From: "Census2000" <census2000(a)ccmc.org> Supreme Court Upholds Imputation Method In Utah Apportionment Case In a close (5 – 4) decision issued this morning, the United States Supreme Court upheld the Census Bureau’s use of imputation to count some people who census takers cannot reach through direct enumeration methods. Utah filed a lawsuit last spring, arguing that the statistical method violated both the Census Act and the Census Clause of the U.S. Constitution, and deprived it of a fourth congressional seat that instead went to North Carolina. North Carolina, which picked up a 13th congressional seat after the 2000 census, intervened in the case in support of the Census Bureau. The Supreme Court ruled that imputation does not violate either section 195 of title 13, United States Code (the Census Act), or the Constitution’s requirement for an enumeration of the population every ten years for the purpose of congressional apportionment. Justice Stephen Breyer authored the majority opinion in Utah et al. v. Evans, Secretary of Commerce, et al. (01-714), in which Chief Justice William Rehnquist and Justices John Paul Stevens, David Souter, and Ruth Bader Ginsburg joined. Justice Sandra Day O’Connor joined Parts I and II of the majority opinion, and filed an opinion concurring in part and dissenting in part. Justice Clarence Thomas also filed an opinion concurring in part and dissenting in part, in which Justice Anthony Kennedy joined. Justice Antonin Scalia wrote a dissenting opinion. “Hot-deck imputation” involves the use of scientific models to assign occupants (or vacancy status) to housing units from which a mailed questionnaire was not returned and census takers were unable to find any residents after six visits. The Census Bureau assigns the number of occupants based on information collected from a similar, nearby housing unit. Imputation added roughly 1.2 million people in 620,000 housing units (less than one-half of one percent of the national population) to the state population totals used for congressional apportionment. (For some households, imputation is used to fill in missing characteristics, such as race, or to add occupants who are not listed on a census form but for whom there is some evidence of residency. A National Academy of Sciences panel reported last fall that Census 2000 included 5.8 million imputations, a disproportionate number of which involved racial minorities, renters, and children.) The bureau first used the method in the 1960 census. Indiana unsuccessfully challenged the technique after the 1980 census, when it lost a House seat to Florida based on imputed population. In oral arguments before the Supreme Court on March 27, Utah’s lead counsel, Thomas Lee, described imputation as a form of sampling that the Court struck down in 1999 (Department of Commerce v. U.S. House of Representatives, 526 U.S. 316). In the earlier case, a 5–4 majority of the Court ruled that the Census Act barred the use of sampling to compile the state population totals used for congressional apportionment. In today’s ruling, the Court noted that imputation differs from sampling in several ways. Sampling involves extrapolating the characteristics of a large group of people from a small one, the Court wrote, while imputation seeks only to fill in missing data for particular housing units, using information from similar housing units that are not chosen randomly. The methodological differences place imputation “outside the scope” of section 195, which refers only to “sampling” (in quotations in the law), the majority concluded. The Court suggested that Congress knew about the use of imputation when it amended the Census Act in 1976 to prohibit sampling in deriving the apportionment counts, inferring that sampling, as used in the statute, was a “term of art.” The Court also held that hot-deck imputation does not violate Article I, section 2, of the Constitution (the Census Clause), which calls for a census every ten years “in such manner as [Congress] shall by Law direct.” The majority rejected Utah’s contention that the Constitution’s reference to an “actual Enumeration” requires the Census Bureau to count every person directly. Instead, the term “actual” referred to apportionment of the Third Congress (which would be seated after the first census was conducted), while apportionment of the First and Second Congresses was based on “conjecture” as to the population of the states then in existence, the Court said. The majority also pointed to other methods the Census Bureau has used since 1800 to add missing people to the count, including asking neighbors, landlords, and postal workers to supply information about households that failed to respond to the census. The Court said it was not deciding the “precise methodological limits” of the Census Clause. However, the use of imputation in this case was part of an effort to reach every household, represented methods of “inference” rather than statistical sampling, involved a “tiny percent” of the total population, was not likely to be subject to manipulation, and was an alternative to “a far less accurate assessment of the population,” and therefore did not exceed the Constitution’s limits, the majority concluded. In addition, the Court rejected North Carolina’s argument that Utah lacked “legal standing” to challenge imputation or other counting methods after the census was completed. North Carolina had asserted that statutes governing apportionment and the post-census certification of seats sent to each state do not allow revisions to be made. The Court said it found no intent to bar judicial relief if a state believes a serious error in the census led to an incorrect apportionment. 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 <terriann2k(a)aol.com>om>. For copies of previous News Alerts and other information, use our web site 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 <Census2000(a)ccmc.org> or 202/326-8700. Please feel free to circulate this information to colleagues and other interested individuals.