Data Quality: Expanded Use of Key Dissemination Practices Would  
Further Safeguard the Integrity of Federal Statistical Data	 
(31-MAY-06, GAO-06-607).					 
                                                                 
In 2003, the Bureau of the Census (Bureau) changed the day and	 
location of the release of its Income and Poverty Estimates. Some
data users believed the change was an effort to suppress	 
unfavorable news and questioned the Bureau's data dissemination  
practices. GAO was asked to assess whether (1) the Bureau adhered
to its dissemination practices for the 2003 and later releases,  
and (2) the Bureau and 13 other federal statistical agencies	 
follow data release practices recommended by the National	 
Research Council (NRC). GAO reviewed the Bureau's dissemination  
process for the 2003 thru 2005 Income and Poverty Estimates.	 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-06-607 					        
    ACCNO:   A54908						        
  TITLE:     Data Quality: Expanded Use of Key Dissemination Practices
Would Further Safeguard the Integrity of Federal Statistical Data
     DATE:   05/31/2006 
  SUBJECT:   Census						 
	     Data integrity					 
	     Government information dissemination		 
	     Income statistics					 
	     Internal controls					 
	     Policy evaluation					 
	     Statistical data					 
	     Policies and procedures				 
	     Transparency					 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Product.                                                 **
**                                                              **
** 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.                                         **
**                                                              **
******************************************************************
GAO-06-607

     

     * Results in Brief
     * Background
     * Release of Income and Poverty Estimates Adhered to Most of t
          * The Bureau's Income and Poverty Estimate Dissemination Pract
          * Release of 2003 and Subsequent Income and Poverty Estimates
          * According to the Bureau, Delays in the Companion Report Caus
          * According to the Bureau, Several Factors Affected the Change
          * Bureau Officials Made Key Decisions on Releasing Income and
          * The Bureau Is Updating Its Practices for Releasing the Incom
     * Governmentwide Guidance Is Being Developed That May Improve
          * Most Agencies' Data Dissemination Practices Generally Adhere
          * OMB Is Developing Governmentwide Data Dissemination Guidance
     * Conclusions
     * Recommendations for Executive Action
     * Agency Comments and Our Evaluation
     * Appendix I: Scope and Methodology
     * Appendix II: Comments from the Department of Commerce
     * Appendix III: GAO Contact and Acknowledgments
          * GAO Contact
          * Acknowledgments
     * Related GAO Reports
          * Quality of Federal Data
               * Order by Mail or Phone

Report to Congressional Requesters

United States Government Accountability Office

GAO

May 2006

DATA QUALITY

Expanded Use of Key Dissemination Practices Would Further Safeguard the
Integrity of Federal Statistical Data

GAO-06-607

Contents

Letter 1

Results in Brief 2
Background 5
Release of Income and Poverty Estimates Adhered to Most of the Bureau's
Data Dissemination Practices 7
Governmentwide Guidance Is Being Developed That May Improve Statistical
Agencies' Data Dissemination Practices 15
Conclusions 22
Recommendations for Executive Action 22
Agency Comments and Our Evaluation 23
Appendix I Scope and Methodology 25
Appendix II Comments from the Department of Commerce 28
Appendix III GAO Contact and Acknowledgments 32
Related GAO Reports 33

Tables

Table 1: Income and Poverty Estimates Have Traditionally Been Released on
a Tuesday or a Thursday at the National Press Club 5
Table 2: Most Statistical Agencies Reported That They Generally Adhered to
the National Research Council's Guidance for Releasing Information 16

This is a work of the U.S. government and is not subject to copyright
protection in the United States. It may be reproduced and distributed in
its entirety without further permission from GAO. However, because this
work may contain copyrighted images or other material, permission from the
copyright holder may be necessary if you wish to reproduce this material
separately.

