2000 Census: Information on the Implications of a Post Census Local
Review Program (Correspondence, 10/13/1999, GAO/GGD-00-9R).

Pursuant to a congressional request, GAO provided information on how
census operations and data accuracy could be affected if the Bureau of
the Census were to include a coverage improvement program known as Post
Census Local Review (PCLR) in the 2000 census.

GAO noted that: (1) the extent to which a PCLR program would affect
other operations already scheduled for the 2000 Census is unclear; (2)
although Bureau officials believe that PCLR could affect the completion
of the Accuracy and Coverage Evaluation (ACE) as well as other
subsequent operations, the results of the 1990 Census suggest that the
effect on the quality of the ACE estimates could be minimal; (3) on the
basis of the 1990 Census, PCLR could add small numbers of housing units
and people to the count if implemented in 2000 and make a small
contribution to increased accuracy; (4) the bureau has not studied the
effect that PCLR could have on the 2000 Census; (5) in terms of
scheduling, there would be no overlaps between PCLR and ACE because
Bureau officials said that if the Bureau is required to do PCLR in 2000,
it would delay the ACE matching and reconciliation process to
accommodate PCLR operations; (6) resolving discrepancies between Bureau
and local governments' housing unit counts could delay ACE matching and
reconciliation activities by 6 weeks, according to Bureau officials; and
(7) although GAO has not comprehensively analyzed the full implications
of implementing a PCLR program in 2000, on the basis of GAO's prior work
on the 1990 PCLR program, discussions with Bureau officials, and review
of the 1990 PCLR program evaluations, PCLR has several potential
benefits and problems.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  GGD-00-9R
     TITLE:  2000 Census: Information on the Implications of a Post
	     Census Local Review Program
      DATE:  10/13/1999
   SUBJECT:  Surveys
	     Data integrity
	     Census
	     Intergovernmental relations
	     Local governments
	     Proposed legislation
IDENTIFIER:  Census Bureau Local Update of Census Addresses Program
	     Census Bureau New Construction Program
	     Census Bureau Postcensus Local Review Program
	     2000 Decennial Census
	     Census Bureau Accuracy and Coverage Evaluation Program

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO report.  Delineations within the text indicating chapter **
** titles, headings, and bullets are preserved.                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
** A printed copy of this report may be obtained from the GAO   **
** Document Distribution Center.  For further details, please   **
** send an e-mail message to:                                   **
**                                                              **
**                                            **
**                                                              **
** with the message 'info' in the body.                         **
******************************************************************

United States General Accounting Office
GAO

GAO/GGD-00-9R

Ordering Information

The first copy of each GAO report and testimony
is free. Additional copies are $2 each. Orders
should be sent to the following address,
accompanied by a check or money order made out
to the Superintendent of Documents, when
necessary. VISA and MasterCard credit cards are
accepted, also. Orders for 100 or more copies to
be mailed to a single address are discounted 25
percent.

Order by mail:

U.S. General Accounting Office
P.O. Box 37050
Washington, DC 20013

or visit:

Room 1100
700 4th St. NW (corner of 4th and G Sts. NW)
U.S. General Accounting Office
Washington, DC

Orders may also be placed by calling (202) 512-
6000 or by using fax number (202) 512-6061, or
TDD (202) 512-2537.

Each day, GAO issues a list of newly available
reports and testimony. To receive facsimile
copies of the daily list or any list from the
past 30 days, please call (202) 512-6000 using a
touch-tone phone. A recorded menu will provide
information on how to obtain these lists.

United States
General Accounting Office
Washington, D.C. 20548-0001
Official Business
Penalty for Private Use $300
Address Correction Requested

                    Bulk Rate
               Postage & Fees Paid
                       GAO
                 Permit No. G100

(410478)

B-283314
Page 2          GAO/GGD-00-9R Post Census Local Review Program

B-283314

October 13, 1999

The Honorable Dan Miller
Chairman, Subcommittee on the Census
Committee on Government Reform
House of Representatives

Subject: 2000 Census: Information on the Implications of a
Post Census Local Review Program

Dear Mr. Chairman:

This letter responds to your request for information on how
census operations and data accuracy could be affected if the
Census Bureau were to include a coverage improvement program
known as Post Census Local Review (PCLR) in the 2000 Census.
Used in the 1990 Census, PCLR gave local governments an
opportunity to review the Bureau's housing unit counts for
their jurisdictions for any discrepancies and suggest
corrections if warranted. Census Bureau representatives and
local officials worked together to resolve the
discrepancies-sometimes by conducting additional field
verification.

