Amicus Curiae Brief — Department of Commerce v. United States House of Representatives

; @ | FILED |

No. 98-404 NOV 3~ 1998

Hite oF THE CLERK

In The SUPREME COURT, U.S.

Supreme Court of the Anited States

October Term, 1998

UNITED STATES DEPARTMENT OF COMMERCE, et al.

Appellants,

v.

UNITED STATES HOUSE OF REPRESENTATIVES, et al.

Appellees.

On Appeal from the United States District Court

For the District of Columbia

BRIEF OF NATIONAL REPUBLICAN LEGISLATORS

ASSOCIATION,

LOCAL GOVERNMENT COUNCIL,

DR. ALAN HESLOP

AS AMICI CURIAE IN SUPPORT OF APPELLEES

E. MARK BRADEN

(Counsel of Record)

SUVARNA U. RAJGURU

BAKER & HOSTETLER LLP

1050 Connecticut Avenue, N.W.

Washington, DC 20036

(202) 861-1500

CLARK BENSEN

Lake Ridge, Virginia 22192

(703) 690-4066

ii

TABLE OF CONTENTS

INTERESTS OF AMICI CURIAE 1

THE NATIONAL REPUBLICAN LEGISLATORS ASSOCIATION vossoseossss.. 1

THE LOCAL GOVERNMENT COUNCII I

Dr. ALAN HESLOP 2

SUMMARY OF THE ARGUMENT 2

I. THE FRAMERS OF THE CONSTITUTION WERE

AWARE OF THE POTENTIAL FOR MANIPULATION OF

CENSUS COUNTS FOR POLITICAL GAIN AND INSTITUTED

SAFEGUARDS TO ASSURE AN OBJECTIVE

ENUMERATION. 3

Il. THE COMMERCE DEPARTMENT'S PLAN TO

CONDUCT A LIMITED POPULATION COUNT AND TO

STATISTICALLY ADJUST THAT RESULT CREATES THE

POTENTIAL FOR SIGNIFICANT MANIPULATION .............. 10

Ill. THE PROHIBITION ON SAMPLING IN TITLE 13

SECTION 195 ALSO APPLIES TO THE DISTRICTING PART

OF THE APPORTIONMENT PROCESS 13

IV. FOR CONGRESS TO HAVE PROHIBITED THE USE OF

SAMPLING IN STATE-LEVEL APPORTIONMENT, BUT TO

HAVE ALLOWED ITS USE AT THE DISTRICTING LEVEL

WOULD BE IRRATIONAL. 19

Vv. THE COMMERCE DEPARTMENT'S PLAN FOR THE

2000 CENSUS IS SO COMPLEX AND SUBJECT TO

INEVITABLE INADVERTENT ERROR THAT THERE IS A

SUBSTANTIAL LIKELIHOOD THAT LEGALLY SUFFICIENT

CENSUS DATA WILL NOT BE AVAILABLE FOR THE

DRAWING OF REPRESENTATIVE DISTRICTS IN TIME FOR

THE 2001 AND 2002 ELECTIONS 22

VI. STATE AND LOCAL OFFICIALS CHARGED WITH THE

CREATION OF NEW REPRESENTATIVE DISTRICTS WILL

BE UNABLE TO CREATE DISTRICTS WHICH MEET A

GENERATION OF THIS COURT'S EQUIPOPULOUS

REAPPORTIONMENT JURISPRUDENCE WITH THE USE OF

ESTIMATION AND POLLING DATA 24

CONCLUSION .28

TABLE OF AUTHORITIES

CASES

Boyer v. Gardner,

540 F. Supp. 624 (D.N.H. 1982) .........cceessseseereeneee 15

Brown v. Thomson,

462 U.S. B35 (1983) ......0...ccsccorsscrescoccssresrsoreransoorees 25

Buias v. Gill,

316 F. Supp. 1285 (D. Haw. 1970) .........seseesereeees 14

City of New York v. Department of Commerce,

739 F. Supp. 761 (E.D.N.Y., 1990) ......escceceervereeeene 4

Davis v. Bandemer,

478 U.S. 109 (1986) ............scecscsccssesssereercenssensesenees 14

Davis v. Mann,

377 U.S. 678 (1964) ....0...cecccesccrsvecsosseccsecoersocesessesses 14

Dixon v. Hassler,

412 F. Supp. 1036 (WD Tenn.), affirmed, 429 US.

QBS (1DTG) o.occcvensececacossvervevercssevseosorscsnossrsvecsooes 15

Gaffney v. Cummings,

412 US. 73S (1973) ......cceccssssecccsssessescssrsessnensesseosees 14

Garza v. County of Los Angeles,

918 F.2d 763 (9th Cir. 1990), cert. denied, 498 U.S.

RGIS CEDIN). cncncencscsssccoscocerscesecsccorcscoseosencosees 15, 27

Gorin v.

775 F. Supp.1 (D. Wyo. 1991) .......:ccceeseeeennneeennnees 15

Johnson v. Mortham,

926 F. Supp. 1540 (N.D. Fla. 1996) .........00 sesso 14

Karcher v. Daggett,

EE Cs Fe CIPO) nccnsctncivescesecnsndetorsesvoseseecsecs 14, -

Lucas v. Forty-Fourth General Assembly Of Colo,

377 U.S. 713 (1964) ......ccceccesccrcserccesesrssereerersceosenses 14

Maryland Commission for Fair Representation v.