United States Government Accountability Office

Washington, DC 20548

May 31, 2006 May 31, 2006

The Honorable Wm. Lacy Clay Ranking Minority Member Subcommittee on
Federalism and the Census Committee on Government Reform House of
Representatives The Honorable Wm. Lacy Clay Ranking Minority Member
Subcommittee on Federalism and the Census Committee on Government Reform
House of Representatives

The Honorable Carolyn B. Maloney House of Representatives The Honorable
Carolyn B. Maloney House of Representatives

Data collected and disseminated by federal statistical agencies are a
critical strategic asset because they provide indicators of the economic
and social well-being of the nation and help inform agencies' rule-making
activities. Given the widespread use and the impact of federal
information, the data itself, the timing of reports, and any news releases
that accompany them must be accurate and objective. Moreover, statistical
agencies must be viewed as credible to secure the public's trust. One way
statistical agencies foster credibility is by maintaining a position of
independence; that is, taking appropriate steps that help ensure their
data products and the timing of their release are free from even an
appearance of political influence. Data collected and disseminated by
federal statistical agencies are a critical strategic asset because they
provide indicators of the economic and social well-being of the nation and
help inform agencies' rule-making activities. Given the widespread use and
the impact of federal information, the data itself, the timing of reports,
and any news releases that accompany them must be accurate and objective.
Moreover, statistical agencies must be viewed as credible to secure the
public's trust. One way statistical agencies foster credibility is by
maintaining a position of independence; that is, taking appropriate steps
that help ensure their data products and the timing of their release are
free from even an appearance of political influence.

Because of the sensitivity of certain statistical information, seemingly
mundane actions, such as the timing of a data release, can be called into
question, as interested parties may perceive the change as a maneuver to
obscure or manipulate the information for partisan purposes. The
Department of Commerce, which has a number of data gathering programs
under its purview, oversees the U.S. Bureau of the Census (Bureau). Among
its many surveys, the Bureau has responsibility to report estimates of the
nation's income and poverty. In 2003, the Bureau changed the day and
location for releasing this information. Because the estimates showed that
poverty rates had risen, some data users believed that the change was an
effort to suppress unfavorable news, and called into question the Bureau's
practices for disseminating such data, and the Department of Commerce's
role in the 2003 release. Because of the sensitivity of certain
statistical information, seemingly mundane actions, such as the timing of
a data release, can be called into question, as interested parties may
perceive the change as a maneuver to obscure or manipulate the information
for partisan purposes. The Department of Commerce, which has a number of
data gathering programs under its purview, oversees the U.S. Bureau of the
Census (Bureau). Among its many surveys, the Bureau has responsibility to
report estimates of the nation's income and poverty. In 2003, the Bureau
changed the day and location for releasing this information. Because the
estimates showed that poverty rates had risen, some data users believed
that the change was an effort to suppress unfavorable news, and called
into question the Bureau's practices for disseminating such data, and the
Department of Commerce's role in the 2003 release.

This letter responds to your request that we review the Bureau's level of
independence in releasing its annual Income and Poverty Estimates over the
last several years. As agreed with your offices, we determined the extent
to which (1) the Bureau adhered to its dissemination practices for This
letter responds to your request that we review the Bureau's level of
independence in releasing its annual Income and Poverty Estimates over the
last several years. As agreed with your offices, we determined the extent
to which (1) the Bureau adhered to its dissemination practices for the
issuance of the 2003 annual Income and Poverty Estimates and subsequent
releases in 2004 and 2005, and (2) the Bureau and 13 other federal
statistical agencies follow data dissemination practices that the National
Academy of Sciences' National Research Council (NRC) recommend in a 2005
report.1

To address the first objective, we reviewed relevant program documents and
interviewed Bureau officials responsible for disseminating the Income and
Poverty Estimates and other data releases. The dissemination process
includes the steps from approval of the report content up to and including
public distribution of the report. For the second objective, we surveyed
key officials at the Bureau and 13 other federal statistical agencies
collecting information on their practices for releasing data. We compared
the 14 agencies' practices to those developed by NRC that are important
for (1) the wide dissemination of data, and (2) maintaining a strong
position of independence. According to NRC, elements within these
practices facilitate the provision of timely, credible, and politically
neutral information into the hands of data users. Appendix I provides
additional details on our scope and methodology. We conducted our work
between March 2005 and April 2006 in accordance with generally accepted
government auditing standards.

