Who Counts? The Politics of Census-Taking in Contemporary America

   by Margo J. Anderson and Stephen E. Fienberg

    Ever since the founding fathers authorized a national headcount as the
    means of apportioning seats in the federal legislature, the decennial census
     has been a political battleground. Political power, and more recently the
    allocation of federal resources, depend directly upon who is counted and
    who is left out. Who Counts? is the story of the lawsuits, congressional
    hearings, and bureaucratic intrigues surrounding the 1990 census. These
    controversies formed largely around a single vexing question: should the
    method of conducting the census be modified in order to rectify the
    demonstrated undercount of poor urban minorities? But they also stemmed
    from a more general debate about the methods required to count an ever
    more diverse and mobile population of over two hundred million. The
    responses to these questions repeatedly pitted the innovations of
    statisticians and demographers against objections that their attempts to
    alter traditional methods may be flawed and even unconstitutional.

     Who Counts? offers a detailed review of the preparation, implementation,
    and aftermath of the last three censuses. It recounts the growing criticisms
    of innaccuracy and undercounting, and the work to develop new
    enumeration strategies. The party shifts that followed national elections
    played an increasingly important role in the politization of the census, as the
    Department of Commerce asserted growing authority over the scientific
    endeavors of the Census Bureau. At the same time, each decade saw more
    city and state governments and private groups bringing suit to challenge
    census methodology and results. Who Counts? tracks the legal course that
    began in 1988, when a coalition led by New York City first sued to institute
    new statistical procedures in response to an alleged undercount of urban
    inhabitants. The challenge of accurately classifying an increasingly mixed
    population further threatens the legitimacy of the census, and Who Counts?
    investigates the difficulties of gaining unambiguous measurements of race
    and ethnicity, and the proposal that the race question be eliminated in favor
    of "ethnic origin." Who Counts? concludes with a discussion of the
    proposed census design for 2000, as well as the implications of population
    counts on the composition and size of Congress. This volume reveals in
    extraordinary detail the interplay of law, politics, and science that propel the
    ongoing census debate, a debate whose outcome will have a tremendous
    impact on the distribution of political power and economic resources among
    the nation's communities.

                                                         Russell Sage Foundation
                                                         $32.50 cloth
                                                         256 pp.
                                                         August 1999

Now available in a new updated paperback edition to cover the 2000 Census, March, 2001.

                                                         Russell Sage Foundation
                                                         $16.95 paperback
                                                         March 2001

To order a copy of Who Counts? directly from the Russell Sage Foundation click here: Russell Sage Foundation.

About the Authors

Stephen Fienberg  is Maurice Falk University Professor of Statistics and Social Science in
the Department of Statistics and Acting Director of the Center for Automated Learning and
Discovery at Carnegie Mellon University, Pittsburgh.

    e-mail: fienberg@stat.cmu.edu

    webpage:  http://lib.stat.cmu.edu/~fienberg


Margo Anderson is Professor of History and Urban Studies at the University of
Wisconsin - Milwaukee, Wisconsin.

    email:  margo@csd.uwm.edu

    webpage: http://www.uwm.edu/~margo

Recent Census Cartoons

For the cartoon by Tom Toles, 1/24/00, go to  http://www.uexpress.com/ups/opinion/cartoon/tt/  and click on Jauary 24, 2000.


Excerpts from Reviews of Who Counts?

One of Choice Magazine's Outstanding Academic Books of 2000:

From Choice, April 2000  by  D. W. Hastings

 The authors review for the past three censuses the tasks of planning the census, its administration, and efforts to
 "accurately" count the population. They also analyze the strategies: the traditional techniques of comparison of
 cohort births with enumerated populations for selected subgroups, and postenumeration sampling techniques--the
 smoothing of sampling errors, "undercount." They assess the decision to use enumeration counts versus
 statistically adjusted values in the final report and in the calculation of funding formulas against the interplay of
 Constitutional requirements, legal directives and constraints, and shifting political policies with turnover in
 administrations. Attention to detail, trenchant analysis, and crisp, lucid prose capture the reader's attention. A
 must for any library serving those interested in politics, history, or demography. Excellent references.

From The Public Interest, No. 141, Fall 2000, "Counting Heads," A review by William Petersen

Anderson and Fienberg's book has a broader theme [Skerry's], "the relative distribution
of political  power and tax dollars triggered by census data."  ...The immense amount of
work in laying a base for Who Counts? is to be commended ....