Tawes,

ST Wes ID eacevsecccccesennssocenscososonsonss et NEN 14

Peterson v. Lindner,

765 F.2d 698 (7th Cir. 1985). .......ccccccecceeeeee enenenennees 21

iv

Reynolds v. Sims,

ECT aE Sa oe 14

Roman v. Sincock,

SIT TN TT 14

Shaw v. Reno,

ET Ce 14

Taylor v. McKeithen,

499 F.2d 893 (Sth Cir. 1974) oo... .ccccccccccccscseseececeees 14

Texas Rural Legal Aid, Inc. v. Legal Services Corp.,

940 F.2d 685 (D.C. Cir. 1991) oooccccccccccccceccceseseseeoes 14

United States v. Wilson,

ce 9

WMCA, Inc. v. Lomenzo,

EE 14

Wesberry v. Sanders,

I 14, 24

Westwego Citizens for Better Government v.

Westwego,

906 F.2d 1042 (Sth Cir. 1990) oo.cccccccccccccccceceeceeeees 15

White v. Weiser,

SG ETE PE: a ae 14

STATUTES

NE A RTE Ser eee See 17, 18,

19

RES SSS Se Oe ee 17, 18,

19

eer reas 2, 24

MISCELLANEOUS

Anderson, David L. "When Restraint Requires

Ce a ee 16

Baker, Gordon E. "The Unfinished Reapportionment

og ETE ae Sees 16

Brown, Lawrence D. Statistical Controversies in

RESET ea 23

Bureau of the Census, Historical Statistics of the

United States, 1789-1945

Vv

Bureau of the Census, 1995 Census Test, Census ,

Block Level Data (CDROM) .........----+-+s-e+s+e00 25

Bureau of the Census, A Century of Population

Se ES Ea 6

Bureau of the Census. 1990 Census of and

Housing Profile, Congressional Districts of the

103rd Comgress ..........--.-sesesseeeeneeseeres iolimasonenessce 25

Cain, Bruce E. The Reapportionment Puzzle .......... 16

Congressional Record, 94th Congress ........-.--------++- 18

Cooke, Jacob E., ed., Federalist Papers .............-------. 7

Department of Commerce, Commerce News ......... 26

Dixon, Robert. Democratic i

i in Law and Politics ............... 16

Farrand, Max, ed., Records of the Federal Convention

OE ATED vnacececcececccccceccsessnssetbseccnnsniantasnanesontonss 6,7,12

General Accounting Office, GAO/GCD-97-142,

Progress Made on Design, But Risks Remain

(July 1997) ........scceceseseeessneneneeensnsenensncenssnenenenss 20

Hogan, Howard. The 1990 Post Enumeration Survey,

Operations and Results ...............---:sesssseneneeeseees 22

Johnson, Samuel. A Dictionary of the English

Language (1773) .....-...s-:-ssssssssnesnsesnesnnesnssnennsens 9

Kraeger, Philip. "New Light on Graunt" .................-. 5

McKay, Robert B. Reapportionment: The Law and

Politics of er SSS 16

Mosbacher. Robert A. Statement Of, On Adjustment

Of the 1990 Cemsus. ...........cccecnersenenenenes 10,20,28

Oxford icti Second Edition............. ll

Riche, Martha Farnsworth. Report to Congress, ..... 26

Stigler, Stephen M. The Hi of Statistics — The

Measurement of U Before 1900 .......... 5

Tarrance, V. Lance, Jr. Summary Report to the

Secretary of Commerce, June 14, 1991 ...... 11,27

Thomas, Stephen J. "Lack of Judicial Direction and

Political Gerrymandering” ............-.----0-00+++s+++ 16

Wolfe, A.B. Journal of the American Statistical

Association, Vol. 27 ...........c.-ccsseceeeesseseneesnsennnees 4

The National Republican Legislators Association is a

voluntary group of elected members of state legislatures. Its

chambers includes representatives from _ legislative

across the nation. Its interest is in the fair and

equitable distribution of political authority within the United

States and within the boundaries of each state. Many of its

mauiaied Gast ee Ranting ef

districts. Its redistricting part of the

apportionment process is dependent upon the timely receipt

of accurate data from the Census Bureau.

I -business, free-market, and traditional values public

policies. LGC is associated with over 1,700 individuals from

38 states who share a desire to keep government small and

support Constitutional principles. LGC's interest is that local

government officials have access to timely and accurate

census data, that can be used for redistricting for counties

cities, towns, school districts and other emmmnent

jurisdicti local gov t

. All parties in this matter have consented to the fili of this

Amicus Curae Brief, at evidenced by leters of consent lodged wih th

brief was not authored in whole or in part by any counsel for

a party. No person or entity, other than the amicus curiae, made a

monetary contribution to the preparation or submission of this brief.

2

Dr. Alan Heslop is the Director of the Rose Institute of State

and Local Government at Clarenyont McKenna College. He

is the Rose professor of government at the same institution.

He has written extensively on issues affecting representation

and has also served as a consultant for numerous jurisdictions

regarding the drawing and redrawing of district lines. He is

closely familiar with the ways in which census data are used

in computerized districtings and redistrictings and has

experience in working with congressional representatives,

state legislators and municipal officials in such processes.

Both as a political scientist and as someone concerned

about the future health of local government, Dr. Heslop has

taken a keen interest in proposals for different kinds of

census sampling and enumeration techniques. In this work,

Dr. Heslop has become very familiar with the uses and

misuses of different kinds of census information.

SUMMARY OF THE ARGUMENT

We recognize the Constitutional imperative of the

federal census as a means by which peaceful and periodic

transfer of political power is accomplished in America. The

Framers of the Constitution knew about the temptations to

accumulate and hold political power and provided numerous

institutional safeguards to keep such temptations away from

participants in our governing process. The very real potential

exists that the numbers from the 2000 census for the

apportionment process, including redistricting, may be

subject to manipulation to further partisan goals. We view

with alarm a plan to deliberately not count at least 10% of the

nation’s population and to add virtual persons to, or subtract

real persons from, the census count, as being violative of

Article 1, § 2, U.S. Constitution.

