<DOC>
[106th Congress House Hearings]
[From the U.S. Government Printing Office via GPO Access]
[DOCID: f:66541.wais]
OVERSIGHT OF THE 2000 CENSUS: EXAMINING THE GAO'S CENSUS 2000 OVERSIGHT
ACTIVITIES
=======================================================================
HEARING
before the
SUBCOMMITTEE ON THE CENSUS
of the
COMMITTEE ON
GOVERNMENT REFORM
HOUSE OF REPRESENTATIVES
ONE HUNDRED SIXTH CONGRESS
SECOND SESSION
__________
FEBRUARY 15, 2000
__________
Serial No. 106-146
__________
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
66-541 CC WASHINGTON : 2000
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 ROBERT E. WISE, Jr., West Virginia
ILEANA ROS-LEHTINEN, Florida MAJOR R. OWENS, New York
JOHN M. McHUGH, New York EDOLPHUS TOWNS, New York
STEPHEN HORN, California PAUL E. KANJORSKI, Pennsylvania
JOHN L. MICA, Florida PATSY T. MINK, Hawaii
THOMAS M. DAVIS, Virginia CAROLYN B. MALONEY, New York
DAVID M. McINTOSH, Indiana ELEANOR HOLMES NORTON, Washington,
MARK E. SOUDER, Indiana DC
JOE SCARBOROUGH, Florida CHAKA FATTAH, Pennsylvania
STEVEN C. LaTOURETTE, Ohio ELIJAH E. CUMMINGS, Maryland
MARSHALL ``MARK'' SANFORD, South DENNIS J. KUCINICH, Ohio
Carolina ROD R. BLAGOJEVICH, Illinois
BOB BARR, Georgia DANNY K. DAVIS, Illinois
DAN MILLER, Florida JOHN F. TIERNEY, Massachusetts
ASA HUTCHINSON, Arkansas JIM TURNER, Texas
LEE TERRY, Nebraska THOMAS H. ALLEN, Maine
JUDY BIGGERT, Illinois HAROLD E. FORD, Jr., Tennessee
GREG WALDEN, Oregon JANICE D. SCHAKOWSKY, Illinois
DOUG OSE, California ------
PAUL RYAN, Wisconsin BERNARD SANDERS, Vermont
HELEN CHENOWETH-HAGE, Idaho (Independent)
DAVID VITTER, Louisiana
Kevin Binger, Staff Director
Daniel R. Moll, Deputy Staff Director
David A. Kass, Deputy Counsel and Parliamentarian
Lisa Smith Arafune, Chief Clerk
Phil Schiliro, Minority Staff Director
------
Subcommittee on the Census
DAN MILLER, Florida, Chairman
THOMAS M. DAVIS, Virginia CAROLYN B. MALONEY, New York
PAUL RYAN, Wisconsin DANNY K. DAVIS, Illinois
MARK E. SOUDER, Indiana HAROLD E. FORD, Jr., Tennessee
------ ------
Ex Officio
DAN BURTON, Indiana HENRY A. WAXMAN, California
Jane O. Cobb, Staff Director
Lara Chamberlain, Professional Staff Member
Esther Skelley, Professional Staff Member
Amy Althoff, Clerk
David McMillen, Minority Professional Staff Member
C O N T E N T S
----------
Page
Hearing held on February 15, 2000................................ 1
Statement of:
Mihm, J. Christopher, Associate Director, Federal Management
and Workforce Issues, U.S. General Accounting Office,
accompanied by Randolph C. Hite, Associate Director,
Accounting and Information Division, U.S. General
Accounting Office; and Robert Goldenkoff, General
Accounting Office.......................................... 13
Letters, statements, et cetera, submitted for the record by:
Davis, Hon. Danny K., a Representative in Congress from the
State of Illinois, prepared statement of................... 50
Hite, Randolph C., Associate Director, Accounting and
Information Division, U.S. General Accounting Office,
information concerning DCS 2000............................ 46
Maloney, Hon. Carolyn B., a Representative in Congress from
the State of New York, prepared statement of............... 11
Mihm, J. Christopher, Associate Director, Federal Management
and Workforce Issues, U.S. General Accounting Office,
prepared statement of...................................... 17
Miller, Hon. Dan, a Representative in Congress from the State
of Florida, prepared statement of.......................... 4
OVERSIGHT OF THE 2000 CENSUS: EXAMINING THE GAO'S CENSUS 2000 OVERSIGHT
ACTIVITIES
----------
TUESDAY, FEBRUARY 15, 2000
House of Representatives,
Subcommittee on the Census,
Committee on Government Reform,
Washington, DC.
The subcommittee met, pursuant to notice, at 2:04 p.m., in
room 2203, Rayburn House Office Building, Hon. Dan Miller
(chairman of the subcommittee) presiding.
Present: Representatives Miller, Maloney, and Davis.
Staff present: Timothy J. Maney, chief investigator; Chip
Walker, communications director; Erin Yeatman, press secretary;
Lara Chamberlain and Esther Skelley, professional staff
members; Jo Powers, assistant press secretary; Amy Althoff,
clerk; David McMillen and Mark Stephenson, minority
professional staff members; and Earley Green, minority
assistant clerk.
Mr. Miller. Good morning. Mrs. Maloney should be here
momentarily, but I think we're ready to begin. And I'll make my
opening statement and Mrs. Maloney will be here certainly in
time for hers.
Good afternoon. Last week we heard from Census Bureau
Director Dr. Kenneth Prewitt. Dr. Prewitt testified that the
activities for the 2000 census were on schedule and, at the
time, no major problems existed. This included an ad campaign
that was running smoothly and hiring that was on schedule.
I want to be clear from the outset about the purpose of
this hearing. The purpose of this hearing is to have the
nonpartisan General Accounting Office, the investigative arm of
the U.S. Congress, give us its professional assessment of where
they believe the Census Bureau is with respect to the myriad of
tasks that must be carried out successfully in the upcoming
months.
I believe it is critical that this Congress not only fully
fund the Census, but fully promote it as well. Yet, at the same
time, this Congress, and specifically this subcommittee, has a
responsibility to conduct oversight of the census and the
spending of almost $7 billion in taxpayer dollars. If the
nonpartisan GAO fully endorses the Bureau's own assessment of
the state of the census 2000, nothing would make me happier.
However, if its assessment differs, this subcommittee must know
and know quickly. The Members of this body are the ones elected
to provide stewardship over the Federal Government. The elected
Members of this body are also the ones ultimately held
accountable by the American people.
The mission statement of the nonpartisan GAO, as stated on
its website, says the following,
The GAO's mission is to help the Congress oversee Federal
programs and operations to assure accountability to the
American people. GAO's evaluators, auditors, lawyers,
economists, public policy analysts, information technology
specialists, and other multi-disciplinary professionals seek to
enhance the economy, efficiency, effectiveness, and credibility
of the Federal Government, both in fact and in the eyes of the
American people.
GAO accomplishes its mission through a variety of
activities including financial audits, program reviews,
investigations, legal support and policy/program analyses. GAO
is dedicated to good government through its commitment to the
values of accountability, integrity, and reliability.
From the outset, this committee has relied on the
professionals at the nonpartisan GAO to provide important
insight into a number of complex operations within the Census
Bureau and elsewhere. Many of those professionals at GAO were
involved in reviewing the 1990 census, including Chris Mihm,
Associate Director, Federal Management and Workforce Issues,
who will be testifying today.
As we get closer and closer to Census Day, April 1, the
interest in the census continues to rise tremendously. This is
certainly evident in the increased coverage of the census by
the news media. Many of these reports are locally oriented,
focusing on this community or that one, this county or that
town, this reservation or that rural community.
I find these stories important. They help to remind me that
to look at the census as a national census, controlled within
the beltway, is very wrong. While the census produces a
national head-count, that head-count is made up of more than
39,000 local governments that stretch from California to Maine,
Alaska to Florida, and beyond.
While Dr. Prewitt, in his testimony last week, said that
hiring is on or ahead of schedule, there does seem to be some
problems in various pockets throughout the country. And while
Dr. Prewitt and Ranking Member Maloney accurately pointed out
that there are going to be problems in an operation this large,
there is still reason to be concerned. If the Navajo
reservation in Arizona is having hiring difficulties, it
doesn't much matter that the Bureau is ahead of its hiring
goals in Miami. Additional workers in Miami are not going to be
flown to Arizona to count the Navajos. Communities that are
doing well do not have the ability to help those communities
that are doing poorly.
Dr. Prewitt also accurately noted that not all news stories
are accurate and not all news stories are highlighting certain
Census Bureau shortcomings. As Dr. Prewitt said later in his
testimony, one of the ways Congress and the American people
would know about a serious problem with the operational plan is
through news reports.
So when Congress has conflicting reports on, for example,
the success of the employment operation, it rightfully turns to
the GAO to shed light on this conflict and, hopefully,
reconcile the matter one way or the other or, at a minimum,
provide Congress with more information to consider.
Beyond the employment issues, which are at an important
stage, the subcommittee will hear today the status of the DCS
2000, the new data capture system. The Bureau expects to
capture nearly 1.5 billion pages of data from approximately 119
million households. These pages will be captured at four data
capture centers where the handwritten forms will be optically
scanned, converted into files, and transmitted to Bureau
headquarters for tabulation and analysis.
