V�

LEXSEE

09WarningAs of: Jun 02 , 2011

THE CITY OF NEW YORK, THE STATE OF NEW YORK, THE PEOPLE OFTHE STATE OF CALIFORNIA EX REL. DANIEL E. LUNGREN, ATTORNEYGENERAL, THE CITY OF LOS ANGELES, THE CITY OF CHICAGO, DADE

COUNTY, FLORIDA, THE U.S. CONFERENCE OF MAYORS, THE NATIONALLEAGUE OF CITIES, THE LEAGUE OF UNITED LATIN AMERICAN CITI-ZENS, THE NATIONAL ASSOCIATION FOR THE ADVANCEMENT OF COL-ORED PEOPLE, MARCELLA MAXWELL, DONALD H. ELLIOTT, JOHNMACK, OLGA MORALES, TIMOTHY W . WRIGHT III , RAYMOND G. RO-MERO, ANTONIO GONZALES, and ATHALIE RANGE, P l aintiffs, an d THE

STATE OF TEXAS, THE CITY OF PHOENIX, ARIZONA, THE STATE OF NEWJERSEY, THE STATE OF FLORIDA, THE C ITY OF CLEVELAND, OHIO, THECITY OF DENVER, COLORADO, THE CITY OF INGLEWOOD, CALIFORNIA,

THE CITY OF NEW ORLEANS, LOUISIANA, THE CITY OF OAKLAND,CALIFORNIA, THE CITY OF PASADENA, CALIFORNIA, THE CITY OF

PHILADELPHIA, PENNSYLVANIA, THE C ITY OF SAN ANTONIO, TEXAS,THE CITY OF SAN FRANCISCO, CALIFORNIA, BROWARD COUNTY, FLOR-IDA, THE STATE OF ARIZONA, THE C ITY OF BALTIMORE, MARYLAND,THE CITY OF BOSTON, MASSACHUSETTS, THE CITY OF LONG BEACH,CALIFORNIA, THE CITY OF SAN JOSE, CALIFORNIA, LOS ANGELES

COUNTY, CALIFORNIA, SAN BERNARDINO COUNTY, CALIFORNIA, THEDISTRICT OF COLUMBIA, THE NAVAJO NATION, THE STATE OF NEWMEXICO, THE CITY OF TUCSON, ARIZONA, THE COUNTY OF HUDSON,NEW JERSEY a n d, THE COUNCIL OF THE GREAT CITY SCHOOLS, Pla i ntiff-Intervenors, v. UNITED STATES DEPARTMENT OF COMMERCE, RONALD H.BROWN, as Secretary of the United States Department of Commerce, MICHAELR. DARBY , as Under Secretary for Econom i c Affairs of the U nited S ta tes Depa rt-men t of Commerce, BUREAU OF THE CENSUS, BARBARA EVERITT BRYANT,as Director of the Bureau of the Census, WILLIAM CLINTON, as P reside n t of theUn ited States, and DONALD K. ANDERSON, as Clerk of the U nited States House ofRepresentatives, Defendants, and THE STATE OF WISCONSIN, and THE STATEOF OKLAHOMA, Defendants- Intervenors. CITY OF ATLANTA, and MAYNARDJACKSON, I ndividually and as Mayor, City of Atlanta, P l a i ntiffs, v. RONALD H.BROWN, as Secretary of United States Depa rtment of Com merce, BUREAU OFTHE CENSUS, and BARBARA EVERITT BRYANT, as Director of t h e Bureau ofthe Census, Defen dants. FLORIDA HOUSE OF REPRESENTATIVES, FLORIDA

STATE CONFERENCE, THE NATIONAL ASSOCIATION FOR THE AD-VANCEMENT OF COLORED PEOPLE, MIGUEL A. DE GRANDY, WTLLYE

DENNIS, MARIO DIAZ-BALART, Dr. CHARLES EVANS, RODOLFO GARCIA,JR., BOLLEY L. " BO" JOHNSON, ALFRED J. LAWSON, JR, WILLIS LOGAN,JR., JOHNNIE MCMILLIAN, ALZO J. REDDICK, PETER RUDY WALLACE,

T.K. WETHERELL, Plaintiffs, v. RONALD H. BROWN, as Secretary of the UnitedStates Department of Commerce, MICHAEL ESPY, as Secretary of Agricu l ture,DONNA E. SHALALA, as Secretary of Health and Human Services, HENRY

C ISNEROS, as Secretary of H ousing and Urban Development, ROBERT B. REICH,as Secretary of Labor, FREDERICO PENA, as Secretary of Transportation, RICH-ARD W. RILEY, as Secretary of Education, and M I CHAEL R. DARBY, as Under

Page 1

822 F. Supp . 9 0 6 , *; 1993 U . S . Di st. 1 ,EXIS 4818, **

Sec r e t ary fo r Economic Affairs of th e United States De partm e nt of Commerce, De-fend a nt s.

88 CV 3474,92 CV 1566,92 CV 2037

UNITED STATES DISTRICT COURT FOR THE EASTERN DISTRICT OF NEWYORK

822 F. Sa pp. 906 ; 1993 U. S . Di s t . LEXIS 4818

Apri l 13, 1993, Dec i dedApri l 1 3, 1 993, Fi l ed

CASE SUMMARY:

PROCEDURAL POSTURE: Plaintiffs, states , cities ,citizens groups and individual citizens, filed an actionagainst defendant federal agencies and officials, andsought a judgment vacating a decision that a censuswould not be adjusted and ordering the adjustment to bemade, and for a judgment allowing them access to cer-tain census information .

OVERVIEW: The Department of Commerce, whichwas respons ible for the 1990 census , was aware that mi -nority groups were undercounted at a higher rate thanwhites . A system was developed to correct the under-counting, but the Secretary of Commerce decided not toadjust the census . In addition, the Department refused torelease certain census information . Plaintiffs fi led anaction to reverse the Secretary's decision and for accessto the withheld information . Judgment was rendered infavor of plaintiffs as to the access to information issue ,and in favor of defendants as to the adjustment issue . TheSecretary's repo rt of the final cens us counts was a finalagency action , and was subject to review for arbitrarinessand capriciousness pursuant to Administrative ProcedureAct $ 706 . U . S . Const . art . 1 , § 2 required an accuratecensus and did not bar adjustment . The Secretary' s de-terminations that the adjusted census count would be nomore accurate than the original and was not constant andcomplete were based on some legitimate concern s andwere not arbitrary and capricious . Because half of thecontested information had been released to Congress , thedeliberative process privile ge no longer applied.

OUTCOME: The cou rt found in favor of plaintiffs as totheir demand for the re lease of information , and found infavor of defendants as to the censu s adjustment issue .

i mputation , poststratum, household , arbitrary and capri-cious, pre- speci fied, practicable , smoothing , counted ,local level , statistically, capricious, variance , homogene-ity, raw , pre - speci fication, deliberative

LexisNexis(R) Headnotes

Constitutional Law > Congressional Duties & Powers >Census > Census & EnumerationGovernments > Federal Government > U.S. Congress[HNl]The Constitution requires a decennial census. U . S.Const . art. I , § 2 , states that the actual enumeration shallbe made every ten years in such manner as the Congressshall by law direct . Congress has, in turn, delegated tothe Secretary of Commerce the duty of taking the censusin such form and content as he may determine , includingthe u s e of sampling procedures and special surveys. 13U.S . C.S . & 141(a). The Bureau of the Census , an agencywithin the Department of Commerce, actually conductsthe census .

Civil Rights Law > Voting Rig/:ts > EnforcementGovernments > Local Governments > Elections[HN2]See 42 U.S.C.S. § 1973a.

Civil Rights Law > Contractual Relations & Housing >Property Rights (sec. 1982) > General OverviewCivil Rights Luw > Voting /t igh ts > Enforcemen tGovernments > Local Governments > Claims By &Against[HN3]The Voting Rights Act, 42 U.S.C.S. § 1973, etseq. , applies only to misconduct by states or their politi-cal subdivisions.

CORE TERMS: censu s, guideline , adj u sted , under-count, adj u st , statistical, accuracy , enume ration , e s timate,b l ock, red istr i ctin g, tap es, apport ionme nt , differenti a l ,

Administrative Law > Judicial Review > Reviewability> Final Order Requirement

Page 2

8 22 F . Supp . 906 , *; 1993 U . S . Di st . LEXIS 4818 , **

[HN4]The Secret ary o f Commerce' s repo rt in g of the c en-sus counts i s a final agency act i on for purposes of anAdministrative Practi c e Act review .

Constitutional Law > Congressional Duties & Powers >Census > Census & Enumeration[HN S ]In making decision s regarding the census, the Sec-retuy of Commerce's interpretation of U.S . Const . art. 1,§ 2 must be cons istent with the con stitutional languageand the constitutional goal of equal representation. Be -cause art. I , § 2 re quires the census to be as accurate aspracticable, the Constitution is not a bar to statisticaladjustment.

the arbitrary and capriciou s s tandard of review . Thejudge , even when he is free, is st i ll no t wholly fre e. H e i snot to innovate at pleasure. He is not a knight-errant ,roaming at will in pursuit of hi s own id ea l oFbeauty or ofgoodness . He is to draw his insp i ration from consecratedprinciples.

Civil Procedure > Judgments > Relief From Judgment> General OverviewConstitutional Law > Congressional Duties & Poivers >Census > General Overview[IIIQ9]13 U.S.C.S. & 141(c) requires the Secretary ofCommerce to provide the states with data to be used inredistricting.

Consttartionrel Law > Congressional Duties & Powers >Census > General Overview[HN6]Although common sense supports a test requiringa good faith effo rt to achieve precise mathematical qual-ity within each State's census , the constraints imposed byU . S . Const . art. 1, § 2, itself make that goal illusory forthe nation as a whole . The specific constraints makingmathematical precis ion illusory on the national level arethe constitutional guarantee of a minimum of one repre-sentative for each State , and the need to allocate a fixednumber of indivi s ible representatives among 50 states ofvary i ng populati ons.

Administrative Law > Judicial Review > Standards ofReview > Arbitrary & Capricious Review[HN7]An agency decision is arbitrary and capricious ifthe agency has relied on factors which Congress has notintended it to consider, entirely failed to consider an im-portant aspect of the problem, offered an explanation forits decision that runs counter to the evidence before theagency, or is so implausible that it could not he ascribedto a difference in view or the product of agency exper-tise.

Administrative Law > Judicial Review > Reviewability> StandingAdministrative Law > Judicial Review > Standards ofReview > Arbitrary & Capricious ReviewEnvironmental Law > Litigation & Administrative Pro-ceedings > Judicial Review[HN8]The Administrative Procedures Act (APA) sanc-tions judicial intervention when the parties feel aggrievedby a final adminis trative ruling. But the APA tightly cab-in s judicial oversight, permitting judicial intrusion onlywhen the admini strative decision abuses reason. It isesse nt ia l to the mai ntenance of j udicial integrity thatcourts rev i e wi ng such d e terminatio ns zealous ly adhe re to

Evidence > Privileges > Government Privileges > Offi-eial Information Privilege > Deliberative Process Privi-legeEvidence > Privileges > Government Privileges > Pro-cedures to Claim Privileges[HNIO]The "deliberative process" privilege protectsfrom disclosure those agency documents which reflectadvisory opinions, recommendations and deliberationscomprising part of a process by which governmentaldecisions and policies are formulated. The privilege is aqualified or discretionary one that turns on a balance ofcompeting policy claims. The privilege does not protectpurely factual material.