Court is not only the method that the Census Bureau may use

to determine the population data for the division of

the Census Bureau's plan, those responsible for drawing the

I THE FRAMERS OF _ THE

CONSTITUTION WERE AWARE OF THE

POTENTIAL FOR MANIPULATION OF CENSUS

COUNTS FOR POLITICAL GAIN AND INSTITUTED

SAFEGUARDS TO ASSURE AN OBJECTIVE

ENUMERATION.

While we can never know with absolute certitude

what each of the Framers of our Constitution had in mind

when the phrase “actual enumeration” was used, their

probable understanding of the history of census taking

provides the logical starting point for any analysis.

4

Historically, a census had been undertaken by one of

two basic means. First was an effort to catalogue every person,

or at least persons meeting certain criteria, e.g., males over a

certain age. The second general means of determining the size

of the population was to count a small subset of the total

population in order to estimate the size of the overall

population.

The notion of institutionalizing a complete nominal

enumeration,’ or a head count, was not unknown at the time

of the Constitutional Convention. However, it was not

standard practice for many governments. "It is commonly

thought that the idea, as well as the execution, of an

enumerative census was started by the United States in 1790.

The truth is that the idea of a complete enumerative

population census was then by no means novel." Wolfe at

357.

"The earliest medieval record of any importance is the

Domesday Survey made in England at the order of William the

Conqueror in 1083-1086 .” Id. at 361. The "names of property

owners were recorded” in the famous Domesday Book. Id.

: A. B. Wolfe, Journal of the American Statistical Association. Vol.

27: December 1932, 357-369 passim, (hereinafter “Wolfe”). See also City

of New York v. Department of Commerce, 739 F. Supp. 761, 762

(E.D.N.Y., 1990) (discussing how Christianity was “founded in a stable -

thanks to the census”).

’ “A population census, in the correct, and specific sense, is a direct

enumeration, preferably on a set date, and by name of each individual in the

census area. A census so made is both enumerative and nominal.” Wolfe at

357.

5

"Lively interest in quantitative data on population was

evinced during the eighteenth century in France and also in

England. In both countries, suggestions for national

:

Statistical sampling, published the first ion estimate of

London based on statistical analysis.” "In 1753, Thomas

Potter introduced a bill in Parliament providing for a general

enumeration." Wolfe at 368.

; A Frenchman, LaPlace, proposed a method in 1786 of

“taking a precise population, but only at a few given places in

a country."" His method of counting was by enumeration,

"carefully enumerating at a given time, the inhabitants of

several communities." Accuracy and statistical adjustment

were key concerns. LaPlace confirmed that the accuracy of

the ratios used to estimate the overall population would be

"more accurate as the enumeration is more extensive.” Stigler

at 164. This method was used in France, in 1802, but was

abandoned later, being replaced by full counts nationwide,

due in part to the concern about the representativeness of the

sample communities. Stigler at 164.

F Philip Kraeger, "New Light on Graunt,” Population Studies,

Journal of Demography, Vol. 42, No.1, (Mar. 1988) p.129.

. Stephen M. Stigler, The History of Statistics - The Measurement

of Uncertainty before 1900, Belknap Press, Cambridge, Mass., (1986) p.

164 (hereinafter "Stigler”).

6

In America, before the Constitutional Convention,

"several of the states composing the young Republic had

formed the habit of making frequent enumerations of their

inhabitants during their existence as colonies." An

awareness of the difference between different forms of

censuses, statistical estimates and an actual enumeration can

therefore be reasonably assumed.

The method that many American states used prior to

the Philadelphia Convention to determine their population is

unclear. Such censuses were apparently undertaken without

sufficient procedural safeguards, as evidenced by. the

disparities in the numbers available at the Convention.’ The

potential for manipulation of the process by the individual

states was on the minds of several delegates.

The Framers provided for a number of institutional

safeguards against manipulation “of the respective numbers.”

First, the Constitution provided that an institutionally neutral

entity, the Federal Government, undertake the actual

enumeration. Edmund Randolph, of Virginia, observed that

' A Century of Population Growth, Bureau of the Census,

Washington, DC (1909), p.2.

' Several sets of population numbers and contribution quotas were

available at the Convention. Max Farrand, ed., The Records of the

Federal Convention of 1787, Revised Edition in Four Volumes, Yale

University Press, New Haven (1966) (hereinafter “Farrand”). Farrand at

1:190: 1:573-4; 111:253; IV:160-1. Estimates for colonial censuses were

“based on materials ranging from relatively complete ‘umeration... to

fragmentary data...” Historical Statistics of the Unitec__ .tes, 1789-1945,

Bureau of the Census, Washington, DC. (1949) p. 16. The debates reflect

several discussions about the numbers and attempts to alter the initial

representation of particular states. Farrand at 1:561, ':63, Il: 623-24.

7

"the census must be taken under the direction of the general

legislature [Congress]. The states will be too much interested

to take an impartial one for themselves." Farrand at 1:580.

However, this precaution alone did not provide a sufficient

safeguard in the minds of the delegates because "... the

accuracy of the census to be obtained by the Congress will

necessarily depend, in a considerable sn gel on the

disposition, if not on the cooperation of the states ...

To solve this problem of having an interested party

involved in preparing the numbers by which political power

would be apportioned, the Framers’ second step was to link

representation with direct taxation. This coupling formed an

important disincentive to the states against manipulation of

the numbers. As Madison observed, "[w]ere their share of

representation alone to be governed by this rule, they would

have an interest in exaggerating their inhabitants. Were the

rule to decide their share of taxation alone, a contrary

temptation would prevail. By extending the rule to both

objects, the States will have opposite interests which will

control and balance each other and produce the requisite

impartiality."*

As stated by another member of the convention, "by

connecting the interest of the states [representation in the

House) with their duty [payment of direct taxes], the latter

would be sure to be performed.” Farrand at I:197 (emphasis

added). However, since the adoption of the Sixteenth

Amendment in 1913, whatever financial disincentive had

Jacob E. Cooke, ed., The Federalist, Wesleyan University Press,

Middletown, CT (1961) p. 371, (hereinafter "the Federalist").

