[House Hearing, 107 Congress]
[From the U.S. Government Publishing Office]
THE CENSUS BUREAU'S PROPOSED AMERICAN COMMUNITY SURVEY [ACS]
=======================================================================
HEARING
before the
SUBCOMMITTEE ON THE CENSUS
of the
COMMITTEE ON
GOVERNMENT REFORM
HOUSE OF REPRESENTATIVES
ONE HUNDRED SEVENTH CONGRESS
FIRST SESSION
__________
JUNE 13, 2001
__________
Serial No. 107-9
__________
Printed for the use of the Committee on Government Reform
Available via the World Wide Web: http://www.gpo.gov/congress/house
http://www.house.gov/reform
U.S. GOVERNMENT PRINTING OFFICE
75-326 WASHINGTON : 2001
____________________________________________________________________________
For Sale by the Superintendent of Documents, U.S. Government Printing Office
Internet: bookstore.gpr.gov Phone: toll free (866) 512-1800; (202) 512�091800
Fax: (202) 512�092250 Mail: Stop SSOP, Washington, DC 20402�090001
COMMITTEE ON GOVERNMENT REFORM
DAN BURTON, Indiana, Chairman
BENJAMIN A. GILMAN, New York HENRY A. WAXMAN, California
CONSTANCE A. MORELLA, Maryland TOM LANTOS, California
CHRISTOPHER SHAYS, Connecticut MAJOR R. OWENS, New York
ILEANA ROS-LEHTINEN, Florida EDOLPHUS TOWNS, New York
JOHN M. McHUGH, New York PAUL E. KANJORSKI, Pennsylvania
STEPHEN HORN, California PATSY T. MINK, Hawaii
JOHN L. MICA, Florida CAROLYN B. MALONEY, New York
THOMAS M. DAVIS, Virginia ELEANOR HOLMES NORTON, Washington,
MARK E. SOUDER, Indiana DC
JOE SCARBOROUGH, Florida ELIJAH E. CUMMINGS, Maryland
STEVEN C. LaTOURETTE, Ohio DENNIS J. KUCINICH, Ohio
BOB BARR, Georgia ROD R. BLAGOJEVICH, Illinois
DAN MILLER, Florida DANNY K. DAVIS, Illinois
DOUG OSE, California JOHN F. TIERNEY, Massachusetts
RON LEWIS, Kentucky JIM TURNER, Texas
JO ANN DAVIS, Virginia THOMAS H. ALLEN, Maine
TODD RUSSELL PLATTS, Pennsylvania JANICE D. SCHAKOWSKY, Illinois
DAVE WELDON, Florida WM. LACY CLAY, Missouri
CHRIS CANNON, Utah ------ ------
ADAM H. PUTNAM, Florida ------ ------
C.L. ``BUTCH'' OTTER, Idaho ------
EDWARD L. SCHROCK, Virginia BERNARD SANDERS, Vermont
------ ------ (Independent)
Kevin Binger, Staff Director
Daniel R. Moll, Deputy Staff Director
James C. Wilson, Chief Counsel
Robert A. Briggs, Chief Clerk
Phil Schiliro, Minority Staff Director
Subcommittee on the Census
DAN MILLER, Florida, Chairman
CHRIS CANNON, Utah WM. LACY CLAY, Missouri
MARK E. SOUDER, Indiana CAROLYN B. MALONEY, New York
BOB BARR, Georgia DANNY K. DAVIS, Illinois
------ ------
Ex Officio
DAN BURTON, Indiana HENRY A. WAXMAN, California
Jane Cobb, Staff Director
Erin Yeatman, Professional Staff Member
Dan Wray, Clerk
David McMillen, Minority Professional Staff Member
C O N T E N T S
----------
Page
Hearing held on June 13, 2001.................................... 1
Statement of:
Barron, William, Acting Director, U.S. Bureau of the Census,
accompanied by Dr. Nancy Gordon, U.S. Bureau of the Census. 7
Voss, Paul, Department of Rural Sociology, University of
Wisconsin-Madison; Linda Gage, California State Census Data
Center, California Department of Finance; Donald Hernandez,
Population Association of America, Department of Sociology,
SUNY-Albany; and Marilyn McMillen, Chief Statistician,
Center for Education Statistics, U.S. Department of
Education.................................................. 41
Letters, statements, etc., submitted for the record by:
Barron, William, Acting Director, U.S. Bureau of the Census,
prepared statement of...................................... 11
Clay, Hon. Wm. Lacy, a Representative in Congress from the
State of Missouri, prepared statement of................... 36
Gage, Linda, California State Census Data Center, California
Department of Finance, prepared statement of............... 52
Hernandez, Donald, Population Association of America,
Department of Sociology, SUNY-Albany, prepared statement of 61
McMillen, Marilyn, Chief Statistician, Center for Education
Statistics, U.S. Department of Education, prepared
statement of............................................... 70
Miller, Hon. Dan, a Representative in Congress from the State
of Florida, prepared statement of.......................... 4
Voss, Paul, Department of Rural Sociology, University of
Wisconsin-Madison, prepared statement of................... 43
THE CENSUS BUREAU'S PROPOSED AMERICAN COMMUNITY SURVEY [ACS]
----------
WEDNESDAY, JUNE 13, 2001
House of Representatives,
Subcommittee on the Census,
Committee on Government Reform,
Washington, DC.
The subcommittee met, pursuant to notice, at 2 p.m., in
room 2247, Rayburn House Office Building, Hon. Dan Miller
(chairman of the subcommittee) presiding.
Present: Representatives Miller, Cannon, Barr, Clay, and
Maloney.
Staff present: Jane Cobb, staff director; Chip Walker,
deputy staff director; Michael Miguel, senior data analyst;
Erin Yeatman and Andrew Kavaliunas, professional staff members;
Daniel Wray, clerk; Tim Small, intern; David McMillen, minority
professional staff member; and Teresa Coufal, minority staff
assistant.
Mr. Miller. Good afternoon. The subcommittee will come to
order.
We will proceed with my opening statement. Mr. Clay is on
his way. If he does not come, then we will start with a video
and then at some stage allow Mr. Clay to have his opening
statement. And I apologize if I get beeped or such; I am in the
process of a markup in the Appropriations Committee just
upstairs, so I just run upstairs and run right back. So I
apologize in advance. The markup was scheduled after this
hearing and we could not change things.
Our census 2000 was a tremendous success. Because of the
hard work and dedication of thousands of Census Bureau
employees around the country, census 2000 was able to reach
more of America's population and the traditionally undercounted
than ever before, and is the most accurate census in our
Nation's history. The hard work of thousands of census
employees and the dedication of thousands of community
volunteers nationwide made census 2000 a success.
As we leave last year's census behind us, it is time to
begin planning for our next decennial census in 2010. One of
the means by which the Census Bureau has proposed to improve
the 2010 census is by implementing the American Community
Survey [ACS] as a replacement for the decennial census long
form. The American Community Survey, if funded by Congress,
will allow for the Census Bureau to conduct a much simpler and
more accurate census. Without the long form, the much talked
about post card census may be closer to reality. Not only will
it be easier for the Census Bureau to conduct, but it will also
be easier and less burdensome for the American people to
respond. A higher response rate will decrease the need for
costly followup field work and significantly reduce the overall
cost of the decennial census operations.
The other major advantage of the American Community Survey
is its ability to provide up to date and timely social,
economic, demographic, and housing data that tells us who we
are as a Nation. If and when fully implemented, the American
Community Survey will be distributed continuously to 250,000
housing units per month and 30 million housing units over a 10
year period. Information collected by the survey will become
available as early as 1 year after it is collected. It will
continuously provide annual data in place of that which is now
available only once every decade. This will allow our Nation's
data users, community leaders, and policymakers to use much
more current information as the basis for the decisions they
will make that will affect all of us.
While full implementation of the American Community Survey
has its definite advantages over the continued use of the
census long form, there are some concerns with the survey that
must be addressed. I hope we can get many of these issues into
the record today so that the Bureau can respond and give
Congress the assurances we need to go forward with confidence.
One of the issues is cost. Based on the Census Bureau's
budget estimates for fiscal year 2003, the year in which the
full implementation of the American Community Survey is
proposed, the survey will not be cheap. The American Community
Survey is projected to cost some $130 million in that fiscal
year. I would like to explore what goes into this estimate and
whether we can expect this figure to change significantly over
the decade.
We must also examine the content of the American Community
Survey questionnaire. The American Community Survey
questionnaire currently being tested asks 69 questions. The
census 2000 long form only asks 53. By what means will the
questions be added or subtracted from the American Community
Survey questionnaire? I believe that without the establishment
of a predetermined and definitive process by which to alter the
American Community Survey questionnaire, the survey has the
potential to become a much more intrusive survey than the long
form is or ever was. This will not be acceptable.
I would also like to explore whether the American Community
Survey will generate the privacy concerns voiced over the long
form. Many of my colleagues' offices here on Capital Hill have
received calls from their constituents wondering just what the
American Community Survey is and why they have to answer it
when they just received and answered their census forms last
year. If responding to the American Community Survey is deemed
mandatory, as is the decennial census, will the privacy
concerns and people's reluctance to answer the long form simply
be redirected at the American Community Survey? And should the
American Community Survey be a mandatory survey like the
census? What are the implications if it were voluntary? Are we
sure that the American Community Survey will not duplicate
other current, ongoing survey work?
Ultimately, we must answer these and other questions in
order to determine whether the American Community Survey is the
best means by which to collect the demographic information
required for implementing our Federal programs and informing
public policy decisions.
It was a little less than a year ago that we began the
process of looking forward to our next decennial census by
holding our first hearing on the American Community Survey. In
the time that has passed since then, however, many questions
remain. This afternoon we meet again to examine the American
Community Survey to try to answer some of the questions and to
determine whether the American Community Survey is the proper
means by which to replace the decennial census long form and
collect the demographic, social, economic, and housing
information that our Nation's data users and policymakers need
to aid their decisionmaking.
With us this afternoon is the Acting Director of the Census
Bureau, Bill Barron, and data users from across the Nation and
our Federal Government. Thank you all for being here today and
I look forward to your testimony.
[The prepared statement of Hon. Dan Miller follows:]
[GRAPHIC] [TIFF OMITTED] T5326.001
[GRAPHIC] [TIFF OMITTED] T5326.002
[GRAPHIC] [TIFF OMITTED] T5326.003
Mr. Miller. Mr. Cannon, do you have an opening statement?
Mr. Cannon. No, thank you.
Mr. Miller. As I say, I think we will go ahead and proceed.
When we get a break we will ask if any of the other Members
have opening statements. I think we have a video first, so we
will go ahead and proceed with the video.
[Video presentation.]
Mr. Cannon [assuming Chair]. Thank you all. I will be
taking over a bit for the chairman who is, as I understand,
going in and out of an appropriations markup of some sort.