COUNSEL: ["*I] O . Peter Sherwood, Corp . Counselof The City of New York, by David B . Goldin, Asst .Corp . Counsel of the City of New York , New York City,for plaintiff The City of New York . Robe rt Abrams,Any. Gen . of The State of New York by Sanford M.Cohen, Asst. Any. Gen , of The State of New York, NewYork City, for plaintiff The State of New York . Daniel E.Lungren, Atty. Gen. of State of California, Charlo tt eSato , Special Asst. Any. Gen . of State of Californ ia , byYeoryios C . Appallas , Deputy Any. Gen . of State ofCalifornia, San Francisco, Cal. , for plaintiff People ofState of Californ ia, ex rel. Daniel E . Lungren , Atty . Gen ..Cravath, Swaine & Moore by Robert S . Rifkind , NewYork City , for all plaintiffs except State of New York ,People of State of Californi a , ex rel . Daniel E . LungrenAtty . Gen. , The State of Texas , The State of New Jersey ,The State of Florida, The State of New Mexico , The Cityof Tucson and the County of Hudson . Arnold & Porter ,Sp. Counsel to Ci ty of New York by Peter L. Zimroth,New York City , for plaintiff, The City of New York .Stein Zauderer, Ellenho rn, Frischer & Sharp , Sp . Coun-sel to City of New York , by Loui s M . Solomon , NewYork City , for plaintiff The City [ * *2 ] o f New York.De Witt W . Clinton , County Counsel o f Lo s Ange l es

Page 3

822 F . Supp . 90 6, *; 199 3 U . S . Di st . LEX I S 4818 , **

County , by Ada Treiger , Deputy County Counse l , Lo sAngeles, Cal ., for plaintiff County of Los Angeles .James K . Hahn , City Atty . of City of Los Angele s, byJess ica F . Heinz, Depu ty City Atty ., Los Ange les C al . ,fo r plaintiff C ity o f Lo s Angeles . Dan Morales , Atty .Gen . of State of Texas , by Javier P . Guadjardo , As st .Atty. Gen . , Austin, Texas, for plaint i ff State of Texas .Robert A. Butterworth , Atty . Gen . of State of Florida byGeorge L . Waas , Asst . Atty . Gen. , Tallahassee , Fla ., forplaintiff State of Florida . Robe rt J. Del Tufo , Atty . Gen .of State of New Jersey, by Michael S. Bokar, SeniorDeputy Atty . Gen ., Trenton, N.J., for plaintiff The Stateof New Jersey . Grant Woods, Atty . Gen . of State of Ari-zona , by Robert B . Carey, First Asst. Atry . Gen ., Phoe-nix, Arizona, for plaintiff State of Arizona. Tom Udall ,Atty . Gen . of State of New Mexico, by Christopher D .Coppin , Asst . Any. Gen., Santa Fe , N.M., for plaintiffState of New Mexico. Herbert Henderson, Acting Gen .Counsel of N.A.A.C .P ., Baltimore , Md., for plaintiffN .A.A.C.P . . Edward A. Hailes, Jr ., Asst . Gen . Counsel ,Baltimore, Md.. Kelly Welsh, Corp. Counsel of [* * 3]City of Chicago , Chicago, Ill., for plaintiff City of Chi-cago, Ill.. John Satalic, Sp. Deputy Corp. Counsel, Chi-cago , I ll.. Clarence A. West , City Atty . of City of Hous-ton , Tex . , for plaintiff City of Houston . Susan Taylor,Asst . City Atty ., Houston , Tex .. Robert A . Ginsburg,Dade County Atty ., Miami , Fla., for plaintiff DadeCounty, Fla . . Gerard Lavery Lederer, Washington, D . C . ,for plaintiff U.S. Conference of Mayors. Frank Shafroth ,Washington , D .C ., for plaintiff National League of Cit-ies. Ruben Bonilla, Gen. Counsel of The League ofUnited Latin American Citizens, Corpus Christi, Tex .,for plaintiff The League of United Latin Ameri can Citi-zens . Ruben Castillo, E. Richard Larson, MexicanAmerican Legal Defense & Education Fund, Chicago,Ill ., for plaintiff Raymond G . Romero . Craig S. Miller ,Dir . , Dep't of Law of the Ci ty of Cleveland, Cleveland ,Ohio, for plaintiff City of Cleveland. Evelyn B . Newall ,Asst . Dir . of Law , Cleveland Ohio . Patricia L . Wells ,City Atty . of the City of Denver, Denver , Col . , for plain-tiff City of Denver . Stan M. Sharoff, Asst. City Atty . ,Denver , Col . . Howard Rosten, City Atty . of Inglewood ,Inglewood, Cal., for plaintiff The City of Inglewood .Barbara [ ** 4] R. Johnson, Senior Deputy City Atty.,Inglewood , Cal. . William Aaron, City Atty. of New Or-leans, New Orleans , La ., for plaintiff The City of NewOrleans . Bruce Naccari, Deputy City Atty. , New Or-leans, La .. Jayne W . Williams, City Atty , of Oakland,Oakland , Cal. , for plaintiff The City of Oakland . PatrickTang, Deputy City Atty . Oakland , Cal . . Victor J. Kaleta ,City Atty , of Pasadena , Pasadena, Cal . for plaintiff Cityof Pasadena. Nicholas Rodrigue z, Deputy City Atty. ,Pasadena , Cal .. T . Michael Mather, First Deputy CitySolicitor of Phi l ad elphi a, Philadelphia, Pa ., for plaintiffThe C ity of Phil ad el p h ia. Lloyd Garza , City AYCy . of San

Antonio , San Anton i o , Tex ., fo r pla i ntiff The C ity of SanAntonio . Ver onica Madrid , Asst . City Atty ., San Anto -nio , Tex .. Louis H . Renne , C i ty Any. of San Francisco ,San Fran c i s c o, Cal. for plaint i ff Ci ty of San Franci s c o.Burk E. Delv enthal , Deputy City AYty . , San Francisco ,Cal .. John J . Copelan , Jr ., County Atty . of BrowardCounty , Fort Lauderdale , Fla . , for plaintiff BrowardCounty. Tony J . Rodriguez, Asst . County Atty . , FortLauderdale , Fla .. Neal M . Janey , City Sol i citor of Balti-more , Baltimore , Md . , for plaintiff City of Baltimore .Frank C . Deer, [ * *5] Chief Solicitor, Baltimore , Md . .Bu rton H . Levine, Asst . Solicitor, Baltimore , Md .. Jo-seph I . Mulligan, Jr. , Corp. Counsel of City of Boston,Boston, Mass ., for plaintiff The City of Boston . Robe rtA. Cohen, Asst . Corp . Counsel , Boston, Mass .. JohnCalhoun, City Attomey of Long Beach, Long Beach,Cal . , for plaintiff City of Long Beach . Donna M. Chris -tensen , Atty. Gen . of the Navajo Nation, Window Rock ,Ariz . , for plaintiff the Navajo Nation . Steven Gonzales,Asst . Atty . Gen., Window Rock, Ariz . . Alan K. Marks ,County Counsel of San Bernardino , San Bernardino,Cal ., for plaintiff County of San Bernadino. DanHaueter , Chief Deputy County Counsel, San Bernadino ,Cal .. Joan R. Gallo , C ity Atty , of San Jose , San Jose,Cal ., for plaintiff The City of San Jose . George Rios,Asst . City Atty., San Jose, Cal .. John Payton , Corp.Counsel of Washington, D.C ., Washington , D . C ., forplaintiff The City of Washington, D . C .. Jacob Walker,Asst . Corp . Counsel, Washington, D .C .. Scarinci & Pelioby Leon S . Segen , Totowa, N . J . , for plaintiff County ofHudson, New Jersey . Katharina Richter , Office of theCity Attomey of Tucson, Tucson , Arizona, for plainti ffThe City of Tucson .

Stuart M . Gerson, [* * 6 ] As st . Atty. Gen ., Anrew J . Ma-loney, U . S. Atty. , Thomas Millet, Michael Sitcov , JasonR . Baron, Judry S . Subar, David M . Glass, Susan L .Korytkowski , Dept . of Justice, Civ . Div . , WashingtonD . C ., for defendants Department of Commerce and fed-eral officials. Burn eatta L . Bridge , Asst . Atty . Gen. ofState of Wisconsin , Madison Wisconsin , for defendantThe State of Wisconsin . Andrews , Davis, Legg, Bixler,Milsten & Price , Sp. Counsel to State of Oklahoma , byGretchen A . Harris , Oklahoma C ity , Oklahoma, for de-fendan t The State of Oklahoma ex rel . Susan B . Lovin g,Atty . Gen ..

JUDGES: McLAUGHLIN

OPINION BY: JOSEPH M. McLAUGHLIN

OPTNION

(*91 iJ MLMORANDUMAND ORDER

McLAUGHLIN , C ircu i t Jud ge '

Page 4

822 F. Supp . 90 6, *; 1993 U . S . Di s t . LEXIS 4 81 8, **

* sitting by des i gn ati o n

Plaintiffs -- state s, citi es, cit izen s' groups, and indi -vidual citizen s and taxpaye rs - - s eek a judgment : (1) va-cating former Secret of Commerce Robe rt Mosbacher' sJuly 15, 1991 decision that the 1990 census would not bestatistically adjusted; (2) o rde ring that such an adjust-ment be made; and (3) allowing plaintiffs to use and pub-licize certain [ * * 71 data generated by the Census Bu-reau , and already produced , subj e ct to a protective order ,to the plaintiffs during this litigation. For the reasons setforth below, the Court holds that the decision againstadjustment shall not be disturbed, but grants the plain-tiffs' request to use and publish the Census Bureau data .The follow ing constitute the Court 's findings of fact andconclusions of law in accordance with Federal Rule ofCivil Procedure 52.

FACTS

Just to recount the facts of this case is arduous ,given its four-year history , the number of part ies in-volved , and the complicated statistical evidence lying atthe core of the dispute . Many of the material facts havebeen set forth in two prior published opinions -- City oNew York v. United States Dept of Comm erce 713 F.Supp. 48 (E .D .N . Y. 1989) ("City of New York 1') , andCitv ofNew York v. Uniled Slates Dep't o� Commerce739 F. Supo . 761 (E. D.N.Y . 1990) ("City of New YorkIX') -- some familiarity with which is assumed .

Census Background

[HNI]The Constitution requ ires a decennial census .Article I , Section 2, Clause 3 states that "th e actual enu-meration [ ** 8] shall be made [every ten years], in suchmanner as [the Congress] shall by Law direct . ° Congresshas , in turn, delegated to the Secretary of Commerce theduty of taking the census "in such form and content as hemay determine , including the use of sampling proceduresand special surveys . " 13 U.S . C . & 14L(aO 19g21 . TheBureau of the Census , an agency within the Departmentof Commerce, actually conducts the census. See 13U . S . C . § (1982) .

The results of the census are used for a galaxy ofpurposes . The federal government u s es them to calculatehow to dispense program funds among the states . Statesuse the counts for political redisCricting . Sociologists andhistorians study them for more esoteric purposes . Noneof this obscures the central truth that the "basic constitu-tional purpose" of the census is "to determine the appor-t ionment of Representativ es among the States . " Carey v .Klutznick 65 3 F .2d 73 2 , 736 (7_ d Cir . 1981) .

The firs t c e ns us o f th e American population was in1 790 . Thomas Jeffe r son , wh o was in charge of it , c om-

plained of an underc ount . There have bee n 20 sub sequentcen s u s e s. Each of them has a l s o [* "9 ] resulted in anundercount. More troubling than the undercount itself,however , i s that raci a l and e thni c minoriti es are under-counted to a greater degree than the population a s awhole . This problem , known antiseptically as the "differ-ential undercount , " has skewed ev ery census since theBureau started measuring it in 1940 .

Because the counts are used to calculate the politicalrepresentation and financial aid to be afforded to a givenarea , the fear that the census may be perpetuating a sys -tem in [ * 912] which those most in need of representa-tion and aid are deprived of both is a major concern.With that in mind , the Census Bureau began, after the1980 census , to develop a method by which both theundercount of the entire population and the differentialundercount could be reduced through a statistical ad-justment employing a "post-enumeration survey"("PES") . This method (and the Department of Com-merce's reaction to it)1Se at the heart of this case .

The 1990 Census

Taking the census has always been a daunting task ,and the 1990 count was no exception. The Bureau beganpreparing in 1983 , seeking to improve the techniques thatit had used in prior censuses . Among other things , it[*"IO] consulted with state and local governments ,planned an extensive advertising campaign, designed amore ethnically inclusive census questionnaire , and in -creased the amount of automation used , includ i ng the useof an automated geographic control system , which as-sured accurate and timely maps and geographic files forthe 1990 census . While the parties may disagree on thequality of the census counts achieved in 1990, the four-step procedure used to conduct the census is largely un-disputed .

First : an address list of housing unit s was compiled .This list was crucial because it indicated eve ry householdin the nation to which the Bureau would send question-naires . Since the Bureau relies on the mail return of thosequestionnaires to count a majority of the population , anaccurate and comprehensive list was vitally impo rt ant . Inconstr ucting the list, the Bureau relied primarily oncommercial mailing lists, supplemented by extensivefi eld research and collaboration with the Postal Service .Then, numerous quality controls were instituted to im -prove the accuracy of the Iist .

Second: census questionnaires were mailed to eachhousin g unit . Householders were as k ed to complete andreturn [ ** I1) the questionnaires to the local census dis-trict office on or before Apr i l 1 , 1 9 90 .' This i s called the" mai] out/mail back " pha s e . The effort to get indiv i dualsto p art icipate in the mail out/mail b ack phase was exte n-s ive. In addit i on t o the Cen s u s Bureau's general advertis-

Page 5

822 P. Supp. 90 6 , *; 1 993 U.S. Di st . LEXIS 48 t 8, **

ing campa i gn , i t a l s o conducted campai gns sp e cificallytargete d at Afi'ic an-Americans , A s ian s, Hispanics , andNative Amer i can s. In addition, the Bureau publishedspec i alize d , fore i gn - lan gu age brochures encouragin gpubl i c participati o n in the c ensus . It al s o maintained a setof toll-free numbers (in eight languages) for anyone whohad que st i on s regarding the census que stionn aire , andevery censu s form advised Spanish speaker s th at theycould call a toll -fr ee 800 number to obtain a censu s formin Spanish. Fin a ll y, the Censu s Bureau employed differ -ent methods in areas where it was believed that the nor-mal procedure would be particularly ineffective. See Sec-retary of the Department of Commerce , Dec ision onWhether or Not a Statistical Adjustm ent of the 1990 De-cennial Census of Population Should be Made for Cov-erage Deficie n c ies Resulting in an Overcount or Under-cnunt ofthe Populaiion, July 15, 1991 (the [ ** 12] "De-cis ion') , at 4- 5-4-6.