, The Federalist at 371-2.

existed at the state and local level was lost to the footnotes of

history.

The Framers’ third affirmative step to limit the potential

inhabitants -- from the Framers’ plan. Freed from the

requirements of an actual enumeration or count, the census

could become just a tool to further the political ends of its

designers, the political party that controls the executive

branch.

The Solicitor General contends that "actual

enumeration" is not meant to restrict Congress’ method of

determining the number of persons to an actual count. The

term census rather than “actual enumeration" was used

throughout most of the Convention. The term "actual

enumeration" was introduced to the convention from the

Committee of Style in the final week of the Convention.

The Solicitor General states that “[njo delegate

suggested that the Committee of Style's use of the words ‘actual

enumeration’ was intended to affect the scope of Congress's

authority..." See Appellants’ Brief at 44. There is no record of

the debates for the Committees of Style or Detail. Generally,

the records of the convention are silent on this issue of drafting

history, but the argument in the Solicitor General's brief makes

an unsavory soup of conclusions from the rocks and waters of

silence and a lack of records.

logical to conclude that the term "actual enumeration" was

purposefully used. In light of the different types and forms of

censuses likely known to the Framers, it is equally as logical to

conclude that actual enumeration was meant to mandate an

actual count of inhabitants.'° The interpretation the Solicitor

General proposes would render the words "actual enumeration"

as meaning exactly the same as census.

When the language of the draft Constitution was

changed from the more general term “census” to the more

specific term “actual enumeration," standard interpretative

analysis demands that the change be given meaning.'’ What

is clear from the Federalist Papers and the debates at the

Philadelphia Convention is an overriding concern for a

permanent and precise standard for representation that would

afford objectivity and avoid conjecture.

A contemporaneous definition of the term also supports this

conclusion. The word “enumeration” is defined in Samuel Johnson's

1773 dictionary as "[t] act of numbering or counting over; number told

out.” 1 Samuel Johnson, A Dictionary of the English Language (1773).

- See ¢.g., United States v. Wilson, 503 U.S. 329, 336 (1992)

(noting the “familiar maxim that when Congress alters the words of a

Statute, it must intend to change the statute’s meaning") (Citing Russello

v. United States, 464 U.S. 16, 23-24 (1983)).

10

Il. ©THE COMMERCE DEPARTMENT'S PLAN TO

CONDUCT A LIMITED POPULATION COUNT AND

TO STATISTICALLY ADJUST THAT RESULT

CREATES THE POTENTIAL FOR SIGNIFICANT

MANIPULATION

The Commerce Department’s plan for the 2000

census to deliberately not count 10% of the nation’s

population and to adjust the initial numbers to reflect

expected norms is subject to manipulation for political ends.

This potential for manipulation is as clear in the 1990's as it

was in the 1780's.

The Secretary of Commerce, in his statement

announcing his decision to not adjust the 1990 census results,

pointed out the multiple scenarios which would have been

possible by adjustment of the 1990 census counts:

Consider the results of two possible adjustment

methods that were released by the Census Bureau on

June 13, 1991. The technical differences are small,

but the differences in results are significant. The

apportionment of the House of Representatives under

the selected scheme moved two seats relative to the

apportionment implied by the census, whereas the

modified method moved only one seat.'”

- Statement of Secretary Robert A. Mosbacher on Adjustment of

the 1990 Census, (July 15, 1991. Federal Register, vol 56:140, p.33582,

(hereinafter “Mosbacher”). "One expert found that among five

reasonable alternative methods of calculating adjustments, none of the

resulting apportionments of the House were the same, and eleven

different states either lost or gained a seat in at least one of the five

methods.” Id. at 33582.

Adjustment, being based upon modeling,'? by its very

nature implies that assumptions are made. Assumptions

themselves imply logical deductions, but also subjectivity.

Subjectivity in such a volatile setting, with political stakes so

high, will result in a succumbing to partisan temptation,

absent the repeal of the nature of Man. "What is unsettling,

however, is that the choice of the adjustment method selected

by the Bureau officials can make a difference in

apportionment, and the political outcome of that choice can

be known in advance." Mosbacher at 33582.

In 1991, the Co-Chairman of the Special Advisory

Panel established by the Secretary concluded that:

It is certainly not hard to imagine that such a

process, especially when cloaked in the

mysteries of statistical complexity, could

easily be corrupted and manipulated,

particularly if it should become accepted

practice and not subject to rigorous public

examination....'*

" Modeling has been defined as “using a simplified or idealized

description or conception of a particular system, situation, or process...

that is put forward as a basis for calculations, predictions, or further

investigations. Oxford English Dictionary Second Edition, Clarendon

Press, Oxford (1989) p. 941.

“. V. Lance Tarrance, Jr., Co-Chairman, Special Advisory Panel,

Summary Report to the Secretary of Commerce (June 14, 1991) p. 29

- in the original) (hereinafter "Tarrance"). See also Mosbacher

at .

12

There are numerous different ways to sample and

adjust the census. These different ways will distribute

political power differently between various groups. Each

potential sampling and adjustment design has its own set of

debatable assumptions, but predictable owtcomes. The

appellants have argued that we must trust the impartial

expertise of the Census Bureau and the Department of

Commerce. Yet, the Framers of our Constitution wisely put

their faith in strict structural limits and not in any belief in the

benevolent nature of man. "[F]rom the nature of man we may

be sure, that those who have power in their hands will not give

it up while they can retain it. On the contrary, we know they

will always when they can rather increase it." Farrand at 1:578-

9.