The GAO has recently released a report on the DCS 2000. The
GAO and the Inspector General's Office are very concerned about
delays and overestimated productivity regarding the operation
of the DCS 2000 system. If the DCS 2000 system does not
function properly, there will be serious problems in providing
the apportionment data to Congress on time, as required by law.
In December, the nonpartisan GAO released a report
outlining its concerns that the Census Bureau was in serious
need of a solid contingency plan. Last week, I was encouraged
to hear a few details about its contingency planning, such as
increasing wages and staying in the field longer than planned
doing non-response followups, but more is needed. Today the
subcommittee hopes to hear more about these reports as well as
future activities of the GAO.
Again, thank you for coming in to testify before the
subcommittee. And now I yield to the ranking member from New
York, Mrs. Maloney.
[The prepared statement of Hon. Dan Miller follows:]
[GRAPHIC] [TIFF OMITTED] T6541.001
[GRAPHIC] [TIFF OMITTED] T6541.002
[GRAPHIC] [TIFF OMITTED] T6541.003
[GRAPHIC] [TIFF OMITTED] T6541.004
[GRAPHIC] [TIFF OMITTED] T6541.005
Mrs. Maloney. Thank you, Mr. Chairman, and welcome to the
witnesses today.
The GAO is Congress' premiere watchdog, responsible for
providing credible, objective, and nonpartisan reports and
evaluations of the programs and management of the executive,
judicial, and, in some cases, legislative branches of
government. Without your work, our jobs as legislators and
overseers of the executive would be almost impossible. So thank
you for all of your hard work.
As I said last week, things seem to be going fairly well.
Recruiting is on track. 520 local census offices are open and
operational. The paid advertising campaign is moving smoothly
into its most active phase. Additionally, the legislation--
wait. I'm ahead of myself. The address list is nearly complete.
Some of the data presented in GAO's testimony indicates that
there may be some localized hiring problems. Although this is
helpful information, the GAO's findings do not affect my
overall judgment that all operations for census 2000 seem to be
on track.
As I also mentioned last week, I believe we need to be
prepared for all contingencies, which is why I've introduced
H.R. 3581. And I'd like to make it bipartisan, Mr. Chairman. I
hope you'll join me on it. This legislation would create a
contingency fund for the 2000 census. If there are problems
with the mail response rate or with the hiring program, funds
need to be available to respond to glitches in a very quick
manner so that the larger job of conducting an accurate 2000
census can be completed on time.
Following on recommendations from the GAO, this legislation
would also expand the labor pool to include active duty
military personnel and individuals who have received buy-outs
from the Federal Government. Additionally, the legislation
would allow recipients of Federal assistance to work for the
Census without a loss of benefits.
These are common sense preventive measures to ensure a high
quality census. I am looking forward to hearing GAO's comments
on my legislation since I tried to respond to the issues they
raised in their December report.
I am also very interested in hearing from you how GAO is
intending to act out its oversight responsibilities, while at
the same time being aware of the total number of watchdogs and
the demands they will be placing on the census at this very
critical time.
As you know, in addition to the GAO, overseeing the census
there is the committee and the committee staff, both sides of
the Census Monitoring Board, the Commerce Department IG, the
National Academy of Sciences Review Panel, and the Commerce
Secretary's Advisory Panels. Each of these groups has important
jobs and responsibilities. It is my hope that these various
oversight bodies have an awareness of each other and their
multiple requests and demands for information.
While we need strong oversight of the census, we need to
make sure that the oversight doesn't get in the way of allowing
the census to do its job. I am very interested in hearing your
thoughts on this issue. I believe that the 2000 census will be
one of the most accurate in our Nation's history, especially
after the raw head-count information is corrected with modern
scientific methods. I am
confident that the extensive planning that the Census Bureau
has done over the last decade and all the hard work of the
Census professionals will pay off with a more accurate count.
Thank you, Mr. Chairman.
[The prepared statement of Hon. Carolyn B. Maloney
follows:]
[GRAPHIC] [TIFF OMITTED] T6541.006
[GRAPHIC] [TIFF OMITTED] T6541.007
Mr. Miller. Mr. Mihm, if the three of you would stand up
and raise your right hands so I can swear you in, we'll begin
with the program.
[Witnesses sworn.]
Mr. Miller. Let the record identify that they all answered
in the affirmative. Mr. Mihm, do you have an opening statement?
STATEMENT OF J. CHRISTOPHER MIHM, ASSOCIATE DIRECTOR, FEDERAL
MANAGEMENT AND WORKFORCE ISSUES, U.S. GENERAL ACCOUNTING
OFFICE, ACCOMPANIED BY RANDOLPH C. HITE, ASSOCIATE DIRECTOR,
ACCOUNTING AND INFORMATION DIVISION, U.S. GENERAL ACCOUNTING
OFFICE; AND ROBERT GOLDENKOFF, GENERAL ACCOUNTING OFFICE
Mr. Mihm. Thank you very much, Mr. Chairman and Mrs.
Maloney. I just want to start out by saying how much I
appreciate your very kind words about the work that the GAO has
been doing and we look forward to continuing to support the
subcommittee in its oversight of the decennial census.
It's a real pleasure to be here to talk about the status of
the census. I'm very fortunate that I'm joined by two of my
colleagues: Randy Hite, who manages a range of GAO work on
Federal technology issues, including technology at the census;
and Robert Goldenkoff, who has day-to-day responsibility for
much of our work that we're looking at the decennial.
My statement today draws upon two recent reports which we
did at the request of the subcommittee in which we discuss some
specific operational challenges that have confronted the Census
Bureau as it moves into the key operations for the 2000 census.
Today I will highlight these challenges. First, achieving the
Bureau's mail response rate. Second, collecting accurate and
timely data from non-respondents. And, third, conducting data
capture operations.
Turning to the first major uncertainty facing the Bureau.
The mail response rate has declined in each of the last two
censuses and the Bureau expects to receive a 61 percent mail
response rate for 2000. To help boost public participation in
the census, the Bureau has instituted an outreach and promotion
campaign that is as ambitious as it is diverse. As Director
Prewitt noted last week, television advertisements already have
appeared on a number of programs and print ads have been placed
in a wide variety of publications. At the local level, the
Bureau has secured partnerships with local governments,
community groups, businesses, and non-governmental
organizations.
However, the Bureau's aggressive outreach and promotion
initiative faces a fundamental challenge. That is bridging the
historic gap between public awareness of the census and the
motivation to respond. This gap has been evident both during
the dress rehearsal taken last year and the 1990 census when
the public's high level of awareness was not matched by a high
mail response rate. In 1990, the Bureau found that about 93
percent of the population reported being aware of the census,
however the mail response rate was just 65 percent. This basic
pattern was also repeated during the dress rehearsal.
With respect to partnerships, the Bureau may have overly
optimistic expectations concerning the resources and
capabilities available at the local level to promote the
census. And here, Mr. Chairman, I completely agree with the
point that you were making. While the census is a national
undertaking, it's implemented locally and, therefore, we have
to look locally for some of our most constructive lessons.
A key element of the Bureau's local partnership effort is
the Complete Count Committee Program, which consists of local
government, religious, media, education, and other community
centers coming together to promote the census. Clearly, as was
discussed at last week's hearing, a number of communities are
aggressively supporting the census. However, the level of
activity and support for the census is likely to vary across
the country, in part because of a lack of resources.
We found that, during the dress rehearsal, the Complete
Count Committees often lacked the money, people, and/or
expertise to promote the census. In part to help, the Bureau
has hired over 600 partnership specialists. However, based on
the dress rehearsal experience, these specialists may be spread
too thin to offer meaningful support. Consequently, it is
unlikely that the Bureau's local outreach and promotion efforts
will be consistently applied across the Nation.
The second major challenge facing the Bureau is the need to
quickly and accurately followup on households that do not mail
back their census forms. Let me just give a sense of the
challenge that the Bureau faces. Let's assume that the Bureau
achieves its 61 percent mail response rate. Obviously, we all
hope it'll be higher than that, but let's just assume that
that's what they get. Census takers will then need to followup
on 46 million households. Completing this workload during the
Bureau's 10 week schedule will be an enormous challenge.
By comparison, during 1990, it took the Bureau 14 weeks to
followup on 34 million households. Thus under the current
schedule and response rate that the Bureau has, the Bureau will
need to followup on 12 million more households in less time in
2000 than in 1990, using essentially the same methodology. And
this is one of the fundamental challenges that the Bureau
faces, again, assuming all of its assumptions work out.
Experience from the 1990 census shows that, as field data
collection drags on, the accuracy of the information collected
tends to decline. This is because people move and others have
difficulty remembering who was residing in their household as
of April 1. As you discussed with Director Prewitt last week,
Mr. Chairman, to complete non-response followup, the Bureau
will collect data from second-hand sources, the proxy sources,
such as neighbors and mail carriers. Not surprisingly, however,
such proxy data are not as reliable as data obtained directly
from household residents.
During the dress rehearsal, although non-response followup
operations were completed on schedule in Menominee County and
Sacramento and 6 days early in South Carolina, the Bureau
collected proxy data at a much higher rate than it had hoped.