It is manifestly impossible to conduct a complete and accurate count of several hundred million people.
The issue is not only how to produce the most complete and acurate count possible but also how
the inevitably imperfect result should be reported.

From The Journal of Official Statistics, Vol. 16, No. 2, 185-190, by Milorad S. Kovacevic

The bookis a unique and rich melange of historical facts and data, reviews of statistical
methods and concepts, newspaper reports and commentaries, excerpts from the lawsuit
filings, expert tstimonies, and court rulings. The authors, although participants in some
of the events surrounding the 1990 and 2000 censes[es], are trying to unfold the story in
a neutral and a nonjudgemental way.  ...[O]verall, the book represents an excellent
review of events that marked census-taking in the second half of the twentieth century
in America.

From Biometrics, Vol. 56, No. 2, June 2000, A review by M.S. Ridout

America has counted its population every ten years since 1790, during
which time the recorded population has increased more than 60 fold.
Regional data are used to apportion various resources, including seats
in the House of Representatives and federal funds.  However, the
census produces a net undercount of the true population, the extent of
which varies with region and race.  There is well-documented evidence,
from the 1940 census onwards, that the undercount is substantially
greater for the black population than for the non-black population.
Inevitably, there have been suggestions that estimates of the
undercount should be used to adjust the census data, and considerable
effort has gone into the development of sophisticated statistical
methodology for doing this.
                                           . . . .

It is refreshing to read a book in which a bureaucratic organization
is not portrayed as bungling and incompetent. Indeed the Census
Bureau, though occasionally falling victim to `computer errors,'
appears commendably thorough in its planning, consulting widely with
outside experts before implementing new procedures.  The bad guys are
rather more stereotypical-politicians, and Republican politicians at
that, putting narrow party interests before the wider good.

Of course, this is far from being a simple issue, and some senior
statisticians have argued against adjustment.  The reader will need to
look elsewhere for the very detailed technical arguments, but the book
does an excellent job of making the essential issues accessible to
non-specialists.  It is also a surprisingly entertaining story, and
one that is fully accessible to non-American readers.

From ISI Short Book Reviews, Vol. 20, No. 1,  April 2000,  A review by David Binder

 Over the past few decades, the U.S. Census has been embroiled
in technical, legal and constitutional, and political controversies.
This book outlines how the U.S. Census has reached such a high
profile.  It details the history of censuses in the U.S., as well as
describes in some depth the controversies associated with the 1970,
1980 and 1990 censuses and the impact on future census-taking.

 As a non-American who has been following some of the
developments in the U.S., I found the book gave a wonderful, clear
exposition of the issues and current situation.  Statisticians not
familiar with these controversies will find the material both curious
and fascinating.  Those who have more familiarity will find that this
book provides a useful summary of this complex situation.

 This book combines Margo Anderson's background as a social
historian and Stephen Fienberg's background in statistical methodology
to create a unique narrative and commentary.

From Choice,  April 2000,  A review by D.W. Hastings

Anderson (history, Univ. of Wisconsin, Milwaukee) and Fienberg
(statistics, Carnegie Mellon Univ.) combine their disciplinary
expertise to depict the bureaucratically complex and highly
politicized process of taking a census to apportion seats in the House
of Representatives and to identify populations qualified for receipt
of federal funds.  The authors review for the past three censuses the
tasks of planning the census, its administration, and efforts to
``accurately'' count the population.  They also analyze the
strategies: the traditional techniques of comparison of cohort births
with enumerated populations for selected subgroups, and
postenumeration sampling techniques--the smoothing of sampling errors,
``undercount.''  They assess the decision to use enumeration counts
versus statistically adjusted values in the final report and in the
calculation of funding formulas against the interplay of
Constitutional requirements, legal directives and constraints, and
shifting political policies with turnover in administrations.
Attention to detail, trenchant analysis, and crisp, lucid prose
capture the reader's attention.  A must for any library serving those
interested in politics, history, or demography.  Excellent

Excerpts From Science, January 14, 2000:  "Do We Have to Count One by One?"   A review by Thomas R. Belin

. . . For each of the last several U.S. censuses, members of minority groups have been
missed about five percent more often than whites. Margo J. Anderson and Stephen E.
Fienberg, in their new book Who Counts?, provide an important contribution to the ongoing
discussion on how to address this persistent "differential undercount." . . .The two
have joined forces and achieved their goal of writing a narrative that provides context and
perspective for future debates.

