View PDF Version

No. 98-564

In the Supreme Court of the United States






Solicitor General
Counsel of Record
Department of Justice
Washington, D.C. 20530-0001
(202) 514-2217

A. The Census Bureau's Plan Is Based On Its Determination, Supported By Abundant Scientific Evidence, That Use Of Statistical Sampling Will Improve The Accuracy Of The Relevant Population Figures
Two forms of statistical sampling are at issue in this case. See Gov't Br. 6-7. First, the Census Bureau intends to employ sampling during the Nonresponse Follow-Up (NRFU) phase of the census by sending enumerators to a substantial and representative sample of households that fail to return mail questionnaires. See 98-404 J.A. 88-92. Second, the Bureau intends to conduct an intensive survey of approximately 750,000 housing units after the NRFU phase of the census is complete. Id. at 92-93. Through a process known as Integrated Coverage Measurement (ICM), the Bureau will compare the results of that survey to those of the initial phase of the census in order to determine population figures for States and political subdivisions nationwide. Id. at 94-98. Each of those forms of sampling is integral to the Bureau's efforts to increase the accuracy of the 2000 census, and each has broad support within the scientific community.1
1. As we explain in our opening brief in No. 98-404 (at 5), Congress in 1991 directed the Secretary of Commerce to contract with the National Academy of Sciences (NAS) to study possible means of improving the accuracy of the census. See Decennial Census Improvement Act of 1991, Pub. L. No. 102-135, 105 Stat. 635 (13 U.S.C. 141 note). The NAS was instructed to consider, inter alia, "the appropriateness of using sampling methods." § 2(b)(1)(C), 105 Stat. 635. The NAS subsequently established three panels, each of which "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." 98-404 J.S. App. 4a.
The NAS Panel on Census Requirements in the Year 2000 and Beyond stated that
physical enumeration or pure 'counting' has been pushed well beyond the point at which it adds to the overall accuracy of the enumeration. Moreover, such traditional census methods still result in a substantial undercount of minority populations. Techniques of statistical estimation can be used, in combination with the mail questionnaire and a reduced scale of follow-up of nonrespondents, to produce a better census at reduced costs.
Modernizing the U.S. Census at 3-4 (B. Edmonston & C. Schultze eds., 1995).2 The Panel to Evaluate Alternative Census Methods "conclude[d] that sampling and associated statistical estimation constitute an established scientific methodology that must play a greater role in future censuses in order to obtain a more accurate picture of the population than is provided by current methods." Counting People in the Information Age at 4 (D. Steffey & N. Bradburn eds., 1994). The panel "endorse[d] the Census Bureau's stated goal of achieving a one-number census in 2000 that incorporates the results from coverage measurement programs, including programs that involve sampling and statistical estimation, into the official census population totals." Ibid.
A third NAS panel (the Panel to Evaluate Alternative Census Methodologies) has issued two interim reports, with its final report expected by the end of 1998. The first report stated that "the use of sampling and other statistical procedures will be fundamental to achieving the goals of reduced costs, increased accuracy, and reduced differential undercoverage." Sampling in the 2000 Census Interim Report I, at 2 (A. White & K. Rust eds., 1996). The second report concluded that "a census without sampling will almost certainly be unacceptable in terms of both quality and cost." Preparing for the 2000 Census Interim Report II, at 6 (White & Rust eds., 1997).3
2. Two of the three NAS panels unequivocally endorsed both of the uses of sampling at issue in this case-i.e., sampling for non-response follow-up (NRFU) and integrated coverage measurement (ICM). See Modernizing the U.S. Census, at 87-96; Preparing for the 2000 Census Interim Report II, at 6.4 Appellees contend that "[i]t is undisputed that sampling for nonresponse follow-up decreases accuracy." Glavin Br. 3; see also id. at 41. That statement is entirely untrue. The Census Bureau believes that NRFU sampling will increase the accuracy of the census as a whole, and its view is supported by abundant scientific evidence.5
NRFU is highly labor-intensive, requiring the Bureau to hire a very large number of temporary enumerators. See, e.g., 98-404 J.A. 100 (Census Bureau's Report to Congress states that "[a]t peak employment, about 117,000 workers * * * would be hired to conduct the nonresponse follow-up" if sampling is utilized, and that "59,000 additional enumerators" will be needed if the Bureau does not employ NRFU sampling).6 Because the use of sampling reduces the number of enumerators required, NRFU sampling "gives reasonable promise that accuracy can be improved and differences in the quality of enumeration can be diminished by concentrating the efforts of better enumerators on the sampled cases." Sampling in the 2000 Census Interim Report I, at 6; see also Preparing for the 2000 Census Interim Report II, at 7.7 Use of sampling also ensures that the NRFU phase can be completed within a much shorter period of time than was required during the 1990 census. By decreasing the time required for the NRFU phase, sampling has the potential to increase the accuracy of the NRFU data themselves, to improve the quality of the subsequent ICM, and to ensure that the census is completed in time to permit the transmittal of state-level population figures to the President by January 1, 2001, as required by 13 U.S.C. 141(b).8
Appellees place substantial reliance (see Glavin Br. 3, 41) on a report issued in 1997 by the General Accounting Office (GAO). The passage on which appellees rely states:
Technically, the most accurate design alternative, according to the results of the Bureau's research, would be to attempt 100-percent follow-up of nonrespondents and use ICM to address accuracy problems. That design could produce slightly improved accuracy in census data, particularly for smaller geographic areas, but would come at a greater cost (approximately $400 million more than that for the Bureau's refined plan). Furthermore, such an option may not be feasible given projected staffing difficulties and, especially, the risk that the Bureau could not complete 100-percent follow-up and ICM by the December 31 deadline for reporting census results for congressional reapportionment.