I April 1, 1990 day is officially entitled "Cen -su s Day , " and is the precise date as of wh ich theCensus Bureau seeks to count the populat i on .

Third: because the return rate of census quest i on-naires is obviously never 100% , and in 1990 was only63%, see Transcript of Trial ("Tr . ") at 1823 , the CensusBureau embarked on an extensive follow-up campai gn .Second mailings were sent to households that failed toreturn the i nitial form, and i n census districts with par-ticularly low return rates, the Bureau remailed censusforms to all residents . Tr. at 1730-31 .

Fourth : when steps 1-3 did not produce a census re-turn from a particular household , the Census Bureau en-gaged in "non-response follow-up," the final stage of theenumeration . During this phase, each non-respondinghousing unit was assigned to a°census enumerator, " anemployee who was directed to make up to six attempts Cocontact a household member to obtain the informationnecessary to complete a census form. If this also [ *X 13)proved fruitless , the enumerator was then required to tryto obta i n basic i nformation on the missing housing unitfrom a reliable source , such as a neighbor or build ing[ M913] manager . Decision at 4 - 7 . Once 95% of a dis-tr ict's operations were completed , a final phas e of non-response fo ll ow-up requ i red enumerators to make onelast-ditch attempt to visit each remaining unresolvedhousehold to obtain as complete an interview as possible .

After the enumeration was completed , post -enumeration "Coverage Improvement Programs" ' wereimplemented, with the result that 5 . 4 million people wereadded to the counts . Decision at 4-7 . The result of all ofthese effo rts was that 249 , 63 2,692 people were countedduring th e 1990 census. Dec isio n at 4 -2.

2 These coverage improvement programs in-c lud ed: (I) a 100 perc ent re-check of vacant , un-inhabitable, or none xi stent units ; (2) the "Wereyou count ed?" advertisi n g campaign to reachpeople who thought they m ight have been missedby the census ; ( 3 ) a paro lee and probationercheck , to set th e names and Census Day ad-dre sses of tho s e people and add them Co the cen-sus i f they had not already been counted; (4) thehousing coverag e check, in which the Census Bu-reau recanvas sed select blocks based on evidenceflush ed out by the automated management infor-mation sy stem ; and (5) the local government re-view program , which provided local governmentswith the opportunity to challenge census countsfor their areas . Dec ision at 4-7-4-9 .

[ ** 141 TheDrfferential Undercount

Despite the herculean effo rts of the Census Bureau,it is undisputed that the 1990 Census was not -- andcould not realistica ll y be - - successful in its goal ofachieving an exact count of the nation's population.Given the nature of the task , it is not s urprising that thecensus fails to count some individuals ( " omissions") andalso adds persons into the count erroneously ("erroneousenumerations") . Tr . at 80-82 .

The "net undercount" is the difference betweenomissions and erroneou s enumerations. It is undisputedthat the 1990 census , like all previous censuses, resultedin a net national undercount. Dec ision at 1- 1. It is simi -larly uncontroverted that A fr ican -Americans and otherminorities have been persistently undercounted to agreater degree than non-Hispan ic whites in all censusessince 1940 when the Bureau began measuring such dif-ferences, and that this anomaly is perpetuated in the 1990census . The difference between the undercount rate fornon-Hispanic whites and that for minority populations isknown as the " differential undercount . " Tr . at 91-92.According to the Secretary , " Blacks appear to have beenundercounted in the 1990 census ( ** l5) by 4 .8%, His-panics by 5.2%, Asian-Pacific Islanders by 3 . 1%, andAmerican Indians by 5 . 0%, while non-Blacks appear tohave be en undercounted by 1 . 7%." Dec i sion at 1 -1.

Possibility ofStatistical Adjustment

The Census Bureau has been aware of the existenceof a differential undercount since the 1950's. The intrac-table problem has been liow to fix it . Following the 1980census , concerns over the persistence of the differen ti alundercount , its deleterious effects on the accuracy ofcensus counts, and the unfair results arising from suchinaccuracy , prompted the Bureau to s ta rt a research pro-gram aimed at developin g stat is tical te chn i ques to ame -l iorate the p ro bl em in th e 1990 cen s u s. ' Tr . at 5 2 5 , 1 29 1 -92. By 1984, th e Bure au h ad develo p ed a timetable for

Page 6

822 F. S upp . 90 6 , ' ; 1 99 3 U. S . D is t LEXIS 4818 , **

interna l Bureau research th at wou l d ultimately lead to adecision wheth e r to adjusi the 1990 cen su s stat istica lly i nan effort to reduce the differe ntial undercount . Two taskforce s were create d to c on s i der the und ercount pro bl emas it related to th e upcoming 1 990 cen s u s: Th e Under-count Steering Committee ( "USC " ) was re sponsible forplanning undercount rese arch and policy dev elopment.The Undercount Research Staff ("URS " ) conducted[ * * 16] the actual res earch . Other divisions at th e Bureaua l so conducted re search on the undercount and the possi -bility of adjustm ent. Tr . at 517 -25, 1292-93 . In addition,the Bureau sought the opinions of outside experts andorgani zations , such as the American Statistical Associa-tion and the National [ * 914] Academy of Science , re-gardin g the possib i lities for adjustment .

3 Discomfiture over the persistent pattern ofdifferential undercount had prompted the Bureauto conduct a Post-Enumeration Program (the"PEP") i n 1980, a survey designed to evaluate thequality of the 1980 census and to estimate the un-dercount, including the differential undercount , atboth national and subnational levels. A lawsuit tohave the 1 980 census adjusted statistically by useof the PEP or another statist i cal techn ique wasunsuccessful . See Cuomo v. Baldrive 674 F.SUDo. 1089 (S D.N.Y . 1987� .

After considerin g the alternatives, the Bureau se ttledupon the PES as the best too] to statistically adjust thecensus through [ *"` 17 ] the use of "dual system estima-t i on" ("DSE") . Tr . at 559-61 . Dual system estimation or ,in more pedestrian terms, "capture/recapture , " is , as rele -vant here, an approach that uses a second measurementto asce rtain the qual ity of the estimate obtained by aninitial measurement , and then uses that information toprovide a purportedly more accurate , dual system esti -mate. ' Here , the original enumeration, the census, wasfollowed by a second measurement, the PES , wh ich at-tempted to meas ure the rate at which people were omit-ted and e rroneously enumerated by the census , in orderto determine a net undercount rate .

4 At trial , the parties explained cap-ture/re capture in . terms of determinin g the numberof fish in a lake. First you capture 1000 fi sh , tagthem and throw them back . Then, you catch an-other 1 00 . If 90 of those have tags , it suggeststhat 90 percent of all the fish in the lake aretagged. If so, then the 1000 fish initially taggedrepresent 90% of all the fish in the lake . Doingthe algebra, the total population of fi sh in the lakei s therefore 1 ,111. Tr . at 41 -4 2.

["* 18 ] Whi le th e Bureau has used p ost -enum e rati on s m veys in a va r i ety o f ways s ince 1 95 0 , i t

has n ev er stati stica l ly adjusted based on DSE . The Bu-reau worked throughout the 1980's to des ign the PES tomake it an effective tool for census adjustment. Tr . at5 7 2. For example , c orrelation bias, which may occurwhen residents become confused by an overlap betweenthe census and the PES , was addressed by distinctlys eparating the two procedures. Tr . at 578 - 82 . Anotherspecies of correlation bias , which ari s es when individualswho have different probabilities of being counted ("cap-ture probabilit i es " ) i n the census are grouped together inthe PES , was reduced by the use of "poststratification."' 1' r. at 205-208 . ' In addition, statistical "smoothing" waschosen to address anomalous results in the PES . 6

5 Poststratification grouped all individuals witha similar likelihood of be ing counted in the cen-sus . These groups, labeled "poststr ata" , were de -fined by age, sex, race , Hispan ic origin , housingtenure (i.e. whether the individual owned orrented a residence), type of place (i.e., centralcity, suburb , outside metropolitan area), and geo-graphic region . Tr. at 513 . This categorization re-sulted in a total of 1 , 392 exhaustive and mutuallyexclusive poststrata . Tr. at 206-07 . In otherwords , each resident of the United States fits intoone , and only one , poststratum .

1 ** r 916 For an explanation of smoothing, see infranote 10 .

By the Spring of 1987 , after much testing and fine-tuning, the Census D i rector, John Keane , had decidedthat the Bureau should proceed with plans to adjust the1990 census data through the use of DSE, if the PESresults met a certain quality standard. Dr. Keane metwith his superior, Robert Ortner , the Under Secretary ofthe Department of Commerce , to tell him that such adecision had been made and that a press conference tothat effect was imminent . Six days later , Keane metagain with Ortner and other Commerce Department offi-cials , who informed Keane that they had decided againstadjustment . Shortly thereafter, Commerce Departmentofficials instructed their Census Bureau officials not todisclose that a decision had been made . Tr . 629-30 , 1330 .On October 30, 1987 the Department of Commerce an-nounced its decision aga inst adjustment, and this lawsuitwas born .

His t ory ofThis Litigation

In November , 1988, plaintiffs sued to enjoin the1990 census, challenging the methodology by which itwould be taken , and seeking to reverse [ **20] the deci-s ion against adjustment. Defendants -- the Department ofCommerc e, it s Secretary , Pres id en t Bush , and other offi-cial s within the Department of Commerce and it s s u b-s idiary, th e Bureau o f th e Cen su s - - moved to d is mi ss th e

Page 7

822 F . Supp . 9 0 6, * ; 1 993 U. S . Di st . LEXIS 4818, **

appl i cation for the injunction . This Court d en ie d the di s-m iss al m oti on, holdin g that the plaintiffs had standing tochalle nge the cen su s on constitutional grounds ; ' theCourt a lso rul e d [ * 91 5 ] th at it would con s ider th eCommerc e Department's decision against adju stmentunder the " arb itrary and capric ious " standard of revi ewof the Admini strative Procedure Act, S U S . C .706 2 A 1982 (the ' APA") . CiCy ofNew York / 713F . Supp . at 54 .

7 While the defendants continued to argue dur-ing pretrial proceedings that this case presented anon-justiciable political question , the SupremeCourt has now rejected this argument, holdin gthat constitutional challenges Co the census meYh-ods employed to arrive at the apportionment arejusticiable . United States Deo't of Commerce v .Montana 118 L. Ed. 2d 87 112 S. Ct 14151R24-26 1992 .

bers submi tted recommendation s to the Secretaryregarding the decision on wh ether to adjust . Stip .at 5 .

[**22 ] The defendant s adopted and promulgatedthe required guidelines , but the plaint i ffs challengedthem as inadequate, and they also sought a declaratoryjudgment that a statistical adjustment would not violatethe Constitution or any federal statute . Defendants coun -tered that the pla i ntiffs ' chall enge to the cen sus presenteda non -justiciable political question . This Court rejectedthe defendants' political question claim , and concludedthat statistical adjustment, per se, would not violate ei-ther the Constitution or the laws of the United States .CitLOfNew York 17 739 F. Sup . at 767-68. This Courtnoted that , while the guidelines were vague , they didsatisfy , albeit just barely, the defendants' obligations un-der the Stipulation. Id. at 770 .

The Post-Enumeration Survey

[ **21] When the dismissal motion was denied, ahearing was scheduled on the injunction . It was set to goforward in the Summer of 1989, when, at the eleventhhour, the parties entered into a stipulation (the "Stipula-tion" or the " Stip ." ) . The Stipulation vacated the Com-merce Department' s 1987 decision against adjustmentand agreed that the new Commerce Secretary, Robe rtMosbacher, would consider de novo and "with an openmind," whether adjustment was warranted. Stip . at 2- 3 .The Stipulation also agreed that the program to gatherthe statistical data necessary for adjustment would pro-ceed, that the Secretary would decide whether to adjustby July 15, 1991 , and that his decision would be consis-tent with certain procedures , includin g the promulgationof "guidelines" articulating what the defendants believedto be the relevant technical and policy considerationsaffecting the decision. It also mandated the creation of aneight-member Special Advisory Panel (the "Panel") a ofstatistical and demographic experts to advise the Secre-ta ry on whether to adjust. Stip . at 4-5 .

8 By agreement, the Secretary chose four mem-bers of the Panel from a list of seven candidatessubmitted by the plai ntiffs , and chose the fourremaining Panel members himself. The Stipula-tion required the Panel members to be "of suchknowledge, judgment and probity that theirjudgment and advice shall be entitled to the ut-most respect by defendants . " Stip . at 5 . The fourpanel members cho sen from the plaintiffs' listwere Eugene P . Ericksen , Leobardo F. Estrada,John W . Tukey, and Kirk M . Wolter . The fourpanel members cho sen unilatera ll y by the Secre-t a ry were Will i am Kru s kal , Michael McGeehee,V . Lan ce Tarrance Jr . and Kenneth M . Wachter .A s req uired by th e St ipulati on , t h e Panel m em-

Following this Court 's decision in City of New YorkI, the Bureau resumed work on its plans to implement thePES , and implemented it in 1990 . In the first step of thePES , the Bureau methodically selected approximately5000 blocks ' in an effort to attain an appropriate samplesize for each poststratum. In February 1990, Bureau em-ployees visited [ * *23] each sample block and listed allthe housing units they found, identifying approximately170, 000 households . In July 1990, Census Bureau inter-viewers returned to each address to obtain informationregarding the residency status of those households onCensus Day . The Bureau found that those blocks con -tained approximately 400,000 people . Tr . at 208. Aftercollecting the PES data, the Bureau matched it to theinformation collected in the original enumeration forthose same sample blocks . From this matching, the Bu-reau endeavored to estimate , for each poststratum , ratesof om i ssion and e rroneous enumeration , and from thesecalculated a net undercount rate for each poststratum. Tr .at 221.