                                Results in Brief

While not all of the Bureau's data dissemination practices are documented,
we were able to determine through discussions with Bureau officials and
review of available documentation, that the Bureau adhered to most of its
long-standing data release practices. In changing the date and location of
the 2003 and subsequent releases of the Income and Poverty Estimates, the
Bureau did depart from its tradition of releasing this information on a
Tuesday or Thursday at a news conference at the National Press Club. That
said, under the Bureau's documented data dissemination practices (1) there
is no requirement for the Bureau to release this information at a
particular location on a given day, and (2) no particular official is
designated authority to choose the release date and location. Also, the
Bureau's documented data dissemination practices for the Income and
Poverty Estimates are contained in a memo from 1985. The Bureau is
updating them, to among other things, reflect current technology.

1Margaret E. Martin, Miron L. Straf, and Constance F. Citro eds.,
Principles and Practices for a Federal Statistical Agency, (3rd ed.)
(Washington, D.C.: The National Academies Press, 2005). The National
Academy of Sciences organized the National Research Council in 1916 "to
associate the broad community of science and technology with the Academy's
purposes of furthering knowledge and advising the federal government."
According to Ms. Citro, the Principles and Practices for a Federal
Statistical Agency report is the signature product of the Committee on
National Statistics (which is a standing committee of the National
Research Council).

Bureau officials stated that the date of the 2003 release was changed from
September 23rd to September 26th for several reasons, including delays in
producing a companion report on supplemental measures of expenditures,
consumption, and poverty. According to Bureau officials we interviewed and
available documents, the Commerce Department's Under Secretary for
Economic Affairs wanted to release the two reports simultaneously to
broaden the public's understanding of social well-being, a decision that
was consistent with the Bureau's ongoing effort to provide alternative
measures of poverty.

In 2004, the Bureau moved up the release of the Income and Poverty
Estimates to August of that year so it would coincide with the release of
data from the American Community Survey.2 According to Bureau officials,
this was done in an attempt to head off the confusing press coverage that
occurred the previous year when estimates from Income and Poverty
Estimates did not always match estimates from the American Community
Survey. In 2004, when the Bureau issued the Income and Poverty Estimates
and American Community Survey numbers at the same time, the Bureau's news
release explained the methodological and other factors that could account
for any differences.

Bureau officials stated that in 2003, the location of the press conference
for the Income and Poverty Estimates was changed from Washington, D.C., to
the Bureau's Suitland, Maryland campus, in part, because the Director of
the Census Bureau wanted to raise awareness that the construction of its
new headquarters building had just started. According to Bureau officials,
future releases of the annual Income and Poverty Estimates are to occur at
the Suitland, Maryland campus in late August.

Most of the 14 statistical agencies we included in our review reported
general adherence to NRC's guidance, important for (1) the wide
dissemination of data, and (2) maintaining a strong position of
independence. Still, there were some noteworthy gaps. On the one hand, for
example, all 14 agencies (in accordance with NRC guidance) had multiple
avenues for disseminating data, released data in a variety of formats, and
had policies to guide what data should be preserved and how they should be
archived. On the other hand, adherence to NRC's practices was not
universal. For example, 6 of the 14 agencies lacked dissemination
practices that promote the regular and frequent release of major findings
from an agency's statistical programs to the public via the media, the
Internet, and other means.

2The Census Bureau's American Community Survey is a monthly sample survey
of 250,000 households that is to replace the long-form questionnaire used
in prior decennial censuses.

The Office of Management and Budget (OMB), in concert with the federal
statistical agencies, is developing a governmentwide directive on the
release and dissemination of statistical products that, according to OMB
officials, parallels NRC's and other generally accepted dissemination
practices. OMB officials indicated the guidance is intended to help ensure
statistical products are policy-neutral, timely, and accurate. To the
extent that statistical agencies appropriately follow these practices, the
directive could promote more consistent adherence to practices that
facilitate broader dissemination of statistical data and enhance its
credibility. For example, OMB's directive could help replace the patchwork
of agency-specific guidance with a more transparent, commonly accepted,
and consistently applied framework for disseminating data. OMB plans to
release the directive for public comment in the spring of 2006.