2000 Census Progress Made on Design, but Risks Remain, GAO/GGD-97-142, at 26-27 (July 1997) (1997 GAO Report). Appellees quote the first sentence of that passage (see Glavin Br. 3, 41), ignoring (1) GAO's apparent agreement that ICM will improve the accuracy of the relevant population figures and (2) GAO's own doubt that full non-response follow-up could feasibly be conducted (at least in conjunction with ICM) given anticipated staffing difficulties and the existence of an inflexible statutory deadline (see 13 U.S.C. 141(b)) for transmittal to the President of complete state-level population figures. The clear thrust of the paragraph as a whole is that while NRFU without sampling might in theory lead to modest gains in accuracy for smaller geographic areas, any effort to implement both full NRFU and ICM would face substantial practical difficulties.9
Appellees contend (Br. 3) that the Bureau plans to utilize NRFU sampling "solely in order save time and money, not because it even arguably increases accuracy." That statement is incorrect and postulates a false dichotomy. The Bureau believes that, if the NRFU phase of the census is viewed in isolation, sampling will neither substantially improve nor substantially decrease the accuracy of the counts. Because NRFU sampling will leave more time, funds, and trained personnel available for the ICM, however, and will permit the ICM to begin nearer in time to the census date, it is expected to improve the overall accuracy of the census.10 Insofar as the correctness of the Bureau's judgment on that point is relevant to the issues presented by this case, that judgment warrants deference from this Court. Compare Wisconsin v. City of New York, 517 U.S. 23 (1995).11
B. Appellees Lack Standing To Bring This Suit
1. Appellees do not defend the district court's assertion (see J.S. App. 9a) that the States of Connecticut, Massachusetts, Minnesota, Missouri, Pennsylvania, and Wisconsin are substantially likely to lose a seat in the House of Representatives if sampling is used in the 2000 census. In a footnote, however, they continue to assert that the State of Indiana (in which appellee Hofmeister resides) "is virtually certain to lose a congressional seat solely as a consequence of the Department's methodology." Glavin Br. 14 n.11. The district court made no such finding. Moreover, as we explain in our opening brief (at 22-24), the Commerce Department introduced affidavits in the district court directly controverting the statistical analysis on which that contention is based. The Department's submission demonstrated in particular that appellees' affiant (Dr. Steven Weber) failed to consider the best available data, since he considered data from 1994 and ignored population estimates from 1997 that cast doubt on the accuracy of the Bureau's projections for the year 2000. See Gov't Br. 22-23 n.11.
Appellees defend their affiant's analysis on the ground that "Dr. Weber's methodology did in fact utilize the most recent projections available at the time of his analysis." Glavin Br. 14 n.11. That argument is specious. The situation might roughly be compared to a transaction in which an investor purchases shares in a mutual fund from a broker who projects that the investment will be worth $150,000 at the end of 1998, and $200,000 at the end of the year 2000. If the shares are worth only $120,000 at the end of 1998, no reasonable investor would continue to rely on the $200,000 projection. Dr. Weber's analysis in effect presumes that so long as no new official population projection for the year 2000 has been issued, other newly-available evidence bearing on the likely accuracy of the earlier projection can safely be ignored. Given the obvious irrationality of that assumption, Dr. Weber's statement that "it is a virtual certainty that Indiana will lose a seat" if sampling is employed, J.A. 65, has no meaningful probative value.12
2. Appellees contend (Br. 10-13) that the appellee Georgia residents will be injured by the Bureau's plan for the 2000 census because (1) the State of Georgia will be entitled to an additional Representative (as compared to its current allotment) in the year 2002 regardless of how the census is conducted, and (2) the Bureau's failure to produce state-level population figures derived without the use of sampling will deprive the State of that Representative. Appellees' claim of "unconstitutional inaction" (Br. 11) as a basis for standing is both farfetched as a predictive matter and wrong as a legal matter. The Bureau's plan contemplates that a decennial census will be conducted; that the Secretary will transmit state-level population figures to the President, see 13 U.S.C. 141(b); and that the President will in turn transmit state-level population figures to Congress, see 2 U.S.C. 2a(a). Under the existing reapportionment statute, "[e]ach State shall be entitled * * * to the number of Representatives shown in the statement" transmitted by the President. 2 U.S.C. 2a(b) (Supp. II 1996). Thus, if Georgia will be entitled to an additional Representative in the 2001 apportionment under any plausible census methodology, the State will receive that Representative by operation of law after the 2000 census is completed.
Georgia will lose that seat only if a federal court subsequently decides that (a) the use of sampling in the decennial census violated the Census Act or the Constitution and (b) the appropriate remedy for that violation is to mandate continued reliance on the 1990 census figures. Although appellees assert that "[t]he only remedial alternative" in that hypothetical suit "would be using the 1990 apportionment," they offer no legal argument in support of that proposition, which is contrary to basic equitable principles. See Gov't Br. 34.13 Appellees have therefore failed to show that implementation of the Bureau's plan is likely to deprive Georgia of a Representative to which it would otherwise be entitled. Appellees also offer no authority for the proposition that Article III's traceability requirement can be satisfied in a particular case on the ground that the plaintiff will likely be injured by the relief awarded in some hypothetical other suit in the future. See id. at 33-34.
3. The gravamen of appellees' claims on the merits is that the apportionment of Representatives among the States is a matter of special significance, and that the derivation of population figures to be used for that purpose is therefore subject to distinct constitutional and statutory rules. For the reasons stated above, however, appellees have failed to show a significant likelihood that any of the States in which they reside will be allotted fewer Representatives as a result of the Bureau's use of sampling than they would receive if sampling were not utilized. Indeed, appellees have all but abandoned any effort to predicate standing on such a claim of injury. Appellees' principal argument with regard to standing (Br. 14-26) is that they will suffer harms unrelated to the apportionment of Representatives among the States-i.e., a loss of intrastate voting power and federal funding-as a result of the Bureau's plan.
As we explain in our opening brief (at 18-21, 24-26, 27-29), appellees have failed to demonstrate either that the States or localities in which they reside will lose population share as a result of the Bureau's use of sampling, or that any such decrease in population share will translate into a diminution of intrastate electoral power or loss of federal funds. Even if appellees could make that showing, however, they could not establish standing to sue. The injuries they allege would not be traceable to the purported illegality in the Bureau's plan, and they cannot demonstrate that a favorable judicial ruling would likely redress those harms.
a. With respect to the issue of traceability, our argument is not (as appellees suggest) that state or federal officials might decline to use "census numbers" (Glavin Br. 23) for intrastate redistricting or for the allocation of federal funds. Rather, our point is that the Census Act explicitly contemplates that census numbers calculated for those purposes should be derived through the use of sampling. Under 13 U.S.C. 195, the Secretary is required to use sampling "if he considers it feasible" for all purposes other than the apportionment of Representatives among the States. As a method for determining the population figures that will be used for intrastate redistricting and federal fund allocation, the Bureau's plan for the 2000 census is indisputably lawful. Even under appellees' interpretation of the Census Act and the Constitution, a violation of law will occur when and only when state-level population figures derived through sampling are used in the apportionment of Representatives among the States. Appellees' alleged intrastate electoral and funding "injuries"-which in fact would be entirely lawful consequences of the Secretary's actions-are not fairly traceable to such a violation.
