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.
1 Introduction Census data collection involves four key steps: (1) the construction of an address frame; (2) an initial process to obtain responses that can be linked to the address frame; (3) a follow-up process to obtain responses from those not cov- ered in the initial process; and (4) a coverage assessment process that estimates the size of the population not covered through the initial and follow-up processes. The design of a census data collection process in essence amounts to decid- ing which methods of identification, enumeration, response, and coverage im- provement should be applied at each of the steps; whether sampling methods (and the corresponding estimation methods) should be used at any of the four steps; and, if sampling methods are used, which methods and at which steps. The two strongest criticisms leveled against the 1990 census are that unit costs increased significantly, continuing a trend that began with the 1970 census, and that the problem of differential undercount by race persisted and possibly worsened, despite a large investment in programs that were intended to improve coverage (see, e.g., U.S. General Accounting Office, 1992~. These criticisms have contributed to the growing momentum and advocacy for fundamental change in census operations. In response to these criticisms, the Census Bureau is considering an unprecedented level of innovation for the 2000 census. As part of that effort, officials at the Department of Commerce and the Census Bureau requested a panel study by the National Research Council's Committee on National Statistics to provide independent technical evaluations of candidate methodologies. 15
6 COUNTING PEOPLE IN THE INFORMATION AGE THE ROLE OF THE PANEL The Panel to Evaluate Alternative Census Methods studied feasible methods for the census not only for 2000, but also for 2010 and beyond. We have a mandate to make recommendations for features of census design that should be investigated and developed for censuses after the next one. Some features of these future designs could and should be tested in the near term and further developed in conjunction with the 2000 census, even though they might not be fully implemented until subsequent censuses. Our deliberations led us to con- sider all demographic data systems, including current estimates, sample surveys, and tabulations of administrative records. The panel had four basic tasks: (1) identify designs to be investigated for the 2000 census; (2) evaluate proposed research on alternative census designs; (3) evaluate the results of the research and the selection of census designs for further consideration, in particular for the series of census tests that begin in 1995; and (4) recommend census designs to be explored for 2010 and succeeding years. The determination of methods for conducting the decennial census must take into account the information requirements placed on the census in its role as a key component of the federal statistical system. Data collected by the statistical system can be classified according to one of three levels of population coverage: (1) basic information-age, race, and ethnic origin to satisfy the requirements of the Constitution and the Voting Rights Act is collected from 100 percent of the population on the decennial census short form; (2) information that must be reliable for small geographic areas and subpopulations-e.g., education, occupa- tion, income- is currently collected from a large sample of the national popula- tion on the decennial census long form; (3) information for which timeliness is more important than geographic detail, such as unemployment rates and statistics on participation in federal entitlement programs, is now gathered by smaller sample surveys. Throughout our study, the panel has recognized the link between require- ments and methods, although thorough examination of census requirements is beyond the scope of this panel. Our cognate panel, the Panel on Census Require- ments in the Year 2000 and Beyond, is charged with assessing the needs for data currently collected in the decennial census. Both studies address issues of meth- odology but approach these issues from different perspectives. In considering alternative methods, the Panel on Census Requirements has been primarily con- cerned with implications for content and possible effects on public acceptance of results. That panel has also undertaken a more intensive review of the current census cost structure. Our primary attention has been given to technical issues of implementation and evaluation of promising methodologies. Our evaluations of alternative census methods emphasize the implications for differential coverage and census costs, but not to the exclusion of other considerations. For example, measures of gross census error are important in