                                                  . . . . . .

The authors provide a concise, yet richly informative, history of the census-adjustment controversy.
Readers learn how, in contrast to the analogous figure for whites, the number of black males who
registered  for the armed services in the fall of 1940 far exceeded estimates based on the 1940 census.
They learn how professional statisticians, both career employees of the Census Bureau and members of
outside review panels,  have overseen innovation in the census. They learn how large cities and other
parties angered by disproportionate  undercounts have aired their cases in the federal courts, which have
taken their  grievances seriously. And they learn how politics inevitably interacts with census-taking,
because of the direct political implications of census results and because politicians have increasingly
recognized that the jockeying for position on the next census begins early in the 10-year census cycle.

Anderson and Fienberg provide similarly concise and helpful reviews of dual-system estimation . . .
and of the political context of the debate over  possible adjustments to the 1990 census. They describe
the legal wrangling surrounding the 1990 Census, particularly the case of City of New York, et al. v.
U.S. Department of Commerce, et al., in which Fienberg testified. After considering expert testimony f
rom both sides, a federal district court judge decided that, although he personally would probably have
opted for adjustment, he could not find the decision against adjustment by then Secretary of Commerce
Robert A. Mosbacher to be "arbitrary and capricious." A federal appeals court overruled this decision on
a 2-1 vote, claiming that the standard for review should be whether the plaintiff's constitutional rights
had been violated. The district court ruling, however, was reinstated by a 9-0 Supreme Court decision
that supported the stronger arbitrary-and-capricious standard of review. In their critique of the Supreme
Court opinion written by Chief Justice William Rehnquist, Anderson and Fienberg note inaccuracies in its
discussion of statistical adjustment methods. Their comments serve as a good example of the colorful
material in the book and provide a cautionary tale regarding the risk of leaving all decisions about
census-taking to nonscientists.

The authors take aim at several "myths" about census adjustment, which they introduce in a prologue
that is not easy to follow without knowing the scientific and political context. Their point is well taken,
however, that many criticisms of statistical adjustments are based on misunderstandings or
misrepresentations. Overall, the authors' short course in the facts and logic underlying adjustments
succeeds in dispelling the myths they cite.  . . . [They] deserve particular credit for their discussion of
racial and ethnic classifications in the census, especially the recent modification that allows respondents
to identify themselves as belonging to one or more ethnic categories. They note that, with conventional methods,
reliable estimates of undercount rates require the definition of fairly sizable poststrata. Thus efforts to eliminate
ethnic classification, or to block the grouping of individuals with similar demographic characteristics, conflict
with efforts to correct for differential undercount. The authors provide an excellent treatment of this concern,
and their book seems to be the first to address the issue.

Anderson and Fienberg omit a number of technical details in exchange for brevity and simplicity. But they
successfully avoid inflammatory rhetoric and offer a wealth of insight. For those interested in understanding
the historical and scientific context of the census adjustment controversy, Who Counts? is absolutely essential

To see the complete review by Belin, click here.

From Population and Development Review, Volume 25 Number 4 / December 1999. A review by Harvey M. Choldin.

The authors . . . represent a uniquely qualified team.  Together they have written a book that takes
advantage of both of their areas of expertise: it presents the historical  background and the unfolding
"current history" of what has happened in the US government regarding the census, and it clearly
explains the statistical techniques along with the arguments that have risen about them.  The
well-edited text moves smoothly between the governmental and the statistical developments,
producing a very readable volume.

Who Counts? is well-organized, full of interesting material, and clearly written.  It is accessible to
students as well as professionals.  It should be of interest to any one interested in censuses,
population statistics, and issues at the nexus of statistics  and politics.  I suspect that the authors
would be pleased if  American journalists, lawyers, and members of Congress, many of whom have
misconstrued the recent issues (sometimes deliberately) would read this book.

From Houston Chronicle,  December 16, 1999.  "Numbers games: Scholars address the politics of the census,"  A review by  Bill Hobby.