b. As we explain in our opening brief (at 32), appellees' alleged intrastate electoral and federal funding injuries also would not be redressable by a favorable judicial decision, since the Census Bureau would remain free to utilize sampling for those purposes even if sampling for apportionment of Representatives among the States were held to be unlawful. Appellees argue (Br. 26 n.26) that if "the Court informs the Secretary that he must count 100% of the population," the Secretary could not rationally decide to employ statistical sampling techniques to supplement that count. But that is simply wrong. Whatever the outcome of this case, the Court will not order the Secretary to "count 100% of the population"-a directive that is universally conceded to be incapable of implementation. Rather, acceptance of appellees' legal theory would require the Secretary to prepare state-level population figures derived without the use of sampling, which figures would be used in the apportionment of Representatives among the States. Nothing in law or in logic would prevent the Secretary from concluding that the use of sampling techniques as a supplement to the methodology used for apportionment would produce more accurate numbers, and that those alternative figures should be calculated for all purposes permitted by law.14
C. The Principal Constraint On The Census Bureau's Discretion Is The Requirement That Its Decisions Be Reasonably Related To The Constitutional Goal Of Equal Representation For Equal Numbers Of People
Under 13 U.S.C. 141(a), the Secretary is directed to "take a decennial census of population * * * in such form and content as he may determine, including the use of sampling procedures and special surveys." Section 141(a) does not by its terms require that any specified percentage of the population must be contacted directly; nor does it specifically state that sampling may be employed only if it improves the accuracy of the final counts. Appellees contend (see Br. 34-35) that if the Census Act is construed to permit the use of sampling for purposes of congressional apportionment, the Census Bureau will be authorized to derive population figures for apportionment through sampling methodologies that demonstrably reduce the accuracy of the final population figures. That objection is misconceived.
The constitutional purpose of the census is to effectuate the requirement that Representatives be apportioned among the States "according to their respective Numbers," U.S. Const. Art. I, § 2, Cl. 3-a requirement that reflects "our Constitution's plain objective of making equal representation for equal numbers of people the fundamental goal for the House of Representatives," Wesberry v. Sanders, 376 U.S. at 18. Although the Constitution does not require "a census that is as accurate as possible," City of New York, 517 U.S. at 17, use of a census methodology that is so inaccurate as to bear no reasonable relationship to the goal of equal representation for equal numbers of people would be unlawful, see id. at 19-20. The standard applied in City of New York leaves the Secretary with broad discretion, but that breadth of authority is scarcely accidental. It results from the facts that (a) "[t]he text of the Constitution vests Congress with virtually unlimited discretion in conducting the decennial 'actual Enumeration,'" and (b) "[t]hrough the Census Act, Congress has delegated its broad authority over the census to the Secretary." Id. at 19 (citing 13 U.S.C. 141(a)). Thus, while there is no bright line between permissible and impermissible uses of sampling, the absence of such a bright line is the natural and inevitable consequence of Congress's decision to entrust the conduct of the decennial census to the sound judgment of the expert agency.
It should also be emphasized that potential line-drawing problems would not be eliminated by a prohibition on the use of sampling. There are (at least in theory) "headcount" methodologies that would be impermissible under the standard articulated in City of New York. The Census Bureau could not, for example, lawfully conduct the decennial census by sending a single enumerator to each State. Nor could the Bureau permissibly decide that non-response follow-up would be conducted in some States (e.g., States with Governors of the President's political party) and not in others. Neither of those methodologies is prohibited by any specific provision of the Census Act, but both would be palpably inconsistent with the goal of equal representation for equal numbers of people.
The fact that the Census Act contains no bright-line standards distinguishing between permissible and impermissible "headcounts" does not mean that the Bureau's authority in that regard is subject to no constraints whatever. It is obviously not an effective rejoinder to appellees' legal theory to say that their interpretation of the Census Act must be wrong because it would permit the Bureau to employ the techniques described above. By the same token, the possibility of hypothetical cases involving unreasonable sampling methodologies provides no ground for construing the Act to prohibit all use of sampling in connection with the apportionment of Representatives among the States.15
D. 13 U.S.C. 195 Does Not Prohibit Sampling In Determining State-Level Population Figures For Apportioning Representatives Among The States
As we explain in our opening brief in No. 98-404 (at 28-31), the opening proviso of 13 U.S.C. 195 exempts the apportionment of Representatives among the States from the generally applicable directive that the Secretary must use sampling if it is "feasible," but it does not impose any prohibition on the use of sampling in that context. Section 195 therefore does not withdraw the authority expressly granted by Section 141(a) to use sampling in conjunction with the decennial census. Appellees argue that (1) an exception to a mandatory directive often constitutes a prohibition and (2) Section 195 does not contain a mandatory directive in any event. Those contentions are without merit.
1. Appellees state that "[t]here are numerous examples in the United States Code of statutes in which an exception from a mandate is a prohibition. See, e.g., 22 U.S.C. § 5605(a)(1); 16 U.S.C. § 45(f); 7 U.S.C. § 79(j)(1); 22 U.S.C. § 2452(a)(1)." Glavin Br. 35 n.35. None of the cited provisions, however, can plausibly be read to divest Executive Branch officials of discretionary authority they would otherwise possess. The exception contained in 22 U.S.C. 5605(a)(1) cannot reasonably be construed as a prohibition.16 16 U.S.C. 45f(c)(2)-the statutory provision to which appellees' citation to "16 U.S.C. § 45(f)" apparently refers-merely reflects the obvious fact that one does not pay market value for land that has been donated.17 7 U.S.C. 79(j)(1) merely reflects a self-evident proposition in the opposite situation-that the Secretary of Agriculture cannot charge a user fee for services he does not perform-and it does not in any event divest the Secretary of discretion he would otherwise possess.18 Finally, 22 U.S.C. 2452a(b) contains an exception to a discretionary authorization, not an exception to a mandate.19
The approach we advocate is based not on any artificial interpretive convention, but on simple common sense. With respect to any hypothetical Executive Branch action, Congress may choose to require the action, to prohibit the action, or to leave the decision to the discretion of Executive Branch officials. Absent very unusual circumstances, Congress's decision not to choose the first alternative does not, in and of itself, logically suggest any preference as between the other two. Where (as here) the pertinent Executive Branch conduct is specifically authorized by another provision of law (see 13 U.S.C. 141(a)), Congress's decision not to require that conduct cannot reasonably be construed to override the statutory authorization.
2. As we observe in our opening brief in No. 98-404 (at 28-29 n.14), Section 195's treatment of sampling for purposes other than congressional apportionment is neither wholly mandatory nor wholly non-directive. Contrary to appellees' contention (Br. 36-40), however, construction of Section 195's opening proviso does not turn on whether the rest of that Section is regarded as something approaching an absolute command or as a more conditional directive to use sampling. For even if Section 195 were regarded as purely hortatory-as nothing more than the expression of a non-binding congressional preference for sampling in doubtful cases-there is no reason to construe the opening proviso as prohibiting the use of sampling for apportionment of Representatives among the States.
With respect to functions other than the apportionment of Representatives, Congress sought to encourage the use of sampling in order to reduce the cost and burden of census activities, even in circumstances in which the Secretary does not believe that sampling would improve the accuracy of the count. See S. Rep. No. 1256, 94th Cong., 2d Sess. 5 (1976) (stating, with respect to the mid-decade census, that "the use of sampling procedures and surveys is urged for the sake of economy and reducing respondent burden"); see also id. at 9, 12, 13; 98-404 Comm. Dep't Br. 30. Given the particular constitutional significance of the apportionment of Representatives, however, Congress could reasonably determine that such a "thumb on the scale" was inappropriate, and that the Secretary should be left free to exercise his own best judgment regarding the appropriate methods for obtaining the most accurate population counts practicable. The fact that Congress declined to codify a preference for sampling in determining the population figures used for apportionment of Representatives does not logically imply that Congress intended to prohibit the use of sampling in that context.
* * * * *
For the reasons stated above and in our opening brief, the judgment of the district court should be vacated and the case remanded with instructions to dismiss for lack of jurisdiction. In the alternative, the judgment of the district court should be reversed.

Respectfully submitted.
Solicitor General

1 In the district court, appellees specifically disavowed any effort to establish that the use of statistical sampling would decrease the accuracy of the relevant population figures. Their memorandum of law in opposition to the government's motion to dismiss stated (at 29 n.11) that "plaintiffs do not seek to involve [the district court] in a battle of statistical experts given that their complaint alleges that the Department's plan violates both federal law and the Constitution irrespective of its purported accuracy." For purposes of this case, the Court must therefore assume the correctness of the Census Bureau's determination that its plan for the 2000 census will produce more accurate population figures than would a census conducted without the use of sampling. As we explain below, that determination is supported by abundant scientific evidence.
2 Dr. Charles L. Schultze was the chairman of the Panel on Census Requirements in the Year 2000 and Beyond. He has since testified on two occasions to reiterate the view that the use of sampling mechanisms can be expected to make the 2000 census both more accurate and less costly. In his most recent testimony he stated: "[T]he bottom line is clear. The careful use of modern statistical techniques to help complete the Census will yield more accurate data for purposes of Congressional apportionment and the other major uses to which the census data is put, and at the same time reduce budgetary cost." Census 2000: Hearing Before the Senate Comm. on Governmental Affairs, 105th Cong., 1st Sess. 54 (1997) (1997 Hearing). Dr. Schultze observed "as an economist" that "it is not often that you run across areas in which you can get both better quality and lower costs, so it is a little bit unique, in a way." Ibid.
3 That report acknowledged that the use of sampling might decrease the accuracy of population counts for very small geographic areas (e.g., census blocks), but stated that error rates for small geographic areas are "not an appropriate criterion for judging the quality of the census." Preparing for the 2000 Census Interim Report II, at 11-12. The report concluded that for geographic levels "hav[ing] important legal, political, or financial implications," the use of sampling "can achieve results that are at least as good as those from a more time-consuming and expensive effort to obtain a completed form for every household." Id. at 12.
The General Accounting Office (GAO) has similarly concluded that "the Bureau's plan [for the 2000 census], if effectively implemented, has the potential for producing a more accurate and less costly census than if only conventional census procedures were used." 2000 Census Progress Made on Design, but Risks Remain, GAO/GGD-97-142, at 22 (July 1997). That report noted the GAO's prior conclusion, reflected in testimony given on October 25, 1995, "that the established approach used for taking the census in 1990 had exhausted its potential for counting the population cost-effectively and that fundamental design changes were needed to reduce census costs and improve the quality of the data collected." Id. at 1.
4 With respect to NRFU sampling, the analysis of the third NAS panel (the Panel to Evaluate Alternative Census Methods) was equivocal. See Counting People in the Information Age at 97-105. Appellees refer to the view of that panel (Glavin Br. 42 n.39) but omit all reference to the two NAS panels that recommended in favor of NRFU sampling.
5 Although attacks on the Bureau's plan to utilize NRFU sampling are a recurring feature of appellees' brief, appellees have not alleged or attempted to prove that use of NRFU sampling will injure them in any way. Appellees do not contend that NRFU sampling will cause the States or localities in which they live to be credited with a smaller percentage of the population than those areas would receive under full NRFU. Rather, appellees' claim of injury is addressed exclusively to the use of sampling in the ICM process and is grounded in the purported similarity between the ICM mechanism devised for the 2000 census and the adjustment methodology considered (and ultimately rejected) for the 1990 census.
6 In October 1995, then-Census Bureau Director Riche testified that
sampling may not be just an attractive cost-saving option; it may be the only option we now have for completing the census. And that's because, historically, we have recruited those large numbers of census takers amongst people who weren't in the labor force, who weren't seeking permanent work. Many of them had previous work experience and skills, so we only had to train them for the technical part of the task at hand. Today, the pool of workers qualified, experienced, and able to work on a decennial census has decreased dramatically.
Oversight of the Census Bureau: Preparations for the 2000 Census: Hearing Before the Subcomm. on National Security, International Affairs, and Criminal Justice of the House Comm. on Government Reform and Oversight, 104th Cong., 1st Sess. 76-77 (1995).
Dr. Schultze has more recently explained that the problems associated with the 1990 census "were exacerbated by the growing difficulty of finding, at reasonable costs, sufficient numbers of qualified Census enumerators. Turnover rose and productivity fell." 1997 Hearing at 131. Dr. Schultze further observed that "[t]he recent 1995 census tests suggest that the difficulty and costs of attracting qualified census enumerators have accelerated." Ibid.
7 Dr. Schultze explained: "[B]y applying adequate resources to a smaller number of sample units in each census tract, a high quality follow up can be achieved for those units and the results extrapolated to the rest of the tract. Even though sampling error is thereby introduced in the extrapolation, the higher quality of the results within the sampling units should offset or more than offset the sampling errors." 1997 Hearing at 132.
8 As one NAS panel explained,
the use of sampling will reduce the field workload and may result in more timely completion of the nonresponse follow-up procedures in the field. This increased timeliness will increase data quality because respondents will typically be giving information to enumerators closer to census day than they would if nonresponse follow-up without sampling were implemented. There will be fewer recall errors and less use of poor-quality "last-resort" information obtained from indirect sources in the final stages of field data collection. In addition, the more timely completion of nonresponse follow-up will permit the coverage measurement survey to be implemented in the field closer to census day, thereby decreasing recall errors and the effect of people who move on the process of coverage measurement and correction.
Preparing for the 2000 Census Interim Report II, at 6-7.
9 The NAS Panel to Evaluate Alternative Census Methodologies has observed that "[a] modest nonresponse follow-up design (i.e., one with a relatively small workload) that is executed as planned is far superior to a more ambitious design that runs short of time or resources." Preparing for the 2000 Census Interim Report II, at 41. Given the existence of substantial doubt regarding the Bureau's ability to complete both NRFU without sampling and ICM in time to meet the statutory deadline for transmittal of state-level population figures, it was surely reasonable for the Bureau to adopt a census plan that leaves some margin of safety, even if an alternative plan might produce slightly more accurate results for smaller geographical areas under a "best case" scenario.