Amici pray for the indulgence of the Court to

consider the operational elements of the Commerce

Department’s plan for the 2000 census. It is important for

this Court to understand exactly what this plan will entail for

the 2000 census and how it differs significantly from the

1990 census and all other prior censuses in U.S. history.

Under the Commerce Department’s plan, the 2000

census will become only a partial census completed through

estimation and adjusted through polling and modeling. The

estimation will attempt to account for the Bureau's

intentional undercount of 10% of the housing units and the

modeling will attempt to adjust for the under counting and

over counting which differs among population subgroups.

As part of a consent decree in litigation before the

1990 census was taken, there was an option before the

Secretary of Commerce either to adjust, or to let stand, the

actual emameration numbers. Following the actual

enumeration, the Secretary carefully reviewed the evidence

13

regarding an adjustment of the 1990 census and, in July 1991

decided not to adjust.

This review option is a critical distinction between the

Department of Commerce's plan for the 2000 census and the

1990 census. For the 2000 plan there will not be even an

attempt at a 100% headcount which might be adjusted

retroactively. The only physical count for 2000 will be up to

90% in each census tract. If the Department of Commerce is

permitted to implement its plan, the 1990 census will have

been the last full census conducted to apportion political

power in our national republic. '*

Ill. THE PROHIBITION ON SAMPLING IN TITLE

13 SECTION 195 ALSO APPLIES TO THE

DISTRICTING PART OF THE APPORTIONMENT

PROCESS

Apportionment is a two-step process involving an

initial allocation of representatives followed by the secondary

construction of districts. Apportionment without districting

is incomplete. This was and still is the widely prevalent

understanding of apportionment. Often, this Court, lower

federal courts, state legislatures, scholars and Congress used

1s

The Solicitor General argues that the House of Representatives’

interpretation of actual enumeration as requiring an actual count "cannot

be reconciled with historical practice" and that each person was not

always individually personally counted. See Appellants’ Brief at 47.

There is one element of these historical practices which distinguishes

them fundamentally from those proposed by Appellants. Aside from an

isolated emergency case after the 1970 census, in each of these non-

individual headcount practices, there has always been some physical,

tangible evidence found to account for each individual number.

14

through the 1950's, 1960's 1970's and 1980's.

The term apportionment has been used in the

inclusive sense by this Court to encompass districting

throughout the development of this Court's reapportionment

jurisprudence. The apportionment revolution was, of course,

a redistricting revolution, starting with Wesberry v. Sanders,

376 U.S. 1 (1964), where this Court invalidated the Georgia

“apportionment” of congressional districts. This Court has

consistently used the ternf apportionment to include

districting, by directly and unequivocally referring to the

drawing of district boundaries as apportionment. While each

of the decisions in the revolution ostensibly ruled on

“apportionment,” in fact, the decisions Gealt with the

districting part of the apportionment process. '*

Also, numerous lower Federal Court cases evidence

that the common usage of the term “reapportionment”

incorporated "redistricting."'’ Federal courts properly have

- See Reynolds v. Sims, 377 U.S. 533 (1964); WMCA, Ine. v.

Lomenzo, 377 U.S. 633 (1964); Maryland Comm'n for Fair

v. Tawes, 377 U.S. 656 (1964), Davis v. Mann, 377 U.S.

678 (1964), Roman v. Sincock, 377 U.S. 695 (1964), Lucas v. Forty-

Fourth Gen. Assembly Of Colo, 377 U.S. 713 (1964). See also White v.

Weiser, 412 U.S. 783, 784 (1973); Gaffney v. Cummings, 412 U.S. 735,

735 (1973); Davis v. Bandemer, 478 U.S. 109, 113 (1986); Karcher v.

Daggett, 462 U.S. 725, 727 (1983); Shaw v. Reno, 509 U.S. 630, 633

(1993).

7 See, e.g., Bums v. Gill, 316 F. Supp 1285, 1293 (D. Haw. 1970)

(approving of a “two-tier apportionment plan” whereby all representatives

thereafter district lines would be drawn within the islands themselves):

Taylor v. McKeithen, 499 F.2d 893, 910-911 (S* Cir. 1974) ("{nJot the

(continue)

15

not distinguished apportionment and districting and in fact,

have historically subsumed redistricting within the overall

process of apportionment. '*

Another way to understand the analogous usage of

these terms is to view them from the perspective of state

legislators. Today 47 states use some form of the term

“apportionment” in their statutes that deal with the drawing

of representative districts.'? Of course, none of these states

(continued)

least of the problems that would be created by representative

apportionment is that the cartographer...cannot draw his meandering

ES aa akal Aid, Inc. v_ Legal Servs. Corp., 940 F.2d 685,

688 (D.C. Cir. 1991) (citing regulation that defined redistricting as “any

effort, directly or indirectly, to participate in the revision or

reapportionment of a legislative, judicial or elective district at any level

of government, including the timing or manner of the taking of a

census”); Johnson v. Mortham, 926 F. Supp. 1540, 1542 (N_D. Fla. 1996)

(finding only a “technical difference” between reapportionment and

redistricting).

as “Fundamentally, redistricting versus reapportionment is a

distinction without a difference,” Gorin v. Karpan, 775 F. Supp.1 430,

1443 (D. Wyo. 1991). “Congressional Districts were

Dixon v. Hassler, 412 F. Supp. 1036, 1038 (WD Tenn.), affirmed, 429

US 934 (1976); a a a ee Se. of

Representatives,” Boyer v. Gardner, 540 F. _ Supp. 624, 625 (D.N.H.