INTRODUCTION 17 evaluating the effectiveness of proposed methods for increasing census response, such as distributing unaddressed questionnaires, offering the option of telephone response, and using special "tool-kit" enumeration methods in certain small geo- graphic areas. We believe that census methodology should strive to minimize not only omissions (that produce undercounts) but also erroneous enumerations (that produce overcounts). All undercounts and overcounts complicate the task of accurately measuring census net coverage. Undercounts and overcounts that arise from definitional problems c.g., a person is erroneously counted in one block and omitted from the correct adjacent block essentially balance and are not of much additional concern. But undercounts and overcounts that are non- uniformly distributed among particular areas or types of people lead to mis- distribution of the estimated population, even when such errors balance at larger levels of aggregation. The concept of total information error over a decennial period is a useful criterion against which to evaluate alternative methods for collecting small-area data that have traditionally been collected for a sample of respondents during the decennial census using a longer questionnaire. Proposals for a large, continuous survey of households and for other methods to improve intercensal estimates (e.g., expanded use of administrative records) should be evaluated by considering the needs for information with cross-sectional versus temporal accuracy. This criterion is driven by information requirements and, once established, can be used to judge alternative methods intended to address these requirements. The panel has had regular contact with several official advisory groups that are reviewing the planning for the 2000 census. Early in 1991, the Department of Commerce established a Task Force on the Year 2000 Census to provide an organizational structure for the examination of issues regarding the 2000 census. The task force comprises a technical committee, a policy committee, and an advisory committee. The technical committee consists primarily of experienced professional staff from the Census Bureau. The policy committee includes repre- sentatives from other federal statistical agencies. The advisory committee mem- bership includes representatives of state and local government, minority, and other interested professional organizations. Initially, the technical committee of the task force worked with the Census Bureau to construct a set of 14 census design alternatives. Each alternative was characterized by one or more unique design components; each was also judged to have the potential to meet the current demands of the decennial census. Of the 14 designs, 6 built on the basic structure of the 1990 census, adding different provi- sions: multiple ways of responding to the census, varying degrees of sampling and statistical estimation, and targeted methods to overcome barriers to enumera- tion. Two designs relied entirely or to a very significant extent on administrative records. Four designs would have collected data on fewer topics than have been covered in recent decennial censuses. Two designs proposed collecting census
8 COUNTING PEOPLE IN THE INFORMATION AGE data in two stages or through continuous measurement in the decade following the census year. The panel's first report to the Census Bureau in December 1992 raised questions about the 14-design approach. Subsequently, the Census Bureau de- cided to remove its original set of 14 alternative census designs from further consideration. Instead, the 1995 census test will evaluate promising components of the original alternative designs. The panel's September 1993 interim report strongly supported this reorientation of the 2000 census planning process. The interim report contained 35 recommendations, many of which suggested design components for inclusion in the 1995 census test (in response to the first element of our charge). We gave particular attention to methods with potential to reduce either census costs or differentials in coverage. Since the preparation of the panel's interim report, the Census Bureau has released two key documents (discussed below) the 1995 Census Test Design Recommendation and the Test Design Plan that identify census design compo- nents to be examined in 1995 and discuss plans for testing and evaluating these components. In this report, we review the progress in 2000 census research and development and, in keeping with our third task, evaluate the selection of design components for inclusion in the 1995 census test. We also study and comment on procedures being developed to implement and evaluate the 1995 census test. We continue to recognize the important considerations of cost and differential cover- age, yet we also discuss issues related to other factors, such as data quality and gross census error. Finally, we comment further on the broader research program beyond the 1995 census test that will inform planning not only for the 2000 census but also for subsequent censuses and for other demographic data systems. Many of our evaluations proceed with the assumption that the content of the census short form in 2000 will not be significantly changed from its 1990 version. We further assume that the primary instrument for collecting short-form informa- tion in the 2000 census will remain the mail questionnaire, the primary mode of data collection for the past four censuses (Goldfield, 1992~. Recent reviews of the statutory requirements for census data (Bureau of the Census, 1994a: appendix 2) indicate that legislative mandates exist for collection of most of the items currently gathered on the decennial long form, although the legislation typically does not mandate the vehicle for data collection. Later chapters of this report discuss nontraditional methods i.e., an administrative records census, a large-scale continuous measurement survey- for collecting information that may be comparable to what is gathered on the decennial census short or long forms. We also discuss, from a technical perspective, the pros and cons of dividing the current long form into a series of intermediate-length ques- tionnaires. The next section of this chapter reviews work done by the Census Bureau as part of its 2000 census research and development program, and the chapter con- cludes with some observations about census planning for the year 2000 and