We are recommending that the Secretary of Commerce direct the Bureau to,
as part of its efforts to update its practices for releasing the Income
and Poverty Estimates, fully document those practices. We are also making
recommendations to the Director of OMB, as OMB completes work on its draft
dissemination directive, to consider how best to address the gaps we
identified between agencies' data dissemination practices and NRC's
guidance, as well as certain questions concerning coverage, documentation,
flexibility, monitoring, and the posting of data.

The Secretary of Commerce provided written comments on a draft of this
report (see app. II). While Commerce neither agreed nor disagreed with our
recommendation for the Bureau to fully document its key data dissemination
practices for releasing the Income and Poverty Estimates, Commerce
reiterated the point we made in our report that the Bureau is updating its
practices for releasing the Income and Poverty Estimates. Commerce noted
that the updated document-which details the dissemination practices for
the Income and Poverty Estimates-is under review. The Bureau plans to
issue it prior to the next release of the Income and Poverty Estimates
expected in August 2006.

The Director of OMB did not have comments on the recommendations to them.
However, OMB officials provided suggestions for technical corrections and
we revised the report to reflect these suggestions as appropriate.

                                   Background

In September 2003, the Bureau broke from its tradition of releasing its
Income and Poverty Estimates on a Tuesday or Thursday at a news conference
at the National Press Club (see table 1). The data were instead released
at a news conference on a Friday at Bureau Headquarters in Suitland,
Maryland. Although the Bureau provided the media and other attendees with
bus service from the National Press Club to Suitland, because the data
showed that poverty levels had risen, some data users expressed concern
that the change in day and location was an attempt to suppress unfavorable
information by releasing it at a more remote location and before a
weekend, when the public tends to pay less attention to the news.

Table 1: Income and Poverty Estimates Have Traditionally Been Released on
a Tuesday or a Thursday at the National Press Club

Year Date/Day                Location            
1996 September 26 - Thursday National Press Club 
1997 September 29 - Monday   National Press Club 
1998 September 24 - Thursday National Press Club 
1999 September 30 - Thursday National Press Club 
2000 September 26 - Tuesday  National Press Club 
2001 September 25 - Tuesday  National Press Club 
2002 September 24 - Tuesday  National Press Club 
2003 September 26 - Friday   Census Bureau       
2004 August 26 - Thursday    Census Bureau       
2005 August 30 - Tuesday     Census Bureau       

Source: U.S. Census Bureau.

The Income and Poverty Estimates, like other kinds of federal statistical
information, provide key measures of the health and well-being of our
society. As a result, the data need to be accurate, timely, accessible,
relevant, and objective. At the same time, according to NRC, the manner in
which agencies release the data is also important, and needs to be free
from even an appearance of bias and political manipulation. Failure to
meet this goal could undermine public confidence in the information and
erode an agency's credibility.

That said, although various guidance and laws have been developed to
safeguard the overall quality of federal data, few governmentwide
provisions directly address the data dissemination process itself, and
agencies have largely been left to develop their own practices. For
example, while OMB's Statistical Policy Directive Number 3, "Compilation,
Release, and Evaluation of Principal Federal Economic Indicators" provides
detailed guidance on the dissemination of data, it only applies to 38
market sensitive principal economic indicators. Statistical Policy
Directive Number 3 is highly regarded in the statistical community because
it provides statistical agencies with comprehensive data dissemination
guidance, requiring agencies to, among other actions, promptly release
data according to an established schedule, and announce and fully explain
any schedule changes in advance.

Under the Information Quality Act,3 OMB was required to issue
governmentwide guidelines that provide policy and procedural guidance to
federal agencies for ensuring and maximizing the quality, objectivity,
utility, and integrity of information disseminated by federal agencies.
OMB's guidelines, issued in final form in February 2002, directed agencies
covered by the act (statistical agencies and most others) to issue their
own quality guidelines. OMB's guidelines imposed certain core
responsibilities on agencies, including incorporating quality into their
information dissemination practices. OMB noted that quality consists of
several dimensions, including objectivity (which focuses on whether the
disseminated information is accurate, reliable, and unbiased in
presentation and substance).

