United States v. United States Department of Commerce, Civ. A. No. 98-0456. Three Judge Court (RCL, DHG, RMU)

Decision Date24 August 1998
Docket NumberCiv. A. No. 98-0456. Three Judge Court (RCL, DHG, RMU)
CourtU.S. District Court — District of Columbia
PartiesUNITED STATES HOUSE OF REPRESENTATIVES, Plaintiff, v. UNITED STATES DEPARTMENT OF COMMERCE, et al., Defendants, and RICHARD A. GEPHARDT, et al. ; LEGISLATURE OF THE STATE OF CALIFORNIA, et al. ; CITY OF LOS ANGELES, et al. ; NATIONAL KOREAN AMERICAN SERVICE & EDUCATION CONSORTIUM, INC., et al., Intervenor-Defendants.

Civ. A. No. 98-0456. Three Judge Court (RCL, DHG, RMU)

United States District Court, District of Columbia.

MEMORANDUM OPINION

ROYCE C. LAMBERTH, District Judge.

Plaintiff, the United States House of Representatives, seeks summary judgment against the Department of Commerce and the Bureau of the Census ("defendants") in this action challenging defendants' plan for the 2000 census. Plaintiff claims that using statistical sampling to supplement the headcount enumeration used to apportion representatives among the states violates the Census Act, 13 U.S.C. § 1 et seq., and Article I, section 2, clause 3 of the Constitution. The House seeks a declaration that statistical sampling is unlawful and/or unconstitutional and an injunction preventing defendants from using statistical sampling in the 2000 census.

Now before the court are defendants' and intervenor-defendants'1 motions to dismiss plaintiff's complaint pursuant to Rules 12(b)(1) and (6) of the Federal Rules of Civil Procedure, and plaintiff's motion for summary judgment pursuant to Federal Rule of Civil Procedure 56. For the reasons stated below, the motions to dismiss will be denied and plaintiff's motion for summary judgment will be granted.

I. BACKGROUND

The Constitution requires Congress to conduct an "actual Enumeration" of the population every ten years "in such Manner as they shall by Law direct." U.S. Const. art. I, § 2, cl. 3; see Wisconsin v. City of New York, 517 U.S. 1, 5 (1996). Congress has delegated broad authority over the conduct of the census to the Secretary of Commerce through the Census Act. See 13 U.S.C. § 1 et seq.; see also Wisconsin, 517 U.S. at 19. The Census Act governs the Census Bureau's (the "Bureau") gathering of economic social and demographic data about the United States, including the decennial apportionment census mandated by the Constitution.

Despite the constitutional mandate to obtain an "actual enumeration" of the population, "no census is recognized as having been wholly successful in achieving that goal." Wisconsin, 517 U.S. at 6 (citing Karcher v. Daggett, 462 U.S. 725, 732 (1983); Gaffney v. Cummings, 412 U.S. 735, 745 (1973)). The 1990 census was no exception. According to the Bureau, "[t]hough better designed and executed than any previous census, the Census in 1990 took a step backward on the fundamental issue of accuracy. For the first time since the Census Bureau began conducting post-census evaluations in 1940, the decennial census was less accurate than its predecessor. . . . the undercount rate of 1.8 percent in 1990 was 50 percent greater than the rate had been in 1980." United States Department of Commerce, Bureau of the Census, Report to the Congress — The Plan for Census 2000, at 2 (revised August 1997) ("Census 2000 Report"). Specifically, the Bureau reports that children, renters (particularly in rural areas), and racial and ethnic minorities were significantly undercounted. Among those the 1990 census missed were 4.4% of African-Americans, 5.0% of Hispanics, and 12.2% of American Indians living on reservations, but only 0.7% of Non-Hispanic Whites. See Census 2000 Report at 3-4.2 This undercounting of certain groups relative to others, known as the "differential undercount," raises the possibility of congressional malapportionment, as jurisdictions with large numbers of undercounted persons may have a greater share of the total population than the census figures suggest.

Concerned about problems with the 1990 census, Congress passed the Decennial Census Improvement Act of 1991, P.L. 102-135, 105 Stat. 635 (1991), which directed the National Academy of Sciences to study "the means by which the Government could achieve the most accurate population count possible," including consideration of "the appropriateness of using sampling methods in combination with basic data-collection techniques." Id. (quoted in Census 2000 Report at 6). The National Academy established three panels to develop a means to achieve greater accuracy for the 2000 census. All three panels concluded that traditional census methods needed to be modified in response to societal changes, and that statistical sampling techniques would both increase the census' accuracy and lower its cost. See Census 2000 Report at 7-8 (quoting the conclusion of the Academy Panel on Methods that "[d]ifferential undercount cannot be reduced to acceptable levels at acceptable costs without the use of integrated coverage measurement and the statistical methods associated with it").