1982); “reapportionment between censuses...” peste ego Cauzeny te

Better Gov't v. Westwego, 906 F.2d 1042, 1045-46 ( Cir. 1990); “the

County has adopted its current reapportionment plan...” stating that

fragmentation has been a “goal of each redistricting scheme since 1959”,

Garza v. County of Los Angeles, 918 F.2d 763, 773 (9 Cir. 1990), cert.

denied, 498 U.S. 1028 (1991).

7 Only Illinois, Kentucky and Washington use the term

redistricting to the exclusion of apportionment.

16

actually apportion representatives between governmental

units. They construct representational districts. State

legislatures continue to use the terms interchangeably. The

National Conference of State Legislatures (NCSL) has a

subgroup of legislators interested in these issues, that only

two years ago changed its name from the Reapportionment

Task Force to the Redistricting Task Force.

A listing of the most preeminent scholars with their

works shows the use of the term apportionment

interchangeably with, or to include, districting. Probably the

most generally recognized work in this area is Democratic

in Law and Politics by

Robert G. Dixon, Jr. (hereinafter Dixon’) Clearly Dixon's

view is that redistricting is a subset of reapportionment.”°

20 See Robert G. Dixon, Jr, Democratic Raneponseten:

Reapportionment in Law and Politics, Oxford University ew

York, (1968) p. 3 (noting “reapportionment revolution”). Robert B.

McKay, Reapportionment: The Law and Politics of Equal Representatior..

Twentieth Century Fund, New York (1965) p. 6 (“In short, it appears thi

the apportionment and districting functions in the state legislative proc ss

will be merged into a single task to be performed by each legislature. ").

Bruce E. Cain, The Reapportionment Puzzle, University of California

Press, Berkeley (1984) p.xi; (noting that [rjeapportionment is a murky

and dimly perceived process for most political observers”). (“The district

lines drawn by reapportionment. ..”) Id. at 1.; Stephen J. Thomas, The

Lack of Judicial Direction in Political Gerrymandering: An Invitation to

Chaos Following the 1990 Census;” Hastings Law Journal, Vol.40:1067

(July 1989) p. 1067 (“Gerrymandering may be accomplished through a

variety of methods, including malapportionment."), Gordon E. Baker.

“The Unfinished Reapportionment Revolution,” Politica!

and the Courts. Bernard Grofman, ed., Agathon m Press.

Gerrymandering |

New York (1990) p. 25 (“the problem of malapportionment in the form of

vast population disparities among districts.") David L. Anderson, “When

Dems RS Se ee ee

continue

17

The congressional understanding oof the

interchangeability of apportionment and redistricting is

reflected in the passage of P.L.94-171, which provided for

the distribution of block-level population data for districting,

13 U.S.C. §141 was amended in 1975 to include “legislative

apportionment” as part of the section heading.

Title 13 U.S.C §195 not only prohibits sampling for

apportionment, but redistricting as well. The addition of

specific redistricting language in 13 U.S.C. §141(e)(2) does

not affect this analysis. The Census Act was amended in

1976 in substantial part to allow for a mid-decade census to

provide timely statistical information for allocation of

funding. Congress would have perceived no need to amend

13 U.S.C. §195 to expressly expand the prohibition on

"sampling" data for apportionment, because there can be

little doubt that Congress in 1976 would have understood 13

U.S.C. §195 to already prohibit “sampling” data in

When analyzed in the context of this general

understanding of apportionment, it is reasonable to view the

13 U.S.C. §195 prohibition on the use of sampling for the

purposes of apportionment, as applying with equal force to

the apportioning of representatives among the states

(apportionment) and to the revision of representation districts

(redistricting).

(continued)

Quo Ante,” Stanford Law Review, Vol. 42:1549 (July 1990) (noting “that

the Court should withdraw from reapportionment reform”).

Section 13 U.S.C. §195 prohibits data from sampling

for congressional apportionment. As the above review

indicates, this use of the term apportionment most logically

extends to redistricting as well. Congress intended sampled

data to be used in only two limited circumstances: (1) to

allow the so-called "long form" data to be collected from a

sample of the entire population; and (2) to allow the use of

sampling for a mid-decade census to provide more current

data for federal aid distribution in between censuses. 13

U.S.C. §195 was amended to strengthen the intent of

Congress to use sampling for supplementary data. 13 U.S.C.

§141 was amended to allay fears that any more recent data,

no matter what the level of accuracy, wheat gyda

challenge any apportionment or districting scheme

together, it is clear that Congress intended that sampled data

be used in neither step of the apportionment process.

- Congressional Record, 94" Congress (April 7, 1976) p. 9786-

9795, passim.

19

The silence of 13 U.S.C. §195 with respect to the

term redistricting reflects the interchangeability of the terms

at the time. Indeed, this interconnection explains the silence.

While the concepts of reapportionment and redistricting

share a common base of population data, the timeframes are

distinct. 13 U.S.C. §141(e)(2) addresses the time element of

census, based upon sampling, could not be used either to

invalidate the current apportionment or to enable challenges

to current redistricting plans.

IV. FOR CONGRESS TO HAVE PROHIBITED

THE USE OF SAMPLING IN’ STATE-LEVEL

APPORTIONMENT, BUT TO HAVE ALLOWED ITS

USE AT THE DISTRICTING LEVEL WOULD BE

IRRATIONAL.

Whatever impact adjustment would have on the

apportionment of political representation among states, the

impact on the apportionment process within states would be

much larger. The reasons behind this statement require only

the most basic understanding of sampling or polling as a

Many national surveys or polls commonly note that

the reported percentages are plus or minus 3-4% for national

analysis. However, estimates of sampling error for lower

levels of geography will be much higher. Whereas several

thousand respondents may be representative of the nation as

a whole, to provide a representative cross-section of even a

state in a national survey requires many more respondents.

20

As a basic statistical principle, the relative degree of

sampling/polling error increases as the size of the population

universe decreases.”