I would like to welcome our first panel and our first
witness, Mr. William Barron. Mr. Barron is currently serving as
the Acting Director of the Bureau of the Census. Prior to
January of this year he was the Deputy Director and Chief
Operating Officer of the Bureau. Before being called to the
Census Bureau, Bill served for almost 30 years at the Bureau of
Labor of Statistics, working his way up from a management
intern through the various positions to serve as Deputy
Commissioner for the last 15 years of his tenure there. Mr.
Barron has received numerous awards and honors for
distinguished and meritorious career civil service. He is known
and respected by his peers for his professionalism and
integrity. And I have had the pleasure of learning these
qualities first-hand.
Bill, thanks for being here today. As is customary, would
you please stand and let me swear you in.
Mr. Barron. If I may, Mr. Cannon, I would like to introduce
my colleague, Dr. Nancy Gordon, who is in charge of our
demographic work. She will be appearing with me today.
Mr. Cannon. Would you mind standing also and taking the
oath.
[Witnesses sworn.]
Mr. Cannon. Let the record reflect that Mr. Barron and Dr.
Gordon answered in the affirmative.
On behalf of the subcommittee, we welcome you here today.
Mr. Barron, if you would like to begin with your opening
statement, you will have 10 minutes.
STATEMENT OF WILLIAM BARRON, ACTING DIRECTOR, U.S. BUREAU OF
THE CENSUS, ACCOMPANIED BY DR. NANCY GORDON, U.S. BUREAU OF THE
CENSUS
Mr. Barron. Thank you very much, Mr. Cannon. I have a
lengthier statement I would like to submit for the record and
just make some summary remarks if I may.
It is a pleasure to appear before you again, Mr. Cannon,
and to testify at this the second hearing that this
subcommittee has held on the American Community Survey. The
subcommittee's leadership in providing a public forum for
discussion of the American Community Survey is very important
and it is greatly appreciated.
Mr. Cannon, the American Community Survey is one of three
key components of the Census Bureau's strategy for re-
engineering the 2010 census. If the Census Bureau has adequate
resources early to pursue this strategy, we can buildupon the
success of census 2000 and take advantage of lessons learned.
Thus, we can reduce the operational risks for the 2010 census,
explore ways to further reduce the undercount and improve
accuracy, and provide more relevant and timely data throughout
the decade, as well as ways to contain costs.
While our strategic plan for the 2010 census is still under
development, we have identified what we believe are three main
components: The first, improving the accuracy of our geographic
data base and our master address file; second, eliminating the
long form from the 2010 census by collecting those data in the
American Community Survey; and finally, re-engineering the
census process through early planning.
Mr. Chairman, in your opening remarks at last July's
hearing on the American Community Survey, you said: ``Today we
are here to begin the process of eliminating the problematic
census long form.'' Mr. Chairman, the process of eliminating
the census long form is now well underway. The American
Community Survey will simplify the 2010 census requirements and
allow the Census Bureau to focus exclusively on the
constitutional mandate for a basic count of the population. It
will provide more current and more frequent detailed data for
small geographic areas, and it will allow the Federal
statistical system to keep pace with ever-increasing demands
for timely and relevant data. The ACS will allow businesses,
Federal policymakers, State, and local, and tribal governments
to make decisions using more current and accurate data, and it
will improve the distribution of Federal funds.
Mr. Chairman, in your letter of invitation you asked that I
address the issue of costs. Our initial estimates of life-cycle
costs demonstrate cost neutrality when we compare the estimated
cost of repeating census 2000 to the estimated cost of a re-
engineered 2010 census, including an American Community Survey,
a geographic system modernization, and early planning. Mr.
Chairman, to achieve cost neutrality, and with further
potential for cost savings, while also providing a rich new
source of local area and national data on an ongoing basis
throughout the decade, is a notable and remarkable achievement.
I do not believe it is an overstatement, Mr. Chairman, to say
that this is one of the most important developments in the
modern history of the Federal statistical system.
Our goal in designing the American Community Survey was to
produce data comparable in quality to the decennial census long
form for the smallest areas such as a census tract. One
decision we had to make was how many years should go into the
moving averages that would replace the long form estimate. We
have decided on a 5-year average for the American Community
Survey that will give more timely data throughout the entire
decade, and will give much better information about change over
time than a once-a-decade measurement could.
Another decision is to determine how much sample is needed
each year so that the 5-year averages would have a sample size
to provide data of sufficient quality. We have chosen a sample
size of 3 million because that will meet our goal of producing
data based on 5-year averages comparable in quality to the
census long form data.
The fact that the American Community Survey sample size and
design will not provide data for the smallest areas until 2008
has led some to raise the concern that the American Community
Survey may be treating rural areas and urban census tracts
unfairly. The Census Bureau takes this concern very seriously.
Indeed, we wish it were possible to begin by providing small
areas with high quality, current data right away. But that
would basically require replicating the decennial long form
every year, and that is not an acceptable option in terms of
costs or burden on respondents. Once again, as we have so many
times in conducting the decennial census, we are faced with the
need to balance competing demands.
The Census Bureau takes even the perception that small
areas are being treated unfairly very seriously and we have
worked with data experts to allay those concerns. Even the
smallest areas will have data 4 years earlier than if we had no
American Community Survey and we included a long form in the
2010 census.
So while concerns have been raised about the data for small
areas, the Census Bureau is confident that the American
Community Survey design is going to yield a major improvement
over the existing situation. We need to understand the glass is
more than half full and to fill it all the way would require
some unacceptable tradeoffs in terms of costs and respondent
burden.
We have designed the American Community Survey to provide
the same quality data as census 2000 for all groups, regardless
of size, and we plan to monitor the survey to make sure that
this is the case on an ongoing basis. Remember, the American
Community Survey does not count the population; it estimates
their characteristics. To get accurate measurements, we need
high response rates from all groups.
We have devoted considerable time to discussing the
question of data for small population groups with our Race and
Ethnic Advisory Committees. Working with them, we will focus on
techniques and strategies to ensure that small population
groups participate in the survey, such as exploring using
language assistance guides, revising the mailing package, and
using public service announcements. The permanent staff of
field representatives will establish ongoing relationships with
the communities they are working in, thereby enhancing trust
and willingness to participate.
The data collected by the American Community Survey will
help Congress evaluate and modify Federal programs and will
provide up-to-date information for congressional districts and
States, as well as smaller areas, enabling services to be
targeted to maximize the impact of available resources at all
levels of Government. The American Community Survey will
provide a critical new source of data that will allow the
Congress to evaluate programs below the State level and to
determine and assess accountability. The up-to-date estimates
from the American Community Survey will benefit, for example,
welfare reform, funding for educationally disadvantaged
children, and programs for the elderly.
The American Community Survey is providing current data
from 21 of its 31 test sites to address real-life issues in
rural and urban communities. In my written statement, I have
provided examples of both Federal and State uses.
In conjunction with the Office of Management and Budget, we
have also established a jointly chaired Interagency Committee
charged with balancing respondent burden with the legitimate
information needs of the Congress and the Federal Government.
The Interagency Committee is working on reviewing the content
of the ACS in a process similar to what we did with the
decennial census long form for census 2000. OMB has asked
relevant Federal departments and agencies to document their
legal requirements for these data, the level of geography that
is required, and for what population groups. This information
is expected to be available to us by the end of August.
The Census Bureau takes questions and concerns about
intrusiveness and privacy very seriously. We are aware of the
time pressures confronting people and of the concerns they have
about privacy and confidentiality. The Census Bureau has a 60
year history, going back to the 1940 census, of working to
reduce the number of questions and the number of households
that would have to answer the longer set of questions. Weighed
against the ever-increasing demands for new questions,
including requests from the Congress and the executive branch,
this is the evidence of the Census Bureau's sensitivity to this
issue.
In summary, Mr. Chairman, as part of a strategy to re-
engineer the 2010 census, the American Community Survey will
improve the way we take the census by eliminating the long
form, simplifying the 2010 census requirements, and allowing
the Census Bureau to focus exclusively on the basic count. It
will provide more frequent detailed data for all geographic
areas regardless of size, so that Congress and Federal agencies
will have up-to-date information to administer and evaluate
programs. And it will contribute to a more efficient
statistical system and allow us to keep pace with ever-
increasing demands for timely and relevant data.
Mr. Chairman, in my more than 33 years of service in the
Federal statistical system, two issues of dominant concern have
been how to provide more current and more frequent small area
data, and how to improve the accuracy of the census population
counts. I believe the plan for re-engineering the 2010 census,
including the launching of the American Community Survey,
addresses both of these important longstanding concerns.
Mr. Chairman, that concludes my testimony and I will be
glad, with my colleague, to try and answer any questions that
you may have.
[The prepared statement of Mr. Barron follows:]
[GRAPHIC] [TIFF OMITTED] T5326.004
[GRAPHIC] [TIFF OMITTED] T5326.005
[GRAPHIC] [TIFF OMITTED] T5326.006
[GRAPHIC] [TIFF OMITTED] T5326.007
[GRAPHIC] [TIFF OMITTED] T5326.008
[GRAPHIC] [TIFF OMITTED] T5326.009
[GRAPHIC] [TIFF OMITTED] T5326.010
[GRAPHIC] [TIFF OMITTED] T5326.011
[GRAPHIC] [TIFF OMITTED] T5326.012
[GRAPHIC] [TIFF OMITTED] T5326.013
[GRAPHIC] [TIFF OMITTED] T5326.014
[GRAPHIC] [TIFF OMITTED] T5326.015
[GRAPHIC] [TIFF OMITTED] T5326.016
[GRAPHIC] [TIFF OMITTED] T5326.017
[GRAPHIC] [TIFF OMITTED] T5326.018
[GRAPHIC] [TIFF OMITTED] T5326.019
Mr. Miller [resuming Chair]. Thank you.
Mrs. Maloney, did you have an opening statement?
Mrs. Maloney. I have questions.
Mr. Miller. OK. Mr. Barron, let me ask a couple questions.
First of all, explain to me how low a level of geographic area
would the data be available and when?
Mr. Barron. The lowest level of data availability, Mr.
Chairman, would be the census tract level. That data would be
available beginning in 2005 if we were able to launch the
survey in 2003, it would then be available on an annual basis
thereafter. There are other data that we could make available
to researchers below the tract level, but the basic unit of
publication, if you will, will be the census tract.
Mr. Miller. I have a question about the questions that are
included in ACS. There are 69 questions included right now in
ACS which is more than we had in our long form. I know the
debate that always took place about trying to add questions;
everybody wants more information. I know the Bureau was always
in the difficult position of trying not to add questions.
Apparently, you have already added some. How will you keep this
from getting out of control and the cost and the response rate
that this has an impact on?
Mr. Barron. Mr. Chairman, I am hoping that when we come out
of this process with the Office of Management and Budget we
will, first off, have a good redefinition and re-examination of
all the questions currently being asked. I am also hoping we
can find a way to partner with the Congress on the
congressional view of the questions we are asking and any needs
for either more or fewer questions that we think reflect the
perspective of the Congress.
