclerect single units. Although the evaluation report con- tains some geographic clisaggregation Breakdowns by state), much of the report has little bearing on the questions we have listed above. Block Canvassing Operation The Block Canvassing Operation (Burcham, 2002) playocl a big role in improving the coverage of aciciresses on the MAF anc! in improving the associated geococting, presumably at considerable expense. Block canvassing proclucec! 6.4 million aciclitions (some 30 percent of which were corrections or completions of acictresses already on the MAF anct some 35 percent of which were in multi-unit BSAs). Among the 6.4 million acictitions, 78 percent of them were valict acictresses for the 2000 census. There were 5.1 million cleletions (of which 48 percent were in multi-unit BSAs) anct 24 percent of them turned out to be valict act- ciresses for the 2000 census. Burcham (2002) provides some mention Tithe Address Listing Operation was used to build an initial address list for geo- graphic areas of the country that were to be enumerated using update/leave method- ology. Between fuly 1998 and May 1999, census field staff went door-to-door in these designated areas, making a list of mailing addresses and locations as they went along. The results from this operation, Address Listing, were then used to assign work during the actual census. In 2000, census field enumerators visited these sites to leave cen- sus questionnaires and logged MAF updates they encountered (hence the update/leave terminology).
76 PLANNING THE 2010 CENSUS of particular areas a large number of LUCA 98 cleletes occurrec! in Cook County, Illinois, clustering of acicis anc! cleletes was fount! in Ver- mont, anct there were many nonresidential cleletes in Los Angeles, for example where local information appears to resolve outlying results. These case studies provide useful lessons for future reference; a good clear of information is broken clown by state. The report cloes touch on our Question 3 about LUCA but acicts little information central to it. LUCA 99 Reports by the Working Group on LUCA (2001) anct the National Research Council (2001a) provide further insight into levels of LUCA participation by size of government and geographic location. However, only the Census Bureau evaluations can provide a picture of LUCA effectiveness by key variables. Owens (2002) provides some information pertinent to Question 3, of which the following is perhaps of most interest to the panel. Par- ticipation rates were higher for larger governmental bodies; 2.2 million LUCA 99 addresses were subject to recanvass with about 76 percent being verified, 18 percent corrected, anct 6 percent cleletect; recanvass- ing itself aciclec! 328,000 aciciresses; some group quarters may have been aclclecl through LUCA 99 and the subsequent recanvassing; some of this information is broken clown by state. List/Enumerate List/Enumerate (Zajac, 2002) aciclec! about 390,000 aciciresses to the MAF in sparsely populated areas of the country, more than 99 percent of which were incluclec! in the 2000 census; a rough estimate of cost per acictress is $50. A fair amount of information is broken clown by state. Evaluation here cloes not provide many answers to questions posec! earlier. General Assessment MAF evaluation work is required if the Census Bureau is to assess targeting methods for the ultimate goal: "to accurately identify local ar- eas with potential MAF/TIGER coverage/quality problems," especially as it concerns the 2004 tests (Waite, 2002~. In this vein, some of the
MODERNIZING GEOGRAPHIC RESOURCES . 77 Census Bureau's forthcoming evaluation studies are of particular im- portance. One of these is the Housing Unit Coverage Study (Barrett et al., 2001) that clears with erroneous enumerations by MAF source; we suggest that the Census Bureau focus on erroneous enumerations by type (e.g., cluplicates). The evaluation reports proviclec! to us generally give volumes and rates of acicts/cleletes that, when of interest, lack sufficient detail to be of use in guicling cost-effective strategies for targeting areas for coverage and acictress list improvement. Waite (2002) mentions the relative stability of the aciciress list in the 2000 census as a tool for . MAF targeting we certainly endorse this thought, but we have not seen what is planned in this regard. Evaluations clo not, as yet, yield much information on the aciclec! cost and benefit of programs. This information is surely crucial to the clecision-making process during the present clecacle. Evaluations and presentations still fall short of alleviating our fear that the process of maintaining and updating the MAF in the near future becomes the default one of acquiring DSFs on some regular schedule, with augmentation from LUCA programs after fielc! verification on a neecI-to-know basis. We have seen too few signs that cost and effectiveness of various acictress sources are unclerstooct on the basis of what transpired in the late 1990s. Such unclerstancling remains an issue of the highest possible priority. Recommendation MAF-4: Consistent with the panel's related recommendations on evaluation studies and the crucial importance of address list issues in conducting the census, the Census Bureau should: 1. strive to fully exploit the information on address sources contained in the MAF Extract in complet- ing 2000 census evaluations and assessing causes of duplicate and omitted housing units; and 2. build the capability for timely and accurate evalu- ation into the revised MAF/TIGER data architec- ture, including better ways to code address source histories and to output data sets for independent evaluation purposes.