More generally, OMB helps ensure that the activities of the statistical
agencies are in line with federal statistical policy by coordinating
agency budget requests and interagency groups working on statistical
issues, issuing statistical standards, and reviewing agency requests to
collect information.

3Consolidated Appropriations - Fiscal Year 2001, Pub. L. No. 106-554, S:
515, 114 Stat. 2763A-153 to 2763A-154 (2000) (44 U.S.C. S: 3516 note).

This report is the latest of several studies we have issued on the quality
of federal data. See Related GAO Reports at the end of this report for a
list of selected products we have issued to date.

  Release of Income and Poverty Estimates Adhered to Most of the Bureau's Data
                            Dissemination Practices

While not all of the Bureau's data dissemination practices are documented,
we were able to determine through discussions with Bureau officials and
review of available documentation, that the Bureau adhered to most of its
long-standing data release practices. In changing the date and location of
the 2003 and subsequent releases of the Income and Poverty Estimates, the
Bureau did depart from its tradition of releasing this information on a
Tuesday or Thursday at a news conference at the National Press Club. That
said, under the Bureau's documented data dissemination practices (1) there
is no requirement for the Bureau to release this information at a
particular location on a given day and, (2) no particular official is
designated authority to choose the release date and location. Bureau
officials stated that the date of the 2003 release was changed from
September 23rd to September 26th for several reasons, including delays in
producing a companion report on supplemental measures of expenditures,
consumption, and poverty that was to be released at the same time. Also,
the 2004 and 2005 estimates were released a month earlier than in prior
years to coincide with the release of data from the American Community
Survey. The documented practices for disseminating the Income and Poverty
Estimates are contained in a memo that is 21 years old so the Bureau is
updating them, to among other things, reflect current technology.

The Bureau's Income and Poverty Estimate Dissemination Practices Are Derived
from Several Sources

The Bureau has several sources of documented, agencywide practices for
disseminating data to the public. For example, in accordance with OMB's
guidelines for implementing the Information Quality Act, the Bureau
developed its own set of quality guidelines that include provisions aimed
at ensuring the objectivity and integrity of its data.4 The Bureau also
has a series of data dissemination practices available on its Intranet
site and it has issued four standards governing the dissemination of data
products,5 including Dissemination of Census and Survey Data Products.6 We
found that the only documented practices specific to the release of the
Income and Poverty Estimates are contained in a 1985 memorandum that was
included as one of several appendixes to the Bureau's Administrative
Manual. The manual provides Bureau policy on the release of data and
guidance for divisions to follow in responding to requests for such
information.

4 http://www.census.gov/quality/quality_guidelines.htm (downloaded Oct. 3,
2003).

5The standard, Dissemination of Census and Survey Data Products, became
effective on May 19, 2005, so it did not apply to the 2003 or 2004 income
and poverty releases. However, it covered the 2005 release and will cover
all future Income and Poverty Estimates releases.

The 1985 memorandum, which was signed by the Director of the Census Bureau
at the time, includes eight broad steps, covering the process for
disseminating the Income and Poverty Estimates. The eight steps include
the time period from approval of the report content up to and including
distributing the report at the press conference.

           1. The Associate Director for Demographic Fields approves the
           final content of the report prepared by the Population Division
           after review by the Statistical Methods Division.
           2. The Public Information Office receives a copy of the final
           content to draft a press release. This draft release is approved
           within the Census Bureau, by the Public Affairs Specialist in the
           Under Secretary for Economic Affairs' office, and by the Commerce
           Department's newsroom.
           3. The report is prepared for camera-ready form.
           4. Camera-ready copy is sent to the printer.
           5. When the completion time for this report is known, the Census
           Bureau establishes the release date and time with Commerce
           Department concurrence.
           6. Approximately 48 hours before report release date and time, the
           Census Bureau briefs the Deputy Secretary for Economic Affairs on
           the principal findings.
           7. The Census Bureau makes the report and accompanying press
           release available to the media on the established date at 9 a.m.
           for 10 a.m. release.7 
           8. The Census Bureau distributes the report and press release to
           the Congress and the OMB at the same time as the media.