    . . . .Margo J. Anderson and Stephen E. Fienberg's  [new book,]Who Counts?, [is]
    invaluable to anyone interested in the politics of the census, reapportionment and
    redistricting. Those issues will be in the courts throughout the next decade. The redistricting
    lawsuits will challenge district lines from congressional seats to commissioner court
    precincts. (Anderson's earlier book on the census was cited in Supreme Court opinions on
    both sides of the issue earlier this year.)

    The most contentious census issue before Congress is the use of a sampling technique for
    apportionment purposes. The issue burst into the national news in the summer of 1997 when
    Republican congressmen amended a bill providing flood relief in the Dakotas to prevent
    sampling. President Clinton vetoed the bill.

    Advocates of sampling argue that it will reduce the undercount of minorities, an increasingly
    pernicious legal problem because of its implications for civil rights and federal aid. Congress
    has authorized sampling for the distribution of federal aid to the states but not for

    Opponents of sampling argue that it is an unscientific technique that will allow the Census
    Bureau discretion to manipulate the numbers in favor of Democrats. (One of the authors' best
    contributions to the debate is a list of myths put forward by sampling opponents.)

    So there may be two sets of numbers -- one for reapportionment, another for federal aid and
    perhaps redistricting.

    Congress reapportions. State legislatures redistrict. Which figures will legislatures use to
    redistrict? The Census Bureau sensibly wants to produce a single set of numbers, and that is
    the recommendation of Anderson and Fienberg.

    Census issues have been much litigated during the past decade. The result is a superb legal
    and historical record of the various disputes and the arguments put forth by each side and the
    actions taken by the various players: President Clinton, members of Congress, the secretary
    of commerce, Census Bureau officials, state and local government officials, and federal judges.

    Who Counts? makes that record readily accessible. It will become the handbook for those

(Also available at:  http://www.swt.edu/cpm/corner/whocounts.html

From a Review by Laurie Snell in CHANCE News 9.01 (November 17, 1999 to January 2, 2000)

Historian Margo Anderson and statistician Stephen Fienberg combine forces to give us a
historical perspective of the Census from the first census in 1790 to the upcoming Census 2000.

The book begins with a history of the Census. The story starts with the founding fathers' establishing
a representative form of government that required a census to make it work. We learn that the political
problems we face today with the census have been with us from the very beginning: Who should be
counted? How should they be counted?


This book will be read by science writers, lawyers, judges and politicians as the inevitable court cases
for the census 2000 approach. We hope that people who read this book will come away with the feeling
that there has to be a better way than fighting it out in the courts to figure out how to carry out this
important and challenging statistical task.

From Booklist, September 1, 1999 A review by Mary Carroll.

    This demanding but instructive study provides legal, political, and scientific
    background on why the current battle over use of statistical sampling in the year
    2000 census has arisen and how possible decisions will play out. Anderson, a
    history and urban studies professor, and Fienberg, a statistics and social science
    professor, served on expert panels helping to plan for the 2000 census. They
    provide a brief history of the census and then address the issue of undercounted
    populations, particularly in the 1970 and 1980 censuses. They cover the lengthy 1980
    litigation in which New York and other historically undercounted jurisdictions
    sought a more complete count, as well as controversy over which racial classifications
    the 2000 census should use. A chapter on dual-systems estimation and other statistical
    approaches demands some mathematical understanding. The book includes a brief discussion
    of the Supreme Court's 1999 decision that sampling cannot be used for reapportionment
    purposes but can, perhaps, be used to meet other census objectives.


The U.S. Census in the News

Commerce Secretary Chooses Unadjusted Census Data After Census Bureau
Recommends Against Statistical Correction

Expert Panel Cites Lack of Time To Complete Analysis; Confirms Net and Differential
Undercounts in Census 2000

March 14, 2001 (prepared by Census 2000 Initiative )

Relying on advice from a Census Bureau committee established last fall under a federal rule
issued by his predecessor, Secretary of Commerce Donald Evans last week decided to
release unadjusted census population counts to the states for "the purpose and only the
purpose of redistricting." The Secretary said in his March 6th announcement that he
followed an "open, reasonable and fair" process and "took full account of the views of

On March 1, a committee of twelve senior Census Bureau professionals recommended the
release of unadjusted census numbers to the states for redistricting purposes. Acting
Census Bureau Director William Barron Jr., who as the agency's Deputy Director also
serves on the Executive Steering Committee for A.C.E. Policy (ESCAP), transmitted the
recommendation, in which he concurred, and accompanying report to Secretary of Commerce
Donald Evans. In accepting the report's findings, the Secretary called the recommendation
"correct and prudent." He also praised the Bureau's staff, saying, "You set ambitious goals
and you exceeded them."