The Bureau's goal was to limit the proportion of the non-
response followup universe workload that was proxy to less than
6 percent. Unfortunately, however, in Sacramento, over 20
percent of the occupied non-response followup households was
enumerated using proxy data and, in South Carolina, 16.4
percent and in Menominee County, 11.5 percent. Compared to the
decennial census in 1990, there was about 6.6 percent of the
non-response universe was proxy. So we're looking at, at best,
about double, based on the dress rehearsal experience.
The Bureau's ability to recruit a sufficient number of
staff is another key challenge. The Bureau plans to fill about
860,000 positions for peak field operations, including 539,000
positions for non-response followup. To fill these positions,
as Director Prewitt mentioned, the Bureau wants to have a pool
of 2.4 million qualified applicants by April 19. The Bureau's
goal was to recruit 45 percent of the 2.4 million qualified
applicants, about 1.1 million people, by February 1.
The Bureau data, as of February 9, showed that, nationally,
the Bureau appears to be well on-track. It had recruited 1.3
million applicants or just over half of its total target.
However, national data masks the fact that the Bureau's
progress in recruiting qualified applicants lags in a number of
locations. As of February 9, 3 of the Bureau's 12 regions,
that's Atlanta, Chicago, and Philadelphia, and 178 of the 511
local census offices, that's about 35 percent, were below the
Bureau's 45 percent benchmark. Although some local census
offices were just a few percentage points below the goal, about
25 of them fell 20 percentage points or more. On the other
hand, and this is the favorable news, of the 333 local census
offices that were ahead of the Bureau's February 1 milestone,
163 of those exceeded it by at least 20 percentage points.
We suggested in our December 1999 report that Congress may
wish to consider legislative actions to modify labor provisions
that could prohibit or financially discourage specific groups
of people from seeking census employment. Proposals in this
regard, as Mrs. Maloney mentioned, are included in her
legislation H.R. 3581.
The third uncertainty I will discuss today is the need for
the Bureau to ensure the effective performance of its data
capture systems. The uncertainty falls into two basic
categories. First, ensuring the operational readiness of the
data capture system known as DCS 2000, which is the system that
each data capture center will use to check in questionnaires
and record census data. And, second, ensuring the readiness of
the data capture operations themselves, including the movement
in the processing of the paper questionnaires.
As we recently reported, the Bureau has made considerable
progress in acquiring and deploying the DCS 2000. However, we
noted that the Bureau was still facing a huge challenge in
delivering the promised DCS 2000 capabilities on time,
primarily because much remained to be done within the very
short time remaining before data capture operations were to
begin. Under the Bureau's current schedule, it has just 9 days
between the conclusion of the last system test and the date the
DCS 2000 must be operational in early March.
In addition, the numbers of yet-to-be-resolved defects in
the DCS 2000 were not yet showing the clear and sustained
downward trend that is expected as systems begin to mature.
Finally, of course, yet-to-be-completed development and testing
activities may surface even more problems.
The Bureau and its DCS 2000 development contractor shared
our concerns that we laid out in the report about the delivery
of the promised DCS 2000 capabilities on time and, in response,
were employing a series of important measures to minimize the
risk and expedite the completion of DCS 2000. The Bureau is to
conduct a final operational test involving all four of its data
capture centers on February 22 through 25 and we will be
monitoring those closely on behalf of the subcommittee.
Mr. Chairman, as I have discussed, despite intensive
efforts, the census still confronts some major operational
uncertainties. Because of these uncertainties, we recommended
in our report that the Bureau develop a contingency plan of
actions that it took to address a lower than expected mail
response rate. We suggested that the Bureau's plan address, at
a minimum, the budgetary scheduling, staffing, and other
logistical implications of collecting data from a larger than
expected number of non-responding households.
That contingency plan, which we believe should be shared
with Congress, could include options and procedures to balance
the pressure to meet census schedules against the need to limit
the use of proxy data. The uncertainties facing the Bureau's
data capture system make the need for a contingency plan, in
our view, even more compelling.
In summary, the Bureau has put forth a tremendous effort to
help ensure a complete and accurate count. It has tested and
retested its design and made significant modifications where
necessary. Nevertheless, substantial challenges to a successful
census remain and, as we have done throughout this decade, we
look forward to keeping the subcommittee informed of the
Bureau's progress and the results. This concludes my statement.
My colleagues and I would be pleased to answer any questions
that you may have.
[The prepared statement of Mr. Mihm follows:]
[GRAPHIC] [TIFF OMITTED] T6541.008
[GRAPHIC] [TIFF OMITTED] T6541.009
[GRAPHIC] [TIFF OMITTED] T6541.010
[GRAPHIC] [TIFF OMITTED] T6541.011
[GRAPHIC] [TIFF OMITTED] T6541.012
[GRAPHIC] [TIFF OMITTED] T6541.013
[GRAPHIC] [TIFF OMITTED] T6541.014
[GRAPHIC] [TIFF OMITTED] T6541.015
[GRAPHIC] [TIFF OMITTED] T6541.016
[GRAPHIC] [TIFF OMITTED] T6541.017
[GRAPHIC] [TIFF OMITTED] T6541.018
[GRAPHIC] [TIFF OMITTED] T6541.019
[GRAPHIC] [TIFF OMITTED] T6541.020
Mr. Miller. Thank you very much for the statement. Let me
start. You were involved with the 1990 census, I believe.
Right?
Mr. Mihm. Yes, sir.
Mr. Miller. What are your impressions of the overall
condition and maturity of the operations planned for 2000, as
compared to the situation that existed prior to April 1, 1990?
Are we better off? Worse off? Where do we stand? How does it
compare?
Mr. Mihm. In important ways, we are better off and in other
ways we're about the same. And let me start off in the ways in
which I think we're a little bit better off.
Certainly, it appears that, at the national level, the
Bureau's recruitment is going better than it did in 1990. They
have a paid advertising campaign this time around. Last time,
they were relying on pro bono, which was a bit of a challenge
because they were showing public service announcements at 2
a.m., when people wouldn't be seeing them. And, as you
discussed with Director Prewitt at the last hearing, the
advertising campaign, as a result, since it's paid, is far more
sophisticated this time around.
They also have extended the number of partnerships this
time around. They have about 55,000 different partnerships. Of
course, not all of them are as important as those that they
have with local governments.
Where they are about in the same state, however, is that
they are still showing pockets of areas where they have
problems in recruiting. And that was an issue that we saw in
1990 and what that lesson told us is that it is very difficult,
once you get behind the eight-ball, to sufficiently recover.
That is, the recruitment problems build on each other and you
end up having staffing problems during the census. We also saw
during 1990 that the use of partnerships, while important, was
also inconsistent across the country and it's something that,
as our work now suggests, they're going to have a similar
problem with this time.
And, finally, I guess a third area where they still have a
challenge is that while the quality and the placement of the
ads is far better this time, in my sense, than it was last
time, we still don't know and the Bureau doesn't know, whether
we have made the critical link between people being aware of
the census and actually being motivated to respond. And those
are the key challenges that the Bureau faced going into peak
operations in 1990. And I see they're pretty much the key
challenges this time around, as well.
Mr. Miller. The hiring process is going fairly well,
reasonably well. I recognize, of course, there are pockets of
problems. That's probably because it was paid advertising, we
think. One of the things you're not too sure of is what will
the overall $100 million of ad buys, you know, do. I'm a big
supporter of the advertising plans. I'm optimistic that's going
to be a big help. But, at any rate, that's kind of encouraging,
to some extent, that it's helping with our hiring in a full
employment economy. I know 1990 was fairly close to a full
employment economy, but not as full as it is right now.
Mr. Mihm. But not like right now.
Mr. Miller. So that's encouraging.
You heard Dr. Prewitt testify that he could not come up
with a contingency plan until he saw which census operations do
not meet expectations. Is this legitimate? Or do you think a
contingency plan for every major obstacle is truly feasible?
And they really do have a contingency plan, don't you think?
That they don't want to make public?
Mr. Mihm. Let me start with the first one and then,
hopefully, I'll be able to dodge the second. [Laughter.]
The first question about the feasibility of a contingency
plan, we would take a different perspective than the director
on that. We think it is important and it's also publicly shared
this with Congress. We saw during 1990, we saw during the dress
rehearsal, that limiting the amount of proxy data is very, very
difficult for the Bureau. And that it becomes an enormous
challenge as operations are going on, just the natural pressure
of ``let's get out into the field and get on with subsequent
operations,'' that they need to step back now and think about
how they're going to control the amount of proxy information,
how they're going to look at the relationships between mail
response and staffing needs and workloads at a localized level
rather than at the national level.
So we think that there is a real need for a looking at a
contingency plan.
Now I didn't mean to be flip about the second part of your
question, Mr. Chairman. I don't know if they have something in-
house. I agree with your opening statement that it was
important. Director Prewitt's statement last week was the first
time that I had heard a public statement from the Bureau that
they would be willing to extend non-response followup
operations if they weren't completed in time. In the past, some
senior people have been quite adamant with us that they would
be done in the 10 weeks and that was it.
Mr. Miller. What assurances does GAO have that the Census
Bureau will stay in the field as long as prudent to get non-
response followup work done? I mean, if he said 10 weeks, we
can finish in 10 weeks by just using more proxy data.
Mr. Mihm. Yes, sir.
Mr. Miller. Is the expectation to use more proxy data
because they are committed to that 10 weeks? If you use 14
weeks, there will be a much smaller non-response number. Is
that going to be their intention? Is that your impression, that
they're going to just use more proxy data, which is not as
accurate, we all agree?