Based upon the results of these congressionally-authorized National Academy studies, combined with ninety years of census-taking experience, meetings with the public in thirty cities, congressional input, and advice from no fewer than six advisory committees, see Census 2000 Report at 9-10, the Bureau developed its master plan for the 2000 enumeration. At issue is the Bureau's plan to use statistical sampling to supplement data obtained through traditional census methods.3

Statistical sampling is best understood as using information derived from a portion of a population to infer information on the population as a whole. The Bureau intends to use sampling in three different phases of the 2000 census. First, the Bureau will use sampling in the Postal Vacancy Check program to verify the United States Postal Service's determination that certain housing units are vacant and to correct for anticipated errors in this designation. Second, the Bureau will use sampling techniques in the Nonresponse Follow-Up ("NRFU") phase of the census. Finally, the Bureau intends to conduct a post-census survey, an operation referred to as Integrated Coverage Measurement ("ICM"). The Postal Vacancy Check sampling plan is not at issue in this litigation. The court will describe the latter two processes briefly.

1. Nonresponse Follow-Up: If all households returned their census forms by mail, NRFU would be unnecessary. However, in 1990 only 65% of households returned their forms, down from 78% in 1970. The Bureau does not expect a substantially higher rate of return in 2000, estimating that even with its innovations, the response rate will be around 67%, or approximately 34 million non-responding households. See Census 2000 Report at 26.

During the 1990 census, Bureau enumerators personally visited all non-responding housing units, with some homes receiving as many as six repeat visits before the Bureau ultimately relied upon proxy data (from neighbors) or imputation data (computer-inferred) to determine the number of persons residing in each non-responding household. The Bureau considers the nonresponse follow-up to be the "most difficult logistical segment" of the census. Census 2000 Report at 27.

Under the planned NRFU program, enumerators will not endeavor to personally contact all non-responsive households. Rather, they will visit a randomly selected sample of nonresponding housing units. The sample is random to ensure that the units chosen "will be statistically representative of all housing units in a non-responding tract." Census 2000 Report at 27-28 (defining a "tract" as having homogeneous population characteristics, such as economic status and living conditions). The percentage of housing units visited will vary with the mail response rate to ensure that enumerators directly contact 90% of the units in each tract. For example, in census tracts in which only 30% of households respond, 6 of 7 addresses will be visited by enumerators, but in tracts with an 80% return rate, only 1 of 2 will be contacted. As to the households in each tract not personally visited, the Bureau will estimate the number of persons residing within those units based upon data collected from the households that received a follow-up visit. The Bureau states that with this sampling technique, enumerators will only have to visit 22.5 million housing units, as opposed to the 34 million they would have to visit without relying upon sampling. See Census 2000 Report at 27.

2. Integrated Coverage Measurement: The second phase of the 2000 census challenged by the House is the ICM, a post-census survey which utilizes "Dual System Estimation" ("DSE") or "capture/recapture" to compensate for any undercount or differential undercount after the initial enumeration is complete. See Census 2000 Report at 29-32; see also Wisconsin, 517 U.S. at 8-9 (explaining how DSE would operate in counting the number of pumpkins in a large pumpkin patch). To conduct the ICM, the Bureau will classify each of the country's 7 million city blocks into categories the Bureau refers to as "strata." These strata will be based on characteristics of the block determined in the 1990 census such as the block's state, racial and ethnic composition, and proportion of renters to homeowners.4 The Bureau will then select blocks at random from each stratum for a total of 25,000 blocks. Based on an average of 30 housing units per block, it will obtain information from approximately 750,000 housing units. That number will establish a representative sample large enough, the Bureau claims, to estimate population totals for each state.

ICM interviewers will interview the residents of the 750,000 housing units in the sample blocks, thereby establishing a roster of Census Day residents independent of the initial enumeration roster. If data collected for a household during the ICM interview differs from data obtained during the original headcount phase, a follow-up ICM interviewer will return to the address to rectify the discrepancy and find the "true" situation. Each person then will be assigned to a unique "poststratum," which is a group...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT