SAMPLING FOR CENSUS 2000: A LEGAL OVERVIEW

CRS Report for Congress
Sampling for Census 2000: A Legal Overview
Updated December 30, 1998
Margaret Mikyung Lee
Legislative Attorney
American Law Division


Congressional Research Service ˜ The Library of Congress

ABSTRACT
This report will give an overview of the legal issues surrounding the use of sampling and
statistical procedures. It will first review the background of “actual enumeration” in the
constitutional authority for the census and of 13 U.S.C. § 195, which some have interpreted
as prohibiting the use of sampling to adjust the census headcount for apportionment, then
discuss the case law interpreting these provisions prior to the plans for the 2000 Census and
the current lawsuits. Next, the history of adjustment efforts by the Bureau of the Census
will be briefly described, followed by a summary of the legal arguments and the judicial
decisions, as of the date of this report, concerning the proposed sampling methods for the

2000 decennial population census. Finally, the report will summarize the legislation in theth


105 Congress concerning the use of sampling in the decennial census. This report will be
updated as circumstances warrant.



Sampling for Census 2000: A Legal Overview
Summary
Sampling and statistical adjustment of the decennial population census taken for
the purpose of apportioning the Representatives in Congress among the States, have
become increasingly controversial during the past two decades and have culminated
in two lawsuits concerning the legality and constitutionality of sampling, Department
of Commerce v. U.S. House of Representatives and Clinton v. Glavin, which were
heard by the U.S. Supreme Court in consolidated oral arguments on Nov. 30, 1998.
The Supreme Court took the case on a direct appeal from the decisions by two three-
judge district court panels that the Census Act prohibits sampling in the
apportionment census. More controversy resulted from the decision by the
Department of Commerce and the Census Bureau to continue with both sampling and
non-sampling scenario preparations in the wake of the district court decisions.
Funding for Census 2000 preparations remains an issue since current appropriations
enactments fund census activities only through June 15, 1999.
For the past two censuses, in 1980 and 1990, the Bureau of the Census has
considered adjustment but has not adjusted the census. Each time, litigation resulted
when interested parties, including state and local governments and minority advocacy
groups, sued for adjustment of the census or for the release of adjusted figures.
Ultimately, in Wisconsin v. City of New York, the United States Supreme Court
upheld the decision by the Secretary of Commerce against adjustment in the 1990
census without directly addressing the constitutionality or legality of sampling. As
the 2000 decennial census draws near, the plans of the Bureau of the Census to use
statistical techniques, including a hitherto untried sampling technique for non-
response, and the apparent Administration support for these plans, has focused
renewed attention on the issues of reliability, legality and constitutionality of these
techniques. Disagreement between the legislative and executive branches about
whether to proceed with plans to use sampling techniques resulted in a compromise,
the creation of a civil action through which any aggrieved person could challenge the
use of sampling in the census for apportionment before a three-judge panel of a
federal district court, on an expedited basis, with a direct appeal to the U.S. Supreme
Court, also on an expedited basis. Two suits have resulted, the aforementioned
Department of Commerce v. U.S. House of Representatives and Clinton v. Glavin.
The compromise also established the bipartisan Census Monitoring Board to observe
and monitor all aspects of the preparation and implementation of the 2000 census.
This report will give an overview of the legal issues surrounding the use of
sampling and statistical procedures. It will first review the background of “actual
enumeration” in the constitutional authority for the census and of 13 U.S.C. § 195,
which some have interpreted as prohibiting the use of sampling to adjust the census
headcount for apportionment, then discuss the case law interpreting these provisions
prior to the plans for the 2000 Census and the current lawsuits. Next, the history of
adjustment efforts by the Bureau of the Census will be briefly described, followed by
a summary of the legal arguments and the judicial decisions, as of the date of this
report, concerning the proposed sampling methods for the 2000 decennial population
census. Finally, the report will summarize the legislation in the 105 Congressth
concerning the use of sampling in the decennial census.



Contents
Introduction ................................................... 1
Background .................................................... 2
A. Constitutional Requirements—the Census Clause.................2
B. Federal Statutory Prohibition Against Sampling..................5
C. Case Law Prior to Dept. Of Commerce v. U.S. House of Representatives and
Clinton v. Glavin........................................7
1. Before Wisconsin v. City of New York......................7
2. Wisconsin v. City of New York...........................11
Plans Of The Bureau For Sampling In Census 2000.....................13
A. A Brief History of Adjustment Efforts by the Bureau of the Census..13
B. Constitutionality and Legality of Census Bureau Plans............15
Legislation Concerning Sampling...................................21
A. Bills in the 105th Congress.................................22
B. Ramifications...........................................27
References .................................................... 27



Sampling for Census 2000: A Legal Overview
Introduction
Sampling and statistical adjustment of the decennial population census taken for
the purpose of apportioning the Representatives in Congress among the States, have
become increasingly controversial during the past two decades and have culminated
in two lawsuits concerning the legality and constitutionality of sampling, Department
of Commerce v. U.S. House of Representatives and Clinton v. Glavin, which were12
heard by the U.S. Supreme Court in consolidated oral arguments on Nov. 30, 1998.
The Supreme Court took the case on a direct appeal from the decisions by two three-
judge panels, of the D.C. District Court and the Eastern District of Virginia,
respectively, that the Census Act prohibits sampling in the apportionment census.
Further controversy resulted from the decision by the Department of Commerce and
the Census Bureau to continue with both sampling and non-sampling scenario
preparations in the wake of the district court decisions. Funding for Census 2000
preparations remains an issue since current appropriations enactments fund census
activities only through June 15, 1999. In the meantime, the recently constituted
Census Monitoring Board has been holding hearings on possible alternatives to
sampling which could help resolve the non-response and undercount problems of the
census, such as the use of administrative records.
For the past two censuses, in 1980 and 1990, the Bureau of the Census has
considered adjustment but decided not to adjust the census. Each time, litigation
resulted when interested parties, including state and local governments and minority
advocacy groups, sued for adjustment of the census or for the release of adjusted
figures. Ultimately, in Wisconsin v. City of New York, the United States Supreme3
Court upheld the decision by the Secretary of Commerce against adjustment in the

1990 census without directly addressing the constitutionality or legality of sampling.


As the 2000 decennial census draws near, the plans of the Bureau of the Census to
use statistical techniques, including a hitherto untried sampling technique for non-
response, and the apparent Administration support for these plans, has focused
renewed attention on the issues of reliability, legality and constitutionality of these
techniques. Doubts about the use of sampling led to legislative activity to study
further, restrict, or prohibit the use of sampling for the apportionment census.
Disagreement between the legislative and executive branches about whether to


11 F. Supp. 2d 76 (D.D.C. 1998), motion for expedited consideration granted, 119 S.Ct.1


27 (U.S. Sept. 10, 1998) (No. 98-404).


19 F. Supp. 2d 543 (E.D. Va. 1998), motion for expedited consideration granted, 119 S.Ct.2


290. (U.S. Oct. 2, 1998) (No. 98-564).


___ U.S. ___, 116 S.Ct. 1091 (1996).3

proceed with plans to use sampling techniques resulted in a compromise. Section
209 of P.L. 105-119, 111 Stat. 2440 (1997), created a civil action through which any
aggrieved person could challenge the use of sampling in the census for
apportionment before a three-judge panel of a federal district court, on an expedited
basis. The panel's decision could be appealed directly to the U.S. Supreme Court,
also on an expedited basis. Two suits have been brought under section 209, the
aforementioned Department of Commerce v. U.S. House of Representatives and
Clinton v. Glavin. Section 210 of P.L. 105-119 established the bipartisan Census
Monitoring Board to observe and monitor all aspects of the preparation and
implementation of the 2000 decennial census.
This report will give an overview of the legal issues surrounding the use of
sampling and statistical procedures. It will first review the backgrounds of “actual
enumeration” in the constitutional authority for the census and of 13 U.S.C. § 195,
which some have interpreted as prohibiting the use of sampling to adjust the census
headcount for apportionment, then discuss the case law interpreting these provisions
prior to the plans for the 2000 Census and the current lawsuits. Next, the history of
adjustment efforts by the Bureau of the Census will be briefly described, followed by
a summary of the legal arguments and the judicial decisions, as of the date of this
report, concerning the proposed sampling methods for the 2000 decennial populationth
census. Finally, the report will summarize the legislation in the 105 Congress
concerning the use of sampling in the decennial census.
Background
The constitutional and statutory language relevant to sampling and statistical
techniques appears to be clear, but nevertheless have been the subject of competing
interpretations which would either permit or prohibit sampling and other statistical
techniques in the census for apportionment. Only two courts, the three-judge panels
in Department of Commerce v. U.S. House of Representatives and Clinton v. Glavin,
have ever decided the issue squarely on point, although several courts have expressed
opinions in dicta.
A. Constitutional Requirements—the Census Clause
The constitutional requirements for the census are simple. Article I, § 2, clause
3, as amended by the Fourteenth Amendment, provides that the “[r]epresentatives
shall be apportioned among the several States according to their respective numbers,
counting the whole number of persons in each state excluding Indians not taxed ....
The actual Enumeration shall be made within three Years after the first Meeting of
the Congress of the United States, and within every subsequent Term of ten Years,
in such manner as they shall by law direct.” The framers of the Constitution provided
for a simple headcount of population. The census was to provide the figures for the
periodic adjustment of the representation and apportionment of representatives
among the states. Originally, it was also to provide the figures for determining the
proportionate share of direct taxes for a state.



An examination of the debates and documents of the Constitutional Convention4
reveals that the wording changed from “census" to "enumeration.” The records of
the debates on the Constitution do not reveal why the wording was changed; the
change appears in one of the various drafts, so it appears to have been an amendment
made in the Committee on Style. Possibly the change was made because of religious
attitudes regarding the taking of a census. The traditional prejudice against the taking
of a census not approved by God, resulting from the biblical account of pestilence
inflicted on Israel because David ordered an unsanctioned numbering, may have
caused them to change the terms used. Many people may still have harbored this
traditional prejudice against the taking of censuses and may have been reluctant to
go along with the idea of a census. The United States was the first country to require5
the regular taking of a census.
Although apportionments are based on the figures from the actual enumeration
of the population, the original apportionment in the Constitution was based on figures
from the various states. There is no clear indication in the record of the debates6
about the origins of the figures used or even of the actual figures used. However,
the figures used may be those given by Charles Cotesworth Pinckney in a speech
before the South Carolina House of Representatives in January 1788 and published
by various newspapers of the time. During the period of the Articles of7
Confederation, the states had taken population counts or estimates to determine the
proportion of the contribution to the general treasury and payment of the war debt for
which each would be responsible. Although the Articles of Confederation were8
amended in 1783 to provide for a triennial census, apparently not all of the states9
complied with this plan for actual censuses. These counts or estimates may have
been the basis of the first apportionment.
So the original apportionment apparently was not based on any actual headcount
of all of the states, but it was based on the best available estimates. The framers of
the Constitution had apportioned the representatives without figures from a census;


The background and history of the drafting of the Census clause of the federal Constitution4
is drawn from the following sources: H. Alterman, Counting People 177-192 (1969); M.
Anderson, The American Census: A Social History 7-13 (1988); M. Jensen, ed., 1 The
Documentary History of the Ratification of the Constitution 148-9, 255-299 (1976); P.
Kurland & R. Lerner, eds., 2 The Founders’ Constitution 102-106, 126-8, 133-4, 139-140
(1987) (this includes excerpts of the debates and reprints of Madison, The Federalist, No.

54, 12 Feb. 1788, and No. 58, 20 Feb. 1788 and Tucker, Blackstone’s Commentaries 1: App.


189 (1803); A.H. Scott, Census, U.S.A. 11-24 (1968); C. Wright & W. Hunt, History and
Growth of the United States Census 12-13 (1900).
Alterman, supra note 4, at 24-26, 174; Anderson, supra note 4, at 10, fn. 3; Scott, supra5
note 4, at 17.
Alterman, supra note 4, at 187-8.6
Jensen, supra note 4, at 298-9.7
Alterman, supra note 4, at 167-8; Kutner, Our Extraconstitutional Census, 68 U. of Detroit8
L. Rev. 117, 118 (1991).
Alterman, supra note 4, at 168-176; Jensen, supra note 4, at 149-150.9

they had negotiated a fair representation on the basis of wealth and population and10
they relied on figures that may have been mere estimates. They did not choose to
direct the taking of a census prior to the initial apportionment, possibly because the
successful conclusion of a constitution would have been impossible without a clear
indication of representation and a nationwide census was impossible prior to the
creation of a federal authority. But although precise population accuracy apparently
was not the paramount concern for the original apportionment, the framers devised
a mechanism to ensure that future representation would reflect the actual differences
in population between the states as accurately as was possible. A legal commentary
of the time noted that the U.S. Constitution was more consonant with the ideal of
democratic representation than the scheme in England which at that time permitted
representation in Parliament which did not reflect the actual relative populations of11
the boroughs because there was no provision for enumeration and reapportionment.
One legal commentator has suggested that a possible reason for the term
“actual” in “actual enumeration” was the intention to distinguish projections of12
potential population growth from actual current population. In the estimates for the
original apportionment, Georgia was allotted more representation than it was actually
entitled to have, because of estimated rapid population growth that would have
justified the allotment of one more congressional seat before the first census and
subsequent apportionment could be held.
The constitutional term, “actual enumeration,” appears to require an actual
headcount. Some of the dictionaries roughly contemporaneous with the period of the
Constitutional Convention define “enumerate” as meaning “to reckon singly” and
“enumeration” as “the act of counting over.” However, one should keep in mind13
that the framers of the Constitution could not foresee the development of statistical
and survey methods which were non-existent at the time. But the Census Clause has
been interpreted by the courts as not precluding the use of sampling techniques for
adjustment of the undercount. No court has addressed the issue of whether the
Census Clause would permit the type of sampling in non-response follow-up
contemplated by the plans for the 2000 census, in which a sample of non-respondents
would receive a follow-up visit resulting in actual contact for 90 percent of the
population and extrapolation of the remaining 10 percent through statistical formulas.
The three-judge district court panels in Department of Commerce v. U.S. House of


Kurland & Lerner, supra note 4, at 101-2.10
Tucker, Blackstone’s Commentaries 1: App. 189 (1803), as reprinted in Kurland and11
Lerner, supra note 4, at 140.
Hearing on Census 2000 Before the Senate Committee on Government Affairs, 10512thst
Cong., 1 Sess. (Testimony by Hon. Stuart M. Gerson, Assistant Attorney General for the
Civil Division under the Bush Administration, April 16, 1997, not yet published by the
Government Printing Office); also noted in the Bureau of the Census, United States
Department of Commerce, Report to Congress—The Plan for Census 2000 52 (originally
issued July 1997, revised and reissued August 1997).
Hamilton, Johnson’s Dictionary of the English Language in miniature (2d American ed.13

1806; 3d American ed. 1810) ("enumerate" means "to reckon up singly," "enumeration"


means "the act of counting over"); Wiggins, The New York Expositor 93 (1822, 1825)
("enumerate" means "to reckon up singly"). Neither dictionary defines “census.”

Representatives and Clinton v. Glavin did not reach the constitutional issue because
they held that the federal statute prohibited sampling of any kind in the
apportionment census.14
B. Federal Statutory Prohibition Against Sampling
Section 195 of Title 13, United States Code, currently provides that “[e]xcept
for the determination of population for purposes of apportionment of Representatives
in Congress among the several States, the Secretary shall, if he considers it feasible,
authorize the statistical method known as `sampling’ in carrying out the provisions
of this title.” The legislative history of section 195 of Title 13 indicates that
Congress originally merely authorized, but did not require, the use of sampling for
economy’s sake, and did not authorize the use of sampling for apportionment. The15
1957 House report states “[h]owever, section 195 does not authorize the use of
sampling procedures in connection with apportionment of Representatives.” The16
legislative history also indicates that, in the 1976 amendment to section 195,
Congress intended to encourage the Bureau of the Census strongly to use sampling
as much as possible and therefore required sampling whenever feasible. The17
exception to this requirement is the census for apportionment. There is no clear
indication in the reports or floor debates about the reason for the exception. In the
House debates regarding the 1976 amendments and legislation for a mid-decade
census, Representative Lott’s remarks show that he considered the mid-decade
population census to be one reason for encouraging the use of sampling, and that he
felt Congress should make clear that the mid-decade census was to be used for
allocation of funding and not for apportionment. However, the clarification he18
sought appears in section 141(e)(2) of title 13 and the sampling/apportionment
prohibition predates any mid-decade census authorization. Congress may have had
qualms about the use of sampling for apportionment in light of the “actual
enumeration” language. The fact that they authorized sampling for everything other
than apportionment indicates they were not convinced sampling was an acceptable
substitute for a headcount. Apparently, they thought it was acceptable for other
aspects of the census, although prior to the enactment of section 195, there was an
actual enumeration of everything on the census. One should note that section19

181(a) permits the use of sampling for the interim population data, which are not the


U.S. House of Representatives v. U.S. Department of Commerce [case name in the lower14
court], 11. F. Supp. 2d 76, 104 (D.D.C. 1998), and Glavin v. Clinton [case name in the lower
court], 19 F. Supp. 2d 543, 553 (E.D. Va. 1998).
S.Rept. 85-698, 85. Cong., 1 Sess. 3 (1957); H.Rept. 85-1043, 85. Cong., 1. Sess. 1015thstthst
(1957).
H.Rept. 85-1043, 85. Cong., 1. Sess. 10 (1957).16thst
S.Rept. 90-1256, 90 Cong., 2d Sess. 6 (1976); H.Rept. 90- 944, 90 Cong., 2d Sess. 617ththth
(1976); H.Rept.90-1719, 90 Cong., 2d Sess. 13 (1976).

122 Cong. Rec. 9787 (1957).18


Congressional Research Service for the Subcomm. On Energy, Nuclear Proliferation and19th
Federal Services of the Senate Comm. On Governmental Affairs, 96. Cong., 2d Sess., The
Decennial Census: An Analysis and Review, 82-86 (Comm. Print 1980) [hereinafter The
Decennial Census].

official apportionment data. Although the legislative history is ambiguous because
an argument can be made that the 1976 legislation supports the notion of sampling
for adjustment, the statement in the 1957 House report and the text of section 195
itself appear to prohibit any sampling for apportionment purposes.
There is also the issue of whether section 195 applies to statistical methods other
than sampling. Although there were statistical methods other than sampling in
existence at the time of the 1976 amendments to section 195, and these methods were
described in hearings before the Senate Committee on Post Office and Civil
Service, section 195 does not explicitly refer to anything other than sampling.20
Therefore, one could argue that the Congress knew of other methods for estimating
the population and by only prohibiting the use of sampling for purposes of21
apportionment, it did not intend to restrict the use of other statistical methods. On
the other hand, in a 1976 report to the House Committee on Post Office and Civil
Service, the Comptroller General of the United States noted that, although several
statistical methods were being developed, methods other than sampling were22
considered experimental or unreliable. Thus one could also argue that the reason
sampling is the only method explicitly mentioned in section 195 is that Congress did
not consider other methods to be feasible and worth mentioning, and did not intend
to permit their use for apportionment. One could also conclude that since the other23
methods were considered inferior to sampling, and would not be permitted where
sampling is not permitted, then Congress intended “sampling” to be a generic term
covering other methods.24
Section 141(a) of Title 13, the United States Code, provides that “[t]he Secretary
shall, in the year 1980 and every 10 years thereafter take a decennial census of
population as of the first day of April of such year, which date shall be known as the
‘decennial census date’, in such form and content as he may determine, including the
use of sampling procedures and special surveys.” This must be interpreted together
with section 195 of Title 13. Prior to the district court decisions in Department of
Commerce v. U.S. House of Representatives and Clinton v. Glavin, the weight of case
law had reconciled the apparent conflict between sections 141(a) and 195 by finding
sampling permissible for adjustment of but not as a substitute for the headcount.
Although some have argued that “census of population” as defined in subsection

141(g) does not mean the census for apportionment, but some general survey,25


Id.20
Id.21
Id.22
Id.23
Id.24
Memorandum of Plaintiff in Reply to Defendants' Opposition to Plaintiff's Motion for25
Summary Judgment at 12, n. 18, U.S. House of Representatives v. U.S. Department of
Commerce, 11 F. Supp. 2d 76 (D.D.C. Aug. 24, 1998) (Civ. A. No. 98-0456).

subsection 141(b) indicates that the term “decennial census of population” means the
census used for apportionment and redistricting.26
C. Case Law Prior to Dept. Of Commerce v. U.S. House of
Representatives and Clinton v. Glavin
1. Before Wisconsin v. City of New York. The decisions of the Bureau of the
Census to not adjust the census in 1980 and 1990 by using statistical methods to
correct the undercount led to a spate of litigation for each census. In the course of
this litigation, the courts indicated that statistical adjustment was not barred by the
“actual enumeration” language, because accuracy of the count was the paramount
consideration. Therefore, if greater accuracy could be achieved by adjustment than27
by a headcount alone, the adjustment would be permitted and arguably required.
Nevertheless, the initial census data had to be based on a traditional enumeration or
headcount. 28
The federal courts, prior to the Wisconsin v. City of New York decision,
emphasized accuracy and sanctioned the use of certain statistical methods for the
purposes of providing for more accurate congressional redistricting, supporting the
idea that statistical adjustment for apportionment is permissible for accuracy’s sake.29
In Wesberry v. Sanders, a redistricting case, the Supreme Court held that the
language of Article I, § 2, meant that “as nearly as is practicable one man’s vote in
a congressional election is to be worth as much as another’s,” noting at the same time
that “it may not be possible to draw congressional districts with mathematical
precision.” A lower court noted that “it is but a short step from the reasoning
employed in Wesberry to the conclusion that census figures must accurately reflect
the populations of each state in order to preserve the efficacy of an individual’s
vote.” In the 1969 Supreme Court decision in Kirkpatrick v. Preisler involving3031
Missouri’s congressional redistricting plan, the Supreme Court, while invalidating
the plan, nevertheless indicated that the use of projected population figures was not


Brief of Petitioner State of Wisconsin at 33, Wisconsin v. City of New York, ___ U.S. ___,26

116 S.Ct. 1091 (1996) (No. 94-1614).


E.g., Carey v. Klutznick, 508 F.Supp. 404, 414 (S.D.N.Y. 1980). “It appears to the Court27
that this language [‘whole number of persons in each state’] indicates an intent that
apportionment be based on a census that most accurately reflects the true population of each
state. We cannot accept the defendants’ position that the phrase ‘actual enumeration’ used
in Art. I, § 2, cl. 3, requires otherwise. Indeed, Webster’s defines ‘actual’ as ‘existing in fact
or reality’ and defines ‘enumeration’ as a ‘listing’ or ‘counting.’ Webster’s Third New
International Dictionary (1971). When combined, these terms require a count of the
population most reflective of the true facts or reality, and thereby supports the conclusion
that apportionment is to be based on census tabulations that most accurately reflect the
population of each state.”

508 F. Supp. at 415.28


376 U.S. 1, 8, 18 (1963).29


Carey v. Klutznick, 508 F. Supp. at 415 (citing Young v. Klutznick, 497 F. Supp. 1318 E.D.30
Mich. 1980).

394 U.S. 526 (1969).31



unconstitutional and that states may properly consider such statistical data if such
data would have a high degree of accuracy. So if modern surveying and statistical32
techniques result in a count that is more accurate than an actual headcount, they may
arguably be permissible even as a substitution for the headcount. The Census clause
also provides that the enumeration shall be made “in such manner as they [Congress]
shall by Law direct.” This language seems to permit Congress to authorize any33
method for the taking of the census, although at least one court has suggested that
there may be a requirement for some type of raw headcount.34
Before the Wisconsin decision and the Second Circuit decision it reversed, the
courts indicated that the standard for reviewing a decision of the Bureau of the
Census regarding the census methodology, such as the extent of using statistical
adjustment methods, was whether the decision was arbitrary and capricious, a
standard under which it would be unlikely that the Bureau’s methods would be
invalidated. 35
In cases where states and municipalities have sought an adjustment to the
headcount for purposes of reapportionment and allocation of federal assistance funds,
the federal courts deemed statistical adjustments to be permissible for apportionment
in order to achieve the greatest accuracy possible. In City of New York v. U.S. Dept.36
of Commerce, the district court concluded that “because Article I, § 2, requires the
census to be as accurate as practicable, the Constitution is not a bar to statistical
adjustment.” The census must be as accurate as practicable to permit redistricting
within the states to be as close as possible to the ideal of “one man, one vote” and to
permit reapportionment of Representatives among the states to be as fair as possible,
given the other requirements for apportionment (at least one representative per state
and congressional districts cannot cross state lines). However, the court warned37
that “it does not follow that any and all forms of statistical adjustments will be38


sanctioned."” The opinion in this case cited Carey v. Klutznick and Young v.
See also Dixon v. Hassler, 412 F.Supp. 1036, 1040-41 (W.D. Tenn 1976), aff’d sub. Nom.32
Republican Party of Shelby County v. Dixon, 429 U.S. 934 (1976); Exon v. Tiemann, 279
F. Supp. 601, 608 (D.Neb. 1967).
U.S. Const., art. I, § 2, cl. 3.33
City of Philadelphia v. Klutznick, 503 F. Supp. 663, 679 at fn. 9 (1980).34
Cuomo v. Baldrige, 674 F. Supp. 1089, 1105 (S.D.N.Y. 1987); City of New York v. U.S.35
Dept. Of Commerce, 713 F. Supp. 48, 54 (E.D.N.Y. 1989); City of Philadelphia v. Klutznick,

503 F. Supp. 663, 677 (E.D. Pa. Cons. Stat. Ann. 1980).


739 F. Supp. 761, 767 (E.D.N.Y. 1990).36


One should note that as far as reapportionment among the states is concerned, the "one37
man, one vote" principle is limited by the requirement that each state must have at least one
representative and congressional districts must not cross state boundaries.

508 F. Supp. at 404.38



Klutznick (reversed on other grounds), as supporting the use of sampling to adjust39
the census, but not to replace the traditional enumeration.
In Carey v. Klutznick, where the state of New York sought an adjustment of the
alleged undercount in the 1980 census, the court noted that the reasoning of Wesberry
led to the conclusion that “census figures must accurately reflect the populations of
each state in order to preserve the efficacy of an individual’s vote.” The court noted40
that although section 195 of Title 13 prohibited the use of sampling for the purpose
of apportionment, section 141 of Title 13 provided for the population census, in such
form and content as the Secretary of Commerce may determine, including the use of
sampling procedures and special surveys. The court was obligated to adopt an
interpretation that would reconcile the two sections rather than declare one null and
void, therefore it concluded that the “sole use of sampling procedures has been
authorized only for purposes other than tabulating census figures for the purpose of
apportionment and that in the area of apportionment where important constitutional
rights are at stake, the Bureau of the Census may utilize sampling procedures but
only in addition to more traditional methods of enumeration.” 41
In Young v. Klutznick, the city of Detroit sought an adjustment of the alleged
undercount in the 1980 census. The court noted that since 1970 there had been no
“simple straight forward headcount” but rather a “relatively accurate estimate of the
population developed through the use of self-enumeration by questionnaire, statistical
techniques, and computer control devices.” The court concluded that the words42
“actual enumeration” do not “prohibit an accurate statistical adjustment of the
decennial census to obtain a more accurate count. Quite to the contrary, Article 1 of
the Constitution, which mandates equal apportionment of representatives, mandates
an adjustment to obtain a figure that more accurately reflects the actual population
of the United States and state and sub-state areas than the adjusted headcount.”43
Finally, the court also held that section 195 does not prohibit the use of statistical
techniques to adjust the population census for apportionment, although “the use of
figures derived solely by statistical techniques” is prohibited. The court seemed to44
treat “sampling” and “statistical techniques” as equivalent and interchangeable. The
court in City of Philadelphia v. Klutznick also concluded that section 195 does not45
prohibit the use of statistical methods to adjust the headcount. Finally, in City of New
York v. U.S. Dept. of Commerce, the district court stated that “[i]t is no longer novel46
or, in any sense, new law to declare that statistical adjustment of the decennial census


497 F. Supp. 1318 (E.D. Mich. 1980); reversed on other grounds by Young v. Klutznick,39th


652 F.2d 617 (6 Cir. 1981) (dismissed for plaintiffs’ lack of standing and unripeness for
judicial review).

508 F. Supp. at 415.40


508 F. Supp. at 415.41


497 F. Supp. at 1330.42


497 F. Supp. at 1333.43


497 F. Supp. at 1334-35.44


503 F. Supp. at 679.45


739 F. Supp. at 761.46



is both legal and constitutional. This Court has already recognized that Article I, §

2, require[s] that the census be as accurate as practicable.” [Cites omitted].47


Although the courts in City of New York v. U.S. Dept. of Commerce, Carey v.
Klutznick, and Young v. Klutznick appear to have concluded that statistical methods,
including sampling, may be permissible for adjustment, these methods probably are
not acceptable as an outright substitute for an actual headcount. In City of
Philadelphia v. Klutznick, in which the city alleged injuries of loss of federal and48
state representatives and federal and state funding because of an undercount, the
opinion appears to support the use of statistical methods to adjust the headcount.
However, the court in that case took the position that the census figures must be raw
data from an actual enumeration, although those figures may be adjusted by4950
statistical methods, including sampling, in order to arrive at a more accurate figure.
The court observed that
[the] principal concern of the Framers in the constitutional provision at
issue was not the manner of conducting the census. Their chief concern
was the apportionment of seats in the House of Representatives. . . . It may
be that today an actual headcount cannot hope to be an accurate reflection
of either the size or distribution of the Nation’s population. If so it is
inconceivable that the Constitution would require the continued use of a
headcount in counting the population. Therefore the Court holds that the
Constitution permits the Congress to direct or permit the use of statistical
adjustment factors in arriving at the final census results used in
reapportionment. . . . Therefore, the Court holds that the Census Act
permits the Bureau to make statistical adjustments to the headcount in
determining the population for apportionment purposes.51
Footnote 9 to this passage states “It is, nevertheless, also reasonable to assume that
another purpose of calling for an ‘actual Enumeration’ was to ensure that
apportionment would be based on numbers more reliable than population estimates.
Therefore, while statistical adjustment factors may be used in arriving at the final
census figures, the Constitution would appear to require that the census at least be
based on raw data obtained by an actual headcount.” In Senate of the State of
California v. Mosbacher, the court noted that “[t]he statutes do not define the52
method by which the Secretary is to do the count, but it is generally expected to be
a head count rather than a mere statistical manipulation through the use of sampling
and other techniques.”


739 F. Supp. at 767.47


503 F. Supp. at 663.48


503 F. Supp. at 679, fn. 9.49


503 F. Supp. at 679.50


Id. (Emphasis in the original).51

968 F.2d 974, 978 (9 Cir. 1992).52th



2. Wisconsin v. City of New York. After both the 1980 and 1990 censuses,


there were lawsuits contesting the decennial population census as inaccurate and
seeking an adjustment of the undercount. Since 1940, there has been an
acknowledged undercount, but until relatively recently, there has been no feasible
method of accurately estimating the undercount and adjusting the actual headcount
accordingly. The litigation after the 1980 census appeared to focus on whether a
statistical adjustment was constitutionally permitted and whether there was a feasible
adjustment method, in addition to raising issues of standing and justiciability. The
decisions after the 1980 census also appeared to accept the “arbitrary and capricious”
standard as appropriate in reviewing the Secretary of Commerce’s decision against
census adjustment. The litigation following the 1990 census appeared to accept the
feasibility of adjustment and focused instead on the issue of the accuracy of the
chosen adjustment method.
Two federal appellate courts, the Sixth and Seventh Circuits, using the “arbitrary
and capricious” standard, held that the Secretary of Commerce had not acted in an53
arbitrary and capricious manner in deciding not to adjust the 1990 census figures.
The decision of whether to adjust was in the discretion of the Secretary of Commerce
who considered whether an adjustment would result in greater accuracy in several
respects. The Sixth and Seventh Circuits apparently accepted the Secretary’s
explanation that an adjustment of the actual count would not result in greater
accuracy with regard to the distribution of population among the states and other
areas measured and might even result in decreased distributive accuracy. Finding
that the decision to not adjust was not arbitrary and capricious, the courts refused to
take sides in a dispute among the experts about the desirability of adjustment, finding
that there were no judicially administrable standards for determining the desirability
of adjustment.
The Second Circuit in City of New York v. United States Dept. of Commerce,54
held that the arbitrary and capricious standard was not appropriate. Instead, after
looking to a line of precedent involving judicial review of intrastate districting
decisions, beginning with Wesberry, the Second Circuit decided that the appropriate
standard was whether the Secretary’s decision to not adjust “was essential to the
achievement of a legitimate governmental interest.” It established a standard of
review that would make it more difficult for the Secretary of Commerce to defend
the decision against an adjustment of the raw headcount figures from the 1990 census
and found that the plaintiffs had sustained their initial burden of proving that the
Secretary had not made a good-faith effort to achieve the most accurate census
possible. The court remanded the case to the lower court for further proceedings not
inconsistent with the opinion. Wisconsin, Oklahoma and the United States each filed


City of Detroit v. Franklin, 4 F.3d 1367 (6 Cir. 1993), cert. denied sub. nom. City of53th
Detroit v. Brown, 510 U.S. 1176, 114 S.Ct. 1217 (1994); Tucker v. United States Dept. Ofth
Commerce, 958 F.2d 1411 (7 Cir. 1992), cert. denied, 506 U.S. 953, 113 S.Ct. 407 (1992).

34 F.3d 1114 (2d Cir. 1994).54



petitions for certiorari. The Supreme Court granted those petitions and consolidated
them; it issued a unanimous decision on March 20, 1996.55
The Supreme Court reversed the Second Circuit decision. Rejecting the
applicability of the intrastate redistricting line of cases with their requirements for
mathematically precise equality and accuracy, it held that the appropriate standard for
review of the adjustment decision of the Secretary of Commerce was whether the
Secretary’s conduct of the census was “‘consistent with the constitutional language56
and the constitutional goal of equal representation,’” and, in light of the broad grant
of authority over the census given to Congress in the Constitution, whether the
Secretary’s decision had “a reasonable relationship to the accomplishment of an
actual enumeration of the population, keeping in mind the constitutional purpose of
the census [equal representation through fair apportionment of the House of
Representatives].” The Court looked to Franklin v. Massachusetts, a case involving57
the method of counting overseas federal employees, and Department of Commerce
v. Montana, a case involving the method of calculating the apportionment, for this
standard of review. The Court noted that it had rejected the applicability of the
standard in the Wesberry line of cases with regard to the congressional apportionment
methods contested in Franklin and Montana. The Court had rejected the Wesberry
standard because it considered congressional reapportionment to be quite different
from intrastate redistricting, because of the reapportionment requirements that each
state have at least one representative and that congressional districts cannot cross
state lines. Therefore, the standard applicable to intrastate redistricting could not
appropriately be applied to congressional reapportionment. The Court found that the
Federal Government’s conduct of the census was even further removed from
intrastate redistricting than was congressional reapportionment. Thus, the Federal
Government was due even greater deference in its census methodology than in its
congressional reapportionment methodology.
With regard to accuracy, the Court observed that “[for] several reasons, the
‘good-faith effort to achieve population equality’ required of a State conducting
intrastate redistricting does not translate into a requirement that the Federal58
Government conduct a census that is as accurate as possible.” The “several
reasons” included the fact that the Census Clause of the Constitution granted
Congress broad authority to conduct the census “in such Manner as they shall by Law
direct” and that further constitutional requirements for apportionment, as mentioned
above, made it impossible to achieve precise population equality among interstate
districts and in determining congressional reapportionment. The Court’s holding
could be interpreted in two ways—(1) the Federal Government is no longer required
to conduct the most accurate census possible or (2) accuracy is still as important as
the cases preceding the Second Circuit decision in the Wisconsin case indicated it is,
but there are no guidelines for accuracy which enable the courts to review the Federal


Wisconsin v. City of New York, ___ U.S. ___, 116 S.Ct. 1091 (1996).55
Franklin v. Massachusetts, 505 U.S. 788, 804 (1992); Dept. Of Commerce v. Montana, 50356
U.S. 442, 459 (1992).
Wisconsin, 116 S.Ct. at 1101.57
Wisconsin, 116 S.Ct. at 1099-1100.58

Government’s determinations regarding the accuracy of particular census
methodologies.
It appears that a stronger case can be made for the second option, since the
passage following the one quoted above goes on to note that the Court sees no way
in which the courts can determine whether the Federal Government has made a good-
faith effort to achieve population equality. The Court compares the situation before
it to that in the Montana case, observing that just as it was impossible for the Court
to choose one apportionment methodology over another, it was impossible for it to
choose one census methodology over another as more accurate. The Constitution and
the intrastate redistricting cases could offer no guidance about what constituted a
good-faith effort, that is, which census methodology was more accurate and therefore
better than another. “The polestar of equal representation does not provide sufficient
guidance to allow us to discern a single constitutionally permissible course.” The59
Court does not appear to be saying that accuracy is not as important as it was or that
the Federal Government has a reduced obligation to maximize accuracy. Rather, it
is saying that a good-faith effort to conduct the most accurate census possible is not
easily measurable with the same mathematical precision as in intrastate redistricting.
Therefore, the Court declined to second-guess the Federal Government’s decisions
about census methodology.
The Supreme Court specifically declined to decide whether the “Constitution
might prohibit Congress from conducting the type of statistical adjustment
considered here" or whether under section 195 of title 13, United States Code,60
Congress had "constrained the Secretary’s discretion to statistically adjust the
decennial census.” In the Wisconsin decision, the Court decided the Federal61
Government was not required to adjust the census by any constitutional standard.
It has yet to decide whether the Federal Government may adjust the headcount or
whether it may use sampling techniques in the adjustment; Department of Commerce
v. U.S. House of Representatives and Glavin v. Clinton, discussed below, present the
opportunity to do so.
Plans Of The Bureau For Sampling In Census 2000
A. A Brief History of Adjustment Efforts by the Bureau of the
Census
The actual headcount that comprises the decennial population census contains
unavoidable errors for a variety of reasons, including the overcounting caused by
multiple counting of an individual or the counting of non-existent persons, and
undercounting caused by the failure of some persons to respond to the census survey.
The census tends to undercount minority groups by a larger percentage than it
undercounts other groups. This so-called “differential undercount” has been a known


Wisconsin, 116 S. Ct. at 1100, quoting Montana, 503 U.S. at 463.59
Wisconsin, 116 S.Ct. at 1101, n. 9.60
Wisconsin, 116 S.Ct. at 1101, n. 11.61

problem since 1940. Because the decennial census data is constitutionally and62
statutorily required for a number of purposes of great public interest, the Bureau of
the Census has studied estimation and adjustment methods at least since 1970, in
addition to methods of improving the accuracy of the headcount itself. Although
persons not actually counted were apparently imputed to the 1970 population
figures, the Bureau appears to have been dissatisfied with the methods. After6364
further study and consideration of the adjustment methods available in the 1970s for
the 1980 census, the Bureau decided not to adjust the 1980 Census because an
acceptably accurate adjustment method was not feasible: “[a]t present, the Bureau has
no sound statistical basis for estimating the true undercount or introducing
adjustments.” Consequently, fifty-odd lawsuits were brought against the Bureau65
in the wake of the 1980 census, most seeking adjustment of the decennial population
figures to increase the population.66
In preparing for the 1990 census, the Bureau of the Census conducted research
aimed at developing an adjustment method. Two task forces were created within67
the Bureau, the Undercount Research Staff and the Undercount Steering Committee
and other Bureau divisions also conducted studies on the undercount and the
possibility of adjustment. The Bureau also sought outside opinions from various
experts and professional organizations such as the National Academy of Science and
the American Statistical Association. The Bureau chose to use the post-enumeration
survey (PES) as the best method of adjusting the census through the use of “dual
system estimation” (DSE) or “capture/recapture.” This method would evaluate the
quality of the actual headcount by measuring the rate at which persons were omitted
and erroneously enumerated by the census to determine the net undercount rate. This
could then be used to calculate adjusted figures.


City of New York v. United States Dept. Of Commerce, 713 F. Supp. 48, 50 (E.D.N.Y.62

1989)[hereinafter NYC v. DOC I]; City of New York v. United States Dept. Of Commerce,


822 F. Supp. 906, 913 (E.D.N.Y. 1993)[hereinafter NYC v. DOC III]; Congressionalthst


Research Service, Improving Census Accuracy, H. Comm. Print No. 100-6, 100 Cong., 1
Sess. 40 (1987).
Young v. Klutznick, 497 F. Supp. at 1321; M. Anderson, supra note 4 at 230;63
Congressional Research Service, supra note 62, at 45 and note 88. All these sources state
that uncounted persons were added to the census headcount used for apportionment by
“imputation” or “synthetic estimation.” Anderson states this in the context of describing
plaintiff’s arguments in Young v. Klutznick. However, the court in City of Philadelphia v.
Klutznick, 503 F. Supp. at 663, stated that despite an undercount in the 1960 and 1970
censuses, the Bureau of the Census did not adjust the census figures to compensate for the
undercount in the results reported to the President for apportionment.
Cuomo v. Baldrige, 674 F. Supp. at 1107.64

45 Fed. Reg. 82872-82885 (Dec. 16, 1980); M. Anderson, supra note 4, at 232;65


Congressional Research Service, supra note 62, at 6; Cuomo v. Baldrige, 674 F. Supp. at

1096; Young v. Klutznick, 652 F.2d 617, 621 (1981).


Cuomo v. Baldrige, 674 F. Supp. at 1091; Congressional Research Service, supra note 62,66
at 6; M. Anderson, supra note 4, at 230.
City of New York v. United States Dept. Of Commerce, 34 F.3d at 1114 [hereinafter NYC67
v. DOC IV]; NYC v. DOC III, 822 F. Supp. at 913-914.

The PES had been used for various purposes since 1950, but had never been
used for adjustment. The Bureau refined the PES by developing various techniques
to reduce and eliminate adjustment problems with the PES. By spring of 1987, the
Bureau of the Census had decided that it should proceed with plans to adjust the 1990
census if the PES results met a certain quality standard. However, other high-level
Commerce Department officials informed the Director that they had decided against
an adjustment. On October 30, 1987, the Department of Commerce publicly
announced its decision to not adjust the 1990 census. Again, this decision led to a
spate of lawsuits, the most notable of which were discussed above.
In 1996, the Census Bureau announced its plans to use sampling for non-
response follow-up and for the Integrated Coverage Measurement (ICM). This
caused a great deal of concern reflected in the legislative response in the 105th
Congress, discussed below. In August 1997, the Bureau of the Census reported on
its plans for Census 2000 as directed by Pub. L. 105-18. The two types of sampling68
contemplated in the plan are the sampling for non-response follow-up and for the
Integrated Coverage Measurement (ICM). Under the sampling for non-response
follow-up, the Bureau will conduct a sample survey of non-respondent households
in each census tract to arrive at a 90-percent response rate for the tract and to
extrapolate data for the remaining 10 percent of each tract’s population. Under the
ICM, an independent sample survey of about 750,000 households will be taken in a
manner to be representative of all fifty states in order to correct the differential
undercount rate and the results will be incorporated to obtain a one-number census.
While the ICM is similar to the Post-Enumeration Survey (PES) contemplated in the
1990 census, the sampling for non-response follow-up is a new technique never
before considered for the decennial census for apportionment.
In addition to efforts to adjust the census for the undercount, the Bureau noted
in its report on plans for census 2000 that the statistical method of imputation has
been used at least since 1940. Under imputation, when an enumerator knew that69
a housing unit was occupied, but could not get information on the number of people
living in the unit, the enumerator would impute the number based on occupation of
similar units.
B. Constitutionality and Legality of Census Bureau Plans
Under the civil action created by P.L. 105-119, two lawsuits have been brought
challenging the use of sampling in the 2000 decennial census, Department of


Bureau of the Census, United States Department of Commerce, Report to Congress—The68
Plan for Census 2000 23 (originally issued July 1997, revised and reissued August 1997),
as directed by Pub. L. 105-18, Title VIII, 111 Stat. 217 (1997). See also Bureau of the
Census, United States Department of Commerce, Census 2000 Operational Plan IX-

17—IX-19 (April 1998).


Bureau of the Census, United States Department of Commerce, Report to Congress—The69
Plan for Census 2000 23 (originally issued July 1997, revised and reissued August 1997).

Commerce v. U.S. House of Representatives and Clinton v. Glavin. District court7071
decisions have been issued in both cases. Both cases were appealed to the United
States Supreme Court, which consolidated them for oral arguments heard on
November 30, 1998. An opinion is anticipated within the next several months,
probably sometime in the spring of 1999. The Census Bureau plans must be resolved
by that time if the 2000 decennial census is to be conducted on schedule.
The decision of the three-judge district court panel in the Department of
Commerce v. U.S. House of Representatives case was issued on August 24, 1998.
The plaintiff, U.S. House of Representatives, argued that summary judgment should
be granted because:72
!The plain text of section 195 of the Census Act prohibits the use of sampling
in the census for the apportionment of the House of Representatives because
Congress did not express an intent to repeal the traditional sampling
prohibition or to grant the Secretary of Commerce such discretion with respect
to such sampling in the apportionment count.
!The plans for using sampling in the 2000 decennial census violate the
constitutional census requirements by using an estimation technique, where
the term “actual enumeration” was understood and intended by the Framers
to mean a one-by-one headcount.
!The Framers required an actual enumeration to protect the census from
political manipulation, which would compromise the integrity of democratic
representation in the federal government, by establishing a precise,
unambiguous basis for the population count for apportionment.
The defendants, U.S. Department of Commerce and others, argued in reply
that: 73
!The language and legislative history of the Census Act show that Congress
delegated to the Secretary of Commerce discretion to use sampling for the
decennial census for apportionment. (The position of the defendants appears
to go beyond the case law discussed above, which suggested that some
sampling techniques to refine and supplement the traditional headcount census
were within the discretion of the Secretary.)


Supra at note 1.70
Supra at note 2.71
Memorandum for Plaintiff in Support of its Motion for Summary Judgment and72
Memorandum of Plaintiff in Reply to Defendants’ Opposition to Plaintiff’s Motion for
Summary Judgment, U.S. House of Representatives v. U.S. Department of Commerce, 11
F. Supp. 2d 76 (D.D.C. Aug. 24, 1998) ( Civ. A. No. 98-0456).
Defendants’ Memorandum in Opposition to the House’s Motion for Summary Judgment,73
U.S. House of Representatives v. U.S. Department of Commerce, 11 F. Supp. 76 (D.D.C.
Aug. 24, 1998) (Civ. A. No. 98-0456).