 "We have achieved a quality count," Secretary Evans said, calling Census 2000 "the most
accurate census ever," with "the lowest undercount [and] the smallest differentials in
history by far." Based on the results of its Accuracy and Coverage Evaluation (A.C.E.)
survey, the Census Bureau reported a net national undercount of 1.2 percent, or 3.3 million
people. It has not yet reported the total numbers of people missed (omissions) and counted
twice or wrongly included (erroneous enumerations), which offset each other to calculate
the net national undercount. The net undercount in 1990, measured by a similar though
smaller quality-check survey, was 1.6 percent, or 4 million people. The Bureau reported an
undercount of 1.2 percent, or 2.8 million people, in the 1980 census, based on its independent
demographic estimate of the population.

In its recommendation, the ESCAP concluded that "there is considerable evidence to
support the use of adjusted data, and that Census 2000 and A.C.E. operations were well
designed and conducted." However, the committee said it was "unable to conclude, based on
the information available at this time, that the adjusted Census 2000 data are more accurate
for redistricting." Further research, the panel said, is "likely" to show that "adjustment based
on the A.C.E. would result in improved accuracy," and to confirm that Census 2000 reduced
but did not eliminate a net national undercount and a "differential" undercount of minorities,
renters, and children.

Dr. Barron noted in his memorandum of transmittal that the "primary reason" for the
conclusion is the "apparent inconsistency in population growth over the decade as estimated
by the A.C.E. and demographic analysis." The committee could not explain the differences,
which "raise[s] the possibility of an unidentified error in the A.C.E. estimate or Census
 2000," in the time available, he said. The demographic analysis figures are "significantly
lower" than the population as measured by the A.C.E. survey, the committee said. A.C.E. is
the Accuracy and Coverage Evaluation program, which included a quality-check survey of
314,000 households designed to measure under- and overcounts in the census. Demographic
analysis is an independent estimate of the nation's population the Bureau produces based
primarily on birth, death, immigration, emigration, and Medicare records.

The ESCAP described the results of "quality measures" designed to evaluate the relative
accuracy of the adjusted and unadjusted numbers. It concluded that while "the adjusted data
are more accurate overall," it had concerns that merited further research, including a
possibility that the unadjusted data for counties with populations below 100,000 were more
accurate. The ESCAP report and recommendation are posted on the Census Bureau's web
site at www.census.gov/dmd/www/escapreport.html.

New Report Evaluating  The Bureau's Decision Not to Adjust

July, 30, 2001

Considerable controversy has followed the decsion by the Bureau to recommend the use of  unadjusted data
for release as part of the PL 94-171 file, which is used for redistricting by the states, among other uses.
The  Democratic and Republican members of the U.S. Census Monitoring Board have issued separate reports,
that are available on their websites: http://www.cmbp.gov/ and http://www.cmbc.gov/, respectively.

For an independently-prepared assessment of ACE and the Census Bureau adjustment decision by the authors
of Who Counts?,  Margo Anderson and Stephen Fienberg, click here.

CQ's Encyclopedia of the U.S. Census

Margo J. Anderson, Editor, University of Wisconsin - Milwaukee
August 2000 * 8 1/2 x 11 * App. 450 pages
Hardbound * ISBN 1-56802-428-2 * $125.00 R

CQ's Encyclopedia of the U.S. Census represents a unique and definitive collaboration
among the leading experts on all aspects of the U.S. Census. Drawing from academia,
government, and the private sector, CQ Press and the editorial board have
commissioned more than 100 concise, definitive articles on the decennial census and
related topics. Subjects covered include:

            Content of the census. What the census tracks, and when it began to ask
                specific questions; how questions are formulated, and factors that affect
                which questions are asked; other types of censuses (housing, economic, and
                agricultural); and other records that supplement the census (tax records,
                 Social Security records)
           Procedure. How the census is planned, advertised, and conducted; state and
                 local involvement. How results are tabulated and stored; how the census uses
                 maps and geographic data.
           Uses of the census. Publications, databases, and electronic products that
                provide census information for statistical and demographic research; archiving
                of census records and genealogical use of census schedules.
           Census history. Census taking in America from colonial times to the present
                and at each census; population trends over time, including changes in family
                composition; racial and ethnic groups; and the social, economic, and
                educational status of the population.
           Politics of the census. Effect of census data on congressional districts and
                funding of federal programs; controversies: from slavery and the three-fifths
                compromise in 1790, to the use of statistical sampling in 2000.