Mr. Mihm. We hope that the experience of the dress
rehearsal is not instructive in this case, in which they got
out of the field on time or even early, but it appears at the
expense of having much higher rates of proxy information than
they had wanted.
One of the critical elements that they need to look at, and
then hopefully would be informing the Congress and keeping the
Congress aware of, is that as they are in the field longer,
there is more of a tendency to use proxy data. There is also
more of a tendency for people who they get from the households
to either be forgetful or to not give the correct information.
So just being in the field a long time is not good. The use of
proxy information is not good, in terms of data quality. A
careful balance needs to be made and they need to be looking at
that right now rather than waiting until everything is really
going on in the census and then on a case-by-case basis be
making those decisions.
Mr. Miller. Thank you. Mrs. Maloney.
Mrs. Maloney. Thank you, Mr. Chairman. I'm a little
confused about the workload for non-response followups that you
talk about in your report. According to your report, with the
expected mail-back response at 61 percent, the Census Bureau
will have to visit 46 million addresses. You go on to point out
that they have to complete interviews with roughly 650,000
households each day. That does sound like an overwhelming task,
but with 500,000 interviewers in the field, that is only 1.3
households a day or 9 a week. That sounds much easier and very
doable. It's my understanding that the Census Bureau's
assumptions about productivity is that each interviewer will
complete about 1 household an hour or about 25 interviews a
week.
Can you explain to me and to the panel why you believe that
it will be difficult for Census interviewers to complete 9
households a month and why that is so different from the 25
households a week in the budget assumptions?
Mr. Mihm. Yes, Ma'am. The point we were making in our
report is just that the very scope of the Bureau's efforts are
enormous. We're actually making two points.
Mrs. Maloney. The greatest peacetime mobilization ever.
Mr. Mihm. That's our mantra. And if there's ever a bigger
one, ``the second greatest peacetime'' will not ring as well.
The point we were making was that, just as the greatest
peacetime operation ever, it's an enormous challenge and, as
you're pointing out, that if they make their assumptions on
mail response, if they make their assumptions on workload, if
they make their assumptions on staffing, we're talking 670,000
cases and it will not be undoable for them. They'll be able to
finish on time or at least the math works that they'll be able
to finish on time.
However, the second point that we were making is that that
is fraught with a whole series of difficult assumptions or, as
we call them, challenges and uncertainties about whether
they'll make the mail response; whether they will get the
enumerator staffing that they need; whether people will be
willing to cooperate with them on a sufficient level. Those are
all the things that, in our view, at least raise the concern of
risk with the census.
But I quite agree with the point that you were making that
the math, in a sense, works out. That is, if they make their
assumptions, they should be able to finish on schedule.
Mrs. Maloney. You've certainly reviewed the Bureau's
assumptions about recruitment, retention, and productivity of
enumerators. If you feel the 10 schedule is too short, where
are these assumptions in error?
Mr. Mihm. The biggest problem that I think the Bureau will
face, and it gets back to the difference between a national and
local examination of the census, is that we know in a
percentage of district offices, that it will take the Bureau
much longer than the 10 weeks in order to finish non-response
followup. The last offices to close, I regret to report, were
in the New York regional office in New York City. Some of them
took 14 weeks.
And the challenge, and I know this won't be news to you,
Ma'am, is that these are also the areas where it is among the
hardest to enumerate. And so you have a snowball, in effect, of
interrelated challenges for the census: poor mail response
rate, hiring difficulties, high workload, large proxy data,
schedule problems. All of those come together in, not
nationally, but in hard-to-enumerate areas and in enough areas
to matter that hamper the overall success of the census.
Mrs. Maloney. Well, I read your report carefully. And,
based on your report, I introduced legislation that would
create a contingency fund of $100 million for census 2000. And
I am hoping that you've had an opportunity to review this
legislation and I'd like your comments. The fund could be
accessed if you run into serious problems. For example, if the
mail response rate dropped significantly, a point that you
continue to raise. My bill also expands the labor pool for 2000
census among certain specific groups, another recommendation
you put forward that I followed up on, along with many of my
colleagues, including active-duty members of the military,
those receiving certain Federal benefits, and Federal retirees
who have received buy-outs.
Mr. Mihm. Yes, Ma'am.
Mrs. Maloney. So I really tried to respond to the
contingency that you felt needed to be there by offering this
legislation and I'm wondering about your comments on it. Does
GAO endorse legislation? [Laughter.]
Mr. Mihm. Generally not.
Mrs. Maloney. Even if it's written mirroring your report?
Mr. Mihm. Well, in this case, the language in the report
based on the staffing was worded, and we were quite careful on
this, is that we understand that there are a series of
competing policy issues with staffing and that's why we offered
it as a matter for consideration for the Congress.
Similarly, with the part dealing with the contingency fund.
Clearly a lower than expected mail response rate has cost
implications for the Bureau. The Bureau has estimated about $25
million per percentage point. We've actually estimated it's
about $34 million per percentage point in direct costs.
Mrs. Maloney. It all has cost implications.
Mr. Mihm. Yes, Ma'am.
Mrs. Maloney. Even if you go past the 10 weeks, there are
cost implications.
Mr. Mihm. Yes, Ma'am.
Mrs. Maloney. So any way you look at it, there are cost
implications.
Mr. Mihm. And the Bureau has needed supplementals in the
past during the decennial census. The best mechanism for
getting them that money, though, is a policy determination,
that is, whether it's a contingency fund, whether it's a quick
supplemental if they need it. It is a policy call that we'll
leave to the Congress rather than engage in.
Mrs. Maloney. Well, I have quite a few more questions, but
my time is up.
Mr. Miller. Mr. Davis.
Mr. Davis. Thank you very much, Mr. Chairman. Mr. Mihm, let
me ask, if you had to give the Census Bureau a letter grade in
planning and preparation for this undertaking, what would it
be?
Mr. Mihm. I think that I would give them a firm ``B.'' I
think that it is unquestioned that they have worked very, very
hard throughout the decade. And I know there's been quite a
contention that we've been involved in as well on the issue of
sampling and the rest, but they've worked very hard throughout
the decade and certainly since the court decision to implement
the best census that they can.
One of our continuing concerns is that the census is a
national undertaking under the Bureau's leadership, but all of
us in other organizations have key roles in making the census
successful. That is a point that Dr. Prewitt made last week.
And so, in many cases, they're dependent upon local
governments. They're certainly most fundamentally dependent on
citizens to step up. And so, while we have had and will
continue to have, no doubt, some criticisms of, operationally,
how the Bureau is doing, fundamentally, our concerns and the
issues that we raise are about things that are really beyond
the Bureau's control.
The economy, as we've been discussing is very, very strong.
And, to the extent that they can get people to work on the
census in this sort of economy, that's really to their credit.
So it's issues such as that. So that's why we can be very
concerned about the census and call it high-risk while, at the
same time, I'd give them a firm ``B.''
Mr. Davis. That they've done a good job and that the other
things are difficult to really handle.
Let me ask, 10 years ago there were allegations after we
finished that there were people who had been counted twice. Are
you satisfied that the likelihood of that happening, in terms
of the preparation this time, has seriously diminished?
Mr. Mihm. We know for a fact that there were people counted
twice, as you mentioned, in 1990. It's every expectation that
there will certainly be a percentage of the people that are
counted twice. It's what the Bureau technically calls
``erroneous enumerations,'' that is people included twice, this
time around.
I think the key to reducing the level of double-counting is
to limit proxy and get out of the field, as soon as possible.
And let me give you just one number that kind of underscores
the issue. People in 1990, people enumerated between January
and April, who are basically people that mailed back their
census forms and other early census operations, had an
erroneous enumeration rate of about 5 percent. That is about 5
percent of those were double-counted. By the time you got
through August to December 1990, the erroneous enumeration rate
climbed to almost 30 percent.
So, basically, toward the latter part of the year, for
every three people you add, you add one person in error. And
that is a real challenge for the Bureau to control that. So, in
direct answer to your question, we have not looked in detail at
the procedures they have in place to guard against erroneous
enumerations this time, other than to urge them to control
proxy data and to get out of the field as soon as appropriate.
Mr. Davis. Would you consider that to be one of the big
concerns? I mean, if there are errors made and especially if
those errors are made in such a way that some advantage might
be given to populations that really don't need the advantage,
would that not be a great concern?
Mr. Mihm. Certainly. Everything we've seen is that the
Bureau does try and limit the level of erroneous enumeration.
We join the Bureau and most others, though, in focusing more
often on the differential undercount, rather than the rate of
erroneous enumeration, if, for no other reason, because it's
higher and more politically at issue.
Mr. Davis. Let me make sure that I understand that. The
double-counting, basically, were individuals who may have owned
two homes or had two residences and may have been counted at
each?
Mr. Mihm. That is certainly a part of it. We can certainly
get for the record, to the extent that the evaluation data is
available--the precise breakdown on this, but it would also
include people who did not have a usual residence and may have
been captured on two different census forms. The Bureau during
1990 had a coverage improvement program dealing with
individuals on parole or probation that had a very high
erroneous enumeration rate as well. But it certainly includes
the group, sir, that you're talking about.