INTRODUCTION 19 beyond. Chapter 2 discusses some important issues related to the first step of the collection process, the creation of an address frame, as well as legal and opera- tional issues. Chapter 3 considers methods with potential for improving response and coverage at various stages of the collection process. Chapter 4 addresses the possible use of sampling and estimation at each stage of the census data collec- tion process particularly in following up households that do not return the mail questionnaire and in measuring census coverage. Chapter 5 discusses the pos- sible use of administrative records in the four collection steps, as well as for current estimates and other demographic programs outside the decennial census. Chapter 6 addresses issues related to alternative schemes that would spread the collection of sample information for small areas and subpopulations over a de- cade, rather than concentrating efforts in a single year, or would involve the use of multiple sample forms in the decennial census. CENSUS BUREAU RESEARCH AND DEVELOPMENT The Census Bureau has responded to the challenges of counting people in the information age by undertaking an ambitious research and development program that reflects an imaginative rethinking of census methodology. This program will lead to the large-scale field testing in 1995 of design components that represent fundamental change from current census practice. The two main areas in which innovation in census design is taking place are: (1) response and coverage improvement and (2) expanded use of sampling and estimation. Research on response and coverage improvement has led to potentially important changes in questionnaire design and implementation. Sampling and statistical estimation methods are being explored to close the remaining differentials in census cover- age while controlling, or even reducing, overall cost. The Census Bureau will test a variety of innovative design features in the 1995 census test. Collection of reliable information in the 1995 census test about the costs and effectiveness of census design components will be essential for their proper evaluation particularly to inform decisions about allocating resources between efforts to improve primary response and efforts to use sampling and estimation methods to correct the counting operation. Evaluation Criteria for the 2000 Census The Census Bureau developed a set of mandatory and desirable criteria for assessing design alternatives, and it has specified that any design being consid- ered for the 2000 census must satisfy all mandatory criteria. Any design that meets the mandatory criteria will then be assessed according to the set of desir- able criteria. Six criteria are specified as mandatory by the Census Bureau for the 2000 census design:
20 COUNTING PEOPLE IN THE INFORMATION AGE 1. not require a constitutional amendment; 2. meet data needs for reapportionment; 3. provide data defined by law and past practice for state redistricting; 4. provide age and race/ethnic data defined by law to meet the requirements of enforcing the Voting Rights Act; 5. protect the confidentiality of respondents; and 6. possess the ability to reduce the differential undercount. Ten criteria are specified as desirable by the Census Bureau for the 2000 census design: i. result in comparative cost-effectiveness with respect to other alternatives under consideration in real terms on a per unit basis; 2. provide small-area data that the census is uniquely capable of providing; 3. provide a single, best set of census results produced by legal deadlines for reapportionment and redistricting; 4. provide an overall high level of coverage; 5. increase the primary response rate to the census; 6. reduce the level of respondent burden; 7. minimize the degree and type of changes needed in federal or state law; 8. consider the reliance on new or unproven methods or capabilities; 9. permit full development and testing of its major design features; and 10. provide opportunities to involve the U.S. Postal Service, state and local governments, national organizations, and other private, nonprofit, and commercial enterprises. We have already identified cost and differential coverage as important con- siderations in the panel's evaluations of alternative census methods. (Later, we describe some basic assumptions about the costs of census-taking that have guided the panel's deliberations.) Reduction of the differential undercount appears in the above list of mandatory criteria, along with constitutional and other legal requirements, underscoring the importance assigned by the Census Bureau and other interested parties to this objective of census reform. The prominence of cost reduction as the first desirable criterion above is also suggestive of the significant efforts being expended to achieve a more cost- effective census design in 2000. The Census Bureau has a very detailed cost model (Bureau of the Census, 1992b) that is used for operational planning. This model has been used to estimate costs associated with several design components being considered for the 2000 census, including conducting follow-up of mail nonrespondents over a truncated period of time or on a sample basis. The cost- effectiveness of other design components, such as the use of special enumeration methods, has yet to be determined. The 1995 census test should produce better information about costs and benefits for such components. Cost estimates for the continuous measurement option should become firmer when the product is more