The use of sampling to improve the accuracy of the

census, if valid at all, can only be true for the highest level of

geography. As one moves down the geographic hierarchy,

the numbers become less accurate. There is basically an

inverse relationship between the population size of a unit

(state, county, city or block) and the accuracy of the data

from sampling.” It was exactly this concern which was a

basis for the Secretary’s concern for the ‘distributive

accuracy” rather than ‘numerical accuracy’ in his decision to

not adjust the 1990 Census. Confidence that any

sampling/poll will accurately represent the actual numbers,

which would have been obtained by a full traditional count,

at the block level disappears.”°

- For example, an error of one person out of 100 persons is a 1%

error. An error of one person out of 10 persons is a 10% error.

- Note that the Bureau's plan involves the use of sampling in two

phases, the Non-Response Follow-Up (NRFU) and the Integrated

Coverage Measurement (ICM). While the ICM phase is designed to

improve accuracy, the Bureau does not claim that accuracy is a concern

for NRFU, which is primarily “to save time and money.” General

Accounting Office, GAO/GCD-97-142, Progress Made on Design, But

Risks Remain (July 1997) p. 44, (hereinafter “GAO”). This element of

the plan for 2000 ignores the fact that the most accurate data are obtained

directly from respondents, which can be only collected through the

traditional means of personal contact. GAO at 26.

” Mosbacher at 33584.

= The Secretary expressed a further concern about adjustment for

small areas. “As the population units get smaller, including small and

(continue)

21

A judicial rule of statutory construction is that it is

presumed that a legislative body enacts rational legislation.”°

Given the disproportionally larger impact on redistricting

within a state, it would be absurd to prohibit the use of

sampling for apportioning seats between the states, but to

allow it for representative redistricting. The only rational

analysis is that if sampling/polling cannot be legally used for

the division of congressional seats between states, it cannot

be used for the subset of that process, representative

redistricting within a state, where its problems and their

effects are indisputably larger.

(continued)

medium sized cities, the adjusted figures become _increasi

unreliable.” Mosbasher at 33583. ingly

26 :

wi hiateee Tere ae nib attmnat to 00 eeteasl,T’ Dusen

v. Lindner, 765 F.2d 698, 705 (7th Cir. 1985) (citing Hodel v. Indi

452 US. 314, 322.(1980)). ) (citing Hodel v. Indiana,

22

V. THE COMMERCE DEPARTMENT'S PLAN

FOR THE 2000 CENSUS IS SO COMPLEX AND

SUBJECT TO INEVITABLE INADVERTENT ERROR

THAT THERE IS A SUBSTANTIAL LIKELIHOOD

THAT LEGALLY SUFFICIENT CENSUS DATA WILL

NOT BE AVAILABLE FOR THE DRAWING OF

REPRESENTATIVE DISTRICTS IN TIME FOR THE

2001 AND 2002 ELECTIONS

Experience provides evidence of this problem.

Review the situation following the 1990 census. The Bureau

given the | of time to review its 1990 post enumeration

survey (PES)’’ revised the national undercount rate

downward in July 1992. “As a result of an error in computer

processing, the estimated national overcount rate of 2.1%

was overstated by 0.4%. ...[a]fter making other refinements

and corrections, the national undercount is now estimated to

be about 1.6%.” Report of the Committee on Adjustment of

Postcensal Estimates, Bureau of the Census, Aug. 1992, p.

15. This overstatement revolved around coding errors for

only 2,000 households in the 1990 post enumeration survey.

However, the impact of this error, from these few

households, was enormous.”* It affected the estimate of the

national undercount by about a million persons.

This inadvertent error, which reduced the national

undercount rate from an estimated 2.1% to 1.6%, would have

changed the apportionment of congressional seats affecting

” The PES was a large poll of 165,000 housing units taken after

the 1990 Census designed to measure the coverage of the census.

* Howard Hogan, “The 1990 Post-Enumeration Survey

Operations and Results.” Journal of The American Statistica!

Association, Vol. 88:423 (September 1993) p. 1047, 1054.

23

several states. For the reasons mentioned above, the effect of

this error would have been greatly magnified in the

redistricting part of the apportionment process.

The modeling schemes proposed by appellants for the

2000 plan will result in each housing unit in the nationwide

survey having an impact far beyond its own infinitesimal

Share of 1 out of 100+ million housing units. Under a

traditional count, the total population for any level is merely

the aggregate of the data from all housing units. Under

appellants’ plan, the totals for every jurisdiction are impacted

by the representativeness of the sample used in the poll. Bad

samples produce flawed numbers.

Considering the complexity of the appellants’ plan,

with its nationwide survey component much larger than

1990, it is reasonable to conclude that the Bureau can not

implement this Rube Goldbergesque machine in a timely

manner without some significant errors creeping into the

process. As one prominent statistician recently observed,

"[t}he results of adjustment are highly dependent on somewhat

arbitrary technical decisions. Furthermore, mistakes are almost

inevitable, very hard to detect, and have profound

consequences." Lawrence D. Brown, et al., Statistical

Controversies in Census 2000, Technical Report, Department

of Statistics, U.C. Berkeley (October 1998) p. 18.

24

Vi. STATE AND LOCAL OFFICIALS CHARGED

WITH THE CREATION OF NEW REPRESENTATIVE

DISTRICTS WILL BE UNABLE TO CREATE

DISTRICTS WHICH MEET A GENERATION OF

THIS COURT'S EQUIPOPULOUS

REAPPORTIONMENT JURISPRUDENCE WITH THE

USE OF ESTIMATION AND POLLING DATA

Both the estimation and modeling phases will be

based upon sampling, or to use the more colloquial term -

polling techniques. These will have known error rates.

Sampling or polling error results from the fact that only a

part of the entire population is asked questions. The principal

issue for any poll is whether the sample chosen is an

appropriate sample that is representative of the entire

population.