Mr. Davis. And, finally, if there were ways to better
handle the external influences. That is, early on we talked
about those influences that the Bureau was not in control of or
could not project as much control of, would you have any
recommendations on how to improve that?
Mr. Mihm. I'm sorry, sir. I'm not capturing, I think, the
essence of your question.
Mr. Davis. I mean, for example, the extent of local
government participation. The extent of other agencies being
involved in assisting to help make sure that the effort is as
widespread, as broadly based, as we could make it.
Mr. Mihm. Certainly one of the recommendations that we have
made to the Bureau is to have realistic expectations for what
local governments are able to supply. Many of the largest
cities have very ambitious complete count efforts and are
really working very hard and have people with one or two
censuses of expertise in this and know at least as much as the
people in Suitland.
However, in other cases, and I think the dress rehearsal
experience bore this out, some of the smaller governments, more
rural governments, don't have individuals with the time or the
expertise or don't have the resources that they can really
devote to the census. Now the Bureau has a longstanding policy
that it doesn't fund these local efforts, however it hires
additional partnership specialists to help out.
We've looked at the spans of controls of these partnership
specialists, compared to what they were in the dress rehearsal
and they weren't able to give adequate support in the dress
rehearsal and there are even much greater spans of control now.
And so I think what we're going to see is, again, it's this
local versus national. We're going to see a very uneven
application of local support and the Bureau's ability to get
local governments and local communities to support the census,
just based on the resources available at a local level.
Mr. Davis. I'm not attempting to put words into your mouth,
but it sounds like you're saying that, in some instances, if
there had been resources to assist the local entity, in all
likelihood, that would have increased the level of
participation, which could have helped to increase, overall,
the level of effectiveness.
Mr. Mihm. I think, sir, at a minimum, what the Bureau could
have done is have more of an outreach effort to these local
governments. We looked at the notebook that it gave to some of
these local governments and it listed page after page what
local governments could do to support the census. There is
virtually nothing on what the Bureau was going to do to support
the local governments.
When we were down in South Carolina during the dress
rehearsal, some representatives of local governments told us
the Bureau came in, they gave us the hats, the T-shirts, and
the coffee mugs. And then went away. And, we need more support
than that, folks. We need the tangible support that you talk
about and we need knowledge. We need to know how to do this
sort of stuff.
So more facilitation and hand-holding would have helped, as
well.
Mr. Davis. Well, I thank you very much. And I'm pleased to
know that I think at least somewhat like the GAO, because I'm
in absolute agreement with you and I appreciate your response.
Mr. Mihm. Thank you, Mr. Davis.
Mr. Miller. We'll have another round, too. I was reading
the newspaper on my way back, flying back on Monday, I think it
was in the Washington Journal, about how the city of Detroit is
putting a lot of the resources of the city into it, but I think
the city of Chicago is making a specific effort to put their
own resources into making sure of that good count. They're
doing advertising and all that. So, you're right, it's going to
be spotty throughout the country, but those that recognize the
critical importance, such as Chicago and Detroit in particular,
they are putting the effort in there.
I was interested about Mr. Davis' first question about
grading the Bureau. I'm glad to see the grade. How would you
grade the contingency plan?
Mr. Mihm. Oh, well. [Laughter.]
If I was being charitable, it would incomplete. But that's
just because I haven't seen it, and so I have no basis to judge
at this point. I mean, the importance of this I can't stress
enough--the importance of making it available or making sure
that Congress gets an opportunity to see what's going on.
Because we know that, as data collection drags on, there will
be enormous pressure to close out offices and move on to
subsequent operations. There needs to be an understanding of
what sort of controls the Bureau has in place so that we don't
close out prematurely, that is, don't go to the proxy data
prematurely. We need to understand the tradeoffs between going
to proxy data versus staying in the field, both the tradeoffs
in cost and in quality and in schedule.
These are all the things that we think that the Census
Bureau should be willing to talk about and not just say, well,
we'll come to you if we need more money.
Mr. Miller. I know that somewhere out there there's a point
of diminishing returns for census enumerators to stay out in
the field and continuously pound away at non-response followup.
I am just concerned that the Census Bureau may decide to
prematurely put valuable resources into their ACE survey
instead of exhausting every available alternative in the field.
Would you comment some more on that?
Mr. Mihm. That is a concern. And, it's not just a concern
in the sense of ACE, but it's a concern in terms of any
subsequent operation that they would do any of the other
additional coverage improvement operations. This would be part
of the contingency plan that I think that they ought to be
willing to discuss with the Congress: If we stay an extra
couple of weeks in any particular area, here's what the cost is
in terms of getting in and starting ACE or it may well be that
there is no cost. Certainly there is no magic requirement that
the field work on ACE has to start nationally at any one
period. Just like the census, the ACE is done locally as well.
So this is the type of issue that they should be talking to
the Congress about and letting you know what are some of the
challenges and tradeoffs that they face. And they should be
doing this ahead of time, rather than tell us later that the
census is in real trouble.
Mr. Miller. I agree. I'm concerned that they haven't shared
with us a contingency plan. I feel there is a contingency plan.
And, as Dr. Prewitt talked about yesterday--about possibly
staying in the field longer or putting more money into this or
how much you pay enumerators, it's part of that process. I
think as far as money since it's over twice as much as the 1990
census, as you point out in your report, I think there's a lot
of cushion in that money to be able to move around and shift it
to those areas that may need the additional resources to
complete them.
But I do have a concern that ACE is driving the close-out
procedures, which would sacrifice a full-enumeration census,
which would be unfortunate. And you share that concern, I
guess, yes?
Mr. Mihm. It is something that we are going to be looking
at very closely as non-response followup operations get
underway, as to what controls the Bureau has in place when they
close out, what the level of proxy information is that they're
collecting in these last offices. During 1990, the areas that
were the hardest to enumerate for them, large urban offices, it
was not uncommon for them to average 20 percent or more proxy
data of their non-response universe and this is a real concern
when you're getting that percentage of the population that's
based on proxy.
Mr. Miller. And the later you get in the field for ACE, the
less accurate ACE can be. If you have to wait 14, 18 weeks, you
know, or whatever----
Mr. Mihm. Yes, sir. They're asking people also, just like
with non-response, you're asking them to recall April 1, in ACE
you're asking them to recall April 1 as well.
Mr. Miller. Right. Despite the Bureau's efforts, the data
capture system may still be at risk. Do you have any
suggestions as to anything more the Bureau can do to diminish
that risk at this late date?
Mr. Mihm. Randy's really the expert on that and I'll ask
him to.
Mr. Hite. I have two responses to that. The first deals
with, at the time that we were looking at DCS 2000, we
identified the high probability that, given the significance of
the events that remained and the nature of the events that
remained, relative to developing and deploying DCS 2000, there
was considerable risk that everything was not going to get done
on time. And, at that time, we spoke to the Census and its
development contractor about how they could address this and
the type of risk mitigation strategies that would be effective.
Both the Census and the contractor were very responsive to this
and, in fact, as we note in our report, put mitigation
strategies in place.
What has happened since we've done our evaluation has borne
out one of our concerns, which is the type of events that
remained, test events, are events intended to identify
problems. That's what tests are designed to do; they identify
problems. They don't demonstrate the absence of problems, but
the presence of them. And what has happened as a result of the
problems that have surfaced recently and that we just became
aware of this past Friday, is that in order to address the
problems, they've decided that they need to modify the system.
And so here we are at the late stage in the development and
implementation process where Census will need to modify the
system, where Census will need to release the software changes,
and then Census will have to test them in the field. And so
what they've done is exasperated a risky situation, because the
test event will occur I believe the 22nd through the 25th, when
these changes will be tested. And that will leave you 9 days to
address any problems that the test will surface. And, as I
mentioned before, tests are designed to identify problems. Nine
days is not a whole lot of time to deal with problems.
Mr. Miller. Is there a better data capture project, 119
million forms, in any other Federal program that can capture
that much data in such a quick period of time?
Mr. Hite. The similar application that comes to mind is the
tax processing systems, because the forms come in within a
certain timeframe and they have to be processed within a
timeframe. It's heavily manual, but it is also heavily
automated, too. That tax processing infrastructure has been in
place for a long time and it's really done, as you know, year
after year. So, in that sense, it is a different situation.
But census data capture, in some ways, is analogous to the
Y2K problem because there, too, we had an immutable deadline
that we had to deal with. And what happened was agency efforts
were pushing further and further back up against the deadline.
And any system development, any system maintenance effort,
unless you change requirements and thus reduce the magnitude of
the task that you're trying to accomplish, the only thing that
can give is the testing process. And what we have here is where
testing is the end of the process, but if problems surface,
they will have to be corrected and then retested, again, to
make sure that they, in fact, the system is performing
correctly.
Mr. Miller. Thank you. Mrs. Maloney.
Mrs. Maloney. You said that the Bureau needs to, ``limit
the use of proxy data and get out of the field as quickly as
possible.'' So if the response rate drops below the
expectation, would it be better to keep enumerators in the
field beyond the planned 10 weeks or would it make more sense
to put more enumerators in early in an attempt to get the total
out there quickly and to get out of the field as soon as
possible? What would be better?