PCLR is not part of the Bureau's current plans for the 2000
Census. However, the Local Census Quality Check Act (H.R.
472), currently pending before Congress, would require the
Bureau to use PCLR for each decennial census taken after the
date of enactment. The Bureau already has two programs that
give localities the ability to ensure a more complete count of
their jurisdictions prior to Census Day. The Census Address
List Improvement Act of 19941 gave rise to the Local Update of
Census Addresses (LUCA) program, which gives local and tribal
governments the opportunity to review and suggest changes to
relevant portions of the Bureau's master address list.
Moreover, a new coverage improvement program called the New
Construction Program is to give local and tribal governments
the opportunity to identify newly constructed housing units.

Enclosure I contains our responses to your specific questions
on the implications of implementing a PCLR in the 2000 Census.
To respond to your questions, we interviewed Bureau officials
responsible for the 2000 Census who were aware of the Bureau's
use of PCLR in the 1990 Census. We also reviewed Bureau
program and evaluation documents and drew from our previous
work on the conduct and results of the 1990 PCLR program.

On August 13, 1999, we requested comments on a draft of this
letter from the Secretary of Commerce. The Secretary provided
us with written comments on September 17, 1999 (see encl.
III), which we address in enclosure II.

We did our work in Washington, D.C., and Suitland, MD, between
May and July 1999 in accordance with generally accepted
government auditing standards.

We are sending copies of this letter to the Honorable Carolyn
B. Maloney, Ranking Minority Member of this Subcommittee; the
Honorable William M. Daley, Secretary of Commerce; and the
Honorable Kenneth Prewitt, Director of the Bureau of the
Census. Copies will be made available to others on request.

Please contact me on (202) 512-8676 if you have any questions.
Major contributors to this letter were Robert N. Goldenkoff
and Victoria E. Miller.

Sincerely yours,

J. Christopher Mihm
Associate Director, Federal Management
 and Workforce Issues

_______________________________
1 P.L. 103-430, Oct. 31, 1994.

                                                 Enclosure I
GAO Responses To Questions From Chairman Miller Concerning The
Use of a Post Census Local Review in the 2000 Census
Page 6          GAO/GGD-00-9R Post Census Local Review Program
The following are responses to the questions received from the
Honorable Dan Miller, Chairman, Subcommittee on the Census,
House Committee on Government Reform, dated April 16, 1999.

Question #1: What effect will the implementation of a PCLR
program have on the implementation of other programs already
scheduled for the 2000 Census?

The extent to which a PCLR program would affect other
operations already scheduled for the 2000 Census is unclear.
Although Bureau officials believe that PCLR could affect the
completion of the Accuracy and Coverage Evaluation (ACE), as
well as other subsequent operations, the results of the 1990
Census suggest that the effect on the quality of ACE estimates
could be minimal. 1

Under the proposed PCLR legislation, H.R. 472, the deadline
for resolving local governments' challenges to census housing
unit counts would be November 1, 2000. However, in order to
meet this deadline, the Bureau believes that PCLR would need
to be completed by October 1, 2000. Otherwise, the Bureau says
there would be insufficient time to accurately review the
incorporation of the PCLR results into the census totals while
adhering to congressionally mandated deadlines. In order to
complete PCLR by October 1, 2000, the Bureau said it would
need to adjust certain census operations, including suspending
coverage improvement operations already planned for the 2000
Census earlier than originally scheduled.

A key component of ACE involves matching data from the ACE
sample to the data obtained during the initial census
enumeration (known as the E-sample). Differences are generally
to be reconciled by an additional interview with a household
member. The ACE matching and reconciliation process is
scheduled to begin October 2, 2000. However, Bureau officials
told us that the Bureau would need to incorporate all PCLR
changes into the E-sample prior to starting the ACE matching
and reconciliation operation. Thus, the Bureau would need to
delay the start of the ACE matching and reconciliation process
until all local governments' challenges were resolved in
November 2000.

