The Panel on Research on Future Census Methods
AS THE 2000 CENSUS APPROACHED, the U.S. Census Bureau requested that the Committee on National Statistics (CNSTAT) of the National Research Council convene two panels, one to provide an independent and comprehensive review of the 2000 census and one to examine census conduct in 2000 with an eye toward the planning of the 2010 census. The Panel to Review the 2000 Census began work in 1998 (National Research Council, 2001a, 2004). Our Panel on Research on Future Census Methods began its operations in 1999 to assist the early planning efforts for the 2010 census, and this is our final report.
1–A CHARGE AND OPERATIONS OF THE PANEL
The Panel on Research on Future Census Methods has as its charge the following:
The panel will review the plans for acquisition, analysis, and evaluation of research data needed to begin planning for the 2010 decennial during the 2000 census. The panel will suggest improvements and preferred approaches, observe the implementation of the 2000 census, suggest pri-
orities for analyzing the census experimental and tracking data, examine census accuracy, evaluate the research program results, and determine appropriate lessons for the 2010 census.
During the course of its work, the panel’s charge evolved into a broad charter to review and advise on the emerging general plans for the 2010 census. This alteration in focus was enacted with the knowledge and consultation of the Census Bureau and through identification of a broader set of tasks in contract modifications with the Bureau. It also evolved naturally due to three factors.
First, this panel was formed very early relative to the census it examines, a factor that makes it unique in the experience of National Research Council panels regarding the decennial census. Our early start—beginning before the 2000 census—provided valuable opportunities to observe the census process, but also created unique challenges. The panel underwent a nearly year-long hiatus in 2000, respecting the heavy demand on the Census Bureau and its senior staff during the active follow-up and processing of the 2000 census. In the summer and fall of 2001, the panel observed another period of relative dormancy at the Census Bureau’s request, as the Bureau’s executive and evaluation staff were committed to intensive research and deliberation over the question of statistical adjustment of 2000 census data for estimated undercount. The intensity of census coverage evaluation research in 2001 also delayed the Bureau’s general program of operational and procedural evaluations of the 2000 census, which was to be the focus of the panel’s activities under its original charge.
Second, the Census Bureau made an early start on active planning for the 2010 census. At the end of 2000, the Bureau sought the panel’s assessment of the emerging “three-legged stool” plan for 2010 (consisting of the new American Community Survey, modernized geographic resources, and early planning and testing, and discussed further in Chapter 2). Since that time, the developing 2010 census plan and its major initiatives have been the primary areas of concentration in the panel’s work. In cooperation with the Bureau, full-panel meetings during 2001 and 2002 were largely replaced by meetings of five
working groups of panel members: Address List Development, American Community Survey, Enumeration Methods, Technical Infrastructure, and Coverage Evaluation and Statistical Infrastructure. Each of these working groups dealt with aspects of the developing 2010 census plan, and each was assigned liaison staff from the Census Bureau. The broad classification of topics across the working groups provides the basic structure for this final report.
Finally, the role and charge of our panel was influenced by the simultaneous operation and coordination of our panel and the Panel to Review the 2000 Census. The two panels were regularly apprised of each other’s work during their tenures; members of our panel participated in some of the activities and workshop meetings of the 2000 census panel and vice versa. By their nature, the charges of the panels overlap, with both panels having at their core a mission to review the 2000 census and advise on possible changes. Given the presence of a standing panel to comprehensively review the 2000 census, it was natural for our panel to be (as we have colloquially been known since our founding) “the 2010 panel” and to take as a primary focus the developing plans and initiatives of the 2010 census.
In support of our charge, the panel conducted a variety of activities. The panel met in plenary session nine times during its tenure, but more often in small-group settings. Members of our panel joined the Panel to Review the 2000 Census to visit data capture centers, regional census offices, and local census offices during the conduct of the 2000 census. The two census panels held one joint meeting in March 2003 to hear final results from the Census Bureau’s Accuracy and Coverage Evaluation Program and the Bureau’s decision on statistical adjustment of census data for use in generating postcensal population estimates.
In addition, the two census panels jointly established a working group to evaluate the Local Update of Census Addresses (LUCA) Program, in which state, local, and tribal governments were able to review address lists or block-level population counts for their areas and suggest revisions. The Working Group on LUCA completed its report to both panels in early 2001 (Working Group on LUCA, 2001).
1–B PREVIOUS REPORTS OF THE PANEL
In February 2000, the panel issued its first interim report, Designing the 2010 Census (National Research Council, 2000a), based on early information gleaned from the panel’s first two meetings. The report focused on priorities for the evaluation program for the 2000 census and argued in particular for the creation of a Master Trace Sample, collating information from many census operational databases for a sample of addresses.
In December 2000, the panel heard the Census Bureau’s first presentation of its preliminary 2010 census strategy (the “three-legged stool” approach described in Section 2-C), and offered early feedback on the general strategy in a letter report to acting census director William Barron in February 2001 (National Research Council, 2001c).
The panel’s second interim report, Planning the 2010 Census, was released in July 2003 (National Research Council, 2003a). In that report, the panel focused on three substantive areas: development of a technical infrastructure, modernization of geographic resources, and implementation of the American Community Survey (ACS). The report also provided brief comments on the plans for the 2003 and 2004 census tests.
1–C OVERVIEW OF THIS REPORT
This final report of the Panel on Research on Future Census Methods builds from and extends the material from the two interim reports that preceded it. Indeed, the text and structuring of some portions of the report are adapted directly from those reports, with appropriate revisions; we have chosen this approach (rather than incorporating the material through citation) in order to make this report as self-contained a document on 2010 census planning as possible.
The report is structured in three parts. Part I provides background and an overview of the Census Bureau’s general plan for the 2010 census. Following the present chapter’s synopsis of the panel’s charge and activities, Chapter 2 describes the basic 2010 census strategy that has been developed by the Bureau. More
significantly, Chapter 2 summarizes conclusions about the risks inherent in the 2010 census plan.
In Part II, we turn to specific issues in the design of the 2010 census as the plans have been developed to date. Chapter 3 discusses the Census Bureau’s plans to modernize its geographic resources. Chapter 4 examines a particularly crucial piece of the Bureau’s plan for 2010: implementation of the American Community Survey and, with it, elimination of the census long form, and we suggest critical research priorities for the survey. Chapter 5 explores enumeration and data-processing methods for the 2010 census, focusing primarily on the Census Bureau’s plans to use portable computing devices for nonresponse follow-up enumeration. Our discussion on enumeration methods also provides commentary on the need for attention to the Census Bureau’s handling of special places and group quarters and to the enumeration of hard-to-count populations. We also describe two elements of the data-processing stage that emerged as concerns in the 2000 census: unduplication and imputation. Chapter 6 describes the Census Bureau’s attempts to more effectively model its technical infrastructure—work that is critical to making sure that all the activities and plans described in the preceding chapters are efficient and supported by a reliable technological base. Finally, Chapter 7 briefly discusses the need for coverage measurement and evaluation in the 2010 census.
In Part III, we return to high-level themes of 2010 census planning. Chapter 8 discusses the Census Bureau’s research and evaluation program, noting the need for a comprehensive evaluation program and for better exploitation of resources such as the Master Trace Sample. Chapter 9 discusses priorities for census testing, outlining what we believe to be crucial elements of a proof-of-concept test in 2006 and a dress rehearsal in 2008.