Mr. Mihm. Mrs. Maloney, it would always make more sense for
the Bureau to put as many enumerators in the field as early as
they possibly can. They are going to try and hire well over
500,000. If they can get 600,000, that's better. If they can
get more than that, that's always better. And this has been the
Bureau's traditional position, is that they will say this is
how many positions we have, but if they get two applicants who
are qualified, pass the test, pass the background check,
they'll split a position into two positions in order to get
people in there. However, that often does not happen because of
the hiring problems that they have. But, nevertheless, it is
always better for them to get as many people in as possible so
that they can get out of the field with as complete data as
possible.
Mrs. Maloney. Following up on the labor market, you noted
in your testimony that 25 LCOs fell short of their recruitment
goals by 20 percentage points or more. You also note that 163
offices exceeded the 45 percent goal by 20 percent or more. To
me, that sounds pretty good.
I'm curious about two or three things, though. Are there
generally any similarities about the offices with recruitment
problems? Are there procedures in place to address the
shortfall areas, like intensified promotion or maybe sharing
recruits among different LCOs, if they are close enough?
Finally, are there any recruiting problems in New York City?
[Laughter.]
And, I might add, Chicago or Florida? [Laughter.]
Are we 1 of those 25 LCOs, any of us?
Mr. Mihm. Well, Chicago is one of the three regions that is
having the most problems. 25 of its district offices, to each
region there's usually 40 to 45 district offices, so 25 of its
LCOs did not make the February 1, 45 percent threshold. While
New York City as a region made the threshold, it has had some
problems. Eleven of its LCOs were below the 45 percent
threshold. The Atlanta region, which was another one of those
regions, unfortunately covers Florida as well, and there are
pockets of problems in Florida.
You asked if there is a consistent lesson, which is a
tendency that urban areas are the ones that have, typically,
the hardest problems recruiting. We saw that in 1990. We saw it
in dress rehearsal. We're seeing it again in 2000.
Among the things that the Bureau is doing is that they are
intensifying recruitment efforts. In some locations, they are
studying the possibility of raising some wage rates. Dallas is
the region that has done among the very best in terms of its
recruitment. And while the census is always local, there are
some real lessons learned there going on in trying to replicate
some of the lessons that Dallas has done in terms of its
recruitment effort. Apparently one of the things that a number
of the regions have learned from Dallas is how important it is
to really support the regional recruiters and the local
recruiters to give them some additional training and help. And
so there is that leverage that's going back and forth.
Robert, you're closest to the field. Do you have anything
that you want to add in particular?
Mr. Goldenkoff. Yes, we can talk specifically about this
issue. Let's first mention some things that are specifically
being done in the Dallas region. Postcard mailings they said
have been very helpful. Extensive training of recruiting
assistants who enhance the message of the importance of the
census. There's a lot of activity going on with the partners.
There have been websites set up and State and local governments
have put up websites to assist in recruiting. There's a 1-800
recruiting hotline. So there have been a number of procedures
put into place to facilitate the recruiting.
Mrs. Maloney. OK. Your report, Mr. Mihm, also says that the
dress rehearsal data suggests that the Bureau's outreach and
promotion program may have only a modest impact on the mail
response rate--that the advertising that was done during the
dress rehearsal was extremely minimal and really not very
professional--certainly nothing like the very professional
program in so many different languages and nationalities and
ethnic groups that the $100 million-plus campaign now has
going. Don't you think that this nationwide campaign may have
more of an impact than it did on the dress rehearsal?
Also the activity that I'm hearing from my colleagues, some
of them are organizing marches and handing out literature. One
has developed her own contract that she's handing out to every
constituent: I pledge I'll fill out my form. My own personal
favorite is the census in the schools. I am convinced if we
could meet with each superintendent and get them to put that
into the schools that that would increase dramatically.
One member had a great idea they shared with me the other
day. The whole CHIP program, the enrollment of children in the
health care plan that is way behind expectations, partnering
with them as they're enrolling these young people, also
enrolling and reminding their parents about the importance of
filling out the census form.
Maybe it's because I'm living it every day. I see all this
activity and all these ideas and I'm more optimistic of
generating knowledge and a desire to fill out that form and
send it back. But you are not particularly optimistic in your
report, based on the dress rehearsal. But I don't think that's
very indicative of what we have going on in the field now.
Mr. Mihm. We're hopeful, with you, that all of these
additional efforts will make a difference. And I completely
agree. We tried to capture this in the report, that there is a
lot going on for the 2000 census that obviously was not going
on for the dress rehearsal.
Mrs. Maloney. Even an ad during the Super Bowl.
Mr. Mihm. Yes.
Mrs. Maloney. The test that everybody remembered. Fill out
your form. Don't leave it blank. So, I mean, I think there's a
tremendous amount of effort out there.
Mr. Mihm. I happened to be watching the Super Bowl with my
daughter who's 9 and she picked on why are they having
classrooms in the janitor's closet. Will we have to do this,
Dad? We're in Fairfax and I said, no, honey, they put you all
in trailers out back. That's the Fairfax issue. [Laughter.]
Mrs. Maloney. Your daughter said that?
Mr. Mihm. No, I was the one who told her they put them in
trailers. [Laughter.]
But she was very concerned. So they've hooked into 9-year-
olds at least on this. And so it's clearly a larger, more
persuasive ad campaign. The unknown, and this was the point
that we were trying to make in the report and I think the
Bureau is wise in not saying that, oh, the paid advertising
campaign is going to give us this much of a bump in the mail
response rate. What is unknown is whether we've broken that big
historical pattern, the difference between awareness and
motivation. We're certainly hopeful, as is everyone else, that
as people get more aware, this time around they will be
motivated.
But in the dress rehearsal, we had very, very high levels
of awareness in 1990--very, very high levels of awareness--it
just didn't translate into a mail response. And that's going to
be the critical juncture.
Now the Bureau's ad campaign is moving into its second
phase. The first phase that Dr. Prewitt talked about last week
was just awareness, letting us all know that the Census is
there. And now we're really getting into the intensive
motivational part. You noticed the kick-off yesterday and I
didn't get a chance to see the TV this morning, but I
understood there was going to be something this morning in New
York and elsewhere on this.
Mrs. Maloney. You can't walk down the streets in New York
and not see a sign to apply for a census job. I mean, it is
literally everywhere. They have it up in stores. They have it
everywhere.
Mr. Mihm. That's great news.
Mrs. Maloney. I'm really impressed. Maybe they're doing it
just around my home and my neighborhood. [Laughter.]
Because they know I'm on the case. But, believe me----
Mr. Mihm. We'll check for you. [Laughter.]
Mrs. Maloney. I'm really, really impressed. I mean, you
see, if we don't make it through this next election, we'll have
a job out there. [Laughter.]
But my colleague has a lot of important questions to ask.
Mr. Miller. Well have another round, too. Mr. Davis.
Mr. Davis. Only one. Actually, I want to get back to the
individuals who are homeless, who have no stability in terms of
addresses to followup with. And, given the fact that we're
going to be into spring, which means the weather will hopefully
be good, it really increases the numbers of people who move
about without any place that they call home or where you can go
looking for them to actually find them. Have you seen any
evidence of the Bureau's ability to reach those people to find
them?
Mr. Mihm. Certainly, sir, the Bureau is undertaking a far
more serious effort in 2000 than in 1990. That's not the right
word. I don't mean to imply it wasn't serious in 1990. It's a
far better designed effort this time around. In 1990, they had
a single night that they called S Night in which they went out
and they attempted to count people on streets and they went to
shelters and attempted to count people here. This time it is a
far more intensive effort covering a couple of nights in which
they're going to be out looking for people. We plan to be
monitoring that and be prepared to report back to the
subcommittee on how it's going.
But, as you know, the thesis behind your question is right
on about how difficult it is to get these people because the
challenge the Bureau faces is that, especially people on the
street, is that many of them don't want to be enumerated. And,
you know, as we hear each year when, unfortunately, when some
people freeze to death because they won't even go to a shelter,
you can imagine how difficult it gets, then, to get them to be
willing to talk to an enumerator.
And so the Bureau really faces a very, very difficult task
in getting the people without traditional housing.
Mr. Davis. Well, I'm really pleased to know that because,
while the numbers in many instances may not appear to be that
great, and in some instances they're not, but in others I think
that they're quite substantial. And that every effort has got
to be made to try and reach those individuals. Because even
when we start talking about the return of resources, obviously
these are the communities and these are the people who need
those resources the most, trying to help them out of the
situations that they're currently in. And so I'm very pleased
to know that and I thank you very much.
Mr. Mihm. One thing that it's important that we all keep in
mind, Mr. Davis, is that the Bureau does not release a count of
the homeless population. They release a count of people where
they happen to reside. Some people live in streets. Some people
live in shelters. Some people live in other housing. And they
leave it up to others to, if they wish, to come up with a
homeless count on that.
And the point there is that it is important that we not
just take the number of individuals enumerated in streets or
shelters and automatically assume that that is the total of the
Nation's homeless population. It could be quite larger.
Mr. Davis. Thank you. Thank you very much, Mr. Chairman.
Mr. Miller. We'll continue the questions. I have a short
one. The 61 percent, what is your projection?
Mr. Mihm. Well, we don't have an actual projection, Mr.
Chairman, but let me tell you about the source of a little bit
of our concern on that. The Bureau met its mail response rate
goals during the dress rehearsal. They're always lower for a
dress rehearsal. It's usually in the 50's, about 55 percent.