On an ongoing basis, I am hoping we could establish some
sort of interagency committee, perhaps with permanent
congressional involvement, to look at this on an ongoing basis
so that we can maintain a consensus as to how many questions we
should ask or not ask.
Mr. Miller. Is this going to have the potential for
eliminating any other surveys or forms or any duplicative
reports that would fix cost but also get more accurate
information?
Mr. Barron. I think on an ongoing basis, once the survey is
fully established, we can look at that. I think for now, Mr.
Chairman, what we are learning is that agencies are seeing this
as a way to expand and improve their information. I know just
recently we received some information from the Bureau of Labor
Statistics indicating that with the data that would be
available from the American Community Survey they are going to
be able to greatly enhance their program of local area
unemployment estimates which currently is developed on the best
data the BLS has but that it is not very detailed data by area.
So, for now we are hearing more about ways to improve the
accuracy of other datasets. I think down the road we will have
to turn to the question of are there things we can eliminate.
Right now we have not identified any candidates.
Mr. Miller. The Current Population Survey, how does that
relate and what are the duplication possibilities there?
Mr. Barron. Right. The Current Population Survey is
collected by the Census Bureau for the Bureau of Labor
Statistics. It is the survey that provides the official Federal
Government measures of labor force activity, including
employment and unemployment. It is a national survey of about
50,000 households. It is designed to measure month-to-month
change in unemployment and other labor force characteristics.
It does focus on labor force activity.
The American Community Survey is going to have a labor
force component. But the American Community Survey is far more
detailed in terms of its geographic reach, if you will. It is
also designed to provide estimates on an annual basis. So the
BLS is viewing the Current Population Survey and the American
Community Survey, and I think I would agree with this, as
complements. They are going to be able to use the data from the
American Community Survey to greatly improve the local area
unemployment estimates that they are required to produce for
purposes of distributing job training funds. They are going to
be able to greatly enhance the data quality of those estimates.
Right now, they have, as I said a moment ago, sort of a paucity
of data to develop these monthly estimates.
Mr. Miller. For 2010, the post card census is what we are
talking about, is that right, if ACS goes forward and is
working?
Mr. Barron. Mr. Miller, I think we are very close to a post
card census in terms of content. I do not know what particular
mail instrument we would use to send it out to people, but we
are essentially talking about the short form. I do not know
whether we tested whether that would actually fit on a post
card or not. I would have to check on that. But in terms of
content, we are talking about a greatly reduced census.
Mr. Miller. I just received a report the other day about
the cost. I just received it yesterday so I have not had a
chance to really fully evaluate it. But the projected cost for
fiscal year 2003 is $131 million. Would you care to comment
about this report on the life-cycle cost which like $500
million less total cost if we----
Mr. Barron. Yes, Mr. Chairman. First, let me apologize. We
spent a lot of time working on that document and it is a very
sensitive matter. It took us a lot of time to make sure all the
appropriate bases were touched in terms of getting that
document up here to you. I regret that I was not able to get it
here sooner.
I think it is a very important document. It basically is
laying out the fact that if we were just to take census 2000
and use some standard assumptions about inflation, Federal pay,
and pay for information technology contracts, and things like
that, if we compare the cost of taking census 2000 and moving
it out 10 years inflated by assumptions for those basic types
of costs, it is going to approach $12 billion. And if we are
able to re-engineer the census, starting with early planning,
starting with an improved and technologically enhanced master
address file process, and, of course, eliminate the long form,
than in terms of annual appropriations, we think there would be
a cost avoidance of about half a billion dollars.
What we are also achieving though, Mr. Chairman, in having
done that is we would have an ongoing set of new data never
before available except on a decennial census basis, an ongoing
set of products providing a rich dataset--an ongoing video, if
you will, of what is happening to America in terms of all the
characteristics that are collected on the long form. So you
could look at it as cost neutral but with a tremendous benefit
in terms of the amount of data provided. So that is basically
what the life-cycle document is setting forth for you.
Depending on whether you look at net present value or cost
avoidance, it is either cost neutral, that is according to net
present value calculations, or a savings of about a half a
billion if you look at funds that would not have to be
appropriated in the annual appropriations process. So I think
it is a very important finding.
Mr. Miller. I think it was a very interesting document.
Mrs. Maloney.
Mrs. Maloney. Thank you, Mr. Chairman.
Welcome, Mr. Barron. Mr. Barron, can you document the
number of times and the reasons for the Census Bureau putting
out data from the American Community Survey that subsequently
had to be retracted because of errors?
Mr. Barron. From the American Community Survey?
Mrs. Maloney. Yes.
Mr. Barron. No, I cannot, Mrs. Maloney. Let me ask my
colleague, Ms. Gordon, if there are any such documents that we
have.
Ms. Gordon. I am not aware of retracting data from the
American Community Survey. The one circumstance I think that
you might have heard about would be information for Bronx
County in New York where, because we did not have the American
Community Survey in the last decade, our population estimates
were not able to take advantage of that kind of information and
so the use of the population estimates for that particular
county resulted in data that we thought was not as accurate as
we would like. And so those data have sort of a warning label
on them. But the data for all of the other sites that we have
released we think are really quite good and there have been no
concerns that I know of expressed about them.
Mrs. Maloney. OK. Mr. Barron, could you tell me under what
authority are you withholding information from the U.S. Census
Monitoring Board? Their enabling legislation clearly states
that: ``Each co-chairman of the board, and any members of the
staff who may be designated by the board under this paragraph,
shall be granted access to any data, files, information, or
other matters maintained by the Bureau of the Census or
received by it in the course of conducting a decennial census
of population which they may request subject to such
regulations as the board may proscribe in consultation with the
Secretary of Commerce.''
Mr. Barron. I guess no one would ever give me an authority
that complicated because I would not be able to understand it.
But I think what you are referring to, Mrs. Maloney, is that as
we enter into the next phase of the process of looking at
whether adjustment would improve estimates from census 2000, we
were attempting to replicate the same process that I think we
used very successfully earlier in the year where we provided
access to the data that we were looking at on a real time
basis. At the same time we were looking at it, we provided
access to the National Academy of Sciences, to the Congress,
and to the Monitoring Board. So we thought it would be a good
practice to try and replicate the same thing.
Now, since all of the issues that you raise now were not
raised then, I am sort of surprised. But we are simply trying
to be open about what we are doing. We prefer to have people
look at the data that we are looking at at the same time. We
are also aware that the Monitoring Board will be having to
issue reports early in the fall and we would like to help them
do that. We would like them not to publish local area data,
specific area data until we have. That is basically the concern
that we have. We would like to focus on doing our work and not
get caught up in a lot of external debates about local data
until we have finished our work. And that is what we are trying
to achieve.
Mrs. Maloney. The Monitoring Board, I believe there are
members here from the Monitoring Board, would be glad to give
you a list of data that they would like the information. Under
law, they are entitled to it. I would like you to provide the
committee a legal memorandum that explains under what authority
you are withholding any information. It is against the law.
Mr. Barron. Well, I will go back and ask my attorneys to
see if they can defend me on this.
Mrs. Maloney. Thank you very much. I have a few more
questions. In your letter, Mr. Barron, to myself and Mr. Clay,
dated April 24, you stated that the ``Census Bureau is
preparing a plan for examining demographic analysis, the ACE,
and Census 2000 which will be available in the next month.''
Can the committee get a copy of this plan?
Mr. Barron. I do not think the plan exists right now. But
the answer is, absolutely. When it does exist, we would be glad
to provide it to the committee. It should be available very
shortly.
Mrs. Maloney. So you are saying that the plan does not
exist?
Mr. Barron. I know that a plan has been discussed
internally and it is being modified. As soon as there is a
public plan, we would be glad to give it to you.
Mrs. Maloney. So, in your letter of June 8, you stated that
it did not yet exist, and now you are saying that it still does
not exist. Four months after the decision not to go forward
with the corrected data, you do not even have a plan done to
review the differences. Is that correct?
Mr. Barron. We have spent a lot of time identifying
problems that came out of the last set of ESCAP deliberations,
and we do have a plan for that, those kind of data are being
established. In terms of a plan for how we will conduct our
review over the summer, that is another stage of planning that
we have not yet completed. It will be done soon. I am confident
that by the time the fall arrives we will have examined all of
the issues that arose in our initial set of ESCAP deliberations
and, hopefully, we will arrive at a recommendation that will be
acceptable to everyone.
Mrs. Maloney. Well, as you stated, you still do not have a
plan after 4 months. Then why are you embarking on a project to
reconfigure the post-strata of the ACS if you have not even got
a plan to reconcile the differences between the corrected and
uncorrected data and the demographic analysis?
Mr. Barron. I am not familiar with a project to restructure
post-strata, Mrs. Maloney. We have a lot of work under way to
assemble the data that we need to continue our analysis. We
have not produced a plan on the actual conduct of that
analysis, but that is something that I think we can do in
relatively short order. A lot of effort is going into
developing some way to understand the data sets that we have. I
think the Bureau, by October, will have a fine report on this
issue.
Mrs. Maloney. And finally, the Census Bureau has told the
subcommittee that it is conducting a study to identify
duplicates in the group quarters population. However, the
Census Bureau has made no effort to measure people missed in
group quarters and has no intention of doing such a study.
Congress has repeatedly asked the Census Bureau to pay
attention to people missed in the group quarters population and
has been repeatedly ignored. How do you justify this one-sided
approach to measuring error in the group quarters population?
Is this a search for the politically correct number, or are
there other instances where the Bureau tries to assess the
level of duplicates and does not count those missed?
Mr. Barron. I am not sure I completely understand the
question, Mrs. Maloney. We are looking at group quarters and
will, in fact, soon be issuing the short form data and be
working with State and local officials. That is probably the
most effective, nonpartisan, unbiased, open way to assess group
quarter data quality since everyone in the country will have
the data and will be able to assess it. So I do not know really
how to respond to the comment that we are doing something that
is not open and straightforward, but I regret you feel that
way.
Mrs. Maloney. Just to get to the facts. Are there other
cases or instances, past or present or historically, where the
Bureau tries to assess the level of duplicates and does not
count those missed? That is a reasonable question.
Mr. Barron. I am afraid I cannot really answer that.
Mrs. Maloney. Could you have your team look and try to get
the answer?
Mr. Barron. Sure.
Mrs. Maloney. I would be glad to meet with you and go
further in with it. We have a vote right now.
Mr. Barron. I would be glad to meet with you, too.
Mr. Miller. I believe we will have time. The second bell
has not gone. We have a vote on the floor, so we will have to
run out shortly.
Mr. Cannon.
Mr. Cannon. Thank you, Mr. Chairman. Mr. Barron, I
appreciate your being here, again. Of course, the basic
constitutional purpose of the census is the apportionment of
congressional seats. You and I are in a maelstrom over the
difference between North Carolina's count and Utah's count.