!The plans for the 2000 decennial census include an effort for actual
enumeration by contacting every household through the extensive mail-out
and supplementary distribution plans.
!The decision by the Secretary of Commerce to use sampling is consistent with
the constitutional language and the constitutional goal of equal representation,
the standard of review for executive branch decisions concerning census
methodology, affirmed by the U.S. Supreme Court in Wisconsin v. City of
New York. The language “actual enumeration” does not define the type of
evidence sufficient to establish that particular persons exist for the purpose of
the enumeration. The traditional headcount techniques produce population
counts that are no more an “actual enumeration,” as defined by the plaintiffs,
than are the counts that would result from the sampling techniques planned by
the Census Bureau.
The defendants argued for dismissal of the case on the grounds that:74
!The House’s challenge to the sampling plans for the 2000 decennial census is
not ripe because its alleged injuries from the plans are neither impending nor
imminent; ongoing legislative activity could still produce a definitive
resolution of the sampling issue for the 2000 census. The House’s suit is
really an attempt to obtain an advisory opinion, in contravention of
constitutional Article III requirements of an actual case or controversy.
!The House lacks standing to bring this civil action because it has not
established that it will suffer a legally cognizable injury.
!Permitting the House to bring this civil action violates the separation-of-
powers doctrine because only the executive branch has the function of
initiating litigation to protect the national interest in the proper administration
of federal law.
The defendants’ go on to reiterate the arguments on the merits which they proposed
in their arguments against the plaintiff’s motion for summary judgment.
The plaintiff’s argued in reply that:75
!The House’s challenge is ripe for adjudication because the harm threatened by
the planned use of sampling in the 2000 decennial census is not speculative
but is realistic. The remote possibility that the sampling impasse between
Congress and the Administration will be resolved does not render the
threatened injury speculative.


Defendants’ Memorandum in Support of Motion to Dismiss and Defendants’ Reply74
Memorandum in Support of Motion to Dismiss, U.S. House of Representatives v. U.S.
Department of Commerce, 11 F. Supp. 2d 76 (D.D.C. Aug. 24, 1998) (Civ. A. No. 98-0456).
Opposition of Plaintiff to Defendants’ Motion to Dismiss, U.S. House of Representatives75
v. U.S. Department of Commerce, 11 F. Supp. 2d 76 (D.D.C. Aug. 24, 1998) (Civ. A. No.

98-0456).



!The House satisfies the standing requirements under Article III because it has
a cognizable interest in the timely receipt of information required by federal
statute and in the gather of such information in a constitutional and legal
manner. Further, it has a concrete and particularized institutional interest in
its lawful composition, in the protection of its institutional integrity, and in
fulfilling its constitutional duties to ensure that an actual enumeration is taken
and that an apportionment is conducted in accordance with that enumeration..
!This civil action does not pose a separation-of-powers problem because the
House is not exercising an executive branch function by initiating litigation
to protect its concrete institutional interests in preserving the integrity of its
composition.
The three-judge panel of the district court held that the House had standing to
bring the lawsuit, the issues presented were ripe for adjudication, dismissal on an
equitable basis was not called for, the doctrine of separation of powers was not
violated. On the merits, the panel held that, reading sections 141 and 195 together76
and considering the text, legislative history and cannons of statutory construction, the
use of sampling in the decennial census for the purposes of apportionment of the77
House of Representatives violates the Census Act. Having decided the issue on
statutory grounds, the court did not address the constitutional grounds, in accordance
with the doctrine of constitutional adjudication that constitutional interpretation
should be avoided where an issue can be resolved on statutory grounds alone.78
The court agreed with the plaintiff that it had standing because it would suffer
a concrete and particularized injury in not receiving information statutorily required,
the numbers necessary for apportionment, and because it has a concrete and79
particularized interest in its lawful composition, which would be compromised by an
unlawfully conducted census. The court found that the suit was ripe because it80
disagreed that the injury was speculative. It found that the injury did not occur only81
when the President actually transmitted the census figures to Congress, but that the
injury complained was the alleged procedural violation, the use of sampling alleged
to be in violation of the Federal Constitution and federal statutes. The court
disagreed with the defendant’s contention that even if standing and ripeness
requirements were satisfied, the court should decline to involve itself in an essentially82
political dispute between the Administration and Congress. In light of the
jurisdictional invitation extended by the executive and legislative branches in the


U.S. House of Representatives v. U.S. Department of Commerce, 11 F. Supp. 2d 76, 9776
(D.D.C. Aug. 24, 1998).
Id. at 104.77
Id. at 104.78
Id. at 86.79
Id. at 87.80
Id. at 90-93.81
Id. at 94-95.82

statutory creation of the civil action, the court believed it should retain jurisdiction.
In holding that the House’s suit did not violate the separation-of-powers doctrine, the
panel analogized the litigation to ensure the receipt of the proper information
required for apportionment to litigation permitted in the past under the doctrine to83
obtain information by subpoena.
With regard to the merits, the court essentially agreed with the plaintiff’s
statutory interpretation in all respects. It found that, absent an express intention to
eliminate the 1957 prohibition on sampling in the census for purposes of
apportionment, the 1976 amendments to section 195 of the Census Act could not be
read to permit the Secretary of Commerce to use such sampling at his discretion,
particularly when such sampling would be a departure from a 200-year tradition.84
The panel concluded that the apparent general grant of authority to use sampling
techniques contained in section 141 of the Census Act is subject to the specific
directive of section 195, prohibiting such techniques for apportionment numbers.85
Perhaps most notably, the court expressly disagreed with other courts which had
rejected the view that the Census Act prohibits statistical sampling, at the least in a
limited manner to adjust traditional enumeration techniques. Since the sampling86
considered for the 1990 census and used in the post-enumeration survey after the
1980 census involved only adjustment and not non-response sampling to extrapolate
10 percent of the population, the district court could have adopted a position agreeing
with the other courts, but distinguishing the plans for the 2000 census from the
sampling techniques considered under earlier censuses.
The Administration filed its notice of appeal from the district court decision on
August 25, 1998, and the jurisdictional statement on September 4, 1998. The
Administration and the House of Representatives filed a joint motion for expedited
review of the case on September 9, 1998, which was granted by the U.S. Supreme
Court when it noted probable jurisdiction on September 10, 1998.
The district court filings in the Clinton v. Glavin case reflect arguments similar
to those in the House of Representatives case. In their motion for summary87
judgment, the plaintiffs additionally argued that the plans for sampling are
unconstitutional because the Department of Commerce is using sampling for cost-
saving reasons and not just to improve accuracy. The defendants did not appear to88
respond to this specific point, instead discussing the overall constitutionality of the


Id. at 96.83
Id. at 100-102.84
Id. at 103.85
Id. at 98.86
Memorandum of Law in Support of Plaintiffs’ Motion for Summary Judgment,87
Defendants’ Memorandum in Opposition to Plaintiffs’ Motion for Summary Judgment,
Memorandum in Support of Defendants’ Motion to Dismiss, Plaintiffs’ Memorandum of
Law in Opposition to Defendants’ Motion to Dismiss, inter alia, Glavin v. Clinton, 19 F.
Supp. 2d 543 (E.D. Va. 1998) (Civil Action File No. 98-207-A).
Memorandum of Law in Support of Plaintiffs’ Motion for Summary Judgment at 65-68,88
Glavin v. Clinton, 19 F. Supp. 2d 543 (E.D. Va. 1998) (Civil Action File No. 98-207-A).

Administration's decision to use sampling. In their motion to dismiss, the defendants
argued that injuries alleged by the individual plaintiffs were too speculative and that
the plaintiffs could not establish a causal connection between the use of sampling and
the injuries. The alleged injuries differed from those claimed by the House of
Representatives as an institution and included claims of vote dilution by private
citizens, loss of federal funding by local governments, and delay in the ability of an
individual member of Congress to assume his seat in 2002 if reelected. The89
plaintiffs responded that they had alleged sufficient facts to establish threatened and
imminent injuries, arguing that courts have consistently recognized the standing of
individual plaintiffs to challenge census methodology that may result in a dilution of
their voting strength and in economic injury.90
The three-judge district court panel in Clinton v. Glavin issued its decision on
September 24, 1998. The decision was similar to that in the House of
Representatives case: the court found that the case was ripe for review, the plaintiffs
had standing, and the Census Act prohibited sampling in the census for the
apportionment of the House of Representatives. The court noted that the Department
of Commerce had acknowledged the finality of its decision to use sampling and
disagreed with the defendants' argument that because Congress could yet take
legislative action to prevent the use of sampling, the case was not ripe for review.
The court held that the fact that future legislative action could moot the issue did not
shield agency action from judicial review. With respect to standing, the court found91
that the plaintiffs did not have to prove with mathematical certainty the degree to
which they would be injured by sampling and agreed with the plaintiffs that they had
shown that they would suffer injury and that the courts had consistently recognized
the standing of plaintiffs making allegations similar to those in this case. The court92
held that the plaintiffs did not have to await the realization of the threatened injury
to seek relief, that the alleged injuries were fairly traceable to the defendants' use of93
sampling, and that the judicial remedy sought would redress the injuries. With
respect to the merits, the court held that section 141 of the Census Act provided the
general authority for the use of sampling in various purposes of the census other than
congressional apportionment, while section 195 provided the specific authority
qualifying the general authority by prohibiting sampling for the congressional
apportionment. Lastly, the court, having decided the issue on clear statutory94
grounds, held that there was no need to reach the constitutional grounds.95


Memorandum in Support of Defendants’ Motion to Dismiss at 30-44, Glavin v. Clinton,89

19 F. Supp. 2d 543 (E.D. Va. 1998) (Civil Action File No. 98-207-A).


Plaintiffs’ Memorandum of Law in Opposition to Defendants’ Motion to Dismiss at 21-90

47, Glavin v. Clinton, 19 F. Supp. 2d 543 (E.D. Va. 1998) (Civil Action File No. 98-207-A).