The use of polling and estimation in the census results

in a measurable error. The error reflects the fact that the

sample which was used was only one of many possible

samples and may not have been representative of the entire

universe of persons. This error measurement quantifies the

degree of difference between the numbers generated by a

polling technique versus numbers generated by a complete

count. The smaller the number of persons in the sample or

poll, the larger the potential for a greater percentage of error.

Herein lies a problem with respect to the drawing of

district boundaries and equal representation. This Court has

repeatedly made clear that in order to comply with Article 1,

§ 2, U.S. Constitution states must draw congressional

districts as nearly equal in population “as is practicable.”

as Wesberry, 376 U.S. at 8.

(continue)

25

This Court and lower federal courts have invalidated many

congressional redistricting plans with relatively small

population variances. See e.g., Karcher v. Daggett, 462 U.S.

725, 742 (1983) (striking down a New Jersey plan with a

population deviation of 0.6984%). Following the 1990

census, 10 states enacted congressional apportionment plans

with a population deviation of 0 or 1 person’ and although

this Court has generally afforded states greater latitude in

creating state legislative districts, states re still

constitutionally required to make “an honest and good faith

effort” to create population equality among districts. Brown

v. Thomson, 462 U.S. 835, 842 (1983). Given this Court's

limited tolerance for population deviation, if the data used to

construct the districts have a known error rate larger than

permitted population deviations, how is the line drawer to

proceed?

The results of the 1995 Test Census held in Oakland,

California are illustrative of the problem. For a small census

block in Oakland, a reported estimate of 100 persons would

mean that a full count would have found between 72 and 128

persons, an error due to sampling of 28%.*! What is the

population figure to be used for city council redistricting?

(continued)

* Another six states only missed this status by a very few persons.

1990 Census Population and Housing Profile: Congressional Districts of

the 103rd Congress, Bureau of the Census, Washington, D.C., CPH-

L117, (n.d., 1993) p. 1-8.

This estimate reflects a 95% confidence level. 1995 Census

Test, Census Block Level Data, CDRom, Bureau of the Census,

Washington, DC, CD95-CENTEST (August 1996) passim.

26

For its test Census, the Bureau calculated the average

error for the census blocks as follows: 18.3% Paterson, New

Jersey, 12.6% for Oakland, and 25.2% for selected Parishes

in Northwestern Louisiana. These error rates were termed

"quite substantial" by the Bureau. Martha Farnsworth Riche,

Report to Congress, The Plan for Census 2000, revised

August 1997, p. 45-8, (hereinafter “Riche”).

The relevant level for error analysis is not just the

completed congressional district, as this Bureau now

maintains.** If one is crafting local representation districts,

each census block's error must be considered.*? The census

block is the building block for most representative districts

around the nation. It is also the lowest common denominator

for the entire census. If the lowest level data are suspect,

equal population local representation districts cannot be

drawn with this data.** Local officials will be faced with the

use census block data for their city, town or township, which

they may know, from direct personal observation, are wrong.

as Riche at 46-7.

“ "Margins of error are critical for small areas.” Commerce News,

CB91-214 (June 5, 1991). Adjustment would be implemented et the

block level.

ne These error rates may be compounded in individual

representative districts. The error rate in some blocks may reflect an

undercount, in others an overcount. These errors might “balance out”

nationally if the underlying assumptions and the execution are correct.

They will not “balance out” in representative districts. Redistricting is

not a random process. Blocks are frequently specifically assigned to a

district based upon demographic characteristics. Districts are typically

small enough and sufficiently homogenous that they may well be

composed substantially of blocks with the same kind of error, magnifying

that many users of census daia rely on data for small

geographic areas, the report stated that this would be "where

the ship called accuracy or ‘closer to the truth’ will founder on

the shoals of reality.” Id.

It takes no active imagination to foresee the numerous

pieces of litigation that will be spawned by using sampling in

the redistricting process. How is a line drawer to reconcile

these significant error rates? What data should any court use

when reviewing future redistricting cases?’’ The appellants’

plan for 2000 will aggravate uncertainty in the already

complex and litigious apportionment process. The timeframe

under which new districting plans must be enacted is

extremely tight in several states holding fall 2001 legislative

elections. These plans must be finalized a few short months

“s While equal representation has frequently been measured by

overall population deviation among districts, alternative perspectives are

possible. See Garza v. County of Los Angeles, 918 F.2d 763, 778 (1990).

Judge Kocinski concurring and dissenting.

28

after the block-level data first become available in April of

2001 and include states which must submit plans for pre-

clearance under Section 5 of the Voting Rights Act.

An “actual enumeration,” based upon physical

evidence of the existence of persons, and not conjectural

estimates,** provides the best available data for the creation

of representational districts. Méillions of persons will be

counted as some fraction above or below a whole person.”’

CONCLUSION

Amici share the concern of the Framers that partisan

manipulation of the census should be avoided. The federal

census is the largest participatory event of our American

government. The goal should be to encourage every

American to be counted, not discounted or left uncounted.

Amici pray this Court in the interest of judicial

economy, as suggested by the Solicitor General, to resolve

the Constitutional issues with respect to the “actual

enumeration" of the "respective numbers." If the present

litigation is resolved solely on the grounds of statutory

construction and is limited solely to the apportionment of

congressional seats between the states, there will be even

more litigation following the 2000 census with respect to the

redistricting part of the reapportionment process. Clearly,

” This Court has held that “if a state does attempt to use a measure

other than total population or to “correct” the census figures, it may not

do so in a haphazard, inconsistent, or conjectural manner.” Karcher 462

U.S. at 732, n. 4 (citing Kirkpatrick v. Preisler, 394 U.S. 526, 534-35

(1969)).

» Mosbacher at 33636.

November 3, 1998