Utah is missing a seat. So if you would not mind, I would like
to revisit some of those issues that we have spoken about
before.
The last time you were here I asked you several questions
related to disproportionate counting of Americans overseas in
the 2000 census and how that affected the State of Utah, and
missionaries of the Mormon Church, in particular. If you
recall, I asked specifically about how the Bureau is
progressing on a report and then a final plan on how to count
Americans overseas in the 2010 census. At the time, I found the
progress a little disappointing. But let me pose that full
question to you again.
Chairman Miller included in the Bureau's fiscal year 2010
appropriations a requirement that the Secretary of Commerce
``Submit to the Congress no later than September 30, 2001, a
written report on any methodological, logistical, and other
issues associated with the inclusion in future decennial
censuses of American citizens and their dependents living
abroad for apportionment, redistricting, and other purposes.''
What progress has been made on that report since our last
hearing with you, if any? And given the tremendous and
immediate interest in this issue, might that report be given a
greater priority by the Bureau?
Mr. Barron. I think we have made a lot of progress, Mr.
Cannon. We recently had a briefing up here, and I would like to
have staff come back and meet with your staff because I
understand it was not a convenient time for your folks and we
want to make sure that your folks are involved in it. I think
we have made a lot of progress in identifying the issues that
we see in trying to construct an accurate count of Americans
overseas. One of the big issues, for example, is whether we
could rely on administrative records to do that, whether that
would be from a perspective of folks who are very interested in
this number, and whether that would be sufficient. We are also
interested in trying to reach a consensus on uses and whether
it would be satisfactory to identify people who sort of ``self-
nominate'' themselves as being an American overseas, or do we
have to go through some further degree of proof to determine
exactly who they are and why they are there and that sort of
thing.
At any rate, we have made a great deal of progress. I do
not know if it is possible to speed up the September 30th
report. I will look into that and get back to you. I think
maybe the first thing to do might be to get with your staff and
brief you on what we have done.
Mr. Cannon. Thank you. We will take you up on that. But
even if the report is on time, proceeding as you are now, how
long do you think before there is a final plan for counting
overseas Americans in the next census?
Mr. Barron. Well, we are probably a pretty good length of
time away from having a final plan, Mr. Cannon. I think the
issues are daunting. Another thing we need to do, and I think
we agreed to do this in the briefing we held up here last week,
is to meet with the groups that are representing the folks who
live overseas to see what sort of reaction we can get from them
in terms of the issues that we have identified. So I do not
want to commit to a timeframe. It depends on whether we can get
a consensus on the type of enumeration we can conduct and how
that number would be used.
Mr. Cannon. Let me just jump on to the next question. It
looks like you are going to spend about $131 million on the
American Community Survey this next fiscal year. Can you give
me a rough estimate of how much the Bureau is spending this
fiscal year to put together the report and plan for counting
overseas Americans?
Mr. Barron. I would have to provide that to you for the
record, Mr. Cannon. It is a very small amount of money relative
to the budget request for the American Community Survey.
Mr. Cannon. I appreciate your responses to this line of
inquiry. I remain concerned that the Bureau is neglecting this
core responsibility and devoting its resources to projects
outside the core mission, it is a paramount mission in the
Constitution, while leaving unresolved these really difficult
issues which we have been dealing with for 70 years. So I would
appreciate your getting back to us on some of those things, and
look forward to having our staff meet with you.
Thank you, Mr. Chairman. I yield back.
Mr. Miller. Thank you. Mr. Barr, we have a vote but if you
would like to proceed before the vote.
Mr. Barr. Thank you. Is this the survey that we are talking
about, the American Community Survey?
Mr. Barron. I believe so, sir, yes.
Mr. Barr. This copy is 24 pages long. Is that correct?
Mr. Barron. I think that is correct. Yes, sir.
Mr. Barr. How many questions including subparts are
contained in this?
Mr. Barron. I believe there are 65 questions.
Ms. Gordon. It is approximately that many. You asked about
including subparts, in many questions, for example, asking
about how much you pay in a mortgage, we need to first ask do
you have a mortgage. So we do not have a tally of the subparts
available right at the moment. But we could certainly provide
it to you.
Mr. Barr. It would be interesting. I could go through and
count them all up. It is an awful lot more than 65. Do you have
any concern that this is awfully intrusive?
Mr. Barron. We are very concerned about it, Congressman.
Mr. Barr. Then why are you asking it?
Mr. Barron. Well, the basic reason, Congressman Barr, is
that there is in back of each one of these questions a
legislative requirement by a Federal agency. I want you to know
that we are working with the Office of Management and Budget
this summer to review each and every one of those requirements
to make sure that it is there and to assess that and to see if
the question could be restructured. But, no, we are very
worried about that. It is our staff who go out and, in this
case, talk to people face-to-face about filling out the survey.
So we want it to be as acceptable to the American public as we
can make it.
Mr. Barr. What if somebody just does not want to fill all
this out. Is there anything he can do about that?
Mr. Barron. Well, one of the issues that has been raised is
whether this should be conducted with mandatory reporting. Our
initial thinking, although we want to work with the Congress on
this, is that we think as part of the decennial census it
should be mandatory reporting.
Mr. Barr. All of this information?
Mr. Barron. Yes, sir. That is consistent with the approach
to conducting the collection of the long form on the decennial
census which this is replacing.
Mr. Barr. But there are an awful lot of concerns raised
about that.
Mr. Barron. Indeed, there were.
Mr. Barr. And this just perpetuates it.
Mr. Barron. It does, but we also think, and we will do the
review to make sure that we have got this down to the bare
minimum, we also have found that when----
Mr. Barr. This is not the bare minimum, is it?
Mr. Barron. I do not know, sir. I think that each one of
those was looked at prior to the 2000 census and it may be
terribly close. So I do not want to lead you astray and make a
promise to you that I cannot come close to keeping. I think
each one of those questions----
Mr. Barr. So 24 pages of detailed questions with numerous
subparts might be the bare minimum?
Mr. Barron. There is a legislative requirement that----
Mr. Barr. You are starting to smile. You cannot say that
with a straight face, can you?
Mr. Barron. The reason I was smiling is I think that some
of the----
Mr. Barr. You are smiling because there is no way that it
can legitimately be maintained that this is the bare minimum
information that the Government needs to get a handle on how
many people are in this country.
Mr. Barron. No, these are population characteristics, not
numbers of people. I think some of the length is coming from
the fact that we do ask a set of questions for each person.
That makes it longer.
Mr. Barr. So what is it specifically that you are going to
do to pare this thing down?
Mr. Barron. First we are going to meet with the Office of
Management and Budget which has asked every Federal agency to
examine the questions that they say are required by law in this
form and to explain back to the Office of Management and Budget
is this true or is it not true. We are going to examine that,
we are going to document it, and then I hope we can come back
up to Capitol Hill and share that with folks up here so that
they understand that this is the situation that we are dealing
with.
Mr. Barr. And when you do that you will not just look at
the number of questions, but all of these cockamamie subparts.
Some of these questions go on for columns.
Mr. Barron. I promise you that we will look at all the
cockamamie subparts.
Mr. Barr. Thank you. And I know that you are concerned
about this, Mr. Chairman, and I look forward to working with
you to address the very serious privacy concerns that we have
with this sort of detailed information project. Thank you.
Thank you, sir.
Mr. Miller. On the mandatory issue, is that a determination
that Congress will ultimately make on whether this is
mandatory, or will OMB or Census Bureau issue the decision?
Mr. Barron. I do not know who has the ultimate authority on
that, Mr. Miller. Given the obvious sensitivity, we would come
up and talk with----
Mr. Miller. What impact will it have on response rates and
all that?
Mr. Barron. That is a worry. Our sense both from talking to
our staff who actually goes out and knocks on doors as well as
the several times when this has been tested in the past, the
sense is that if it is not mandatory the response rates will go
down and costs would go up and accuracy would deteriorate.
Obviously, that is a very serious concern to us. The life-cycle
cost document we have provided you assumes, that the ACS has
got a sample size now that is right at the cusp of what is
going to meet the important objectives that we think need to be
met to provide local data. If it gets cut further, we would be
very worried. Similarly, therefore, if response deteriorated
further, we would be very worried.
Our concern is maintaining response and maintaining the
ability to provide accurate data.
Mr. Miller. We have to go vote right now. But one question,
and I remember seeing the report on the 2000 census about all
the long form questions and the documentation, if it is going
to be something that we can legislatively do to reduce
questions and if they are not essential, I think we need to
revisit them. I know the only question that was added since the
1990 census was one that was added in the welfare reform about
grandparents. So that is the type of thing that is mandated by
Congress that I think maybe we need to revisit.
We will stand in recess for a quick vote.
[Recess.]
Mr. Cannon [assuming Chair]. The subcommittee will be in
order.
Mr. Clay, do you have some questions that you would like to
ask?
Mr. Clay. Yes, I do, Mr. Chairman. Mr. Chairman, if you
would allow me to request unanimous consent to submit an
opening statement.
Mr. Cannon. Without objection, so ordered.
[The prepared statement of Hon. Wm. Lacy Clay follows:]
[GRAPHIC] [TIFF OMITTED] T5326.020
[GRAPHIC] [TIFF OMITTED] T5326.021
[GRAPHIC] [TIFF OMITTED] T5326.022
Mr. Clay. Thank you. Mr. Barron, let me first thank you for
being here today and for your testimony. I am pleased that we
are able to focus on the American Community Survey in a
bipartisan spirit of inquiry. I hope you will take my questions
in that spirit and not as an attack on the ACS.
As we approach the full funding of the ACS in the fiscal
year 2003, Congress must determine whether this expenditure is
in the best interest of the Government. We must ask if we are
going to invest $130 to $140 million a year in our statistical
infrastructure is the ACS the best investment we can make.
There are proposals before the Congress to create a registry of
violent deaths, similar in structure to the birth registration
system. Others are urging Congress to improve the collection of
information on the service sector. In fact, the Census Bureau
is urging Congress to improve collection of information on
electronic transactions. Still others would have us improve
collection of information on the environment, or on energy
supply and consumption, or on the supply of fish in the ocean.
Our questions today are to help us make the judgment of whether
we should fund the ACS or not.
Would you please tell us why you think that funding the ACS
is the best investment we can make today in the Federal
statistical system.
Mr. Barron. Sure, Mr. Clay, I would like to try. I think it
would be the best investment for the Federal statistical system
because it is going to be a smart investment, an investment
that is going to have to be made in 2010. In other words, the
plan that we are proposing, when you look at all parts of it,
not just the conduct of the American Community Survey, but the
fact that if we are able to launch it completely, it would
replace the long form. If you look at the cost of a re-
engineered census--which we can do if we start now to plan it--
if we improve our way of assembling a master address file and
use new technology, and we are way behind local areas in fact
in terms of use of technology, and if we can replace the long
form with an ACS, we actually have a proposal that is cost
neutral. And while I know there are a lot of important
statistical needs in the other statistical agencies, and I know
from my own personal experience that is a very serious problem,
I think one advantage we have over them is that we have a cost
neutral proposal to do something that, in fact, is a
constitutional mandate.