10 The Census Bureau has a limited time to conduct the decennial census, since federal law requires that the population be determined as of April 1 of the census year and that complete state-level population counts be transmitted to the President within nine months after that date. See 13 U.S.C. 141(a) and (b). The Bureau is also restricted to the funds appropriated by Congress. See U.S. Const. Art. I, § 9, Cl. 7 ("No Money shall be drawn from the Treasury, but in Consequence of Appropriations made by Law."). Increases in the time and money devoted to the NRFU phase of the census necessarily entail decreases in the time and money available for the ICM phase. The Bureau's recognition of that elementary fact is not remotely comparable to a decision not to count "any people over 70 years old." Glavin Br. 43 n.40.
The 1997 GAO Report explained that "[t]he effects of the proposed sampling for nonresponse follow-up and of ICM procedures on the success of census operations and the quality of the resulting data need to be viewed in combination to be meaningful." 1997 GAO Report at 21. The Report stated that the decision to use NRFU sampling "is unlikely to significantly change (either improve or decrease) census accuracy," but emphasized that "ICM is unlikely to be successful unless preceding data collection efforts, in particular nonresponse follow-up, are completed on schedule." Ibid.; see also id. at 52 (noting that timely completion of NRFU is "crucial to the success of [ICM] * * * because the Bureau must provide the final census data for reapportionment and redistricting purposes by legislatively mandated deadlines").
11 Contrary to appellees' suggestion (see Br. 3, 41), the 1997 GAO Report is strongly supportive of the Bureau's overall approach. Based on the results of the Bureau's research, the Report stated that
the new statistical methods the Bureau proposes to use in the 2000 Census would likely produce results that appear more accurate or more equitable according to at least three broad criteria: (1) better average levels of error, (2) error distributions compressed closer to the average levels, and (3) an apparently better cumulative error distribution. For example, state population totals, which are the basis for apportioning seats in the House of Representatives, not only show lower error rates on average but also show much less variation in the error rates among states when compared to the undercount rates in the 1990 Census.
1997 GAO Report at 26. The Report observed as well that "Bureau evaluations of the 1990 Census indicated that simply adding more conventional counting operations was not effective in eliminating or reducing differential undercounts of areas and population groups that have been hard to enumerate accurately." Id. at 57-58; see also notes 3 and 10, supra.
12 Appellees suggest (Br. 5-6) that if their own suit is found not to be justiciable, no party will ever have standing to challenge the Bureau's decision to employ statistical sampling, even after the census is completed. Of course, "[t]he assumption that if [appellees] have no standing to sue, no one would have standing, is not a reason to find standing." Valley Forge Christian College v. Americans United for Separation of Church & State, Inc., 454 U.S. 464, 489 (1982) (quoting Schlesinger v. Reservists Committee to Stop the War, 418 U.S. 208, 227 (1974)). But in any event, a potential plaintiff would have a substantially greater likelihood of demonstrating that the use of sampling had shifted a seat in the House of Representatives if the suit was filed after the completion of the census. The plaintiff might, for example, compare the apportionment resulting from the official census figures with the apportionment that would have resulted if the Bureau had not undertaken ICM. See note 13, infra.
13 The court in such a hypothetical future suit would instead be guided by the principle that relief should be tailored to place the parties in the position in which they would have been in the absence of a violation-here, if sampling had not been utilized in conducting the 2000 census. The data derived from the initial contact with every household, as well as the total of those figures and the additional data compiled from the in-person NRFU, will be available from the Census Bureau for that purpose. Moreover, since appellees have not alleged that NRFU sampling will injure them, the court in the hypothetical future suit could fashion a remedy by determining what apportionment of Representatives would have occurred if ICM sampling had not been used. Those data, unadjusted by the ICM process, will be available from the Census Bureau.
14 If this Court holds that population figures derived through sampling may not be used in apportioning Representatives among the States, the Census Bureau will presumably conduct full non-response follow-up of all households that do not return mail questionnaires. Population figures derived through mail questionnaires and NRFU would then be used for interstate apportionment of Representatives. Appellees offer no reason to suppose, however, that the Bureau would forgo the use of ICM as a means of producing more accurate population counts to be used for other purposes. The Bureau's Report to Congress characterizes ICM as "the most critical" undertaking "[o]f all the innovations to improve accuracy in Census 2000." J.A. 92. As we explain above (see pp. 5-6, supra), the Bureau might well be unable to complete full NRFU and ICM in time to satisfy the January 1, 2001, deadline established by 13 U.S.C. 141(b). That deadline, however, applies only to "[t]he tabulation of total population by States * * * as required for the apportionment of Representatives in Congress among the several States." Ibid. The temporal constraint imposed by Section 141(b) would not prevent the Bureau from employing full NRFU and ICM to produce the population figures used for purposes other than apportionment.