More than 80 contributors have provided their unique expertise to create this
book. All articles include references for further reading; many also point the
reader to online resources for census information. For a comprehensive
history of census-taking, for complete and accessible explanations of how the
census operates, and for political controversies as current as the 2000 census,
turn to CQ's Encyclopedia of the U.S. Census. It will be a trusted addition to
any library for years to come.

For more information click here.

Related Reading Material

Who Counts? chronicles the story of the 1990  and 2000 decennial censuses and efforts to use sample to adjust
census results for differential undercount.   Related material appears in:

    Margo Anderson and Stephen E. Fienberg,  "To Sample or Not to Sample?  The
    2000 Census Controversy," Journal of Interdisciplinary History, XXX: 1
    (Summer 1999), pp. 1-35.

    Margo Anderson and Stephen E. Fienberg, "Who Counts?  The Politics of Census-Taking
     in Contemporary America," Items, 53: 1 (March 1999), pp. 4-6 (excerpt from Who Counts?);
    also available at http://www.ssrc.org/march99.htm .

and  a series of papers including :

    Margo Anderson and Stephen E. Fienberg, "Partisan Politics at Work:  Sampling and the
    2000 Census,"  PS , December 2000.

    Margo Anderson,and  Stephen E. Fienberg, "History, Myth Making and Statistics:
    A Short Story about the Reapportionment of Congress and the 1990 Census."
    PS, December, 2000.

in the December 2000 issue of PS.  These and papers by Lynne Billard and Thomas Brunell plus
various commentaries and replyare avaliable at:  http://lib.stat.cmu.edu/~fienberg/PSExchange.html

Final reports for two recent census projects funded by the Donner Foundation are now available:

    Margo Anderson and Stephen E. Fienberg (1999) :"The History of the First American
    Census and the Constitutional Language on Censustaking: Report of a Workshop."
    Technical Report No. 698, Department of Statistics, Carnegie Mellon  University.
    [See also the brief report at  http://wwics.si.edu/WHATSNEW/NEWnews/census.htm].

    Beth Osborne Daponte, Stephen E. Fienberg, Joseph B. Kadane and Duane L. Steffey  (1999).
    "Sampling and Census 2000 -- Methodological Issues."  Technical Report No. 697,
    Department of Statistics, Carnegie Mellon  University.

 These reports are available for downloading in both ps or pdf formats at

For a response to criticisms of the adjustment approaches in 1990 and 2000 see:

     Anderson, M., Daponte, B.O., Fienberg, S.E., Kadane, J.B.,  Spencer, B.D., Steffey, D.   (2000).
     Sample-based adjustment of the 2000   census--A balanced perspective," Jurimetrics,  40, 341-356.

For details on Cemsus 2000, a good place to begin is with materials and evaluations from the
U.S. Bureau of the Census available at their website:  Census 2000.

A brief capsule of the Census 2000 story can be found in

    Margo Anderson and Stephen E. Fienberg, "Census 2000:  Politics and Statistics,"
    PBK, The Key Reporter, 65, (No. 4),  (Summer 2000), 5-6.

We have provided an overview of the March 2001 Census Bureau Decision not to adjust the 2000 census
data in a recent technical report:

    Margo Anderson and Stephen E. Fienberg, (2000) ``Counting and estimation: Methodology
    for improving the quality of censuses.  The U.S. 2000 Census adjustment decision.''  Paper
    presented at the International Conference on Quality in Official Statistics, Stockholm, Sweden,
    May 14-15, 2001.  Department of Statistics, Carnegie Mellon University, Technical Report No. 746:


A website that provides complementary background  and information as well as regular updates on
related political activities: Census 2000 Initiative.

See alo the census informational webpages of the Census Monitoring Board and Rep. Carolyn Molony.

This page was last modified on  8/06/02.