So I think we have some important advantages that need to
be considered as we discuss this proposal with you.
Mr. Clay. Thank you for that. We will hear from witnesses
later today one of the primary purposes of the ACS is to
provide small area data that will replace the data currently
collected as a part of the decennial census. The Census Bureau
has decided to provide small communities with data which are
somewhat less precise than the long form in exchange for 5 year
averages updated each year. Can you explain to us why you
believe this is advantageous to local governments?
Mr. Barron. Well, it is a tradeoff that we are making, Mr.
Clay. But it is a tradeoff that we think is providing data that
is of good quality, slightly less in terms of measures of
accuracy, but very comparable to the data that is available
from the long form, in terms of sampling error. In terms of
nonsampling error, the fact that we are going to have the data
collected by an experienced enumerator, and we will be able to
follow up--we think there are some important data quality
advantages in that process. And the fact that we will have an
ongoing stream of data is an important advantage.
So we think that relative to providing data once a decade--
by the time the local area folks receive it it is often 12
years old--we think that this has some very powerful
advantages.
Mr. Clay. It is my understanding that there are no new
funds requested in your 2002 budget to improve the demographic
analysis estimates for the State and county estimates program.
Have you considered reprogramming some of the remaining
decennial census funds to improve these estimate programs in
2002?
Mr. Barron. We have and we are still looking at it. We have
not made a final decision. We are also looking at future budget
cycles, but that is beyond the scope of what I could talk about
today.
Mr. Clay. When do you think you will make a decision?
Mr. Barron. I think as we get into the summer and we go
through the next set of deliberations that the Executive
Steering Committee on Adjustment Policy needs to go through, I
think we in the Census Bureau are going to come out of that
process with a better insight as to what we have in terms of
the demographic analysis system. It needs to be improved.
Whether we need more resources or can use some existing
resources in the short run is really the issue we can look at
in the summer. It does need to be improved.
Mr. Clay. Thank you. Thank you, Mr. Chairman.
Mr. Miller [resuming Chair]. Mr. Barron, thank you very
much for being here today. I am glad we are planning ahead for
2010. It just seems like we have not even finished all the data
for 2000 and we are immediately planning for 2010, which is
exactly what has to be done. So thank you for the leadership
you are providing at the Bureau. Thank you for being here.
Mr. Barron. Thank you, Mr. Chairman.
Mr. Miller. We will take a short recess till the next panel
comes up.
[Recess.]
Mr. Miller. Our next panel consists of Mr. Paul Voss, who
represents the Department of Rural Sociology at the University
of Wisconsin at Madison; Ms. Linda Gage, representing the
California State Census Data Center; Mr. Donald Hernandez, who
is the Chair of the Population Association of America; and Ms.
Marilyn McMillen is the Chief Statistician for the Center for
Educational Statistics at the Department of Education.
As is the procedure here in this particular committee, we
have you sworn in. So if you would all stand and raise your
right hands for the oath.
[Witnesses sworn.]
Mr. Miller. The witnesses have all answered in the
affirmative.
We will begin with Mr. Voss. Welcome. If you would proceed
with your opening statement, please, sir. And if you see me get
up and leave, I have just been notified there is a vote
upstairs on the appropriation committee. I apologize in advance
for that. And we will probably have another vote on the floor I
am guessing in another hour or something like that.
STATEMENTS OF PAUL VOSS, DEPARTMENT OF RURAL SOCIOLOGY,
UNIVERSITY OF WISCONSIN-MADISON; LINDA GAGE, CALIFORNIA STATE
CENSUS DATA CENTER, CALIFORNIA DEPARTMENT OF FINANCE; DONALD
HERNANDEZ, POPULATION ASSOCIATION OF AMERICA, DEPARTMENT OF
SOCIOLOGY, SUNY-ALBANY; AND MARILYN MCMILLEN, CHIEF
STATISTICIAN, CENTER FOR EDUCATION STATISTICS, U.S. DEPARTMENT
OF EDUCATION
Mr. Voss. Thank you, Mr. Chairman. Could you tell us how
much time approximately you would like us to take.
Mr. Miller. I think we would like to hold it 5 minutes. But
your full statement will be made a part of the record.
Mr. Voss. Five minutes. Thank you, sir. Mr. Chairman, Mr.
Clay, members of the committee, I appreciate your invitation to
be here today and to offer my comments on the subject of the
American Community Survey. Specifically, I have been asked to
reflect on any implications the ACS might have on the quality
of data for rural areas and small population groups, and I will
mostly confine my comments to that topic. I am going to skip
over roughly the first half of my prepared remarks. That was
the part that was quite complimentary to the Census Bureau. I
am now going to skip to the other half. [Laughter.]
Mr. Miller. We will include all your statements in the
record.
Mr. Voss. Basically, I work in an academic research setting
focusing on rural areas. I am engaged on almost a daily basis
on data analysis and in providing data assistance to the
hundred of rural communities, rural agencies, rural small
businesses in my State. I also mentioned in my prepared
testimony that because of that particular interest in rural
areas, small places, small population groups, I was, for a
time, an early critic of the initial plans for the ACS. But
having now spent considerable time evaluating the evolving ACS
procedures and recognizing its potential ability to yield
timely and useful data for rural areas and small places, I have
pretty much now reached my peace with this new initiative. With
the changes in the ACS design that have been implemented over
the past several years, and having first-hand awareness of the
Census Bureau's willingness to listen and respond to the data
user community, I now believe the ACS does have the potential
to meet rural information needs over the course of the decade
better than does the traditional census long form.
However, the durability of my peace with the ACS is
contingent upon the Census Bureau's ability to base rural ACS
data on a sufficiently large sample for the data to have a
level of statistical precision similar to that provided by the
census long form sample. This has been the goal of the ACS all
along. But in this regard, it is my present option that the ACS
is beginning to fall short of this goal.
In my view, the ACS, as currently moving forward in this
critical period of testing and evaluation, is extraordinarily
fragile. The over-sample for small places, which I mentioned
earlier in my testimony, has been reduced significantly from
that discussed by the ACS team 3 and 4 years ago and is
substantially below that used for the census 2000 long form
sample. I give two highly specific illustrations in my written
testimony, but here let me summarize.
Not all that many years ago, the ACS team at the Bureau was
projecting that ACS estimates would have levels of uncertainty
around 25 percent larger than corresponding long form
estimates. That is a substantial difference in precision. Yet,
regrettably, current plans at the Census Bureau are now aiming
for uncertainty levels around 33 percent larger than comparable
long form estimates. Can the ACS still meet its goals with
these sampling fractions and these levels of estimate
uncertainty? I confess to having considerable anxiety on this
question.
Certainly there are efficiencies that can be gained by fine
tuning the sampling and estimation procedures. But my biggest
fear as I testify before you today is that the Census Bureau,
in its desire to convince the Congress that a fully implemented
ACS is cost neutral over the long haul, has the potential of
not asking you for enough money to actually fully support this
important and exciting initiative. Any further reduction in
funding, to a level below what I suspect is the Census Bureau's
already too modest goal, could well place the quality of small
area data from the ACS outside the range of acceptability to
the small area data user. And such an outcome likely could
revert user preferences away from a weak ACS and back to the
traditional census long form.
Now let me be very clear. I am a supporter of the ACS and I
do not wish to see that happen. But my fear, if ACS data
decline any further in reliability, any further in precision,
is that a groundswell could develop around the notion that
statistically more precise data available only once each decade
are preferable to less precise data provided on a continuous
basis. Or said another way, timely data are important, but only
if they are reliable, only if they meet certain minimum levels
of precision. In its sampling design for the ACS, and in a
world of tradeoffs, in a highly responsible effort to contain
both costs and respondent burden in this initiative, the Census
Bureau has already sacrificed some of the statistical precision
that communities of all sizes have come to appreciate in the
census long form data. This weakens the utility of the data for
small villages, for city tracts, for block groups, and for
neighborhoods. Any further weakening will likely be the
beginning of the undoing of this exciting data innovation.
My fervent hope, then, for a sound ACS, as it moves into
full implementation in 2003, is that the risk of truly ``full''
implementation be tried; that the Census Bureau continue to
work with its partners in the data user community and with its
congressional partners to ensure sufficient funding for the ACS
actually to do what it promised almost a decade ago to do--to
meet the continuing data needs of all of America's communities,
to provide such communities with annually refreshed,
statistically reliable data for the small areas that make up
these communities, and thereby to enable America's communities
to make better decisions for their people and to use their
limited resources more responsibly.
Thank you, Mr. Chairman, for this opportunity to speak with
the subcommittee. I would be happy to take questions if there
are any.
[The prepared statement of Mr. Voss follows:]
[GRAPHIC] [TIFF OMITTED] T5326.023
[GRAPHIC] [TIFF OMITTED] T5326.024
[GRAPHIC] [TIFF OMITTED] T5326.025
[GRAPHIC] [TIFF OMITTED] T5326.026
[GRAPHIC] [TIFF OMITTED] T5326.027
[GRAPHIC] [TIFF OMITTED] T5326.028
Mr. Miller. Thank you. We will proceed with the statements.
As I said, everyone's written statements will be included in
the record.
Next we have Ms. Linda Gage from the California State
Census Data Center, California Department of Finance.
Ms. Gage. Mr. Chairman and members of the subcommittee,
thank you for the privilege of appearing before you today to
represent the State of California in discussions about the
American Community Survey. The State of California, along with
its councils of governments, city, county, and tribal
governments, relies on high quality census data. Data from the
decennial census not only help determine the amount of funding
the State receives from various Federal programs but also
supports a myriad of decisions throughout each decade on the
allocation of funds and resources throughout the State. The
data further support needs assessment for State programs, site
and size of service locations, and program evaluation.
It is our goal that official data published about the
people of California, and used in policy and funding decisions,
be as current, complete, and accurate as possible. Since the
early 1950's the State has invested in an independent
demographic research program to annually update the population
and housing counts of our jurisdictions to allow us to more
equitably distribute State subventions and to plan and budget
public services based on current demographic data. The State
also devotes considerable support and expertise to the Census
Bureau's decennial census, and other demographic programs, to
aid the collection and estimation of complete and accurate
information about the State's residents.
We support the full development and rigorous evaluation of
the American Community Survey as a method to collect and
provide more complete and more current demographic information
between censuses. At this time we feel it is premature to
endorse the ACS as the preferred method for collecting long
form data in 2010.
Our primary interests are in the prognosis for the full
development of the survey, the plan and timeline for evaluation
of the survey data, and the determination of the role of the
ACS in the 2010 census.
Full development of the ACS is contingent upon adequate
funding, maintenance of a current and comprehensive master
address file, and successful implementation of the survey for
the next 7 years.