INTRODUCTION 21 clearly defined and a small-scale prototype survey is in operation. Similarly, further experience with the statistical use of records from administrative data systems should lead to more reliable estimates of the cost of an administrative records census. Some relevant criteria are not explicitly identified in the above lists: in considering proposed design innovations, the Census Bureau must address poten- tial problems with erroneous enumerations. The 1990 census had approximately 11 million erroneous enumerations (the largest number recorded to date). As noted earlier, many of these errors result from minor definitional problems and essentially balance at larger geographic areas of interest, but errors that are nonuniformly distributed across the population can reduce the accuracy of census results. Without careful implementation, such innovations as the use of multiple response modes and new fostering procedures could exacerbate the problem of gross census errors. Aggressive research will be needed to develop techniques to prevent erroneous or duplicate enumerations during a census with multiple re- sponse modes or new fostering procedures. One-Number Census With regard to the third desirable criterion above, the Census Bureau has developed the concept of a "one-number census" that would provide "the best possible single set of results by legal deadlines, . . . based on an appropriate combination of counting, assignment, and statistical techniques" (Miskura, 1993~. In this definition, counting refers to the full array of methods used for direct contact with respondents, including mail questionnaires, personal visits, and tele- phone calls. Assignment refers to the use of evidence from administrative records to add people to the count for a specific geographic location without field verifi- cation. Statistical techniques for estimation include imputation procedures, sam- pling during follow-up of nonrespondents, and methods for measuring census coverage. The Census Bureau has expressed its commitment to pursue a one-number census for the year 2000, based on the integration of specific counting, assign- ment, and estimation methods to be determined by the 2000 census research and development program. This commitment is reflected in the decision not to adopt a dual-strategy approach for the 1995 census test. The one-number approach thus represents a departure from the methodology of the 1990 census, in which two sets of population totals were produced, with and without corrections based on coverage measurement, and an ex post facto decision was made about whether to accept the corrected totals. This decision proved to be controversial because it occurred in a highly politicized environment in which interested parties per- ceived themselves as winners or losers, depending on which set of numbers was chosen. Associated with the one-number census is the principle of integrated cover
22 COUNTING PEOPLE IN THE INFORMATION AGE age measurement, the premise of which is that the three components of a one- number census are designed to complement one another in order to produce accurate results by legal deadlines. That is, the results from measurement of coverage will be fully integrated into the official census estimates (Miskura, 1993). The definition of the one-number census does not imply a relaxation of the standards for census documentation, nor does it preclude the release or use of partial or preliminary data that is, intermediate calculations in the process of combining information obtained from counting, assignment, and estimation meth- ods. The fundamental change is that an appropriate methodology for integrated coverage measurement is established before the census is earned out, with the recognition that results at intermediate stages (e.g., before incorporating results from the coverage measurement program) cannot be regarded on scientific grounds as viable alternatives to the final, best set of official population totals. (Chapter 4 provides further discussion of the one-number census and includes several recommendations regarding documentation requirements for the 2000 census.) In our interim report (Committee on National Statistics, 1993b:3 1), the panel expressed strong approval of the one-number census concept. We reiterate that approval by including below the text of the recommendation that appeared in the interim report: · We endorse the Census Bureau's stated goal of achieving a one-number census in 2000 that incorporates the results from coverage measurement pro- grams, including programs involving sampling and statistical estimation, into the official population totals. We recommend that research on alternative method- ologies continue in pursuit of this goal. The panel's position on this issue is similar to the view expressed by the National Research Council's Panel on Decennial Census Methodology convened prior to the 1990 census (Citro and Cohen, 1985:17~: Most important, the panel argues for balance between efforts to achieve a com- plete enumeration and efforts to improve the accuracy of census figures through adjustment procedures. The panel believes that adjustment cannot be viewed as an alternative to obtaining as complete a count as possible through cost-effec- tive means. The United States has a long tradition of a census as a complete enumeration in which it is a civic responsibility to participate in the census process. The panel believes that it is important to continue this tradition and important that census methodology strive for a complete enumeration via count- ing procedures, including the use of cost-effective special coverage improve- ment programs. However, the panel also believes that the ultimate goal of the census should be the accuracy of the census figures. The evidence is over- whelming that no counting process, however diligent, will in fact enumerate everyone.