Glavin v. Clinton, 19 F. Supp. 2d 543, 547 (E.D. Va. 1998).91
Id. at 547-548.92
Id. at 540-550.93
Id. at 550-553.94
Id. at 553.95

Since neither district court addressed the constitutional issue, the interpretation
of the “actual enumeration” language in the Census Clause has yet to be addressed
judicially.
The Administration filed its appeal from the district court decision on
September 25, 1998, and the jurisdictional statement on October 2, 1998. On
October 9, 1998, the U.S. Supreme Court noted probable jurisdiction, granted a joint
motion for expedited consideration by the Administration and the plaintiffs, and
consolidated the two lawsuits for oral argument. Oral arguments were heard on
November 30, 1998, with the Justices questioning the Administration's position96
with regard to the meaning of the constitutional language of "actual enumeration" and
questioning the plaintiffs concerning their standing. In oral arguments, the plaintiffs
and defendants disagreed as to whether sampling would result in greater accuracy,
constitutional and legal issues aside. They also disagreed as to whether statistically
adjusted figures could be used for the purposes of intrastate redistricting and funding
allocation, even if they could not be used for the apportionment of the House of
Representatives. Since this latter issue was not squarely before the Court, it could
remain effectively undecided regardless of what the Court decides with regard to
whether sampling is prohibited in the census to apportion the House of
Representatives. The Court could decide by March 1999 whether plans for sampling97
in the 2000 decennial census are legally permissible.
Legislation Concerning Sampling
In response to the plans of the Bureau of the Census to utilize sampling
techniques for non-response and for integrated coverage measurement, there were
various attempts in the 105th Congress to prohibit the use of sampling in the
decennial census for apportionment of the House of Representatives, to establish
guidelines for the use of sampling, or to otherwise provide for some legislative
control over the use of sampling. Neither the bills favoring sampling as a mode of
achieving greater accuracy for a headcount executed in good faith, nor the bills
favoring an absolute prohibition on sampling in the census for apportionment, were
enacted. The prevalent provisions in the first session were the compromises of
P.L.105-119, which resulted in the creation of the civil action and the Census
Monitoring Board and which provided for the use of funds to prepare and plan a 2000
census without the use of sampling. These allowed the Bureau of the Census to
continue with its sampling plans to a degree, but still left the ultimate decision with
regard to sampling methodology in the hands of either the U.S. Supreme Court or the
Congress at some later date. The notable legislation from the second session was the
Omnibus Consolidated and Emergency Supplemental Appropriations Act for
FY1999, which provided for decennial census funding through June 15, 1999. The
reports for this legislation appear to envision resolution of the sampling issues by the


The oral argument transcript for Department of Commerce v. U.S. House of96
Representatives (and Clinton v. Glavin) is available on Westlaw under the citation 1998 WL

827383.


Linda Greenhouse, High Court to Hear Case on Sampling for Census, New York Times97
A16 (September 11, 1998 Late Edition).

spring of 1999, but do not address the issues explicitly. Meanwhile, pending the98
U.S. Supreme Court decision in the House of Representatives case, the
Administration has announced that it will continue with its dual track plans for both
sampling and traditional enumeration censuses, despite the express disapproval and99
concern of House Republican leaders.
A. Bills in the 105th Congress
H.R. 776, the Decennial Census Improvement Act of 1997, introduced by
Representative Carrie Meek and referred to the House Government Reform and
Oversight Committee, would have required the Bureau of the Census to contact every
household directly by mail or in person and would have permitted sampling as a
substitute for direct contact in a particular census tract only after direct contact has
been made with at least 90 percent of the households in the tract. Further, the Bureau
would have been required to make greater use of state and local government agencies
and non-governmental groups to reduce the undercount and to report to the
congressional oversight committees on the measures it plans to take to accomplish
this. The bill would also have taken other measures to promote the recruitment of
temporary census employees for the decennial census.
H.R. 1178, introduced by Representative Carolyn Maloney and referred to the
House Government Reform and Oversight Committee, would have amended section
195 of title 13, the United States Code, to make clear that sampling may be used in
order to improve the accuracy of the decennial population census for apportionment
of the House of Representatives. But sampling can only be used in conjunction with
a good faith effort using non-sampling techniques, whether the sampling occurs
before, during, or after the enumeration and only to the extent that the Secretary of
Commerce deems it necessary to achieve a more accurate census. The requirements
for reporting to the congressional oversight committees would have been amended
to include a report on the methodology for the census, including a determination and
evaluation of the types of surveys or statistical methods proposed. The bill would
also have amended section 141 of title 13, United States Code, to require that the data
on population or population and housing characteristics, which is used to determine
the allocation of federal benefits to state, county or local governmental units, must
be collected as part of the decennial census and at the same time as the collection of
data for purposes of apportionment. Such data may be collected through sampling
where appropriate.
H.R. 1220, introduced by Representative Thomas Petri and referred to the
House Government Reform and Oversight Committee, would have amended section
141, title 13, United States Code, to make clear that no sampling or other statistical
procedure may be used to determine the total population by States for purposes of the
apportionment of the House of Representatives.


Infra at notes 109 and 111.98
Barbara Vobejda, Census Plans Proceed Despite Court Ruling, Washington Post A2399
(Sept. 7, 1998 Final Edition).

Title VIII of H.R. 1469, the 1997 Emergency Supplemental Appropriations Act
for Recovery from Natural Disasters, and for Overseas Peacekeeping Efforts,
Including Those in Bosnia, as passed by Congress and vetoed by President Clinton
on June 9, 1997, would have prohibited the use of sampling techniques in the
decennial census for the apportionment of the House of Representatives. The
provisions concerning the 2000 decennial census would have amended section 141
of title 13, United States Code, to make clear that no sampling or statistical
procedure, including any statistical adjustment, could be used in the population
census for apportionment of the congressional Representatives among the States.
Further, the bill would have prohibited the use of funds under any act for any fiscal
year in the preparation and planning for the use of sampling or any other statistical
procedure, including adjustment, in the decennial census for apportionment of
Representatives among the States. The provision would have stated the findings of
Congress that the Constitution and section 195 of title 13, United States Code,
prohibit sampling, and that Congress is committed to funding constitutional census
activities. Title VIII of H.R. 1469 emerged from conference report for H.R. 1469.100
The House version of H.R. 1469 had not contained any provision concerning the use
of sampling in the decennial census. Section 303 of the Senate version, which
inserted the text of S. 672 as an amendment, would have prohibited the use of any
funds for fiscal year 1997 to make irreversible plans or preparations for the use of
sampling. The absolute prohibition on sampling and statistical techniques in the
decennial census for apportionment proved to be controversial and was a contributing
factor to the presidential veto.
S. 672, the Supplemental Appropriations and Rescissions Act of 1997, was
reported from the Senate Appropriations Committee and, as noted above, its text was
inserted into H.R. 1469 as an amendment. As reported, section 303 of the bill
originally prohibited the use of any funds for fiscal year 1997 to plan or prepared for
the use of sampling in the 2000 decennial census. Senator Hollings, with others,
proposed Amendment No. 231, which softened this language by prohibiting the use
of any funds in fiscal year 1997 to make “irreversible” plans or preparation for the
use of sampling or any other statistical method, including any statistical adjustment,
in taking the 2000 population census for apportionment of the House of
Representatives among the States. This was adopted May 7, 1997.
Section 202 of H.R. 1755, the Supplemental Appropriations and Rescissions
Act of 1997, introduced by Representative David Obey and referred to the House
Appropriations and House Budget Committees, would have prohibited the use of any
funds for fiscal year 1997 by the Commerce Department to make irreversible plans
or preparation for the use of sampling or any other statistical method, including any
statistical adjustment, in taking the 2000 decennial census of population for the
purpose of apportionment of the House of Representatives among the States.
H.R. 1871, the 1997 Emergency Supplemental Appropriations Act for Recovery
from Natural Disasters, and for Overseas Peacekeeping Efforts, Including Those in
Bosnia, was passed by Congress and signed by the President on June 12, 1997,