The survey is designed to publish annual 1-year estimates
for areas of 65,000 or more population beginning in 2001. This
is fewer than 2 percent of our cities and 24 percent of our
counties. The survey would produce annual 3 year averages for
areas between 20,000 and 65,000 population beginning in the
year 2006. That is only 6 percent of our cities. And annual 5
year averages for areas and population groups of less than
20,000 population beginning in the year 2008. This is over 92
percent of our cities and 43 percent of our counties. On the
current schedule, with no delays or shortfalls, the ACS will
not be fully implemented with the data published until 2008.
The plan for collecting long form data in the 2000 census
was to distribute a separate questionnaire to roughly 1 in 6
housing units nationally. We heard this morning that the ACS is
not designed with the same sampling rate for 2000 through 2010.
If the sample size is smaller, is cut due to funding
shortfalls, or remains static as population and housing growth
occurs in our Nation, the data produced by the ACS may not be
of sufficient quality to substitute for the 2010 long form.
The sample size for the ACS and the effect of lower sample
sizes on the quality of data need to be specified in advance of
a decision to endorse the ACS.
Concerning the evaluation of the ACS, the State of
California has a longstanding concern about the accuracy of the
Census Bureau's intercensal estimates of the State's
population. They have been consistently lower than the
independent estimates produced by the State and less accurate
than the State's estimates when compared to decennial census
counts. Since the 2000 census data were released, we have
additional concerns that the Bureau also underestimates the
national population. The Bureau's estimate for Census Day was
6.9 million persons lower than the number counted in the
census, a 2.5 percent underestimate. If the ACS is not
controlled to accurate population estimates, the long form data
produced will be seriously flawed. Evaluation of the
intercensal estimates is a critical component in the evaluation
of the ACS data.
We are concerned that success in the 31 comparisonsites and
in the Nation's largest jurisdictions will be encouraging but
not definitive. They may form a sufficient base to suggest the
potential, but not to demonstrate the ability, of the survey to
collect high quality small area long form data across the
country 9 years from now.
We are concerned about how data that are released from the
ACS in the years 2006 and 2008 can be evaluated for accuracy
since they will be so far beyond the 2000 census. We are
concerned about whether these jurisdictions will have the same
coverage and quality as the 2000 decennial census and as the
data published for larger cities and counties.
There are case studies and anecdotes to suggest the
usefulness of the ACS; however, a continuous and systematic
evaluation is needed.
We recommend that continuing the successful partnerships
created in the 2000 census process and expanding them to assist
the Census Bureau in planning and evaluating documented usage
and promoting the ACS.
As the role of the ACS is determined for the 2010 census,
the dominant issues are cost, coverage, quality, and
confidence. We recommend that the 2010 census planning include
a contingency for a long form questionnaire until a positive
decision to use the ACS can be made.
And we strongly recommend that a decision date, along with
milestones and critical measurements, be established and
monitored to support a recommendation and decision to use the
ACS. It should be monitored annually for variables, identified
in advance, that are critical to its success. Such critical
measurements include cost, sample sizes, response rates, data
quality, and the status of the master address file.
It is our hope that an ACS that is appropriately funded for
full development will improve the 2010 census and meet the
Census Bureau's goals to provide annual timely information to
States and local governments.
We offer our continued assistance in evaluating the
procedures and results of the 2000 census, the ACS, and in
planning the 2010 census.
We certainly want to thank members of the subcommittee for
their continuing oversight of census programs and for the
opportunity to testify today. I would be pleased to answer any
questions that you have.
[The prepared statement of Ms. Gage follows:]
[GRAPHIC] [TIFF OMITTED] T5326.029
[GRAPHIC] [TIFF OMITTED] T5326.030
[GRAPHIC] [TIFF OMITTED] T5326.031
[GRAPHIC] [TIFF OMITTED] T5326.032
[GRAPHIC] [TIFF OMITTED] T5326.033
[GRAPHIC] [TIFF OMITTED] T5326.034
Mr. Miller. Thank you.
Next, Mr. Hernandez, the Population Association of America,
from the Department of Sociology at State University of New
York in Albany. Mr. Hernandez, welcome.
Mr. Hernandez. Thank you, Mr. Chairman, and members of the
subcommittee, for this opportunity to present the position of
the Population Association of America and the Association of
Population Centers on the subject of the American Community
Survey. I have submitted written testimony to the committee
which I will summarize this afternoon.
The PAA and APC strongly commend the Census Bureau for
initiating the American Community Survey in 1996 and for
vigorously pursuing its development. The PAA and APC strongly
endorse the ongoing development and evaluation of the ACS. The
ACS is a potentially cost-effective alternative to the
decennial long form beginning in 2010. But to realize its
potential, the ACS must be subject to a thorough and scientific
review with respect to its content, design, and estimation
protocols.
The most important condition that must be met if the ACS is
to be successful is that the ACS must be fully funded for each
year during the present decade. This raises the question of
what sample size is required for the ACS if it is to provide
timely, high quality data for local areas. With continued
population growth, a constant sampling fraction implies that
sample sizes would increase. Therefore, we recommend that the
Census Bureau develop a sampling plan for the next decade that
takes into account population growth, and that it develop a
budget reflecting sampling needs for each year. We also
recommend that the Congress take necessary actions to assure
full funding for the ACS during successive years.
This brings us to five issues involved in evaluating the
quality and usefulness of ACS data: the organization of a full-
scale evaluation; weighting and intercensal estimates; topical
content; evaluation of test data; and response rate. In view of
the limited time available today, I will discuss the first two
of these, the organization of the full-scale evaluation and
weighting and intercensal estimates.
First, in view of the complexity and magnitude of the task
of evaluating the ACS and the substantial expertise available
outside the Census Bureau, we recommend that the Bureau
implement the following potentially fruitful mechanisms for
organizing the evaluation. First, it should convene a standing
committee of persons from within and outside the Bureau to
propose innovative evaluative approaches and analyses of
existing and future ACS data. Second, it should create a
mechanism for identifying and funding researchers both within
and outside the Bureau to conduct these analyses. Third, it
should convene an annual conference devoted to the ACS
research, where these and other researchers share their
analyses and discuss data quality, idiosyncracies in the data,
experiences when sharing data with local community leaders, and
so forth. Fourth, it should publish these research results to
foster wide distribution and comment. Fifth, it should develop
a formal mechanism for making changes to the ACS in light of
these research findings and experiences.
In view of the need for an evaluation that spans the years
remaining in this decade, we recommend that the Census Bureau
develop and promulgate specific, measurable benchmarks that
will allow the Bureau and the Nation to judge whether the ACS
is moving successfully toward the goal of providing high
quality data with long form content. These benchmarks should
include both the technical quality of the survey and the costs
and benefits of the ACS relative to long form data collection.
We also recommend that the Census Bureau report annually to the
Congress on the ACS, and whether it is meeting technical and
cost-benefit standards that would justify replacing the long
form on the decennial census in 2010.
One of the most challenging technical issues for the Bureau
will be developing effective weighting and estimation
procedures. The ACS, like other Bureau surveys, must apply
weights to the results from a sample in order to derive
population estimates for various social and economic
characteristics. These weights are based on intercensal
estimates developed by annually updating decennial census data
with results from demographic analyses. But the quality of
these intercensal estimates deteriorates over the course of the
decade. Moreover, Census Bureau comparisons of 2000 Census
results with intercensal estimates strongly suggest that the
quality of the migration component of the demographic analysis
has declined during the past decade. The national statistical
system is not adequately measuring either the number of
emigrants leaving the United States or the number of immigrants
in specific categories which are growing in importance.
We commend the Bureau for planning to improve its
intercensal estimates, both by feeding ACS results back into
its procedures for updating intercensal estimates, and by
improving the international migration component of its
demographic program. As the process of developing these
procedures begins, we recommend that the Bureau cast a wide net
in seeking approaches that might prove effective. In
particular, we recommend that the Bureau consider introducing
new questions in the ACS to identify and estimate the number of
foreign-born persons in various categories who reside in
various communities in the United States. We want to emphasize
that improving intercensal estimates and hence sample weights
for local areas is essential to the success of the ACS.
The Census Bureau plans full-scale annual ACS data
collection in 2003. We commend the Bureau for its innovative
plan to use 5 year moving averages as the foundation for
estimates for small geographic areas and populations. This
approach implies that a full-scale evaluation of ACS data for
the smallest geographic areas cannot begin until the date for
the full 5 years between 2003 and 2007 are collected and
processed. We are confident of the Bureau's capacity to make an
assessment of whether the statistical properties of the ACS are
comparable to long form census data. But the evaluation of ACS
data must also include considerable attention to the utility of
the data, a judgment that can be made only by decisionmakers,
planners, and scholars who use data for specific purposes. We
recommend, therefore, that Federal, State, and local agencies,
as well as private sector users, be included among those
conducting evaluations of the quality and utility of the ACS
when the full 5 year moving average results become available.
Not until evaluations are complete in 2008 or later will
the Nation have the information required to know the quality of
the ACS. We recommend, therefore, that the Bureau continue to
plan a 2010 census which includes full-scale long form data
collection. We judge the marginal cost of planning for long
form data collection in the 2010 census to be small compared to
the potential social and economic costs that would accrue if
the ACS were not successful and if long form data were not
collected in 2010.
A third possibility should also be considered; namely, the
continued collection of ACS data and collection of long form
data in the 2010 census. This might be the best decision if,
for example, the ACS data are judged to be of acceptable
quality and substantial value for States, metropolitan areas,
and other large population groups, but of unacceptable quality
for smaller geographic areas and populations.
A fourth possibility should also be explored seriously--an
experiment in the 2010 census that includes both ACS data
collection and long form data collection in some areas in order
to permit a direct comparison of results between the ACS and
the long form.
Our recommendations are aimed at an accurate, well-run, and
responsive ACS that will meet the diverse and changing needs of
policymakers.
Thank you again for the opportunity to present the PAA and
APC position on the American Community Survey. I would be happy
to answer any questions that you may have.
[The prepared statement of Mr. Hernandez follows:]
[GRAPHIC] [TIFF OMITTED] T5326.035
[GRAPHIC] [TIFF OMITTED] T5326.036
[GRAPHIC] [TIFF OMITTED] T5326.037
[GRAPHIC] [TIFF OMITTED] T5326.038
[GRAPHIC] [TIFF OMITTED] T5326.039
[GRAPHIC] [TIFF OMITTED] T5326.040
Mr. Miller. Thank you.
Dr. McMillen.
Ms. McMillen. Good afternoon. Thank you for the opportunity
to participate in this hearing. In discussing my agency's use
of data from the decennial census long form and our anticipated
uses of data from the American Community Survey, I would like
to focus on four areas this afternoon: the statistical
reporting on critical topics in education; the ways the ACS can
help enhance our current data collection capacity; the ways the
ACS can help enhance the utility of ongoing data collections;
and the importance of good data to ensure fair and equitable
distribution of funds for American education.