On the basis of the Bureau's experience in 1990, however, it
appears that the Bureau's contention that PCLR should be
completed prior to starting the ACE matching and
reconciliation process might be too rigid. In 1990, during a
coverage evaluation operation called the Post Enumeration
Survey (PES), the Bureau collected PCLR and other coverage
improvement data past the start of the matching and
reconciliation process and was able to incorporate portions of
it to improve the quality of PES estimates.
Although the Bureau cut short plans to incorporate all "late"
coverage improvement data into the original count in order to
complete PES on schedule, a Bureau evaluation of the 1990 PES
showed that excluding the remaining data had a small effect on
the PES estimate of accuracy and would have decreased the
undercount only slightly if it had been included. In 1990, the
overall effect of including the late data was to reduce the
net estimated undercount from 2.08 percent to 2.01 percent, or
0.07 percentage points.

Although ACE, like PES, is to measure census coverage,
according to the Bureau, ACE results are to be an integral
part of the census results, whereas PES was used for
evaluative purposes. As a result, the Bureau believes that
adding any operations late in the census cycle could
jeopardize the quality of census data.

Question #2: Will the implementation of a PCLR program
increase or decrease the accuracy of the census counts before
any adjustment is applied due to a coverage and evaluation
survey?

On the basis of the 1990 Census, PCLR could add small numbers
of housing units and people to the count if implemented in
2000 and make a small contribution to increased accuracy.
According to Bureau evaluations, in 1990, 80,929 housing units
were added as a result of PCLR. This represented a 0.08
percent addition to the housing unit count. Of those
additional housing units, 58.7 percent were occupied, 29.6
percent were vacant, and 11.7 percent were deleted by later
census operations because they were determined not to be
housing units. In terms of people, PCLR added about 124,900
individuals (0.1 percent) to the population count. However,
according to Bureau officials, evaluations of 1990 late census
data collection activities (which included PCLR) indicated
that in addition to adding correct enumerations, these
activities generated a high rate of duplicate or erroneous
enumerations. Before adjustment, these duplicate or erroneous
enumerations have the effect of reducing the net undercount,
but not of improving the overall accuracy of the census.

Question #3: If the Bureau projects any such decrease in
accuracy, are there any specific scientific studies that back
this assertion?

The Bureau has not studied the effect that PCLR could have on
the 2000 Census.

Question #4: Given the fact that the ACE will only be
conducted in 10,000 out of 5,000,000 populated census blocks
and significant PCLR corrections for missed housing units
should only be reported for a fairly small number of blocks
(somewhere around 50,000 to 100,000 out of 5,000,000 blocks),
what effect will a PCLR have on the timely field
implementation and analysis of the ACE?

The effect that PCLR would have on the timely implementation
and analysis of ACE could not be determined because, according
to Bureau officials, information on the extent of overlap
between 1990 PES and PCLR blocks is not readily available for
comparison purposes.

PCLR could delay the ACE matching and reconciliation process
if the Bureau requires that all PCLR data be included in the
ACE E-sample. However, as we note in our response to Question
#1, this requirement might be too stringent because, on the
basis of 1990 evaluations, there would not be a significant
reduction in the PES estimate of accuracy if the PCLR data
were not used in ACE.

Question #5: What are the actual overlaps between the two
processes?

In terms of scheduling, there would be no overlaps between
PCLR and ACE because Bureau officials said that if the Bureau
is required to do PCLR in 2000, it would delay the ACE
matching and reconciliation process to accommodate PCLR
operations. Resolving discrepancies between Bureau and local
governments' housing unit counts could delay ACE matching and
reconciliation activities by about 6 weeks, according to
Bureau officials. As noted in our response to Question #1, the
PCLR operation would not be scheduled to end until November 1,
2000-4 weeks after the ACE matching and reconciliation
operation is scheduled to begin October 2, 2000. The Bureau
estimates that ACE matching and reconciliation operations
would be delayed by this 4-week period plus at least an
additional 2 weeks for incorporating the PCLR data into the
census count-thus causing a delay of at least 6 weeks.