Appellees suggest (Br. 26) that a judicial decision leading to such a "two-number census" would at least partially redress their injury, since the Census Bureau would be prohibited from making the same uses of sampling that it currently contemplates. But appellees' alleged injury is based solely on the Bureau's intent to use ICM, not on the anticipated effects of NRFU sampling. See note 5, supra. There is consequently no reason at all to suppose that a judicial order resulting in a two-number census would leave appellees in a better position than they would occupy if the Bureau's current plan were implemented.
15 For essentially the same reasons, there is no merit to appellees' constitutional argument (Glavin Br. 47-48) that all forms of conjectural estimation will be permissible if this Court accepts the Commerce Department's interpretation of the phrase "actual Enumeration." Nor is there any basis for their suggestion (id. at 47 n.47) that the writings of Joseph Story support appellees' argument that apportionment of Representatives must be based on a "headcount." The first passage, from which appellees offer a partial quotation, states: "The third and remaining principle was to apportion the representatives among the States according to their relative numbers. This had the recommendation of great simplicity and uniformity in its operation, of being generally acceptable to the people, and of being less liable to fraud and evasion than any other which could be devised." 1 Joseph Story, Commentaries on the Constitution of the United States § 634, at 450 (1873 ed.) (emphasis added). The italicized language quoted by appellees refers to the requirement that the apportionment of Representatives be based on "numbers"-not (as appellees suggest, see Br. 47 n.47) to any particular "method" of ascertaining those numbers. The second passage on which appellees rely comes at the conclusion of Story's discussion of the difficulties of devising an appropriate mathematical formula for apportioning Representatives among the States after the census has been completed and final population figures determined. Id. § 678, at 475-476; compare United States Dep't of Commerce v. Montana, 503 U.S. 442, 448-456 (1992). Story asserted that the "rule"-i.e., the apportionment formula-"ought to be such that it shall always work the same way in regard to all the States, and be as little open to cavil or controversy or abuse as possible." 1 Story, supra, § 678, at 476. That passage likewise has nothing to do with the method by which the relevant population figures are ascertained.
16 Section 5605(a)(1) addresses the situation where the President has determined, pursuant to 22 U.S.C. 5604(a)(1), that a particular foreign government has used chemical or biological weapons. Section 5605(a)(1) provides that "[t]he United States Government shall terminate assistance to that country under the Foreign Assistance Act of 1961 [22 U.S.C. 2151 et seq.], except for urgent humanitarian assistance and food or other agricultural commodities or products." Section 5605(a)(1) cannot plausibly be construed to prohibit the termination of food or agricultural assistance, if the appropriate Executive Branch official concludes that termination of such aid is warranted and if termination is otherwise permitted by law.
17 Title 16 does not contain a Section 45(f). Section 45f(c)(1) authorizes the Secretary of the Interior to "acquire lands and interests in lands" in Mineral King Valley "by donation, purchase * * *, exchange, or transfer." Section 45f(c)(2) states that "[e]xcept for so much of the property as is donated, the Secretary shall pay to the owner the fair market value of the property on the date of its acquisition." Any prohibition against paying federal funds for donated property comes not from the statutory "except" clause, but from the Appropriations Clause of the Constitution, Art. I, § 9, Cl. 7.
18 Section 79(j)(1) applies to grain inspections undertaken by the Secretary of Agriculture. It states that "[t]he Secretary shall, under such regulations as the Secretary may prescribe, charge and collect reasonable inspection fees to cover the estimated cost to the Secretary incident to the performance of official inspection except when the official inspection is performed by a designated official agency or by a State under a delegation of authority." The immediately succeeding provision of the statute addresses the Secretary's entitlement to fees based on inspections undertaken by a designated official agency or by a State, and provides that the Secretary may collect fees from either to cover supervisory costs incurred by the Secretary. 7 U.S.C. 79(j)(2). Section 79(j)(1) and (2) therefore function as a coherent whole; the proviso in Section 79(j)(1) does not operate to prohibit the Secretary from exercising authority granted by another provision of the statute.
19 Section 2452a(b) states: "The President is authorized to transfer to the appropriations account of the United States Information Agency such sums as the President shall determine to be necessary out of the travel accounts of the departments and agencies of the United States, except for the Department of State and the United States Information Agency, as the President shall designate." 22 U.S.C. 2452a(b) (emphasis added). The italicized proviso is an exception to the President's general discretionary authority ("The President is authorized") to transform funds from "the travel accounts of the departments and agencies of the United States." The fact that Section 2452a(b) contains the word "shall" does not make the proviso an exception to a mandate.



No. 98-564