H.Rept. 105- 119, 105 Cong., 1 Sess. 124-125 (1997).100thst

becoming P.L. 105-18, in the wake of the veto of H.R. 1469. In lieu of the101
sampling prohibitions in H.R. 1469, title VIII of P.L. 105-18 directed the Commerce
Department to submit to Congress, within 30 days of its enactment, a comprehensive
and detailed plan outlining its proposed methodologies for conducting the 2000
decennial census and available methods to conduct an actual enumeration of the
population. This report had to include a list and explanation of all statistical methods
that may be used in the census, a list of statistical errors which could occur as a result
of the use of such statistical methods, the estimated error rate down to the census
tract level, a cost estimation showing cost allocations for each census activity plan,
and an analysis of all available options for counting hard-to-enumerate persons
without using sampling or other statistical techniques. The Bureau of the Census
submitted this report in July 1997, then submitted a revised report in August.102
The conference report of H.R. 2267, the Departments of Commerce, Justice,
and State, the Judiciary, and Related Agencies Appropriations Act, 1998, originally
reported from the House Appropriations Committee, and of S. 1022, its counterpart,
originally reported from the Senate Appropriations Committee, contained a
compromise which became part of P.L. 105-119. Title II of the House bill provided,
inter alia, for $100,000,000 for expenses in conducting the decennial census, to be
available until expended, but prohibited the use of any of these funds for activities
related to the use of sampling or any other statistical procedure, including statistical
adjustment, in the population census for the apportionment of the House of
Representatives. A further $281,800,000 would have become available for expenses
in conducting the decennial census only upon the enactment of a law authorizing the
methods for conducting the 2000 decennial population census for the apportionment
of the House of Representatives among the States. These funds would have remained
available until expended. The House report indicated that the committee was
concerned about the controversial sampling techniques contemplated by the Bureau
of the Census, and felt that the Administration should resolve the issues of reliability,
legality and constitutionality surrounding these techniques before the expenditure of
significant additional resources.103
In the House report, additional views were submitted jointly by Representatives104
Alan Mollohan, David Skaggs, and Julian Dixon. They expressed the belief that
the language of the bill represented a ban on sampling, concern about placing such
restrictions on the Bureau of the Census, to which the Administration had expressed
strong opposition, and hope that all concerned would cooperate to assure the most
accurate census. Accordingly, Representative Mollohan submitted Amendment No.
20, which would have struck the reported appropriations provisions for the decennial
census activities, made the entire $381,800,000 available without subsequent
authorization, and provided instead that none of the funds made available in the act
for fiscal year 1998 could be used to make irreversible plans or preparations for the


111 Stat. 158 (1997).101


Bureau of the Census, United States Department of Commerce, Report to Congress—The102
Plan for Census 2000, originally issued July 1997, revised and reissued August 1997.
H.Rept. 105- 207, 105 Cong., 1 Sess. 65 (1997).103thst
Id. at 205.104

use of sampling or any other statistical method, including sampling, in the 2000
decennial population census for the apportionment of the House of Representatives
among the States. Additionally, the amendment would have created the Board of
Observers for a Fair and Accurate Census, which would have observed and
monitored the preparation and implementation of the 2000 decennial census to
determine whether the process was manipulated in any way so as to bias the results
in favor of any geographic region, population group, or political party, or on any
other basis. The Board would have paid special attention to the design and
implementation of any sampling techniques and statistical adjustments used in the
census for the apportionment. Ultimately, H.R. 2267 as passed by the House made
the entire $381, 800,000 available, but created a civil action to challenge the
sampling plans of the Census Bureau. If a civil action were initiated, no funds could
be used for any statistical method until that method was judicially determined to be
authorized by the Constitution and federal statutes.
Section 209 of S. 1022, as reported and passed by the Senate, prohibited the use
of any funds made available in the bill in fiscal year 1998 to make irreversible plans
or preparation for the use of sampling or any other statistical method, including any
statistical adjustment, in taking the 2000 population census for apportionment of the
House of Representatives among the States. S. 1022 provided $354,800,000 for the105
year 2000 Decennial Census Program. The Senate Appropriations Committee
remained concerned about the use of sampling. The Commerce Department
indicated to the committee that the funding that provided by S. 1022 would be
sufficient to cover the cost of the 1998 dress rehearsal for census 2000, with or
without sampling. The Senate struck the text of H.R. 2267 and substituted the text
of S. 1022.
Sections 209 and 210 of the version reported out of conference and enacted106
as part of P.L. 105-119 created the civil action and the Census Monitoring Board,
respectively. Under section 209, any aggrieved person injured by the use of sampling
or any other statistical method in the decennial census for apportionment could bring
suit before a three-judge panel of a district court. The use of any statistical method
included use as part of a dress rehearsal or other simulation of a census in preparation
for the use of such method, in a decennial census for apportionment or redistricting
of Members in Congress. Section 209 further provides that the report ordered by title
VIII of P.L. 105-18 and the Census 2000 Operational Plan are deemed to constitute
final agency action regarding the use of statistical methods in the 2000 decennial
census, thus making the question of their use in such census sufficiently concrete and
final to be reviewable immediately in a judicial proceeding. An aggrieved person
includes (1) any resident of a State whose congressional representation or district
could be changed as a result of the use of a statistical method challenged in the civil
action; (2) any Representative or Senator in Congress; and (3) either House of
Congress. The decision of the district court panel is directly appealable to the U.S.
Supreme Court. Any agency with authority to carry out the decennial census may
seek a declaratory judgment in a civil action concerning the constitutionality and
legality of the use of a statistical method in the decennial census for apportionment.


S.Rept. 105-48, 105 Cong., 1 Sess. 63 (1997).105thst
H.Rept. 105-405, 105 Cong., 1 Sess. 43-46 (1997).106thst

The two lawsuits brought pursuant to section 209 were noted above. Section 210
establishes a bipartisan Census Monitoring Board whose function is to observe and
monitor all aspects of the preparation and implementation of the 2000 decennial
census. This Board is currently constituted and carrying out its functions.
P.L. 105-277, the Omnibus Consolidated Emergency and Supplemental
Appropriations Act for FY1999, included the conference agreement for Commerce,
Justice, State and Judiciary appropriations for fiscal year 1999, which contained
$1,026,963,000 in funding for decennial census programs and $4,000,000 in funding
for the Census Monitoring Board created by P.L. 105-119. However, the Act107
provides for funding only through June 15, 1999. H.R. 4276 as reported by the108
House Committee on Appropriations and as passed would have provided
$951,936,000 for decennial census programs, $103,433,000 above the budget request
and $566,049,000 above the amount for fiscal year 1998. However, of this109
amount, $475,968,000 would only have been available immediately for operations
through March 31, 1999. An additional $475, 968,000 would have been released
only after March 31, 1999, and only after the President requested such release and
submitted, by March 15, 1999, revised cost estimates for the completion of the 2000
census, and after a further enactment authorizing the release of such funds. The
Census Monitoring Board would have been funded with $4,000,000. An amendment
offered by Representative Mollohan, which would have struck the limitation
reserving half the funds until a further enactment releasing them and would have used
the National Academy of Sciences to resolve statistical methodology issues, was110
defeated. S. 2260, as reported by the Senate Committee on Appropriations and
passed, would have provided the requested level of $848,503,000 without111
restriction. The Senate vitiated passage of S. 2260 and then passed H.R. 4276 after
amending it by inserting the text of S. 2260. The House disagreed with the Senate
amendments and conferees were appointed. Ultimately, the Commerce, Justice and
State appropriations were rolled into H.R. 4328 and became part of the Omnibus
Consolidated Emergency and Supplemental Appropriations Act for FY1999.
B. Ramifications
If either legislation or a judicial decision ultimately results in the clear
prohibition or restriction of sampling and other statistical procedures to adjust the

2000 decennial population census, the issue will be settled with regard to


Omnibus Consolidated and Emergency Supplemental Appropriations Act for FY1999,107
Pub. L. No. 105-277, Division A, § 101(b), title II, 112 Stat. ___ (1998) (reprinted at 144
Cong. Rec. H11317-318 (daily edition Oct. 19, 1998)). See also H.R. Conf. Rep. No. 825,th

105 Cong., 2d Sess. (1998).


Omnibus Consolidated and Emergency Supplemental Appropriations Act for FY1999,108
Pub. L. No. 105-277, Division A, § 101(b), title VI, § 626, 112 Stat. ___ (1998) (reprinted
at 144 Cong. Rec. H11334 (daily edition Oct. 19, 1998)). See also H.R. Conf. Rep. No. 825,th

105 Cong., 2d Sess. (1998).


H.Rept. 105-636, 105 Cong., 2d Sess. 70-72 (1998).109th

144 Cong. Rec. H7185, H7213 (Daily Ed. Aug. 5, 1998).110


S.Rept. 105-235, 105 Cong., 2d Sess 78-80 (1998).111th

apportionment of Representatives among the States. However, some commentators
have suggested that litigation may move to the states, with lawsuits challenging state
redistricting plans based on unadjusted census data. These commentators argue112
that the States have an obligation to use the most accurate data available and thus
there may arguably be an obligation on the part of the States to use adjusted data,
even to the point of using “available” methodology to adjust the figures for a State,
even if the Bureau of the Census does not do so.
References
CRS Report 98-321, Census 2000: Sampling as an Appropriations Issue in the 105th
Congress, by Jennifer D. Williams.
CRS Report 97-137, Census 2000: The Sampling Debate, by Jennifer D. Williams.
CRS Report 98-209, Appropriations for FY1999: Commerce, Justice, and State, the
Judiciary, and Related Agencies, coordinated by Edward Knight.


Samuel Issacharoff and Allan J. Lichtman, The Census Undercount and Minority112
Representation: The Constitutional Obligation of the States to Guarantee Equal
Representation, 13 Rev. Litig. 1 (1993); Christopher M. Taylor, Note: Vote Dilution and the
Census Undercount: A State-by-State Remedy, 94 Mich. L. Rev. 1098 (1996). Both cite
Burns v. Richardson, 384 U.S. 73 (1966), for the proposition that States are not required to
use census data in redistricting and Kirkpatrick v. Preisler, 394 U.S. 526 (1969), for the
proposition that States may use data other than the census data where other data provide a
demonstrably more accurate basis for redistricting.