INTRODUCTION 23 A key design issue in achieving a well-balanced census in 2000 will be the allocation of resources between four major steps in census data collection: con- struction of the address frame, the initial counting operation, nonresponse follow- up, and integrated coverage measurement. The allocation decision will be a critical point in development of the 2000 census design, and information obtained from the 1995 census test should provide a stronger basis for that decision. Plans for the 1995 Census Test The 1995 census test represents the culmination of the Census Bureau's research and development program for the 2000 census, although research activ- ity will continue throughout the decade. Plans for the 1995 census test have been laid out with increasing specificity in a series of Census Bureau documents- including the Design Alternative Recommendations (DARs) released in May 1993, the 1995 Census Test Design Recommendation (TDR) released in August 1993, and the 1995 Census Test Design Plan released in February 1994. The Test Design Plan is a refinement of the TDR that reflects more technical work and comments offered during critical review of the August 1993 document. Further details of the 1995 census test plans are being laid out in a series of operational requirements documents (ORDs) and evaluation requirements documents (ERDs) that will eventually lead to detailed operating specifications. The 1995 census test will be carried out at four sites: Oakland, California; Paterson, New Jersey; New Haven, Connecticut; and six rural parishes in north- western Louisiana. The following methods are scheduled to be examined in the 1995 census test. 1. The Use of Sampling and Statistical Estimation to Reduce the Differential Undercount and Census Costs. Follow-up of nonrespondents to the mail ques- tionnaire will be conducted on a sample basis. Two different sample designs will be tested: (1) a unit design, in which a sample of the nonresponse cases in each block is visited and (2) a block design, in which all nonresponse cases for a sample of blocks are visited. In addition, the Census Bureau will test a new method for integrated coverage measurement that separately estimates the num- ber of persons missed because their housing unit was not enumerated, the number of persons missed within enumerated housing units, and the number of erroneous enumerations. 2. Coverage Questions for Complete Listing of Household Members. The Census Bureau is engaged in ongoing research on household roster questions and analysis of results from the Living Situation Survey and related cognitive re- search on residence rules. To the extent possible, findings from these studies will be incorporated into the development of questionnaires and procedures for the 1995 census test, including the reinterview for integrated coverage measurement. 3. Making Census Questionnaires Widely Available. The Census Bureau
24 COUNTING PEOPLE IN THE INFORMATION AGE proposes to place unaddressed questionnaires in accessible locations (e.g., post offices, convenience stores), particularly in areas inhabited or frequented by his- torically undercounted populations. 4. Real-Time Automated Matching to Improve Census Coverage. To sup- port the new census design proposed for the 1995 census test, the Census Bureau will develop, to the extent possible, an automated, interactive, and real-time record linkage and matching system. 5. Targeted Methods to Count Historically Undercounted Populations and Geographic Areas. The Census Bureau will develop a planning database to identify small geographic areas in which there are major enumeration barriers and to support other census operations. Special enumeration (tool-kit) methods will be developed and applied during the 1995 census test. 6. Mailout of Spanish-Language Questionnaires. Based on results of the Spanish Forms Availability Test, the Census Bureau will mail Spanish-language forms to linguistically isolated Spanish-speaking communities. Also, census questionnaires and promotional materials will be translated into predominant Asian languages if appropriate in the 1995 census test sites. 7. Counting Persons with No Usual Residence. The Census Bureau will implement a new daytime enumeration method that involves visits to service providers (e.g., shelters, soup kitchens). This method is being tested as a replace- ment for the nighttime street enumeration procedures used on "S-Night" in the 1990 census. 