Question #6: Other than the cost factors that have already
been projected by CBO, what other benefits or problems do you
envision with the incorporation of a PCLR in the 2000
Decennial Census?

Although we have not comprehensively analyzed the full
implications of implementing a PCLR program in 2000, on the
basis of our prior work on the 1990 PCLR program,2 discussions
with Bureau officials, and review of 1990 PCLR program
evaluations, PCLR has several potential benefits and problems.
Among its potential benefits, PCLR could do the following:

ï¿½It could allow local government officials a last chance to
review and correct housing unit counts within their
jurisdictions before the census is over. In 1990, 9,847 out of
39,198 eligible governments (or 25.1 percent) participated.

ï¿½It could correct some errors in the Bureau's files showing
the exact geographic location of an address (known as
geocodes) and delete nonexistent housing units from the
Bureau's master address list. In 1990, 198,347 housing units
were geographically transferred, and 101,887 housing units
were deleted from the Bureau's records.

ï¿½It could add people and housing units to the overall census
count.

ï¿½It could identify pockets of missing addresses.

Among PCLR's potential problems:

ï¿½It has unknown implications for timely and accurate
completion of other census operations because of an uncertain
volume of challenges from local governments. The 1990 volume
of challenges was relatively low-local governments formally
challenged the housing unit counts of 270,650 out of 6.5
million blocks (4.2 percent). Nevertheless, according to
Bureau officials, adding PCLR, or any other unplanned
operation, to the operational requirements of the 2000 Census
at this time could cause serious delays and undermine the
ability of the Census Bureau to complete an accurate and
timely census. The Bureau believes that the risks of adding a
new operation, when weighed against the relatively few
additions to the census count that likely would come from
PCLR, argues against adding PCLR to the 2000 Census.

ï¿½It could create logistical problems. For example, printing
maps for local governments asking to participate in the
review, on an ad hoc basis, would be an added operation and
expense.

ï¿½If it follows the 1990 pattern, there will be lower
participation among middle and smaller-sized governments
because they lack the available resources of larger cities.

ï¿½It is more expensive than some other Post-Census Day coverage
improvement programs in terms of housing units and people
added to the final census count. In 1990, the $9.6 million the
Bureau spent on PCLR translated into $118.67 per housing unit
added and $76.89 per person added. By comparison, the costs of
some other Post-Census Day coverage improvement programs are
shown in table I.1. However, when measured by the total number
of corrections made, PCLR costs drop to $25.19 per housing
unit corrected. In 1990, PCLR added 80,929 housing units,
deleted 101,887 housing units, and corrected 198,347 geocoding
errors for total housing unit corrections of 381,163.

Table I.1: Cost of 1990 Post-Census Day Coverage Improvements
Coverage             Total cost    Per housing      Per person
improvement                         unit added           added
Vacant/delete/m     $67,589,000         $69.05          $44.90
overs check
Recanvass           $14,684,000        $105.97          $82.40
Post census          $9,604,000        $118.67          $76.89
local review
Primary              $2,880,000 Not Applicable           $8.22
selection
algorithm
review
Parolee/probati      $5,800,000        $232.00          $12.95
oner coverage
improvement
Other                $6,319,000        $253.76          $10.38
search/match
forms
Source: Bureau of the Census.
_______________________________
1 ACE is a program that measures coverage error in the census
through an independent survey of a sample of households.
2 Decennial Census: Status of Housing Coverage Check and
Postcensus Local Review Programs (GAO/T-GGD-90-63, Sept. 25,
1990).

                                                Enclosure II
Agency Comments and Our Evaluation
Page 9          GAO/GGD-00-9R Post Census Local Review Program
The Secretary of Commerce provided written comments on a draft
of this letter. He stated that Commerce had fundamental
concerns about the position we are taking in response to
Chairman Miller's questions.

Specifically, the Secretary commented that our response
appears to indicate that PCLR would affect only the ACE
survey. However, he stressed that PCLR, or any other unplanned
operation, could corrupt many other 2000 Census operations in
addition to ACE because all of the systems and procedures in
place for the census are highly integrated.