9 As used here, "block" means a square block;that is , all the buildings on four streets forming asquare . Tr . at 209 . The Census Bureau made a listof the more than 5 , 000 ,000 blocks in the UnitedStates and then selected approximately 5 ,000 thatthey believed fairly contained representativesamples of' the 1 , 392 poststrata . Tr . at 208; Deci-sion at 4-11-4 - 12 .

[* * 24] The Bureau used these results to develop an"adjustment factor" for each poststratum , i. e . , the numberby which the population count as indicated by the censushad to be multipli ed so that the entire census would re-flect the variations found in the PES . Accordingl y,[ "9 1 6] th e 1 ,3 92 poststrata resu l t e d in 1 , 392 co rrespond -

Page 8

822 P. S upp . 906 , *; 1993 U . S . Dist. LEXIS 4 818 , **

ing adju stment fac to rs. One further statistical twist to theu se of the PES was th e employment of " smoothing . " '°After smoothing, the Bureau used the smoothed adjust-men t factors to produce adju s ted counts down to theblock lev e l, which were th en aggregated to providepopulation estimates for c ities , counties, states, and thenation. Tr . at 224 -2 5 ; Dec isio n at 4 - 18 .

10 Smoothing is a statistical procedure used toreduce th e e ffects o f sampling error . More par-ticul arly, it seeks to reduce the difference be-tween the results from the PES sample and the re-sults one would receive if one were able to surveythe entire population. Smoothing in the 1990 cen-sus took place as follows . First , the 1,392 raw ad-justment factors with corresponding raw vari-ances (measures of sampling error) were com-piled . The Bureau then employed pre-smoothing,or "modelling the variance , " in an attempt to im-prove the accuracy of the estimates of the vari-ances of the raw adjustment factors. Tr. at 796.Once modelling the variance was completed foreach raw adjustment factor, a regression was per-formed . This regression moved the raw adjust-ment factor for each poststratum towards a typi-cal value by an amount depending on the sam-plin g error associated with that particular post-stratum . Thus, where a particular raw adjustmentfactor had a small variance (i.e., where the sam-pl e was very large) , it would be moved only asmall amount , whereas raw adjustment factorswith larger variances tended to be moved , more .Carrier variables relating to raw adjustment fac-tors were selected in an effort to give the best es-timate of the typical value. Tr . 807 . These carriervariables included the same characteristics thatdefined the post-strata , such as age , sex, race,owner/renter, and other characteristics such asmail return rate . The end result is that the 1,392raw adjustment factors became 1 , 392 smoothedadjusted factors . The census count for eachpostratum group was then multiplied by itssmoothed adjustment factor and adjusted censuscounts were produced . Tr . at 788-89; Decision at4 - 17-4-18 .

[" * 25] A number of quality control checks weremade to test the results of the PES. First, the Bureauconducted or commissioned more than twenty formalreseaich proj ects , called "P - Studies," to study the poten-tial sources of error within the PES. The results of theseP-Studie s regardin g part icular sources of error were thencombin ed in the " tota l error model" that summari zed theoverall quality of the PES data. Tr . at 65 2- 59 . "

1 I The major po tenti al s ources of er ro r arisingfrom the PES included: miss in g data, poor qual ityof the reported Census Day addres s l i st , fabrica-tion , match i ng e rror , mea surem ent of erroneousenumerations , balancing the estimates of grossovercount and gross undercount , co rrelation bias,small area estimation, and late c e n s us data . Tr . at570-73 .

The final result of the PES was that the census enu-meration was estimated to have undercounted the popu -lation by 5 ,269,917 , or 2 . 07% . In terms of the differentialundercount, the PES indicated that the census under-counted Hispanics by 5 .2%, A frican -Americans [ * *26]by 4 . 8% and Asian/Pacific Islanders by 3 . 1% . The PES-calculated undercount for non-A fr ic an-Americans was1.7% and 1 .2% for non-Hispanic whites, with a totalnational undercount of 2 . 1 % . "

12 A recent "discovery of computer errors andsome statistical changes have reduced the esti-mates of an undercount to 1.6 percent, about thesame as in 1980 . " Felicity Barr inger, U. S Popu-lation Passes 265 Million, Bureau Says , N . Y.Times , December 30, 1992 , at A 1 2. As one of thewitnesses testified here, "statistics is never havingto say you're ce rtain . " Tr . at 1922.

The Bureau also conducted a number of "loss func-tion analyses" to compare the quality of enumerationcounts to the adjusted counts. A loss function analysis isa systematic way of assessing the consequences flowingfrom a particular decision . In the context of the adjust-ment decision , the Bureau used loss function analysis todeterm ine whether the adjusted data were expected to bemore accurate than the unadjusted data . Tr. at 1941-42 .This Court is satis fi ed [**271 that for most purposes thePES resulted in a more accurate - - or to be statisticallyfashionable , a less inaccurate -- count than the ori ginalcensus .

The Secretary 's Decision and The Trial

Prior to reaching his decision, Secretary Mosbacherreceived the recommendations of the eight Panel mem-bers. Perhaps not surprisingly , the Panel was deadlocked :the four members selected from the plaintiffs' list rec-ommended in favor of adjustment, while the four mem-bers chosen unilaterally by the Secretary recommendedagainst it . Decision at 1-3. The USC voted 7-2 in favor ofadjustment . Id. The Under Secretary of Commerce forEconomic Affairs and the Administrator of the Econom-ics and Statistics Administration [*917] voted aga instadjustment . Defendants' Exhibit 1 at 898 . F inally , th eDirector of the Censu s, Dr . Barbara B ryant recom -mended i n favor of adjustment, but acknowledged that"there i s n o perfect tr uth as to the s i ze and distribution of

Page 9

822 P . Supp. 906 , *; 1 9 9 3 U. S . Di st . LEXIS 4 8 18 , **

the population, " and that " adjustment is an issue aboutwhich reas onabl e men and women and the best statisti-cians and demographers can disagree . The minori tyviewp oint expre s sed in the Censu s 13ure au ' s r eport ...illu s trates thia " Id. at (" " 28 1 I 118- 19 . "

13 Dr . Bry ant's comments in a year-end inter -view are enli ghtening. See Barringer, supra note12 . In that interview " she said [that ] while the sta-tistical tools were available to make these ad -justments for small geographical units , the neces -sary tools to double-check the findings were in -adequate. In the face of legal scrutiny she said,this made a decision to adjust untenable . " She isalso quoted as stating that "'every number has tobecome defensible , "' and "'when you say - - youknow how to do it but you can't prove its right orwrong -- then it's no longer defensible. If it were-n't for the problem that we had to defend it incourt, there would have been a strong inclinationto have adjusYed . . . . "'

Dr. Bryant also opined that she believed thatan adjustment would have improved the accuracyof counts at "'The national and state levels , the biglevels,"' but that the PES results were " ' very in-conclusive' when used for smaller subdivisions ofthe population . " Id

als o analy zed at length the conclu s ions that the Secre taryreached in the Decision. Plaintiffs' direct case cons i stedof the testimony of nine w i tnesses , including all four ofthe plaintiffs' d es i gnees to th e Panel. It a lso included thei ntroduction of hundreds of exh ibits and numerous depo-sition tran s cript s from other witnesses .

De fendants' evidenc e was s imilarly grand in scope .They presented five expe rt witness es , including onePanel member . They also introduced the deposition tran-s cripts of oth er witnesses and numerous exhibits. Ofthese, Exhibit I , denominated as the Administrative Re-cord by the defendants, and skeptically dubbed "the so-called Administrative Record" by the plaintiffs, containsover 12 , 000 documents and occupies 18 , 000 pages. Thetrial transcript exceeds 2,600 pages.

DISCUSSION

Plainti ffs allege that the Secretary's decision not toadjust the census count violates the APA, the Constitu -rion , and the Stipulation. " They also argue that the proc-e ss the [ * 918] Secretary used to make his decision wasa sham . [ * *31] 1 6 They seek a order directing the Secre-tary of Commerce to make the adjustment and they askfor permission to use Census Bureau data provided tothem by the defendants during the course of this l itiga-tion under a protective order , and to release that data tothe public .

On July 15 , 1991 , in accordan ce [**29] w ith theStipulation, Secretary Mosbacher went on national tele-vision to announce his decision not to adjust. Simultane-ously, he produced the Decision, a 178-page report giv-ing the reasons for his decision . The decision revitalizedthe case and discovery resumed . Claiming that the Secre-tary's decision violated the Constitution, the APA, andthe Stipulation , the plaintiffs requested a trial. More spe-cifically , they alleged that the administrative record prof-fered by the Secretary as the basis of his decision is aself-servin g, post-hoc compilation of documents assem-bled for the purpose of strengthening the defendants'liti gation position and that the Secretary's decision wastai nted by partisan political influence. Over the defen-dants' objection , this Court ordered a trial, which con-sisted almost exclusively of expert testimony in the fi eldsof demographics and statistics , and continued for thi rteentrial days . "

14 Before trial , two other cases presenting theidentical issue in this case were Cransferred andconsolidated with this action -- City ofAtlanta v .Mosbacher, 92-CV-1566 ; Florida Hous e of Rep-res entatives v. Franklln , 92 -CV -2037 .

[**3 0] The expert witnesses ezpressed th e ir opin-i o n s as to wheth er the Secret ary con s idered all the factorss pec ifi e d i n th e guid e lin es in mak i n g hi s decision , and

15 Plaintiff Hudson County, New Jersey, alsoclaims that the decision against adjustment vio-lated the Voting Rights Act , which provides that:

[HN2]No voting qualification orprerequisite Co voting or standard,practice , or procedure shall be im-posed or applied by any State orpolitical subdivision in a mannerwhich results in a denial orabridgement of the right of anyc iti zen of the United States to voteon account of race or color, or incontravention of the guarantees setforth in sect ion 1973 b( f) (2) of thi stitle , as provided in subsection (b)of this section.

42 U.S .C . & ]973a (1982) (emphasis added) . TheCourt rejects this claim because it i s close tofr ivolous . By its plain language , [HN3]the VotingRights Act applies only to misconduct by statesor their political subdiv i sions . See Senate oLCa [i-fornia v . Mosbacher, 968 F .2d 974 979 (9th Cir .1 992) (argument that Voting Right s Act contem -plate s s u i t s again st th e fed e ra l government is " s e-

Aage 1 0

822 F . Supp. 906 , * ; 19 9 3 U. S . Di st . LEXIS 4 818 , **

verely flaw ed " ) ; Tu ck er v, Un it ed States Dep't ofCommerce 9 58 F.2d 1411 141 4 (7th Cir . ) ( "Theplaintiffs cannot b e s erious in arguing that the re-fir s al to adju s t th e h eadcount violates the VotingRights Act . "), cert . denied, 121 L. Ed . 2d 3 3 2 , 61U S L .W . 3334 , 1]3 S . Ct . 407 (1992 ) .

[**32]16 Pla i ntiffs a s sert that Secretary Mosbacherwas clo s ely aligned with the Republican Partyand , therefore , never ser i ously considered ad -justment in the belief that adjustment would favorDemocratic politicians . They also argue that con-tacts made by then-White House Chief of StaffJohn Sununu and a member of his staff to Com-merce Department officials other than Mr. Mos-bacher tainted the decision . I have reviewed theseallegations in detail . While it does appear thatMr . Sununu and his subordinates expressed theircontempt for adjustment to Department of Com-merce personnel, I cannot , on the record beforeme , conclude that such contacts represented im-proper influence . Moreover , the plainti ffs' attackon the integrity of Mr. Mosbacher - - who wasnever a party to these conversations -- does notwarrant extended di scussion here .

the Clerk o f the Hou se. The Cou rt reason ed that because" Ch ere i s no statute that rules out an instruction by thePresident to the Secretary to reform the census , evena fter the data i s [ s icJ submitted to him , " th e Secre tary'sdecision as to how foreign fe deral employee s are countedis "like the ruling of a subordinate offi cial ," and, there-fore, not final for purpo s es of APA review, Id. at 2774(citation omitted).

Defendants believe that the same rationale that ledthe Supreme Court to reject the APA claim in Franklin v .Massachusetts, an apportionment cas e, applies withequal vigor here . I disagree. The Supreme Cou rt held thatthe Secretary 's acts in conducting the census and report-ing the counts to the President were not "final," for pur-poses of challenging apportionm e nt . That case did notinvolve a situation where , as here , plain ti ffs challengethe counts as they are used for intr a-state redistrictingand for federal fund allocation See City of New York I713 F. Supp. at 50 . Neither of these purposes requires theSecretary [ * *35 1 to transmit the counts to the Presidentbefore publishing them or transmi tt ing them to censusdata users . 11 [ * 919] [HN4 ]The Secretary's repo rt ing ofthe counts for those purposes , accordingly , i s finalagency action for purposes of APA review . A s JusticeStevens explained in Franklin :

1. The APA Standard ofReview - Final ity

The standard by which the Court reviews the Secre-tary's decision not to adjust should be stated at thethreshold . At a previous stage in this liti gation , this Courtannounced that "the arbitrary and capricious standard asset [ **33] forth in § 706 of the APA wi ll guide my re-view of the Secretary's determination . " City o New YorkI. 713 F . Supp at 54 .