8. Respondent-Friendly Questionnaire Design and Implementation Meth- ods. The census forms used in the 1995 census test will have a format designed for ease of response rather than ease of processing. The Census Bureau will test a full mail implementation strategy, including a prenotice letter, an initial ques- tionnaire, a reminder card, and a replacement questionnaire for those who have not replied by a predetermined date. The envelopes used for some questionnaire mailings may bear a message indicating that response is required by law. 9. Automation of Data Collection. The 1995 census test will include a telephone network to support questionnaire assistance and the use of touch-tone menu selections with voice recognition and voice recording options. Computer- assisted telephone interviewing (CATI) will be used as a primary response option and as a mode for nonresponse follow-up. Pen computers will be used during address list and map updating. 10. Using the Postal Service to Identify Vacant and Nonexistent Housing Units. Vacant units and nonexisting addresses will be identified earlier in census operations using information supplied by postal carriers from their experience with delivery of prenotice or questionnaire mailings. Adopting this procedure will eliminate one of the two enumerator visits conducted in past censuses by the Census Bureau, and the remaining visit can occur earlier in the process-during the check-in of mail returns and prior to nonresponse follow-up. 11. Data Capture System for the 2000 Census Using Electronic Imaging.
INTRODUCTION 25 The Census Bureau plans to develop electronic imaging technology capable of scanning respondent-friendly census forms, using optical mark sensing and pos- sibly optical character recognition software. A production prototype will be built to capture data from at least one of the questionnaire form types used in the 1995 census test. 12. Cooperative Ventures. The Census Bureau will explore limited joint venture options in the course of operational planning for the 1995 census test. The Census Bureau and the Postal Service are working to reach an agreement for continuous updating of the Master Address File and for identification of vacant and nonexistent housing units. There may also be opportunities to work coopera- tively with local governments to improve the address list and to obtain adminis- trative records for coverage improvement. 13. Collecting Sample Data Using Multiple Sample Forms. The Census Bureau plans to test a prototype matrix sampling design in the 1995 census test. Matrix sampling involves the use of two or more sample forms in which most respondents will be asked only a subset of all the sample questions that are asked. The Census Bureau also considered experimenting with delayed sample- form follow-up in the 1995 census test. Under such a procedure, only short-form information would be gathered during nonresponse follow-up, after which sample-form data would be obtained via telephone or personal visit for a sub- sample of households that had initially received a sample long form. Because of the operational complexities involved in developing multiple CATI instruments and coordinating this effort with integrated coverage measurement, the Census Bureau has decided not to introduce delayed sample-form follow-up into the 1995 census test design, although the concept will be examined further to deter- mine its merit for future testing. The operational constraints on the 1995 census test underscore the impor- tance of learning as much as possible from other research. For example, simula- tion studies using 1990 census data can investigate the effects of truncating nonresponse follow-up operations at different points in time, using different rates of sampling nonrespondents for follow-up, and applying different coverage measurement methods (see Chapter 4 for further discussion). Similarly, not all methods need to be tested in large-scale field settings. To ease experimental complexity, certain methods might be excluded from large-scale field testing in 1995, when such an exclusion would not disrupt the research and development program or if smaller experiments (e.g., questionnaire research, discussed in Chapter 3) conducted simultaneously with the 1995 census test will provide useful information. Other Activities Two activities related to 2000 census planning are being conducted indepen- dently from the 1995 census test. First, through the policy committee of the Task