Our draft pointed out the possibility that PCLR could affect
other census operations in addition to ACE. In response to
Question #1, for example, we noted that PCLR could affect the
completion of both ACE and other subsequent census operations.
Moreover, in our response to Question #6, we noted that PCLR
has unknown implications for timely and accurate completion of
other census operations because of an uncertain volume of
challenges from local governments. However, because the
Secretary emphasized this point in his comments-amplifying on
remarks made to us by Census Bureau officials during our
initial audit work-we added language to our response to
Question #6 to more fully reflect Commerce's concern that
adding unplanned operations at this point in the census cycle
could cause serious disruptions.

The Secretary also noted that although the proposed PCLR
legislation would require the Bureau to complete PCLR by
November 1, 2000, Commerce believes that PCLR would need to be
completed by October 1, 2000. Otherwise, there would not be
sufficient time to accurately review the incorporation of the
PCLR results into the 2000 Census tabulations while adhering
to congressionally mandated deadlines. The Secretary noted
that to complete PCLR by October 1, 2000, the Bureau would
need to consider various adjustments to 2000 Census
operations, including suspending other coverage improvement
operations earlier than planned. Commerce believes that this
could put the quality and integrity of census data at risk.
Moreover, Commerce stated that the coverage improvement
operations already planned for the 2000 Census have been shown
to be more effective in improving the accuracy of census data
than PCLR.

As noted in our draft, during the 1990 Census, the Bureau
collected PCLR and other coverage improvement data past the
start of the matching and reconciliation process and was able
to incorporate portions of it to improve the quality of PES
estimates. Also, as noted in our response to Question #6, one
factor that would affect timely completion of already
scheduled operations is the number of challenges from local
governments in the PCLR program. In 1990, that volume was
relatively low. However, we added language in our response to
Question #1 to reflect Commerce's position that PCLR may have
to be completed earlier than the proposed legislation's
deadline in order to incorporate PCLR results into the census
count.

Also, we recognize that the Census Bureau has included several
coverage improvement operations in the 2000 Census that,
according to 1990 data, appear to have added to the housing
unit and population counts. In this regard, table I.1 compares
the cost of several 1990 Post-Census Day coverage improvement
programs.

The Secretary further commented that we did not acknowledge
that two operations already planned for the 2000 Census-LUCA
and the New Construction program--will provide local
governments with opportunities to review and suggest
corrections to the Bureau's address list. These programs did
not exist during the 1990 Census, and, thus, a PCLR program
has less potential to make a positive contribution to the
quality of the census than was the case in 1990.

Our draft did acknowledge that the Bureau already has two
programs that give local governments the ability to ensure a
more complete count of their jurisdictions. The second
paragraph of our letter to Chairman Miller discusses the
Census Address List Improvement Act of 1994, which required
the Bureau to give local governments the opportunity to review
relevant portions of the Bureau's master address list. The
Bureau was responding to this mandate when it created LUCA. We
added language to our letter to better reflect this. Our
letter also notes how the Bureau's New Construction program is
to give local and tribal governments the opportunity to
identify newly constructed housing units.

The Secretary also commented that the environment in which the
Bureau is conducting the 2000 Census is very different from
the 1990 Census, noting, for example, that the 1990 PES was
initially designed for evaluative purposes, while the ACE is
to be integral to the census results. The Secretary also
stated that during the 1990 Census, several operations were
added late in the decennial cycle, including the method of
handling late additions we referred to in our response to
Question #1. The Secretary commented that the 2000 Census
design will not include such operations, because the risk of
failure is too high.

We added language to our response to Question #1 that
clarifies Commerce's position on these points. However, we
note that although the 1990 PES may initially have been
designed for evaluative purposes, the PES was actually done
with the intent of providing adjusted census figures for
possible use. In this sense, the PES, like ACE, was seen as an
integral part of the 1990 Census.

Enclosure III
Comments From the Secretary of Commerce
Page 12GAO/GGD-00-9R Post Census Local Review Prog
ram

*** End of Document ***