Defendants now contend that the plaintiffs' claimunder the APA and, w ith it , this Court 's decision to re-view the Secretary's decision under the arbitrary and ca-pr i cious stand ar d, have been vitiated by the SupremeCourt's recent decision i n Franklin v Massachusetts 120L Ed . 2d 636 I 12 S Ct . 2767 (1992). There , Massachu-se tt s challenged the Census Bureau's method for count-ing federal employees serv ing overseas, alleging that itwas arbitrary and capricious , and, as such, a violation ofthe APA. Massachusetts also assert ed that the methodviolated the constitutional requirements for conducting adecennial census and damaged it becau se it changed thecongressional apportionment, moving one representativefrom Massachuse tts to Washington . Id. at 2770 .

Refusing to address the APA claim, the SupremeCourt concluded that the Secretary ' s determination wasnot "final" because , in the context of apportionment, theSecretary s imply repo rt s the results of the census to thePres ident, who in turn [ %*3 4) transmits the apport ion-ment for each state in th e House of Representatives to

Even in the Court's view, the Secretary'sreport of census information to recipientsother than the President would certainlyconstitute "final agency action . " TheCourt 's decision thus appears co amount toa pleading requirement . To avoid the barto APA review that the Court imposes to-day, litigants need only join their appor-tionment challenges to other census-related claims . Notwithstanding theCou rt's novel reading of the statute, inview of the Secretary 's insistence on uni -tary census data , relief on any censusclaim would yield relief on all otherclaims .

Franklin 112 S . Ct. at 2783 n14 (Stevens , J . , concur-ring) .

17 With respect Co redistricting, 13 U . S.C . §14I c provides , in pert inent part, that :

Tabulations of population for th earea s identified i n any plan ap -proved by th e Se c retary s ha l l be ..rep orted to the G ove rn o r o f th e

State inv o lv ed an d to th e offi c e rsor pub lic bod i es hav i ng res pon si-

Page 11

822 F. S upp . 90 6 , *; 1 9 9 3 U. S . Di st. LE XI S 48 1 8, '*

bility for le gis lati ve apport i onm entor di strictin g of such State ....

Id. With resp ect to the plaintiffs' c laim based onalloc ation of federal fund s, the following statutesprovide for d irect repo rting of cen sus data by theSecretary of Commerce , without the Pres ident ei -ther acting as an intermediary or retaining finaldiscretionary authority to report the counts : 4 2U.S C . & 9831 et seq, (Head Start program); 42U.S . C. S 702 (Maternal and Child Health Ser-vices Block Grant); 42 U.S .G & 5632 (JuvenileJustice and Delinquency Prevention Program); 42U . S C . & § 3024, 3028 b(Programs for OlderAmericans); 23 U . S.C. & 104(b}(6) (HighwayPlanning and Construct i on) ; 49 U . S.C . § 1607a(Urban Mass Transport ation Capital and Operat-ing Assistance programs) .

[ ** 36] Accordingly, I adhere to my earlier decisionthat the APA governs the Secretary's decis i on . Hence ,the question for review is , as the plaintiffs have pith i lystated, "whether the Secretary's application of the deci -sion guidelines, as construed in light of constitutionalrequirements, to reject the [adjusted] counts is arbitr aryand capri c i ous . " 's Plaintiffs' Brief at 148 .

IS Plaintiffs also contend that the Secretary'sdecision was arbitrary and capricious independ-ently of the Stipulat i on , because it conflicted witha Department Organization Order in which theSecretary delegated authority to conduct the Cen-sus to the Director of the Census Bureau . De-partment of Commerce Organization Order 35 -2A , August 4 , 1975 , as updated July 24, 1987 . 1find this argument unpersuasive. lUhile the Secre-tary did delegate his statutory duty to take the de-cennial census, he also required the Director ofthe Census Bureau to "report and be responsibleto the Assistant Secretary for Economic Affairs,"a position which subsequently became the UnderSecretary for Economic Affairs. Id ; IS U . S . C. &1503a . The Secretary further directed the UnderSecretary for Economic Affairs to "exercise pol-icy direction and general supervision over... theBureau of the Census . " Department of CommerceOrganization Order 10-9 , § 4 . 03 , June 26 , 1984 .Thus , while delegatin g the operational responsi -631ity necessary to prepar e and conduct the cen-sus, the Commerce Department retained the au-thority to control policy direction , to exerc i se de-c i sion-making authority in significant Bureaumatters, and to superv ise the Bureau in th e exer-ci se of it s ce n s u s- taking task .

[*"`37] II. The Constitutional Requirements

In Franklin v. Massachusetts, th e Supreme Courtrem inded u s th at [HN5]in making decis i on s regardingth e cen su s, "the Se cre tary 's in terp reta tion [of Art. I , § 2,cl . 3 ] [must b e ] c on s istent with th e cons t i tutional lan -guage and the constitutional goal of equal representa-tion . " 11 2 S . Ct . a t 277 7. The language of the Constitu-tion i s beguilingly simpl e: "The actual enumeration shallb e made . . . in such manner as [ th e Congress ] shall byLaw direct ."

While the defendants contend that the phrase "actualenumeration" bars adju stment , I have previously con-cluded "that because Article 1, § 2 requires the census tobe as accurate as practicable, the Constitution is not a barto statistical adjustment . " Ci o New York II 739 F.Supp. at 767 ; cf. Kirkpatrick v Preisler , 394 U . S. 526 ,530. 22 L. Ed . 2d 519. 89 S . Ct 1225 (1969) ("the wholethrust of the 'as nearly as practicable' approach is incon-sistent with adoption of fixed numerical standards whichexcuse population variances"); Wesberrv v. Sanders . 376U S 1 7-8 I ] L . Ed 2d 4 81, 84 S . Ct . 526 (1964) ("asnearly as is practicable one [ * *38 ] man ' s vote in a con-gressional election is to be worth as much as another's") .

The defendants also claim that judicial scrutiny ofthe Secretary' s decision for accuracy is inappropriateafter the Supreme Cou rt' s recent decision in UnitedStates Dep't cf Commerce v . Montana 118 L . Ed . 2d 87112 S Ct. 1415 C19921 . In Montana, the state chall engeda federal statute governing the method by which Repre-sentatives are allocated to the states because it resulted ingiving Montana only one congressional seat , although itspopulation was significantly hi gher than that of the aver-age [ * 920] co n gressional district in the nation. TheCou rt rejected the challenge, not ing that [HN6]"althoughcommon sense suppo rts a test requirin g a good faith ef-fort to achieve precise mathematical quality within eachState, the constraints imposed by Article I , § 2, itselfmake that goal illusory for the nation as a whole." /d at1429 (emphasi s in orig in a l) (citation omitted) . The spe-ci fi c constraints making mathematical precision illuso ryon the national level were " the constitutional guaranteeof a minimum of one Representative for each State , " and"the need to allocate [**39] a fixed number of indivisi -ble Representatives among 50 states of vary i ng popul a-tions ." Id.

I reject the government's argument that Montan amandates a depa rt ure from my earlier conclusion that theSecretary of Commerce must conduct the census in amanner to render it as accurate as practicable . First, theconstitutional cons traints that warranted departure fromthat standard in Montan a are not presen t here . S econd , inMont an a, th e Court n ote d that Art,.I -§-8 c l. 18 o t th eConstitution "expressly a utho ri ze s Congress to e n act

Page 1 2

822 F . Supp. 906, *; 19 93 U. S . Dist . LEXIS 48 18 , **

le g islation that 's hall be ne cess ary and proper ' to carryout its del egated re sponsi bilitie s." Id. H e re , no con s titu-tional prov is ion require s s imil ar deference to the Secre-tary's decision. Final ly , th e Montana case involved achallenge to a c en s u s pro cedure only as i t related to ap-port ionment, not as it related to intra-state redi s trie ti n g.Here, by contrast, the decision on whether to adju s t th e1990 census had profound effects on intra-s tate redis-tricting becau s e the adjusted counts would change notonly national and s tate population figures, but the count sfor political subd i visions within states, s uch as cities andcounties . Because [ * *40 ] the implications of the Secre -tary's decision at issue here are fundamentall y differentfrom the federal statute at issue in Montana, I adhere tomy earlier conclusion that the Secretary must assure thatthe census be as accurate as practicable.

The conclusion that the Secretary must provide themost accurate census practicable however , does not, leadinexorably to the conclusion that a decision against ad-justment is therefore unconstitutional . In decidingwhether the Secretary' s decision was arbitrary and capri-cious in light of the requirement that the decision providethe most accurate census practicable, the Cou rt must turnto the Secretary's consideration of the guidelines, whichhelp to illuminate the meanin g of both "accuracy" and" practicabili ty. "

III . Th e Guidelin es

[HN7 ] An agency decision is arbi tr ary and capricious" if the agency has relied on factors which Congress hasnot intended it to consider, entirely failed to consider animportant aspect of the problem, offered an explanationfor its decision that runs counter to the evidence beforethe agency, or is so implausible that it could not he as-cribed to a difference in view or the product of agencyexpert ise . " Motor Vehicle Mfrs Ass 'n v. State Farm Mut.Auto Ins Co 463 U . S 29 43 77 L Ed 2d 443 103 S.Ct. 2856 (1983� . [**41] Here , the analytical scaffoldingfor review of the Secretary's decision i s established bythe guidelines promulgated in accordance with the Stipu-lation. "

19 A discu s sion of how the guidelines wereformulated, considered , and ultimately, promul -gated may be found in the earlier opinion inwhich I rejected a challenge to their s u fficiency .Cite New York I! 739 F . Supp. at 769 & n . 9

The Stipulation provided that the Secretary retainedall authority and decision -making power, " includingwithout limitation th e decision whether or not to adjustthe 1990 Decennial Censu s. " St ip . at I. It also requi redthe defendant s to " d evelop and adopt guideline s art iculat -ing what de fendant s bel ie v e are the relevant techn i caland no ntech ni ca l stati sti c a l and po l i cy ground s fo r de ci-

s ion on whether to adjust the 1990 Decennial Censuspopulation courts." Stip , at 3 . Accordingly, the defen-dants promulgated the following eight final guidelines toserve as the gr i d again s t which th e S e cretary 's decision( ** 42 1 mu s t be measured :

1 . The Census shall be considered the most accuratecount of th e population of the Un ited States , at the na-tional, state, and local level , unless a adjusted count isshown to be more accurate . The criteria for accuracyshall follow accepted statistical [ * 921] practice andshall require the highest level of professional judgmentfrom the Bureau of the Census. No statistical or inferen-tial procedure may be used as a substitute for the Census.Such procedures may only be used as supplements to theCensus.

2. The 1990 Census may be adjusted if the adjustedcounts are consistent and complete across all jurisdic-tional levels: national, state , local, and census block. Theresulting counts must be of sufficient quality and level ofdetail to be usable for Congressional reapportionmentand legislative redistricting, and for all other purposesand at all levels for which census counts are published.

3 . The 1990 Census may be adjusted if the estimatesgenerated from the pre-specified procedures that willlead to an adjustment decision are shown to be more ac-curate than the census enumeration . In particular , thesee stimates must be shown to be robust to variations inreasonable [ * *43] alternatives to the production proce-dures , and to variations in the stat i stical models used togenerate the adjusted figures .

4 . The decision whether or not to adjust the 1990Census should take into account the effects such a deci-sion might have on future census effort s.

5 . Any adjustment of the 1990 Census may not vio-late the United States Constitution or Federal statutes .

6 . There will be a determination whether co adjustthe 1990 Census when sufficient data are available, andwhen analysis of the data is complete enough to makesuch a determination . If sufficient data and analysis ofthe data are not available in time to publish adjustedcounts by July 15 , 1991 , a determination wi ll be madenot to adjust the 1990 Census .

7. The decision whether or not to adjust the 1990Census sha ll take into account the potential disruption ofth e process of th e orderly transfer of political representa-tion likely to be caused by either course of action.

8 . The ability to art iculate clearly the basis and im-plications of the decision whether or not to adjust shallbe a factor in the decision . The general rationale for thed e ci si on will be clearly stated. Th e technica l docum enta-ti o n [ * *44 ] l y ing beh i nd the adju stment de ci si on s hall b e

Page 1 3

822 F . Supp . 906 , *; 199 3 U. S . Di st . LEXIS 4818 , **

i n keeping with profess ional standards of the s tati s ticalcommunity .