26 COUNTING PEOPLE IN THE INFORMATION AGE Force on the Year 2000 Census, the Census Bureau has undertaken a review of the statutory requirements for census data collection. Current findings from that review (Bureau of the Census, 1994a) suggest that only 6 of 59 topics that were included in the 1990 census lack a legislative mandate for collection. Based on this review, the panel's deliberations have not assumed dramatic reductions in the content requirements currently being met by the decennial census. The Census Bureau is also continuing its development of a prototype system for continuous data collection involving a large, monthly, national survey to produce frequent estimates based on moving averages. Such a system for con- tinuous measurement could potentially meet many needs for timely and accurate information about small areas and small groups within the national population. Current development plans call for a prototype survey to begin in fall 1994 at several geographic sites. Chapter 6 of this report contains a thorough review of Census Bureau research on continuous measurement and considers the extent to which a continuous measurement survey might satisfy the legal requirements for the data noted above. PLANNING FOR THE 2000 AND FUTURE CENSUSES The Costs of Census-Taking The panel's consideration of census costs in this report and in its interim report have been guided by a number of assumptions about the relative costs of alternative methods. First, nonresponse follow-up, particularly in its later stages, is clearly one of the most expensive parts of the census. Second, coverage improvement programs also add significantly to census costs. A previous Na- tional Research Council report (Citro and Cohen, 1985) documented the cost per case of coverage improvement programs used in the 1980 census. Eliminating inefficient coverage improvement programs and redesigning nonresponse fol- low-up-for example, to incorporate use of the telephone or sampling methods- should achieve cost savings. Another general principle is that gathering information by mail is cheaper than doing so by telephone, and gathering information by telephone is cheaper than doing so by sending an enumerator to conduct a personal visit. One implica- tion of this principle is that innovations in questionnaire design and implementa- tion that improve mail response will save money. Similarly, the growing cost advantages of the computer over human labor suggest that further automation of census operations, such as record matching, should also produce cost savings. Because of the panel's charge and the nature of our study, we did not attempt to conduct more precise cost-benefit analyses of current or proposed methods. We expect that the 1995 census test will provide more current and reliable infor- mation on which to base cost-benefit judgments. New cost information will be particularly useful in assessing the value of tool-kit enumeration methods, non
INTRODUCTION 27 response follow-up sampling, MAF/TIGER updating, new uses of administrative records, and the continuous measurement prototype. The cost information ob- tained from the 1995 census test, plus improved cost modeling capabilities, should permit cost-benefit analysis as recommended by the earlier panel on census meth- odology (Citro and Cohen, 1985~. Goals for the 1995 Census Test The 1995 census test is of critical importance to the goal of an improved and more efficient census in the year 2000. Because of the extensive operational planning that must occur prior to 2000, the 1995 census test represents the major opportunity to investigate fundamental reform without jeopardizing the integrity of the 2000 census. It is essential that adequate resources are invested in planning and executing this mid-decade test. Otherwise, the 2000 census will have a design very similar to that of the 1990 census, with the risk of continually rising unit costs or an inadequately tested design that risks lost demographic informa- tion and population counts of unknown or inferior quality. The 1995 census test should be structured to provide specific information to answer a limited and well-defined set of questions about alternative census meth- ods To the extent feasible, controlled experiments should be carried out, al- though the panel recognizes that operational pressures will limit the experimental complexity of the 1995 census test. In particular, the test should include evalua- tion components that provide a basis for assessing cost-effectiveness. Recommendation 1.1: In assessing.the design innovations included in the 1995 census test or other research and development, the Census Bureau should place great emphasis on cost-benefit analysis as part of the overall evaluation leading to implementation decisions for the 2000 census. Requirements for evaluating new data collection methodologies in the 1995 census test should include information on such characteris- tics as cost, yield, and gross error that are needed to inform cost-benefit judgments. We note that the year 2000 research and development staff became part of the decennial management division in summer 1994. It will be particularly important during this transition to ensure the continuity of the 2000 census plan- ning process, so that the research, development, and evaluation activities proceed in an integrated, coherent, and effective manner. We also believe it is important to maintain perspective on the role of the large-scale, mid-decade test in the decennial census research and development program. Previous panel studies by the Committee on National Statistics (Citro and Cohen 1985:21) have criticized the Census Bureau's program for placing too much emphasis on field testing over other kinds of research, including further analysis and simulation studies based on existing data. The 1995 census test, the