But they met that using a second questionnaire, which, for
a variety of reasons, primarily because of public confusion and
matching problems, they decided not to pursue for the 2000
census. Now what they did is that led them to reduce their
expected mail response rate for 2000 from 65 percent to 61
percent.
Our point is, and we try to make this clear in the December
report, that the bump that they got from this second
questionnaire during the dress rehearsal was actually much
greater than 4 percentage points. They got in some cases I
think it was between 4.5 or a minimum of 4.5 but it went up to
as much as 15 percentage points in some locations.
And so it's a real concern to us and we haven't seen from
the Bureau an articulation of: we understand we got a huge hit
out of the second questionnaire. We're not using the
questionnaire. Here's how we think we're going to make up the
difference between what that second questionnaire would have
given us.
Mr. Miller. Compared to 1990, how many enumerators are they
projecting? Do you know?
Mr. Mihm. They are projecting to hire 500,000----
Mr. Miller. But in 1990, what was it?
Mr. Mihm. In 1990, they ended up, I think they had
positions for 370,000 or thereabouts and ended up hiring,
because of turnover, well over 500,000. I'll have to get you
precise numbers on that, sir.
Mr. Miller. One more quick question, a question to followup
Mr. Davis on the double-counting issue, is the DCS capable of
reducing that and minimizing that compared to 1990? The
computer, as far as the double-counting? If a college student
is counted twice, we want to avoid that, of course. I think
they're better prepared to do that. How would you rate their
ability?
Mr. Mihm. The reason we were kind of passing the microphone
back here is that it's not so much a DCS issue. It's an issue
dealing with their match rules. We have not looked at the match
rules this time around, but we'd certainly be willing to take a
look at those and get back to you with that information.
Mr. Miller. I was just thinking in terms of being a little
more sophisticated, computerwise, for 2000. They should have a
better ability, I would hope----
Mr. Mihm. They should have.
Mr. Miller [continuing]. To catch----
Mr. Mihm. One would hope, but we'll have to report back to
you on that one.
Mr. Miller. Would the GAO be investigating and evaluating
the Bureau's ACE plans?
Mr. Mihm. Your office has made it clear that, as has Mrs.
Maloney's, once this hearing is over, they want to sit down
with us and continue discussions that we have begun about what
we're going to be doing for 2000 and beyond. That is certainly
something that we expect that you and Mrs. Maloney will ask us
to look at.
Mr. Miller. Do you have any idea how far along the Bureau
is in planning for the ACE? And have you any indication as to
when they will have a complete operational plan? And do you
feel they are where they should be on a preparation timeline
for ACE?
Mr. Mihm. We haven't looked at ACE directly in quite some
time. I can report that a couple of weeks ago, the National
Academy of Sciences held a fairly major symposium. I think Dr.
Prewitt mentioned it in his last hearing, I know there were
members of your office that were there. The NAS brought in just
about anyone who is anyone on the issues of statistical
adjustment, both pro and con.
One of the common themes that I heard coming out of that
was that the Bureau really does need to start locking down some
of the procedures that it's going to use for the ACE and for
adjustment. I detected a bit of a tone of frustration from some
of the experts there that it's time for the Bureau to start to
move beyond. ``Well, here's one option, here's another
option,'' and actually get into, here's what we think we're
going to do.
Mr. Miller. Please outline the GAO's plans for future field
investigations into the decennial census activity.
Mr. Mihm. I obviously want to stress, sir, that this is
subject to your approval and Mrs. Maloney's approval. I mean,
we work at the behest of this subcommittee.
What we have planned to do is to look at--well, we've made
contacts with people in the regional offices. We're then going
to be looking at a subset of local census offices, probably in
the neighborhood of 20 to 30, somewhere in there. It depends on
resources. It depends on where our colleagues in the Inspector
General's Office are going to be to make sure that we minimize
any disruption and any overlap and appropriately leverage off
of what they're doing. And we're hoping to be----
Mr. Miller. Does that mean you'll get to different places
or will you try to go to the same places?
Mr. Mihm. We're hoping that we can ask a consistent set of
questions in different places. We are also sensitive, though,
that we also have different reporting requirements. And this is
one of the issues that we get into with the monitoring board
and others is that we report directly to the Congress. They
have other constituencies or other things that they have to
report to.
What we're going to be doing is looking to get into the
field and visit these local census offices at two points. One
within the next couple of weeks before peak census operations
begin. And, second, as I alluded earlier, toward the end of
non-response followup to get a real sense on how are we doing
on close-out; what sort of pressure, if any, is being applied
to let's get out of the field early; are we doing everything we
can to get full enumeration with the final cases or are we
closing up prematurely.
Mr. Miller. How many people do you have, approximately,
assigned or will have assigned to the census issue over these
next couple of weeks?
Mr. Mihm. We're very fortunate in GAO that we operate using
an approach to matrix management and so it's, directly, working
full-time on census work, we probably have about eight people,
as well as a couple of--Randy is very kind to devote a lot of
his time and his staff on information technology issues.
Colleagues in a different part of the office did the report for
you and Mrs. Maloney on the budget scrub last year and they've
devoted resources as well. We're quite confident that we will
be able to meet any requests that you give us.
Mr. Miller. Let me ask one more question and then I'll be
finished. The 2010 census, are you all looking at--I meant to
ask Dr. Prewitt this--preparing for 2010, running some tests?
Are you looking at that at all or do you have any comments
about it?
Mr. Mihm. We will be looking at it. One of the things that
I learned coming out of the 1990 census--or there's actually a
couple of things. One is the importance of conducting
appropriate tests during a live census that point to the next
census. And then the second thing is the importance of starting
early with your census planning effort. There was a big problem
of some controversy here in that the Bureau's planning efforts
for the 2000 census really didn't tee up for Congress a lot of
the key issues until relatively late in the decade, causing the
Census Bureau to have to rethink its approach.
One of the things that the Bureau is exploring--and we're
certainly going to encourage them in this regard. We encouraged
them last time as they were preparing for 2000--is looking at
the use of administrative records. Either to, at the broadest
extreme, to help with the basic enumeration and to even its
subsets, to help with some of the coverage improvements, to
programs to try and look for missing elements.
One of the highest undercounted groups are children under
5. And so there are opportunities to use administrative
records, whether it be school records or anything or Head Start
records. There are all sorts of policy and privacy concerns in
there. There are all sorts of technical issues and records
matching. But this is the time in the decade to start thinking
toward 2010 for issues such as these.
Mr. Miller. I agree. I think, you know, whether it's the
WIC program--there are a lot of programs--or the Indian
reservations in undercounted areas, there are a lot of
administrative records that I would think could be useful and I
know they don't use them now except for our military and such.
But I am hopeful for that.
Mrs. Maloney.
Mrs. Maloney. Certainly. Thank you. Thank you, Mr.
Chairman.
One of the conclusions you've reached in this report is
that the Bureau's estimated 61 percent mail response rate in
2000 may be optimistic for two reasons. The first is the
decision not to employ a second mailing as was done in the
dress rehearsals. And let's clarify this please, you're not
suggesting that the Bureau change its operational plan to
include a second mailing, are you?
Mr. Mihm. No, Ma'am. Not at this point, no.
Mrs. Maloney. But in your report, you suggest that the 6
percentage point reduction and the estimated mail response rate
from 67 percent to 61 percent may not have been large enough
since evaluations of the dress rehearsal indicate that it may
have been responsible for a greater percentage of responses
than that.
Mr. Mihm. Yes, Ma'am. We are not advocating that they go
back to a second questionnaire, in part because we didn't have
the time to fully evaluate whether or not there was the
matching issue in the scope that they said it would be. But
what we were pointing out in this report is that the second
questionnaire gave them a sizable bump in the mail response
rate during dress rehearsal. They took some reduction in the
anticipated mail response for 2000. It doesn't appear, though,
that they took as much as they should have.
And so our question for the Bureau is, where are we going
to make up the difference? During the dress rehearsal, it gave
you 8 or more percentage points. You reduced your 2000 expected
much less than that. Where are you going to make up the
difference? And that's what our concern is.
The only thing that, and I don't want to waste your time
here, but the other thing that we wanted to point out is that,
for the Census, it's only a 1 or 2 percentage point difference
that can be a real challenge. You heard Dr. Prewitt mention
that last week, that he can probably handle 60, 61, but if he
starts getting to 59 percent, he gets in a heap of trouble in a
hurry. Each percentage point is another 1.2 million cases. And
so we don't need a catastrophic event, which certainly nobody
wants, in order to be in a very difficult place in a hurry.
Mrs. Maloney. But, please, you compared results from the
nationwide testing which the Bureau conducted into a second
mailing with the results from the dress rehearsal. Explain how
it affected your analysis.
Mr. Mihm. I'm sorry, Ma'am. The nationwide test?
Mrs. Maloney. Yes. The nationwide testing which the Bureau
conducted, you put that into the second mailing with the
results from the dress rehearsal. Can you put those two numbers
together?
Mr. Mihm. What we did is we took it--no, we took the second
mailing--the percentage of households that responded during the
second mailing in the two principal locations, that is the
Sacramento and South Carolina. We subtracted out of that, out
of there overall mail response rate, the people that responded
due to a second mailing and got another mail response rate. And
rather than it being in the 50's, it was typically, therefore,
down in the 40's mail response rate. Again, comparisons between
the dress rehearsal and the census must be made with caution.