Citv of New York !l 739 F. Supp . aY 769 (emphas i s inori g in a l) .

mi s s people who are difficult to surv ey, whileDA, which relies solely on an examination of re -cords , will not. Dec ision at 2 - 10 .2 1 Some examples of thi s problem cited by theSecretary are :

"Most o f th es e g u i del ine s are embroidered with anaccompanying ' e xplanation :" Id. The Decision discuss edeach of the guideline s in detai l , and concluded that num-bers 1 , 2 , 3 , 4 and 7 militated against an adju stment ,whil e numbers 5 , 6 , and 8 did not tilt either way . Theplaintiffs argue that the conclusions reached by the Sec-retary with respect to guidelines 1, 2 , 3 , 4 and 7 are theresult of i mplausible assumptions, unwarranted specula-tion , and mi suse , misstatement, and disregard of the evi-dence .

Guideline OneGuideline One , establi she s the point of departure for

analysis of the adjustment question. It mandates that theactual count be considered the most accurate count of thepopulation " at the national state and local level, unless anadjust ed count is shown to be more accurate ."

To te st the accuracy of the adjusted counts againstthe actual enumeration , the Decision referred to a popu -lation measurement technique that the Census Bureau ,had used , [** 45] Demographic Analysis ("DA"). Dec i-s ion at 2-9 . DA estimates the population, and the sub-populations of particular groups , through administrativerecords such as birth and death cert ificates , and immigra-tion statistics. Id. The Secretary conceded "that the PESand DA estimates are not far apart in a statistical sense,"but found "some important and puzzling differences , "which "lead to sharply different conclusions" and raisesome "question" as to " whether the adjusted fi gures aremore accurate than the census count even at the nationallevel . " Id at 2-10 2,_35 .

Among the specific probl ems that the Secretarynoted were : (1) that the overall undercount rate inferredfrom comparing the [ M 922) actual enumeration to DAi 1.85%) is s maller than that inferred from the PES(2. 07%) , a result contrary to intuitive expectation ; a 0 (2)that, at the national level , there were instances where aPES-based adjustment would move sub-population totalsin the oppo s ite direction from that indicated by DA; "(3)that PES would add 1,055 , 82 6 more females than DAi ndicates should be added; and (4) " that all groups ofblack males (except those aged 10-19) are substantiallyundercounted by the PES [" * 46] relative to DA . " De ci-sion at 2 - 10 .

20 According to t he S e cretary, DA would nor-mall y be expected to revea l a hi gh e r undercountrate when compared to t h e Cen s u s than the PESwo u l d , beca u se the PE S and Cen s u s will both

An adjustment based on the PESwill add 180 ,318 non-black malesaged 10 - 19 , while the DA indi-cates 136 , 908 should be de[eted --a difference in the wrong directionof 317 ,226 .

An adjustment based on thePES will de lete 91 ,631 males overthe age of 65, while DA indicatesthat 192 , 950 should be added -- adifference in the wrong directionof 284,541 persons .

An adjustment based on thePES will delete 245,253 femalesover the age of 45 while DA indi-cates 146, 255 should be added -- adifference of 391,508 persons inthe wrong direction .

Decision at 2-12 (footnotes omitted).

[ ** 47] In addition to a compari son with DA, theSecretary also discussed a number of other statis ticaltechniques that were used to gauge the accuracy of thePES counts when compared to the census results . Heconceded that the PES-adjusted estimates might reflectmore accurately the total population, and the racial andethnic subpopulations of the country, and that "at theState and local level . ... the adjusted figures tend to betoo h igh, but generally closer in numeric terms to the truepopulation than the census counts which tend to be toolow . " Id. at 2-1. He concluded , however, that "the lossfunction analysis and hypothesis tests that have beenprepared by the Census Bureau to date, although of un-certain reliabili ty, do support the superi or accuracy of thecensus counts versus the adjusted figures when we con-sider distributive accuracy -- or fairness -- and use rea-sonable estimates of the error variance of the alternative[PES-based adjustment] . " Id. at 2-2.

He also expressed concern that there was li ttle or nodirect evidence that the adjusted counts led to greaterdistributive accuracy at local levels . On that basis, theSecretary concluded that Guideline One militated [ **48]against adjustment because "acceptance of adjustedcounts as more accurate requires not only that the countsthemselves be shown to be more accurate, but that th edistribution of those counts acro ss th e United State s re-

Page 14

822 F . Supp . 906 , " ; 1993 U . S . Dist. LEXIS 4 818 , **

fleet more accurate ly the distribution of the populati on ."Id. at 2-8 ( emphasi s in o ri ginal) .

As support for these concerns, the Secretary di s -cu ssed a Census Bureau lo ss function ana lys is thatmeasured the number of individual s tates whose popula-tion would be made less accurate by adjustment than byus i ng the c ensus count . As conducted by the Bureau , theloss funct i on indicated that 21 states' population shareswould be made le ss accurate by adjustment . However,when the Secreta ry employed a statistical variance to-ward the low end of the acceptable range envisioned bythe USC , he found that the proportional shares of 28 or29 states would be worsened by adjustment. Id at 2-30.The Secretary similarly expressed his trepidation thatthere was insufficient evidence to support the greaterdistributive accuracy of the adjusted counts at the locallevel . Id.

tion pro cess . First, he noted that while th e rate s o f impu -tation in the P and E sampl es " were low - - 1.7% and2 .1 % respectively -- weighted up to the national popula-tion th ey repre sented almo st nine m i llion people , a num-ber almost twice as l arge a s the n e t nati onal undercount .Decision at 2 -16. Second , in noting the high correlationbetween imputation ratios and undercount ratios , hestated that "the strata for which there is more doubt aboutthe quali ty of the adjusted data because of imputationtend to be the same strata for which a adjustment wouldresult in large increases in the population." Id. Finally,the Secretary noted that the assumptions in the i mputa-tion models were largely untested . Decision at 2-17 . Hisconcern was exacerbated by his respect for the researchof Panel member Kenneth Wachter, which indicated thatflaws in the imputation model could render the adjustedcounts "significantly in error . "

In his considerat i on of Guideline One , the Secretaryalso expressed serious concern over the [* * 49] method-ology by which the PES was taken , and the manner inwhich the adjusted counts were tabulated . " He was par-ticularly discomfited by the manner in which unresolvedcases in the PES were tre ated because the Bureau had todetermine whether [ * 923] people found in the PESwere also found in the cens us i n order to compute dualsystem estimates for the poststrata . Such determ inationswere made by "matchin g " census forms to PES forms forthe same household . A household survey in the PES thatwas " matched" to the census record of that residencemeant that there was no error in the census enumerationof that household . A non-match meant an undercount .Decision at 2-16 . Because there were cases where in-complete census and PES forms made such matchingimpossible, the Bureau was forced to employ a mathe-matical model to impute enough miss ing characteristicsto enable it to make a match determination. Even afterthat imputation was complete , there were people found inthe PES for . whom it was impossible to determinewhether they matched people counted in the census , andvice-versa . In those cases a different set of formulas wasu sed to impute match status .

2 2 Among other thin gs , the Secretary was trou-bled by the effects that erroneous enumerations inthe c ensus , correlation basi s , and fa ilure in thePES total e rror model could have on the adjustedcounts . Ic1 at 1-17 -23 . The proof at trial, how-ever, has made it clear that these matters were pe-ripheral to the Secretary's conclusion underGuidel i ne One , and therefore do not merit sig-nificant di s cussion here .

[**50] Th e Secreta ry concluded that, "in general ,mi ssi n g da ta were no t found to be a ser i ou s problem , "bu t i de nt i fi e d s evera l a reas of conc ern with th e imputa-

23 In the jargon of the DSE , the "P sample"represented the group surveyed by the PES . The"E sample" represented the people living in thesame household as the P sample as counted bythe census . Decision at 4-12 -4-13 .

[ ** 51 ] Based on all the foregoing, the Secretaryconcluded that there was simply not enough convincingevidence to support a finding that the adjusted countswould lead to greater distributive accuracy than the cen-sus counts, and therefore that the guideline "weighed infavor o£a decision not to adjust . " Id at 2-36.

The plaintiffs assail this conclusion on severalgrounds . First, they argue that the Secretary misused DAbecause that technique has historically been most accu-rate as a "yardstick of the census in terms of nationalundercount and as a measure of differential undercountbetween demographic groups , " but "is much less reliablein its ability to estimate an undercount rate for a speci fi cgroup in a particular census , " as the Secretary attemptedto use it . They contend that the Secretary's focus on dis-crepancies between PES and DA undercount rates forcertain, specific groups is an attempt to obfuscate the factthat , as the Secretary himself admits, the "detailed analy-sis shows that the PES and DA estimates are not far apartin a statistical sense." Decision at 2-10,

With respect to the Secretary 's professed concernover distributive accuracy , the plaintiffs contend [ **5 2]that the Secretary's invocation of a loss function thatmerely counted up the number of states whose popula-tions would be made less accurate, regardless of thegreater aggregate accuracy of the adjustment , and with-out reference to the extent that counts are made less ac-curate, is statistically in suppo rtabl e. They al s o argue thatthe Secretary's rejection of numerous loss function analy -s es performed by the Burea u s upporting th e s uperi oraccuracy of th e adju s te d co un ts, and h is p utativ e co n cern

Page 15

822 F . Supp . 906 , *; 1 99 3 U. S . Dis t . LEXIS 48 1 8, **

with the technical aspects of the PES are i rrati o na l atb est , and disingenuous at wors t .

I have reviewed in some detail the Secretary' s con -clus i on that Gui deline One militate d aga i nst adju stmentand the plainti ffs' arguments to the contrary. While theplainti ffs have made a compelling attack on the Dec ision,and the Secretary has conceded that the objective criteri aused to measure the [ * 924] adjusted counts show agreater numeric accuracy at the national level and thatthe Census Bureau estimates of dis tri butive accuracymarginally favor the adjusted counts, I find that Secre-tary's conclusion under Guideline One was neither arbi-trary nor capricious.

The Secretary's decision to focus on distributive,[ *M 53] rather than numeric , accuracy was consonantwith the constitutional goal of assuring the most accuratecensus practicable, given the census's function as a stan -dard by which to distribute political representation andeconomic benefits . In that regard, I find that the Secre-tary's use of a loss function that considered the numberof states whose populations would be made less accurateby adjustment to be appropr i ate . Similarly , the Secre-tary's concern th at " with respect to places under 100 ,000population , there is no direct evidence that adjustedcounts are more accurate" was legitimate , given Guide-line One ' s requirement that the adjusted counts be shownto be more accurate at the local level. Decision at 2-30 .

Plaintiffs' attack on the Secretary for subject ing thetests favoring adjustment to unrealistically rigorous s cru-tiny misconstrues Guideline One, which clearly statesthat "the Census shall be considered the most accuratecount of the population of the United States, at the na-tional , state , and local levels , unless an adjusted count isshown to be more accurate. "Citv ofNew York 11 739 F.Supp , at 769 (emphasis added). Thus, plaintiffs' [ ** 54 ]failure to illustrate affirmatively the superior accuracy ofthe adjusted counts either (1) at any level mentioned inGuideline One, or (2) for any reasonable definition ofaccuracy , is sufficient to support a finding that Gui delineOne favors use of the original census counts .

Turning to the Secretary's focus on the mechanics ofthe PES and the use of imputation , I find this considera-tion appropriate . Because the PES , like any sample sur-vey, 1 " rests on an assumption to begin with - - that thepo rt ion sampled is identical to the population as a whole-- placement of additional assumptions into the modelcaused by imputation is a fair basis for escalating skepti -cism. While the log ic of the Secretary's conclusion re-garding Guideline One is not overpowering , neither canit be characterized as arbitrary or capric i ou s.

24 See the d isc uss i o n of capture/recapture innote 4, s upra.

Guideline Two

Guideline Two state s that adjustment may be madeonly i f th e adjusted count s are "cons i stent and completeacro s s a ll [**55] jurisdictional levels: national , state ,local and cen s us block ." Th e guideline also requires theadjusted counts to "be of sufficient quality and level ofdetail to be usab le for ... a ll .. . purpose s and at all lev -els for which census counts are published . " City of NewYork 17 739 F . Supp . at 769 (emphasis in original) . TheSecr etary r ecogn i zed that " the adjusted figures . . . areconsistent acros s all jurisdictional levels and of sufficientdetail for all purposes," but nevertheless concluded thatGuideline Two militated against adjustment because ofthe questionable quality of the adjusted counts. Decisionat 2-45.

In surmising that the counts were of debatable qual-i ty , the Secretary homed in on the "homogeneity assump-tion" in the construct of the 1 , 392 poststrata . He wastroubled that the "adjustment process rests on the as-sumption that persons in each poststratum are homoge-neous with respect to their probability of being missed bythe census, i.e., their capture probabili ty ." Decision at 2-39 . Conceding that many experts did not find this as -sumption problematic , and that at broad levels such asthe national and state levels the [* M 56] assumptioncaused no serious problems , the Secretary ultimatelyconcluded that " ' loc al heterogeneity is a serious problemfor adjustin g the 1990 census"', and that "'[the] evidenceindicates that a substantial portion , possibly a majority ,of relative counts for district-size units can be madeworse off by adjustment . "' Decision at 2-42 (quotingReport of Special Advisory Panel Member Kenneth Wa-chter, at 26) .