28 COUNTING PEOPLE IN THE INFORMATION AGE 1990 census, and other recent Census Bureau operations are rich bodies of infor- mation that may yield answers to key research questions in the coming years if these resources are acknowledged and fully examined. Milestones for 2000 Census Planning The Census Bureau's current schedule calls for a final decision in December 1995 on the fundamental design of the 2000 census. Further operational develop- ment and refinement of the design will continue with a variety of small-scale special purpose tests in 1996 and 1997 and will conclude with a census dress rehearsal in 1998, the results of which will inform plans for the 2000 census operation. As noted above, simultaneously with methodological development, the Cen- sus Bureau is engaged in a process to determine the content of the 2000 census- i.e., what information should be collected and specifically how should questions be phrased to collect this information. A test of census content is scheduled for fall 1996, and the Census Bureau will submit its proposed content for the 2000 census to Congress in spring 1997. A third activity with implications for both census methodology and content is the Census Bureau's program to develop a prototype continuous measurement survey (see Chapter 6~. The current schedule for this program calls for a decision in September 1997 about whether to retain the decennial long form for the 2000 census or replace it with a continuous measurement survey. The requirement that the Census Bureau submit in early 1997 proposed content for congressional review and approval may complicate the reaching of this decision point. At present, it is unclear whether this requirement would be met by furnishing a list of topics to be included in the 2000 census or whether the means of collecting data on these topics must also be determined by early 1997. Longer-Term Census Research and Development The panel's letter report (Committee on National Statistics, 1992) included the following two recommendations: The Census Bureau should initiate a separate program of research on administrative records, focusing primarily on the 2010 census and on current estimates programs. The research program should be funded separately from the 2000 census research and development activities, but there should be close liai- son between them. · The Census Bureau should undertake a planning study, in collaboration with other agencies and contract support as needed, that would develop one or more detailed design options for a 2010 administrative records census. The study would have two major goals: to identify the steps that would need to be taken,
INTRODUCTION 29 early in this decade, to make a 2010 administrative records census possible and to set the stage for a national debate on the desirability of an administrative records census. The study, or at least its initial phases, should be completed during the current fiscal year. In the fiscal 1995 budget currently before Congress, the Census Bureau has included an item titled "Research for 2001 and Beyond," and the item also ap- pears in the fiscal 1996 budget at this time. We believe these are positive steps. However, we caution that the budget for a long-term staff should be independent of the funding cycle for short-term research and development work on the next decennial census (see Bradburn, 19933. Consideration should be given to revis- ing organizational structures to minimize the extent to which short-term and long-term research divisions would compete for personnel and other resources. Chapter 5 includes a discussion of the long-term research that is needed to develop new, potentially cost-effective uses of administrative records for statisti- cal purposes in the decennial census and other demographic programs. We note in Chapter 6 that the Census Bureau has established a continuous measurement development staff to pursue the research agenda for creating a system of continu- ous data collection. Perhaps this newly created staff will serve as a model for mobilizing resources to pursue longer-term decennial census research projects at the Census Bureau.