Turning first to reporting. My agency, the National Center
for Education Statistics, is congressionally mandated to
collect, collate, analyze, and report full and complete
statistics on the condition of education in the United States
and other nations. To meet this goal, we collect and
disseminate data on all aspects of American education from pre-
school through adult education. Our collections range from
universe or census surveys of basic data, to cross-sectional
and longitudinal sample surveys, to assessments of student
performance. While our universe surveys collect and report
basic data at fine levels of geographic detail, most of our
sample surveys where we get our rich data are restricted to
national and, in a few cases, State-level data.
In addition to our collections, we frequently use Census
sample survey data from the Current Population Survey to report
on a variety of population measures, such as: drop-outs,
educational attainment, limited English proficiency, usually
all at the National level.
One of the most frequent requests we hear from our data
users is a request for more data at State and local levels. We
have used the decennial census data to help fill this gap.
Since 1980, data from the decennial census have given a once-a-
decade snapshot of the economic and social demography of
individual school districts across the Nation. This Decennial
School District Project produces a special set of tabulations
that describe the attributes of children, families, and
households in school districts. These tabulations are then
combined with school district education data that is collected
by the Center to give a more complete profile of the education
enterprise in the United States.
We ask then, how can the American Community Survey help us
with these data? In a number of cases the overlap between the
ACS long form and the CPS questions will allow us to drill down
into more detailed levels of geography on key items. Just as
one brief example. The annual NCES report on dropouts draws
heavily upon data from the Current Population Survey. The
ability to describe the young adult population out of school
without a diploma or the equivalent at the State or school
district level would be a major complement to the national data
that are in this annual report.
In a different type of application, estimation models can
be used to combine detailed data from the CPS with data from
the ACS to create reliable estimates for small geographic
areas. Each of our reports using national level CPS and
decennial census data would benefit from the availability of
more current and more geographically detailed data on education
and other population characteristics.
I would like to go back now to the decennial data. Even
though our Decennial School District Project provides detailed
demographic, social, and economic data, that project has been
criticized because the data will be old by the time they are
released in 2003. If we implement the ACS beginning in 2003, it
will provide the opportunity for us to obtain data for large
school districts as early as 2006 and for all school districts
by 2008, with annual updates after that. Instead of waiting a
decade for the contextual data from the census long form, we
will have the capacity to have these data updated annually. We
see great value for our own uses and our users community in the
annual availability of these State, county, and school district
level data.
I would like to talk briefly about the ways the ACS can
help enhance our current data collection capacity. There are a
number of important education topics that are of interest to
researchers and policymakers that involve relatively small or
difficult to identify populations. As one example, there has
been growing interest over the last decade or so in home
schooling. While all reports suggest that this phenomenon is
growing, it is still a rare enough event that it is difficult
to measure with a typical household-based sample survey. Other
examples of populations that can be difficult to measure
include: pre-school learners, a topic of great interest to
educators at this point, children and adults with limited
English proficiency, Native Americans, and recent immigrants.
How can the American Community Survey help in this arena?
We see great promise in using the ACS as a means of expanding
the range of topics about which we collect data. The proposed
sample size of the ACS ensures that sufficient numbers of
households containing these rare populations could be
identified throughout the decade. More extensive, targeted
surveys could then be conducted in the households apart from
the ACS using the households that are identified with the
characteristic of interest.
Looking next briefly at the ways the ACS can help enhance
the utility of our ongoing data collections. Sample surveys are
likely to yield differences in estimates of basic population
characteristics and are likely to have some under-
representation of hard-to-enumerate demographic groups.
Differences in estimates can be a source of confusion for some
of our data users. One solution is to use the best estimates
available for the population characteristic and post-stratify,
or control, the population to these estimates.
How can the American Community Survey help us here? The
official intercensal population estimates that you have heard
about this afternoon are developed from the previous decennial
census and are used for the controls in many surveys. In fact,
we use them in some of our surveys. Once the ACS is fully
operational, the methods that are used for official intercensal
population estimates will be able to incorporate data from the
ACS to improve these estimates that we use as population
controls in many surveys.
And I would like to turn last to the topic of funds
distribution. As the statistical agency within the Department
of Education, we are asked to help prepare and run programs for
the allocation of education funds. Within the Department of
Education, $12 billion are distributed, in whole or part, based
on school district level estimates of the number of children
ages 5 to 17 in families below the poverty level. These data
are currently only collected on the census long form. In 2001,
$8.6 billion of that $12 billion were appropriated for title I
grants to local educational agencies. A number of other large
formula programs also allocate funds based on a State's share
of title I or on census poverty data.
Beyond the Department of Education, the distributions of
funds amounting to another $9 billion from other Federal
sources are also tied directly to census long form data. In
addition, some States also use the census long form data as a
component of their individual compensatory education formulas.
This use of the long form data is critical to the education
enterprise.
Again, looking at how the ACS can help us. Title I of the
Elementary and Secondary Education Act requires the use of
updated estimates of children ages 5 through 17 in poverty at
the school district level using the most recent census data
approved by the Secretary of Education. Currently, the Census
Bureau uses modelling techniques that were reviewed and
recommended for use by the National Academy of Sciences to
produce these counts. The availability of ACS average annual
estimates of poverty data at the county and school district
level has the potential for improving the estimates of counts
of children in poverty that serve as the basis for the
distribution of billions of education dollars.
Mr. Chairman, I appreciate this opportunity to discuss the
importance and utility of data from the American Community
Survey to the education enterprise in America. I am submitting
for the record a document prepared by my agency that identifies
the potential use of ACS data in fulfilling funds distribution
and reporting responsibilities as specified in law for the
Department of Education.
I would like to conclude by reiterating that NCES and the
Department of Education have an ongoing need for data currently
collected through the decennial census long form. We believe
that if the American Community Survey becomes a reality we will
have more current data at a finer level of geographic detail to
use in a variety of important education applications.
Mr. Chairman, I thank you for your consideration and
welcome the opportunity to provide additional information to
you and the subcommittee, if you should desire. Thank you.
[The prepared statement of Ms. McMillen follows:]
[GRAPHIC] [TIFF OMITTED] T5326.041
[GRAPHIC] [TIFF OMITTED] T5326.042
[GRAPHIC] [TIFF OMITTED] T5326.043
[GRAPHIC] [TIFF OMITTED] T5326.044
[GRAPHIC] [TIFF OMITTED] T5326.045
[GRAPHIC] [TIFF OMITTED] T5326.046
[GRAPHIC] [TIFF OMITTED] T5326.047
[GRAPHIC] [TIFF OMITTED] T5326.048
[GRAPHIC] [TIFF OMITTED] T5326.049
[GRAPHIC] [TIFF OMITTED] T5326.050
[GRAPHIC] [TIFF OMITTED] T5326.051
[GRAPHIC] [TIFF OMITTED] T5326.052
[GRAPHIC] [TIFF OMITTED] T5326.053
[GRAPHIC] [TIFF OMITTED] T5326.054
[GRAPHIC] [TIFF OMITTED] T5326.055
[GRAPHIC] [TIFF OMITTED] T5326.056
[GRAPHIC] [TIFF OMITTED] T5326.057
[GRAPHIC] [TIFF OMITTED] T5326.058
[GRAPHIC] [TIFF OMITTED] T5326.059
[GRAPHIC] [TIFF OMITTED] T5326.060
[GRAPHIC] [TIFF OMITTED] T5326.061
[GRAPHIC] [TIFF OMITTED] T5326.062
[GRAPHIC] [TIFF OMITTED] T5326.063
[GRAPHIC] [TIFF OMITTED] T5326.064
[GRAPHIC] [TIFF OMITTED] T5326.065
[GRAPHIC] [TIFF OMITTED] T5326.066
[GRAPHIC] [TIFF OMITTED] T5326.067
[GRAPHIC] [TIFF OMITTED] T5326.068
[GRAPHIC] [TIFF OMITTED] T5326.069
[GRAPHIC] [TIFF OMITTED] T5326.070
[GRAPHIC] [TIFF OMITTED] T5326.071
[GRAPHIC] [TIFF OMITTED] T5326.072
[GRAPHIC] [TIFF OMITTED] T5326.073
[GRAPHIC] [TIFF OMITTED] T5326.074
[GRAPHIC] [TIFF OMITTED] T5326.075
[GRAPHIC] [TIFF OMITTED] T5326.076
[GRAPHIC] [TIFF OMITTED] T5326.077
[GRAPHIC] [TIFF OMITTED] T5326.078
Mr. Miller. Thank you. Let me thank all of you for being
here today. I did read your statements last night and thank you
for the effort you went to to prepare those statements.
You are basically a cross-section of the data user
community. Dr. McMillen is from the Federal Government side.
How would you rate, and how do you think your peers rate, the
cooperation that the Census Bureau has used in developing ACS,
that you are allowed to provide input and they are adjusting it
accordingly? Do you and Dr. Voss feel comfortable from the
small data concerns that both you raised? Does it need to be
more? Do you have other specific recommendations of what the
Bureau can do better to keep close working relationships with
the data user community?
Mr. Voss. Well I will start with a brief answer. Thank you,
Mr. Chairman. I think that line of communication has been
superb and I say that on the basis of first-hand experience. In
my testimony I give three or four examples of where
conversations between data users and the staff developing the
ACS within the Bureau led to changes that improved the ACS
design, and those lines of communication remain open. I am very
pleased with that level of communication. I as a data user know
who to call within the Census Bureau; I do not go through a
large hierarchy of permissions to talk with the staff, I know
who those people are. We have good conversations and I think
out of those conversations comes for the data user community a
better understanding of some of the difficulties the Census
Bureau is facing, and on their part, some of the needs of the
data user community. I rank their cooperation in that regard
very highly.
Mr. Miller. Would someone else care to make a comment?
Ms. Gage. I would, Mr. Chairman. I would like to concur
with what Dr. Voss has said. Perhaps it is because the survey
is in development, but the ACS staff has been excellent in
seeking user input and accepting it.
Mr. Miller. Dr. Hernandez, you had a comment?
Mr. Hernandez. I would certainly concur in that judgment as
well. I think the Census Bureau is very open and responsive to
the needs and interests of the user community. I would like to
reiterate that in view of the complexity and magnitude of the
evaluation task before the Census Bureau for the ACS that we do
have specific recommendations for creating a process which
would maximize the amount of information that the Bureau
receives and help to ensure that the Bureau makes the best
possible decisions in the evaluation process and in providing
information back to us as users and to the Congress.
Mr. Miller. For the other statistical agencies within the
Federal Government there has always been a good relationship, I
think.
Ms. McMillen. Yes. Yes. Most of the statistical agencies
sit on an OMB-organized committee that participates on a
regular basis in meetings.