In reaching this conclusion, the Secretary worriedthat because members of an individual posts tratum mighthave a different likelihood of being undercounted,thereby debunking the homogeneity assumption, general-izing the undercount rate of those counted in [ * 925J thePES to all members of that poststratum might seriouslyinterfere with the accuracy of the count for some censuspurposes , including redistricting . He discussed two stud-ies conducted by the Bureau that addressed the homoge-neity assumption , and which the Bureau had re lied on inconcluding that individuals within each poststratum weresufficiently uniform to warrant such an assumpt ion ,dubbed the "P12 " and "P15" studies . Decision at 2 -38-2-40 . The Secretary opined that the Bureau's evidence fromtho s e studies [ * "57] was "mixed . "

He was al so concerned about the adverse conse-quences that a failure in the homogeneity assumptioncould have on adjustments at l o c a l lev e l s, noting thatbecause there were only 5000 sample blocks, mos t jur is -

Pa ge 16

822 F . Su p p . 906, *; 1 99 3 U. S . D is t . LEXI S 4 8 18, **

dictions would be adju sted based on data gath ered e l se-where . Dec ision at 2 - 4 3 .

The plaintiffs brand the S e cretary's concern abouthetero geneity as " unreaso nabl e." They asser t that b e-cause perfect homogeneity is utterly unattain able in theworld of survey sampling, th e relevant question iswhether a departure from the homogeneity assumptionhas an impo rtant impact on the measurement . They con-tend diat because the pert inent Census Bureau studiessuppo rted the homogeneity assumption , and particularlybecause the P12 study confirmed that the populationsubgroups defined for the PES are sufficiently uniform tobe usable for adjustment , there was sufficient homogene-ity to warrant the conclusion that the adjusted counts leadto improvement .

Plaintiffs' argument i s rejected . While they havemade a strong showing that the adjusted counts are moreaccurate than the ori ginal counts for most purposes forwhich the census is used, the Secretary 's concern thatheterogeneity [ ** 58a may lead to less accurate counts atlocal levels used for redistrictin g appears reasonable.Plainriffs' contention that the Secretary was effectivelyrequired to bite the bullet an d i gnore the problem thatresidual heterogeneity posed, once the Bureau had con-cluded that there was sufficient evi dence to support thehomogeneity assumption , ignores the guidelines' man-date that the Secretary determine that the adjusted countsbe usable for all purposes for which census counts arepublished. Clearly , there is some likelihood that residualheterogenei ty wi ll have an adverse effect on the censuscounts when used for redistricting . This is enough tosupport the Secretary's conclusion that Guideline Twomilitates against adjustment . Accordingly, I find that hisconclusion was not arbitrary or capricious .

Guideline Three

Guideline Three requires that the PES and other ad-justment procedures be "pre-speci fi ed" and that the esti-mates they generate be "shown to be robust to variationsin reasonable alternatives to the production procedures ,and to variations in the statistical models used to gener-ate the adjusted figures . " Citv of New York II 739 F .Supp. at 769 . [ ** 59] The Secretary advanced two argu-ments as the basis for his conclusion that GuidelineThree militated against adjustment: (1) that the actualconduct of the DSE did not proceed sufficiently in ac-cordance with a pre-specified plan ; and (2) that cert ainstatistical techniques and assumptions were not suffi-ciently "robust" to support adjustment . " Dec ision at 2 -54-2-55.

25 In the world of survey sampling, "robust-ness" desc r ib es the integrity and reasonablen es so f the results achieved by a p art i c ular statistical

technique. Robu stne ss is determ ined by exp os ingsuch statistical techniques to variation s in the as-sumpt i on s underlyin g them .

On the fir s t point, the Secretary recounted variou sdecisions that Bureau employee s made after the pre-specification of the PES , including cho ices about thes election of carri er variable s during the regression analy-s is in the smoothing proce s s . Decisio n at 2 -47 . He notedthat one member of the Panel who voted for adjustmenthad conceded that certain pre-specified procedures had[ ** 60 1 changed during the enumeration process and hadaffected the PES. Id. (citing Report of Panel MemberWolter, pp. 9-10) . The Secretary agreed with Wolter'sultimate conclusion that the decisions to change pre-specified procedures made during the enumeration andthe PES were treated with a high degree of professional-ism and al so acknowledged that the PES could not havebeen completely pre-specified, [ * 926] but expressed hisdiscomfort with the deviations as follows:

Although I b e lieve that the dec i sions [todeviate from pre -specifi ed procedures]were made for sound professional reasonsin the 1990 censu s, using these adjustmentmechanisms opens the possibility for ma-n ipulation of future post enumeration sur-veys i n ways that are unavailable in tradi-tional census procedures. This weighsheavily against an adjustment of the cen-sus .

Decision at 2 -4 8 (emphasis in original) .

With respect to the robustness of the results whensubjected to alternative statistical models required byGuideline Three , the Secretary concluded that "the re-sults of the adjustment procedure are broadly robust at anaggregate, national level." Id. at 2-54. However , he foundthree questionable [ ** 61] areas where the adjustmentmethods concerned him : (1) imputation ; (2) poststratifi-cation; and (3) the use of smoothing procedures .

The Secretary concluded that the imputation wasstatistically robust, but expressed a fear that variation s ofthe assumptions underlying the imputation could have aneffect on the apportionm e nt of the House of Representa-tives . Id. at 2 - 48-2-49 . With respect to poststratification,the Secretary observed that if poststratification had rec-ognized the state of re sidence rather than the census divi-sion of residence as a factor , three states would have hadsignificantly different counts . Id at 2 -49 . Finally, mov-ing to th e robu s tnes s of smoothing, th e Secretary con-cluded that th e numerou s d e ci s i on s and techniqu es i n-vo lved i n the two-stage pro c ess, includi ng t he disc re tio n-a ry s electi o n o f c arr i er var i ab l es, le d t o a n impermissibly

Page 17

822 F . Supp . 906 , *; 1993 U. S . D is t . LEXI S 4 8 1 8, **

hi gh level of un cert ainty to employ the adju sted count sas a bas is for reapportionment . Id. at 2 - 49-2 -5 4 . In sh o rt ,the Secretary s tated that the l ack of compre hensive pre-sp e cification , th e p oss ib i li ty it ra i sed for future politicalmanipulat i on , and th e uncert a i nty associated w i th th e useof extensive stati stical assumpt ions [**62 ] in the ad-justment proc ess l e d him to find that Guideline Threemil i tate d again st adjustment .

In th e ir attack on this conclusion , the pla i ntiffs firstargue that the Secretary' s concern over political manipu-lation of future censuses because of the lack of pre-specification is an inappropriate basis for making a de-termination under Guideline Three . While I tend to agre ewith that argument, I read the Secretary's discussion offuture political manipulation as merely an explanatorynote, underscoring why he thought that pre-specificationwas so significant . Because Guideline Three clearlymandated pre-specification , the Secretary's well-suppo rted conclusion that the procedures were not ade-quately pre-specified supported his conclusion under thisguideline .

Plaintiffs also argue that the Secretary required animpo s sible degree of pre - specification because some ofthe decisions to be made, including deci sions relating tothe smoothing process , were highly dependent on data tobe collected during the PES, and therefore could nothave been completely pre - specified . This argument ig-nores the fact that certain techniques were pre-specifiedand then changed later . See Report of Special Advisory[ M *63] Panel Member Kirk M Wolter at 1 0 . The Secre -tary's conclusion that pre-specification did not occur ascontemplated by the guidel i ne was justified .

Plaintiffs also belitt le the Secreta ry 's concern thateven small changes in any of the assumptions underlyingthe statistical procedures of the adjustment could lead toa different apportionment of the House of Representa -tives. Deprecating his conclusion that this seriouslycompromised use of the adjusted counts , they contendthat because the Secretary has conceded that smallrh;inge s in the, ce nsus methodology can move Houses eats as readily as small changes in the PES methodol -ogy , his concern under Guidel ine Three is illusory .

I disagree . The plaintiffs' reliance on the imperfec-tions in the census to blink at s imilar unce rtainties in theadjustment procedure misses the point that, under therubric of the guidel i nes , the adjusted counts must satisfycertain criteria , regardless of whether the original enu-meration could survive exposure to similar criteria. Itmust be remembered that under Guideline One, the pre-sumption of accuracy runs in favor of the original cen su s[* 927 ] c ount . B ecau s e th e Secretary's concerns over pre-specification [ ** 64 ] and the robustness of adjustmentdat a were l egitima t e , I find that th e S ecretary ' s co n c lu -

si on under Guid elin e Three wa s not arb i trary or capri-cious .

Guideline Four

Guideli ne Four counse l s that " the decision whetheror not Co adjust the 1990 Census should take into accountthe e ffects such a decision might have on future censuseffo rts . " Ci��New York li 739 F . Supp at 769. Withthi s in mind, the Secreta ry stated that he " did not findcompe ll in g evidence in e ither direction regarding theeffects of a decision on future individual motivations . "Decision at 2-58 . Weighing the effects that an adjustmentmight have on the effort s of state , community , civic , andinterest group leaders , the Secretary was concern ed that"an adjustment [would] remove the incentive that thesepublic officials and groups currently have to provideactive suppo rt in achieving a complete count. " Id at 2-59 . The Secretary found "unpersuasive" the contentionthat , even with an adjustment, local officials would retaina strong i ncentive to gather data , and "[found] no evi-dence indicatin g that local support would decrease as aresult of a decision not to adjust the [ ** 65] census." De-cision at 2-59 .

He went on to conclude that a decision to adjustcould hinder the operations of the census in other ways,including disincentive s for Congress to provide funding ,and for enumerators to pursue their task energetically ,and the possibili ty that adjustment could be distorted forpart isan political purposes in future censuses . Id. at 2-60.Balancing all these fears , the Secretary concluded "thatan adjustment would adver s ely affect future census ef-forts to a greater extent than any adverse effects of a de-cision not to adjust . " Id, at 2-61 .

Plaintiffs argue that it is futile to fret over censusesin the year 2000 and beyond in considering whether ornot to adjust the 1990 census. This argument blithelyignores the express mandate of Guideline Four that theeffect of the Secretary's decision on future censuses beconsidered. While I recognize that Guideline Four cre-ates a potential tension with the constitutional require-ment that the census be as accurate as practicable , underthe circumstances of this case , that tension is minimal.Accordingly , I find that the Secretary's conclusion re-garding Guideline Four was neither arbitrary nor capri-cious .

[ * "66] Guidelines Fiv e & Six

Because the Secretary's conclusions based on Guide-lines F i ve and Six are not challenged by the plainti ffs , Iwill only say that the conclusions reached by the Secre-tary in the Dec ision sufficiently considered tho s e guide-lin es. ' O

Page 18

822 F . Supp . 906 , '; 1993 U . S . Dist. LBXIS 4 818 , **

26 Guidel ine Five provides that adjustmentcann ot violate th e Constitution or any federalstatute . Th e Secretary concluded that because heh ad r e ach ed a de c i s i o n n ot t o adju s t based onother factors, "legal cons ideration s did not pr o -v i d e a basis" for hi s decision . Decision at 2 -65 .

c luded that , ev en if it were tru e that adju stm ent wouldres ult in a fairer distribution of funds , this considerationwould pale in comparison to the dis rupti on of politicalrepresentation that would en su e from a de ci s ion to ad-j ust , because " adjustment would not re s ult in significantshift s in those funds." Id at 2-75 .

Guidel i ne Six mandate s that i f adju ste dcounts could not be publish ed by July 15 , 1 991 , ad etermination would be made against adjustment .Although adjusted counts were ready to be pub-lished by July 15 , 1991, the Secretary had con-cluded not to adjust , and so this guideline becamemoot .

Guide lin e Seven

Guideline Seven provides that the decision whetheror not to adjust the 1990 Census shall take into accountthe potential disruption of the process of the orderlytransfer [ ** 67 1 of political representation likely to becaused by either course of action . " Citv o New York II739 F Supp at 769 . At an earlier stage of this litigation,I rejected a request to vacate Guidel ine Seven, findingthat it , and Guideline Eight , might, " in a con structivefashion, help define the meaning of 'the most accuratecensus practicable , " ' and concluded that, at least to thatextent, they were permissible factors. ld at 771.

In his consideration of Guideline Seven , the Secre-tary noted that the Clerk of the United States House ofRepresentatives had officially ce rt ified to each of thefifty states the number of seats allotted to that state forthe 103rd Congress (convened in January 1993) based oncensus figures released on December 26 , 1990, and that ,as of May 1991 , "some 20 states had already enactedeither or both of their Congressional and State legislativeredistricting plans . " Decision at 2-71 . The Secretary thenwent on to outline [*928] the disruption and delay thatan adjustment would cause , particularly i n those stateswhere adjustment would change their allotted number ofseats in the House of Represenfarives .

[ ** 68] It should be remembered that Congress de-creed in 1912 there be only 435 seats in the House ofRepresentatives. We are , therefore, dealing with a zero-sum game ; when one state gains a seat , another must loseone. If the adju stment were made, Californ ia and Ari-zona , for example , would each gain one seat in theHous e , while Pennsylvania and Wisconsin would eachlose one . Id. at 2 -72. The Secretary envisioned massivelitigation over such a decision.