We got down to a mail response rate in the 40's and said we
saw where the Bureau had taken a reduction in the anticipated
2000 mail response rate. It was just not equivalent to the
percentage point increase that they got from the second mailing
in the dress rehearsal. Our question for the Bureau was, and
remains, is how are we going to make up the difference? Where
does that difference come from? And, you know, it's hoped that
it will be through the ambitious and national ad campaign.
Mrs. Maloney. OK. You mentioned in your testimony that the
contractor for the development of the software is Lockheed
Martin and it has been independently rated very highly. Would
you elaborate on that, please?
Mr. Mihm. Yes, Ma'am. That's Randy.
Mr. Hite. The rating that we're referring to is the
Carnegie Mellon Software Engineering Institute's capability
maturity model, which lays out effective practices that a
mature software development organization would possess. It
rates organizations on a scale of 1 to 5, with 5 being the
highest level of maturity. In this case, Lockheed Martin's
mission systems division, in particular, has recently been
rated as a level 5 organization. So it's a very mature, very
capable, very effective software development organization,
which is a huge plus that the census is doing business with an
organization like that.
Mrs. Maloney. Are many software companies rated as 5? Or is
it unusual to have the higher rating?
Mr. Hite. That's a lofty group of companies. It is not a
large number of organizations that I am aware of that have
obtained that level of maturity.
Mrs. Maloney. And how was the contractor selected? Was it
done through competitive bidding or only stage 5 could apply?
How was it done?
Mr. Hite. I do not have that information. I would be happy
to provide that for the record.
[The information referred to follows:]
The DCS 2000 contractor was selected through a competitive
RFP. A public notice was published in the Commerce Business
Daily. There was no specific mention of a minimum CMM
accreditation level in the RFP.
Mrs. Maloney. Thank you.
You also said that the results from system level tests show
that the DCS 2000 performance targets are being met. What are
some of those targets? Let's get something positive out here.
The positive things that have been met.
Mr. Hite. There are a number of performance measures that
are used to measure how well the system is performing. There
are throughput numbers in terms of the number of forms that are
being moved through the sorters and the scanners, for example.
And while the numbers from the tests that were performed at
Pomona were below the target levels, and I can provide the
precise numbers for the record, if you're interested, these
tests were also repeated in the Phoenix test, which was
completed recently. The Census Bureau has informed us that they
have exceeded the goals with regard to both the sorters and the
scanners, in terms of throughput of the number of forms that
moved through the system.
Mrs. Maloney. In your discussion of the data capture
system, you also indicated that the productivity rates observed
during operational testing in California, Pomona, CA, were
below expectations. Have you examined the data from the other
test sites? And, if so, what were your results?
Mr. Hite. The numbers for the key from image productivity
rates on the part of keyers were below the model numbers that
were expected. And, what has happened as a result of that, is
that changes to the DCS 2000 system have ensued. So, in fact,
what they're doing is modifying the system so that the
workloads that the key from image keyers would be receiving
will be reduced. So here was a case where, not the system, but
the human element of operating the system was not performing up
to expectations. And, to respond to this problem, the solution
has been to modify the system so that Census will conduct a
two-pass read of the forms. And, initially, they'll just
extract the 100 percent data from the forms and then the sample
data will be collected at a later point in time.
Mrs. Maloney. I understand that there will be a four-site
full-load test of the data capture system next week. Would you
explain what is involved in that test and exactly what you're
looking for?
Mr. Hite. At this test, they will be operating all four
centers at the production levels that they expect during the
actual data capture operations. All software, all releases, all
hardware will be in place at all the centers. That's the plan.
Census will be able to simulate actual operations, that is,
post March 6 environment when the data capture centers are to
be operational. They will be operating centers simultaneously.
So not only will it be able to test the performance of the
data capture centers, but also, for example, how well the
centers are transmitting data to headquarters so they can
monitor how well data capture operations are proceeding. So it
will allow them to test the full operation of the system in a
real-live operational environment as we will have to do during
the actual data capture operations.
Mrs. Maloney. What remains to be tested before everything
is fully operational?
Mr. Hite. As recently as Friday, and I believe actually it
was over the weekend, the final software release, which was
software release 23, was sent out to the field. This software
release, along with the releases that preceded it, will be
tested as an integrated set, along with some hardware that has
been added recently, associated with additional disk drives for
storage. These will be tested from February 22 to February 25
as part of this operational test.
So what remains to be tested is not only the capability
that has been deployed here recently, but also the correction
of the problems that have surfaced as part of the Pomona test.
We don't yet have all the information from the Phoenix tests or
the Jeffersonville tests to find out what kind of problems
surfaced.
But what happens is you do these tests, then you identify a
problem and you fix the problem. Then you send out the patches
to the software to correct those problems. Then you test them
to make sure they're operating correctly. This last test of all
four sites will be testing, hopefully, the completed system in
a real-live operational environment.
Mrs. Maloney. My last question: Where will you be on census
day? [Laughter.]
Will you make sure you're counted?
Mr. Mihm. Well, we're going to enumerate before census day.
We're going to send it as soon as we get the form, Ma'am.
[Laughter.]
Mr. Miller. Two more questions. You tell me the data
capture system is--apparently they're going to do a two-path
system now. What are the ramifications of that? That the first
path will be the seven questions is my understanding, and then
they're going to have to rerun all of the long form? Is that
right? What are the ramifications for doing that?
Mr. Hite. You are correct. What they will do is they will
still run the forms through and create the digital image of the
short forms and the long forms. What they'll do differently
concerns the optical scanning of the marks on the forms and the
characters on the forms. They'll only run the seven questions,
the 100 percent data. That's all they will extract. The images,
then, will be stored on some hardware that they've recently
acquired. And then, later on, once they've completed the data
capture of 100 percent data, they will retrieve the digital
images from disk storage and they will extract the sample data
from the long forms. That will be submitted to census
headquarters.
There are issues associated with this, one of which is the
changes to the system associated with the retrieval from disk
storage, rerunning the images and extracting the data. Those
changes have not yet been made to the DCS 2000. Those are going
to have to be made over the ensuing months. So that's an issue.
There are also downstream issues in terms of how the two
pass, if at all, affects the processing operations at
headquarters. I don't know that they do or don't, but that's a
potential issue. What I suspect is that there could be other
issues.
Mr. Miller. What about the delayed release of all that
information? Is that a factor or not?
Mr. Mihm. We're going to be looking into that, sir.
Certainly much of the information on the long form is required
to be collected by separate statute. And we have our attorneys
back at GAO right now going through that and trying to figure
out when that information, by statute, has to be available.
Mr. Miller. Let me ask one more clarification on this
second mailing issue. There's no question that it's too late to
consider a second mailing now or even months ago. But 2 or 3
years ago, with what you know today would your recommendation
have been to do a second mailing if they could have planned for
it 2 years ago?
Mr. Mihm. It certainly would have been worth more
investigation on their part. The initial argument that the
Bureau made in rejecting a second mailing is they said the
public, and they had some data in South Carolina, was confused
by the second questionnaire. And they held up press articles of
people saying they got two census forms. It's difficult for me
to imagine that given as sophisticated as their ad campaign is,
they couldn't have designed a component that says, you're going
to get two census forms. You know, if you've already filled in
the first, don't fill out the second. We all subscribe to many
magazines and the bills say, if your payment crossed this bill
in the mail neglect this.
It was then, later on, though, they began to raise the
issue that they would have trouble matching, because they would
just be completely overwhelmed with these second
questionnaires. Again, perhaps a sophisticated ad campaign
could have reduced the number of duplicate questionnaires that
they would have gotten in and maybe the matching could have
taken care of it. That is certainly something in the type of
the thing that they need to be studying early in the decade for
2010 and not wait until the dress rehearsal, which is supposed
to be the final operational test, to be rejecting such a major
element of the census.
Mr. Miller. Thank you. Let me thank you very much for being
here today. And, as we've asked Dr. Prewitt to come and brief
us on a regular basis, I hope you will come back on a regular
basis in these next few months that are critical. So let me
thank you again for everything you've been doing on the census.
I ask unanimous consent that all Members and witnesses
written opening statements be included in the record. Without
objection, so ordered. In case there are additional questions
the 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 the witnesses submit written
answers as soon as practicable. Without objection, so ordered.
Thank you very much.
Mr. Mihm. Thank you, Mr. Chairman.
Mr. Miller. Adjourned.
[Whereupon, at 3:29 p.m., the subcommittee was adjourned.]
[The prepared statement of Hon. Danny K. Davis and
additional information submitted for the hearing record
follow:]
[GRAPHIC] [TIFF OMITTED] T6541.021
[GRAPHIC] [TIFF OMITTED] T6541.022
[GRAPHIC] [TIFF OMITTED] T6541.023
[GRAPHIC] [TIFF OMITTED] T6541.024
[GRAPHIC] [TIFF OMITTED] T6541.025
[GRAPHIC] [TIFF OMITTED] T6541.026
[GRAPHIC] [TIFF OMITTED] T6541.027
[GRAPHIC] [TIFF OMITTED] T6541.028
[GRAPHIC] [TIFF OMITTED] T6541.029
[GRAPHIC] [TIFF OMITTED] T6541.030
[GRAPHIC] [TIFF OMITTED] T6541.031
-