Mr. Miller. A question I raised with Mr. Barron is whether
this is duplicating, and I know we have to wait to fully
evaluate ACS. But do you envision the possibility, without a
significant increase in the number of questions, of combining--
we keep referring to the Current Population Survey--of
combining other surveys into this? Mr. Barron said in the
report that just came out that it is an $11 billion estimate
for the 2010 census. A lot of money. But one way some of my
colleagues will look at it is to say, well, if we are going to
be able to consolidate some reports. Do you envision that is
possible if this is successful and after we have a chance to
evaluate it? I guess the earliest we can evaluate it is
probably 2008.
Mr. Hernandez, I will start down at your end.
Mr. Hernandez. I think the issue does require substantial
scrutiny. But based on my experience to date, I think it would
be very difficult to eliminate any other particular survey
because of the level of duplication. Although there is some
duplication between the long form and the CPS, for example, the
two data collection systems serve very different purposes. The
ACS, the long form data, provide a lot of very valuable
information for local areas, whereas the national surveys are
focused at the national level and can provide detailed
information for only some of the larger States. Because the
national surveys with which I am familiar, including the
Current Population Survey, go far beyond the ACS or the long
form in their topical content, it is hard to imagine, how it
would be possible to obtain the information that has proven so
valuable from the CPS, if the CPS were eliminating without
drastically expanding the ACS or the long form content, which
would be problematic for a variety of reasons.
Mr. Miller. Anyone else care to comment? Dr. McMillen.
Ms. McMillen. In the case of my agency, we actually are in
a situation where one of our projects is being melded into the
ACS. In 1980 and 1990, we had to undergo a major effort to
remap each decade all the school district boundaries in order
to have the data and then go to special efforts, at a
considerable cost to the Government, to have the Census Bureau
tabulate the data to the school district boundaries. In part,
because of the requirement that we have biannual estimates of
children in poverty to satisfy title I funding allocations, we
now are updating those boundaries on a regular basis. Because
those are there, that information is being incorporated into
the ACS and they are now treating school districts, as they did
in the 2000 census, school districts are being treated as a
level of geography so it is no longer a special tabulation. We
will be getting data now, once the data come on line, we will
get data on a regular basis.
So there is one small example of a project that has been
enhanced. We have better data more often and we will not have
this considerable buildup of cost and effort every decade in
order to get these data.
On the topic of the Current Population Survey, I think they
are very different surveys. The Current Population Survey can
ask questions in much more detail because of the nature of the
survey. It is, albeit over the phone, a personal interview. But
I think there is real value of having the two combined. As I
said in my prepared comments, if you have rich data from CPS on
something like dropouts at the national level but you do not
have district or county or really State level data, then you
can take the variable that overlaps and drill down on that item
to give the detail at the geographic level. That is sort of the
basic way you can use it. I think there also is a lot of
potential down the road as the data become available for doing
modelling at the national level with CPS and then combining
those data with the ACS data to get a better idea of how some
of those things might be occurring at the local level.
Mr. Miller. Ms. Gage.
Ms. Gage. Mr. Chairman, we do not use a lot of the national
surveys because they do not provide small area data. My concern
with trying to combine those data collections would be that the
burden would increase on the ACS and stress the number of
questions that are being asked of the public.
Mr. Voss. I agree with the panelists in what they have
said. You did ask I think right at the end of that set of
questions about when we can sort of evaluate the ACS. I thought
I would respond to that part. Certainly, we do not have to wait
till 2008. The ACS is already under evaluation in a very small
group of counties. The year 2002 will be a very important year
because the evaluation of the ACS-like national supplemental
survey can be compared with the long form. That will be the
first time we will have for the Nation and for States and other
large areas the opportunity to really compare the kind of data
that the ACS delivers under slightly different procedures and
residence rules than does the long form. So 2002 will be
important for large areas from that survey. And then from
around the country the 39 test sites will be able to use the
ACS data that has been gathered in recent years on an intensive
basis to compare against the long form for very different kinds
of counties. They were very carefully selected so that
different issues that arise--for example, the ones that I am
most interested in, counties with highly seasonal populations,
what does it mean when you are looking at data that has been
gathered over the course of a year from a slice in time in
April where Wisconsin counties are not as populated as they are
in the summer.
So those are the kinds of issues we will be able to look at
in 2000. A phase-in will begin in 2003 and there will be data
coming out from the ACS every year. It will only be 2008 before
the full phase-in is brought in place.
Mr. Miller. Mr. Clay.
Mr. Clay. Thank you, Mr. Chairman.
This is I guess for the entire panel. Why is the ACS not as
precise as possible? If somebody could take a stab at that, at
why do you think it is not as precise as it could be. I heard
you, Mr. Voss, make the comment.
Mr. Voss. Let us agree on what we mean by precise. I used
the words ``precise'' and ``reliable'' in my testimony in
several places. And what I basically mean by that is the level
of confidence you have in an estimate that comes from that
survey. In polling terms, an estimate that comes from a poll of
30 percent plus or minus 2 percent is a more precise estimate
than 30 percent plus or minus 5 percent. So it is the level of
confidence, the level of uncertainty in that estimate.
The reason the ACS is not going to deliver estimates as
reliable or as precise as the long form is because the sample
is not as large. Right now, the long form went to roughly, I
cannot remember, 1 in 6 I think.
Ms. McMillen. Seventeen percent.
Mr. Voss. Seventeen percent. Roughly 1 in 6 of the
population. Right now, with 3 million addresses per year, we
are at about 1 in 8 in terms of the sampling fraction. And the
sampling size and the sampling design largely dictates
precision and confidence and the ACS simply is not as large,
Mr. Clay.
Mr. Clay. Thank you. Let me ask Ms. Gage a question. Ms.
Gage, would you prefer the long form versus the ACS?
Ms. Gage. I do not have an opinion at this time. In
California we do have two counties participating in the ACS. I
have done some very rudimentary examination of those data and
as I compare them to the Census Bureau's current estimates that
are controlled by age, gender, and race-ethnicity, they are
faring better on some variables than others. They are not
faring well on total population, they are not faring well on
the race-ethnic distribution in those counties, and they are
not faring well on the age distribution of our younger
populations.
So at this time, without an evaluation and track record of
the ACS, I would prefer a data collection that is comparable
and equitable for all levels of jurisdictions.
Mr. Clay. And a question that may be somewhat specific to
California. It is my understanding that the residency rules for
the ACS require a person to be living at an address for 2
months to be counted. Migrant laborers often are not at a
single address for 2 consecutive months. Do you believe that
the procedures in the ACS are adequate to capture the migrant
labor population in California?
Ms. Gage. I do not know at this time, Mr. Clay. That is
something we are certainly very concerned about. And although
Tulare County was chosen as a county with seasonal population,
there has not yet been enough study of those data.
Mr. Clay. I see. Thank you.
Dr. McMillen.
Ms. McMillen. To add to that. It is 2 months or if you have
no other usual place of residence you are counted at the place
you are at at the time. So that should help with the migrant
labor problem.
Mr. Clay. Well, some migrants do not have a permanent
address.
Ms. McMillen. That is my point. If they do not have a
permanent--it is 2 months or if you do not have a permanent
residence you are counted where you are at the time. So that
should help with that.
Mr. Clay. I see.
Mr. Miller. Let me ask another question about the questions
on the survey. You heard Mr. Barr raise the issue and, as you
know, there was a little controversy when the long form came
out originally. There is always the pressure to add more
questions to get more data. As data users, I am sure you would
love to add a few more questions. But it does affect response
rates I believe, potentially it could affect response rates,
and there is the cost factor. How do you feel about the
controlling of the number of questions asked and the
limitations on that? I think a couple of you would love to have
more questions asked.
Dr. Voss.
Mr. Voss. Well I will try. I found Mr. Barr's questions
interesting. The questionnaire length is very long. And I mean
no disrespect to Mr. Barr in this answer because I agree with
him that it has to be looked at. But the reason that the ACS
form is 24 pages is not because each of us has to answer all 24
pages, it is long in length so that families with many members
will have an opportunity to essentially count all of their
members. In my household where there are two of us, I think I
probably would have to answer, if I answered fully, four or
five of those pages.
So to count pages is not to really condemn the length of
it. I think if the major----
Mr. Miller. It is a privacy question also that Mr. Barr was
raising.
Mr. Voss. Well, I agree. I think I will not try to speak to
the privacy issue. I have several points on this. Let me make
one more. I think it was 1978, I may be off by a year, when OMB
at the last moment slashed the long form by about I think a
third. And I recall the outcry that came from that; that a long
form that omitted that many questions could not meet the
demands of the Federal Government, of the laws that
subcommittees like your own, Mr. Chairman, had put into place.
I think from that experience we learned that cutting even a few
questions is very difficult. And the voices that were heard
came not only to the Census Bureau but to the oversight
committee at the time. It was not an easy time because it was
so late in that decade coming into the 1980 census.
The questions are there because there are laws behind them.
And making the questionnaire briefer is a challenge for
everyone.
Mr. Miller. Is the 2000 census the shortest one for the
long form questions? It is shorter than the 1990 census. Do you
know if it is going to be, of course, we do not have the data
yet, but is that going to cause any problems that we are aware
of right now? I am not sure what questions we may drop, but we
did drop some questions. The only one that was added was the
one about grandparents taking care of grandkids or something
that was a requirement of the Welfare Reform Act.
We do have a vote going on. Let me see if anyone else wants
to add a comment at all, a concluding statement.
Mr. Voss. The questions that were dropped for 2000 were
largely on the housing side. I think that the users of those
data have figured out that there is other ways that they might
make use of them. That is an example of how questions can be
dropped. But a question on marital status was dropped from the
short form and then immediately the Census Bureau was
criticized for having slipped that over to the long form. So it
is a process that must be done in consultation with the
Congress.
Mr. Miller. Let me conclude the hearing by saying, since
there is a vote on the floor, thank you all very much for
coming today and responding to our questions.
Before I conclude, Jane Cobb, sitting to my left, who is
staff director, is leaving at the end of this week and moving
to another part of the Federal Government, over in FEMA,
actually. She has served a decade on Capitol Hill and she has
done a great job. She has been an invaluable member of the
subcommittee and I appreciate her and wish her well at FEMA. I
hope there are no disasters that you help bring about over
there. [Laughter.]
Let me say thank you again for coming.
I ask unanimous consent that all Members' and witnesses'
opening statements be included in the record. Without
objection, so ordered.
I ask unanimous consent to enter into the record the
written testimony of Robert Hunter, executive director of
Hillsborough County, FL County/City Planning Commission; and
Greg Williams, State demographer of Alaska; and Tom Gallagher
of the State of Wyoming's Department of Employment.
In case there are additional questions that Members may
have for our witnesses, I ask unanimous consent for the record
to remain open for 2 weeks for Members to submit questions for
the record and that witnesses submit written answers as soon as
practical. Without objection, so ordered.
Thank you again. The subcommittee stands adjourned.
[Whereupon, at 4:08 p.m., the subcommittee was adjourned,
to reconvene at the call of the Chair.]