October 18, 2001 Census Bureau Says No To Adjustment; Review Finds Duplicates Wipe Out Most of Net Undercount Plus: Academy Panel Issues Report on Census 2000; Upcoming Advisory Committee Meetings; and more. The Census Bureau announced yesterday that it would not adjust Census 2000 data for non-political purposes, such as allocating federal program funds, citing a larger number of previously unidentified duplicates that reduced the net national undercount to "virtually zero in statistical terms." At a press conference in Washington, DC, Acting Bureau Director William Barron said he "concurred with and approved" the recommendation of the Executive Steering Committee for Accuracy and Coverage Evaluation Policy (ESCAP), which transmitted a report to him earlier this week, that unadjusted data should be used for non-redistricting purposes. He also said it would have been "a terrible mistake" to adjust the census counts issued to the states for redistricting last winter, based on the A.C.E. results. The bureau said that evaluation of "considerable new evidence" revealed that the Accuracy and Coverage Evaluation (A.C.E.) survey did not measure "a significant number" of double-counts and other counting mistakes (collectively called "erroneous enumerations"). As a result, the ESCAP found, the A.C.E. overstated the net national undercount by about three million people. The committee reported earlier this year that 3.1 million people were counted twice and an estimated 6.4 million people were missed in Census 2000, for a net national undercount of 3.3 million. The net undercount is the difference between the number of people counted twice or in the wrong place and the number of people missed, as measured by the post-census survey. The bureau issued preliminary revised undercount estimates that take into account the newly identified duplicates, bringing the net undercount down to 0.06 percent from the 1.18 percent reported last March. Higher proportions of racial minorities and Hispanics than Whites were missed in the census, although the gaps were smaller than in 1990. The revised undercount for the Black (non-Hispanic) population is 0.78 percent, compared to the 2.17 percent originally reported; the Hispanic undercount was lowered from 2.85 percent to 1.25 percent. The revised estimate for Whites and all other groups showed a small overcount of 0.28 percent, compared to an initial net undercount estimate of 0.73 percent. The 1990 census had a net national undercount of 1.6 percent, based on a similar though smaller quality-check survey. The ESCAP noted that the "net undercount remains disproportionately distributed among renters and minority populations." The committee said that further research might result in further revised estimates of undercounting or overcounting that could be used to improve the accuracy of the annual population estimates, particularly for harder-to-count demographic groups. At the press conference, Mr. Barron said the Census Bureau must start planning now for the 2010 census. He said he did not believe the bureau could ever complete an Accuracy and Coverage Evaluation program similar to the one fielded in 2000, in time to adjust the detailed census numbers issued to the states for redistricting within a year after Census Day. Congressman Dan Miller (R-FL), chairman of the Subcommittee on the Census, applauded the Census Bureau for "[choosing] the count that accurately shows real people, living in real neighborhoods and communities in a very real nation. We now know with scientific certainty, that a virtual census would have been less accurate than a real count." Referencing the Commerce Department's decision not to adjust the 1990 census based on a similar post-census survey, the chairman said, "We should give up on this twice failed experiment and put this money into improving the real count in 2010." Congressman William "Lacy" Clay (D-MO), the subcommittee's senior Democrat, called the bureau's decision "disappointing and troubling." He said the revised undercount estimates "do not allay the concerns of cities and communities seeking their fair share of funds for critical services. Despite many successes, Census 2000 still missed a lot of people in our poorer neighborhoods." Congressman Clay called on the bureau to release the adjusted numbers to assess the count's accuracy at the local level. Congresswoman Carolyn Maloney (D-NY), a subcommittee member, said the Census Bureau's decision "gave us more questions than answers," including why the bureau "conduct[ed] a search for duplicates but not conduct a search for omissions." She called on the Bush Administration to publicly release the adjusted data. Mr. Barron's statement and the ESCAP report are available through the Census Bureau's web site at <http://www.census.gov/Press-Release/www/2001/webcast10_17event.html>. National Academy of Sciences panel evaluates Census 2000: The National Research Council (NRC), an arm of the National Academy of Sciences, has issued a new report, "The 2000 Census: Interim Assessment." The Panel to Review the 2000 Census, chaired by former Bureau of Labor Statistics Commissioner Janet Norwood, assessed Census 2000 operations, including the Accuracy and Coverage Evaluation (A.C.E.) program and statistical imputation methods, based on information available through August 2001. The expert panel said Census 2000 was "well executed in many respects," and that adequate funding helped to keep operations on schedule. It called the A.C.E. program, designed to measure the net undercount (or overcount) in the census, "well planned and documented... and [generally] well executed." The panel did not offer an opinion on whether Census 2000 data should be adjusted based on the A.C.E., but concluded that the Census Bureau followed its plan for evaluating the census and A.C.E. results, and that the recommendation last March to release unadjusted numbers for redistricting purposes "was justifiable." Imputation, a statistical procedure used to count people about whom the Census Bureau has incomplete or no direct information, was largely responsible for the lower net undercount in 2000, the panel said. According to the report, the 2000 census included 5.8 million imputed people, three times the number imputed in the 1990 census. A disproportionate number of imputed persons were minorities, renters, and children, the panel found, "thus accounting in large part for the reduction in differential net undercount for these groups, relative to non-Hispanic whites, owners, and older people." The State of Utah has challenged the use of imputation in federal court, contending that the Census Act and the U.S. Constitution prohibit the use of statistical methods to compile the state population totals for congressional apportionment purposes. The panel also questioned the accuracy of the Census Bureau's independent population benchmark, compiled from administrative data such as birth, death, Medicare, and immigration records. The Demographic Analysis estimate of 279.6 million was lower than the raw census count (281.4 million) for the first time, and about 5 million below the A.C.E.-adjusted population estimate of 284.7 million. Focusing on the bureau's methods for gauging immigration and emigration, the Academy panel concluded that "demographic analysis should not be used as a standard for evaluation [of census accuracy] at this time." The Census Bureau had cited the discrepancy between the three measurements as a primary factor in its recommendation against releasing adjusted census data last March. Federal agencies and outside experts should research improved methods for estimating legal and undocumented immigration, the panel suggested. The panel singled out four successful innovations in Census 2000: contracting for data processing, with better technology to 'capture' responses on the forms; simplified questionnaires and a direct mail strategy to encourage response; the paid advertising campaign and expanded outreach; and aggressive recruitment of census takers. The higher-than-expected mail response rate -- about the same as in 1990 - was "an important achievement," the panel said, although areas with low mail-back rates in 1990 had similarly low response in 2000. The report also noted the "marked decline" in the mail-back rate for the census long form, which the panel cautioned could affect the quality of those data. The NRC report expressed concern about the quality of the count of people in so-called "group quarters," such as college dorms, prisons, and nursing homes. The panel also recommended further evaluation of procedures to build the national address list (called the "Master Address File," or MAF), to determine if the Local Update of Census Addresses (LUCA) program improved the accuracy of the MAF. While it made sense to seek address information from local and Tribal governments and the U.S. Postal Service, the panel said, "there were problems in execution that may have increased duplicate and erroneous enumerations." The National Research Council is part of the National Academy of Sciences. The report is available through the Internet at <www.nap.edu/books/0309076498/html/. Appropriations (funding) update: Congress is expected to pass a third "Continuing Resolution" to keep federal agencies running at fiscal year 2001 funding levels until lawmakers enact spending bills for fiscal year 2002, which started on October 1st. The new Continuing Resolution will run through October 31. Census Advisory Committee meetings: The Decennial Census Advisory Committee will meet on November 5 - 6, and the five Race and Ethnic Advisory Committees will meet on November 7 - 9, to discuss Census 2000 evaluations (including the ESCAP's work), planning for the 2010 census, the status of the American Community Survey, the findings of the National Academy of Sciences' Panel to Review the 2000 Census, the expanded Census Information Center program, and other relevant issues. Both meetings will take place at the Hilton Alexandria Mark Center Hotel, 5000 Seminary Road, Alexandria, VA (703-845-1010). The sessions are open to the public. New Census 2000 evaluations: The U.S. General Accounting Office, the audit and investigative arm of Congress, has issued two evaluations of Census 2000 operations and management. The reports, "2000 Census: Review of Partnership Program Highlights Best Practices for Future Operations (GAO-01-579, Aug. 20, 2001)" and "2000 Census: Better Productivity Data Needed for Future Planning and Budgeting (GAO-02-4, Oct. 4, 2001)," are available through GAO's web site at www.gao.gov <http://www.gao.gov> or by calling 202-512-6000 (TDD/202-512-2537). Questions about the information contained in this News Alert may be directed to Terri Ann Lowenthal at 202/484-2270 or, by e-mail at <mailto:terriann2k@aol.com>. For copies of previous News Alerts and other information, use our web site www.census2000.org <http://www.census2000.org>. Please direct all requests to receive News Alerts, and all changes in address/phone/fax/e-mail, to the Census 2000 Initiative at <mailto:Census2000@ccmc.org> or 202/326-8700. Please feel free to circulate this information to colleagues and other interested individuals.