Ultimately , the Secretary concluded that GuidelineSeven favored adherence to the census counts . He re-jected th e argument that non-adjustment i s "inherentlydi sruptiv e," as based on th e que s ti on-begging premisethat th e adj u ste d coun ts are more accurate . H e also con -

Plainti ffs assail the Secretary's Guideline Seven con-clusion on two distinct grounds. First, they suggest that iti s disin genuou s for the Secretary to [ ** 69] rely on thefact that the unadjusted counts were already being usedfor reapportionment and redistricting purposes , when theStipulation required that any release of the unadjusteddata before the Secretary 's decision be acwmpanied by anotice advising recipients that they used the data at theirown risk . None of this however, detracts from the factthat Guideline Seven explicitly required the Seeretary toconsider such disruption in deciding whether or not toadjust, Nor does it contradict the simple logic of the Sec-retary's argument that a decision in favor of adjustmenton July 15, 1991, would have disrupted the reapportion-ment and redistricting that was then ongoing .

Plaintiffs also observe that the Secretary's conclu-sion that adjustment would not result in significant shi ftsin federal funds contradicts an earlier sentence in theDecision that city and state population "shares are veryimportant because they determine . .. how large a 'sliceof the pie' of federal funds go to each ci ty and state."Decision at 1-3-1 -4 . Plaintiffs are right. This , however,does not render the Secretary's decision invalid underGuideline Seven, because it involves a matter -- the allo-cation [ ** 70] of federal funds -- only tangentially re-lated to Guideline Seven , the basic thrust of which is theeffect of a decision to adjust "on the orderly transfer ofpolitical representation." Accordingly , while there is anobvious inconsistency in the discussion accompanyingthe result, the plaintiffs have failed to show that the Sec-retary 's conclusion under Guideline Seven was arbitraryor capricious .

Guidelin e Eight

Guideline Eight requires the Secretary to art iculatethe factors relied upon in reaching his decision , an d alsorequires that "the technical documentation lying behind[his] decision sha ll be in keeping with professional stan-dards of the statistical community . " Citv oJNew York 11739 F. Supp. at 769 . Because the plaintiffs do not spe-cifically attack the Secretary's decision under this guide-line, and because the Secretary concluded that applica-tion of this guideline neither favored nor militatedagainst adjustment, I find that the Secretary's Decisioncomplied with Guideline Ei ght .

Page 19

822 P. Supp . 906, *; 199 3 U . S . Dist. LEXIS 4818 , **

Hav ing thu s parsed the guideline s , the Cou rt con-cludes that the Secre tary 's conclusions und e r each guide-line and hi s ult imate d ec is ion against adjustment [ ** 71 1cannot be chara cter ized as arb i trary or capri c i ou s. Thebreadth of the guide line s left the Secretary enormousdiscretion. Plaintiffs have made a powerful cas e that di s-cretion would have been more wisely employed in favorof adjustment. Indeed , were this Court called upon todecide this is sue de novo, I would probably have orderedthe adjustment. " ( * 929j However, it is not within myprovince to make such determinations. The question iswhether the Secretary's dec i sion not to adjust is so be-yond the pale of reason as to be arbitrary or capricious .That far I cannot go .

27 Additionally , I note that in light of recentimprovement in statist i cal tools and the practicalbene fits that the 1990 PES has provided , the useof adjustment in the next census is probably in-evitable .

One of the central tenets of our founding fathers wasthat the role of the jud iciary should be carefully deline-ated , especially when the controversy related to the man-agement of the government . As Hamilton wrote :

The administration of government,[**72] in its largest sense , comprehendsall the operations of the body politic ,whether legislative, executive, or judici -ary ; but in its most usual and perhaps inits most precis e signifi cation, i t is limitedto executive details , and fall s peculiarlywithin the province of the executive de-partment.

The Federalist No . 72 at 450 (Henry Cabot Lodge , ed .,1888)

The writings of Montesquieu and Locke bristle withthe notion of separation of powers . But nowhere is itarticulated more succinctly than in the MassachusettsConstitution :

In the government of this common -wealth , the legislative department shallnever exercise the executive and judicialpowers, or either of them : the executiveshall never exercise the legislative and ju-dicial powers, or either of them: the judi -cial shall never exercise the legislativeand executive powers , or either of them :to the end it may be a government of law sand no t of men.

Mass . Const . pt.l , art.30 (1780) .

True , [HN8]the APA sanctions jud i cial interventionwhen the parties feel aggrieved by a fin a l administrativerulin g. But the APA tightly cabin s judicial oversight,permitt ing judicial intrusion only when the adminis tra-tive decis i on abuses reason. [ ** 73 ] It i s es s ential to themaintenance of judicial integrity that courts revi ewingsuch determinat ions zealously adhere to the arbitrary andcapricious standard of review . Citiz ens to Preserve Over-ton Park Inc v Volpe 401 U S 402 416 28 L. Ed. 2d136, 91 S. Ct . 814 (1971) (When reviewing agency ac -tion under the arbitrary and capricious standard of re-view , "the ultimate standard of review is a narrow one.The court is not empowered to substitute its judgment forthat of the agency."); Hudson Transit Lines v . UnitedStates ICC, 765 F.2d 329 , 336 (2d Cfr. 1985) ("while arev iewing court may not supply the basis for the agency'sdecision , lest it interfere with matters that Congress en-trusted to the executive agency, it will uphold a decisionof less than ideal clarity if the 'path wh ich [the agency]followed can be dis cerned "') (quoting Colorado Inter-state Gas Co v. FPC 324 U S 581, 595, 89 L Ed . 1206,65 . Ct . 829 (1945)); Connecticut Dep't ofChildren &Youth Servs. v. D�artment of Nealth & Human Servs.788 F. Suop . 573. 577 (D .D.C . 1992) ("Under this stan-dard , the Court is not fr ee to substitute its own judgment,[ ** 74 ] but is limited to determining whether the agencyhas considered all relevant factors and whether theagency ' s decision is reasonable and in accordance withthe relevant statute . Under the [APA] , the standard ofreview is highly deferential to the agency . ") . As Cardozohas reminded us, "the judge, even when he is free, is stillnot wholly free. He is not to innovate at pleasure . He i snot a knight-erran t, roaming at will in pursuit of his ownideal of beauty or of goodness. He is to draw his inspira-tion from consecrated principles . " Benjamin N . Cardozo,Nature of the Judicial Process 141 (Yale Univ . Press ,192 1) .

Midst all the sturm und dreag, after all is said anddone , the question before the cou rt distills to this : did theSecretary act reasonably? This , of course , dependsmainly upon the evidence he had before him . In his tes-timony , Dr . Robe rt E. Fay , one of the principal statisti-cians at the Census Bureau (who , incidentaly , voted toadjust) pierced right to the heart of the case : "I told theSecretary that . . . reasonable statistici an s could differ onthis conclusion." Tr . at 1909 . The Cou rt agrees, andtherefore , concludes that the Secretary 's decision not toadjust [ * *75] the 1990 census count was neither arbi-trary nor capricious .

[ * 9 3 0] The PES Tapes

Page 20

822 F . Supp . 906 , `; 199 3 U . S . D ist . LEX IS 48 1 8, **

Pla i ntiffs a lso m ove to vac ate a protective order, is-s ued by Magis trate Judge Ross, governing cert ain com-puter tapes th ey got from the Government during discov-ery in thi s c a s e . These tapes contain the adj usted cen s u sdata at the block l e vel and are the material that wouldhave been re leased to the states if the Secretary had de-cided to adjust . ' a P l aintiffs argue that the Cou rt shouldvacate the protective order because : (1) plaintiffs alreadyposses s th e tapes , and , thu s, release of the data would notviolate any institutional confidence ; and (2) release ofthe tapes i s appropriate under 13 U.S .C . & lAl(c , which[HN9]requires the Secretary of Commerce to provide thestates with data to be used in redistricting .

28 On July 15 , 1991 , the day the Secretary an-nounced his decision, he also released to the pub-lic the adjusted census data at the national , state,county , and city levels , but not the block level .Subsequently the Department disclosed half ofthe adjusted block- level data to Congress .

[**76] When they opposed production of thesetapes before Magis tr ate Judge Ross, the defendants as-serted the "deliberative process" privilege as a basis fortheir refusal. [HN 10]The "deliberative process" privilege" protects from disclosure those agency documents whichreflect 'advisory opinions , recommendations and delib-erations comprising part of a process by which govern-mental decisions and policies are formulated."' Mobil Oi1Corp v Department of Enervv 102 F A D 1 5(N . D N . Y . 1983� (quoting Mobil Oil Corp. v . Depart-ment of Enerev . 52 0 F . Supp. 414 , 416 (N.D .N.Y.I98 1 :

The privilege is a qualified or discretionary one thatturn s on a balance of competing policy claims . See In reFranklin Nat1 Bank Sec Litiy . 478 F. Supp . 577, 582(E. D.N.Y 1979) . The privilege does not protect purelyfactual material. Id. at 581. Magistrate Judge Ross as-sumed arguendo, that the redistr icting tapes reflectedcertain advisory op inions , but found that the benefit to bederived from protecting such information was out-weighed by the benefit to accurate judicial fact-finding[ * *77] that would follow upon production of the tapes .Accordingly , s he ordered the defendants to producethem , but , at the request of the defendants, also entered ap rotect i ve order forbidding public disclosure . The Magis-trate Judge was aware at the time of what she describedas th e "hotly contested Ninth Circuit litigation concern -ing disclo s ure of these same tapes . "

Assembly under the Freedom of Information Act, 5U SC . & 552(a) ("FOIA") . The Department o f Com-merce argued , as it does here , that the data should not bedi s clos ed b ecau se of th e ir pre- d e ci s i onal and deliberativenature .

The district cou rt rejected th at argument and orderedthe Department of Commerce to produce the data.AssembZy of California v United States Dep't of Com-mrce 797 F Supp . 1554 (�.D C�. [* *7 8 ] TheCommerce Department appealed . Agreeing with the dis-trict court's findings that the data were neither pre-decisional nor del iberat i ve , the Ninth Circuit affirmedthe order that the tapes be released . 968 F.2d at 923 .

A directly contrary result was reached by the Elev-enth Circuit in Florida House of Representatives v.United States Dep't of Commerce 961 P 2d 941 (l lthCir. 1992). There, the Florida House of Representativesbrought a FOIA action to compel the Department ofCommerce to release all the adjusted block-level data forFlorida. The district cou rt granted summary judgment forFlorida , and the Department of Commerce appealed . TheEleventh C ircuit reversed , finding that "because the ad-justed census block level data are a subordinate's opinionand reflect the give- and- take of th e deliberative pro ce ss .

the data are deliberative , and in turn , w ithin the scopeof the deliberative process privile ge . " Id at 950 .

Recognizing this split in the circuits , and assumingarguendo, that the tapes reflect certain aspects of thedeliberative pro cess, I believe the Ninth Circuit has thebetter [* * 79] of the argument. Whatever interest [ *931]the Department of Commerce may have had in the confi-dentiality of the block level counts , that interest was se-riously diluted when the Secretary released one half ofall the data to Congress; and , whatever privacy survivedas to the block level data for California was lost fo llow-ing the Ninth Circuit's decision .

Balanced against the slight residuary interest that thedefendants may have in the confidentiality of the blocklevel data is the public ' s interest in full access to judicialpro ceedings, especially where, as here , the dispute hassparked so much public interest . Because I believe thatthe balance weighs heavily in favor of d i sclosure underthese circumstance s , I vacate the protective order an dpermit the plaintiffs to use and to release to the public thecomputer tapes containing the adjusted block- levelcounts .

CONCLUSIONI now have the result of that "hotly contested" Ninth

Circuit case, Assemblv of California v. United StatesDep 't oj Commerce 968 F2d 916 (9th C ir 1992 . There ,the Department of Commerce was a sked to re l e a s e com-pu ter ta pes cont ai ni ng all the block-level cen s us data forCal i fo rni a purs u a n t to a claim by the Ca l ifornia State

To sum up L (1 ) find that the Secretary's decisionnot to adjust the 1990 census does not violate the APA,the Constitution, the Stipulat i on , or any statute; and (2)v acate the protective order governing th e plaint iffs ' u se

Page 2 1

8 22 F . Supp . 906 ,'"; 199 3 U . S . Di s t . LEXIS 4818, **

of the computer tapes contain ing the adjusted block-le velcounts .

While plaint i ffs' counsel [ ** 80] has illustrated tha tadjustment is statistically feasible, and would improvethe quality of the counts for most purpose s while amelio -rating the profoundly disturbing problem of differentialundercount , the Court cannot , on the record before it ,supplant the Secretary's decision .

Finally , I would be remiss if I did not note the mag-ni fi cent contribution that several law firms representingthe plaintiffs have made in presenting this case to the

Cou rt. These firm s, Cravath Swa i ne, and Moore, Arnold& Porter, and Stein , Zauderer , Ell enhorn , Frischer andSharp have devoted unusualtatent and resource s to thiscase on a pro bono ba si s. This i s in th e hi ghest traditionof the Bar . I commend th em .

SO ORDERED .

DATED : New York, New York

April 1 3 , 1993

JOSEPH M . McLAUGHLIN, U . S . C . J .

Page 22