6The Bureau created this document in response to a recommendation
contained in our report, GAO, Data Quality: Census Bureau Needs to
Accelerate Efforts to Develop and Implement Data Quality Review Standards,
GAO-05-86 (Washington, D.C.: Nov. 17, 2004).

Release of 2003 and Subsequent Income and Poverty Estimates

In releasing the 2003 Income and Poverty Estimates, the Bureau adhered to
most of its data dissemination practices. The change in the date and
location of the 2003 and subsequent releases of the Income and Poverty
Estimates was a departure from the Bureau's tradition of releasing this
information on a Tuesday or Thursday at a news conference at the National
Press Club. That said, under the Bureau's documented data dissemination
practices there is no requirement for the Bureau to release this
information at a particular location on a given day. Based on our review
of available documentation and our interviews with officials involved with
the Income and Poverty Estimates, the Bureau followed the steps in the
1985 memo in the 2003, 2004 and 2005 releases, with the exception of the
release time as previously described.

While the Bureau complied with its documented practices for the
dissemination of the Income and Poverty Estimates, they lacked
specificity. For example, clear and specific documentation does not exist
for how and when the release date and location are to be determined for
the Income and Poverty Estimates and who should make those decisions. In
actuality, as discussed in greater detail subsequently, in 2003, the
Director of the Census Bureau chose the location and the Associate
Director for Communications chose the date. However, because this was not
thoroughly documented (the 1985 memo only provides general guidance), it
is unclear to the public who made these decisions and how they were made.

Furthermore, Bureau officials told us that they did not retain any
internal memos or e-mails that documented the decision to change the 2003
Income and Poverty Estimates release, which would have provided evidence
to support the Bureau's narrative of the events leading up to the release.

7According to Bureau officials, in 2003, 2004 and 2005 the media received
the Income and Poverty report as if it were an economic indicator. After
media were seated in the auditorium and the procedures were announced, the
report was then distributed at approximately 10:10 AM. The Internet link
was also opened at 10:10 AM.

According to the Bureau, Delays in the Companion Report Caused Change in Timing
of the 2003 Release

Based on our review of available Bureau documents and interviews with key
Bureau officials, several factors led to the change in the timing of the
release of the 2003 and subsequent Income and Poverty Estimates.

The Chief of the Census Bureau's Housing and Household Economic Statistics
Division8 at the time of the 2003 release of Income and Poverty Estimates,
and other senior officials we spoke to, stated that the 2003 Income and
Poverty Estimates release was different from years past because the Bureau
decided earlier that year to issue the report at the same time as a
multi-agency report on supplemental measures of expenditures, consumption,
and poverty.9 This decision was made before the findings of the Income and
Poverty Estimates report were known. Bureau officials stated that although
the original target date for releasing both reports was September 23,
2003, complications with finalizing the supplemental measures report kept
it from being ready for release on that day.

According to Bureau officials and documents we reviewed, because the
supplemental measures of expenditures, consumption, and poverty report
involved several statistical agencies, there was a different clearance
process than that used for the Income and Poverty Estimates report. As a
result, while the Bureau had completed its review of the latter report,
all the members of a steering committee still needed to review the report
on supplemental measures.

At the same time, based on our discussions with Bureau officials involved
with the Income and Poverty Estimates report, as well as available
documents, the Commerce Department's Under Secretary for Economic Affairs
wanted to release both reports simultaneously in an effort to broaden the
public's understanding of social well-being. The Under Secretary's
decision was consistent with the Bureau's ongoing effort to provide
alternative estimates of poverty, which itself stemmed from a 1995 report
by the National Academy of Sciences that recommended revising how poverty
is measured.

8The Housing and Household Economic Statistics Division is the unit
responsible for the Income and Poverty Estimates.

9Supplemental Measures of Material Well-Being: Expenditures, Consumption,
and Poverty 1998 and 2001 (Washington, D.C.: U.S. Department of Commerce.
Economics and Statistics Administration, U.S. Census Bureau. September
2003). The report was a collaborative effort by members of the Interagency
Working Group on Alternative Measures of Material Well-Being. The group
was composed of members from the following agencies: the Office of
Management and Budget; the Bureau of Labor Statistics in the U.S.
Department of Labor; the Energy Information Administration in the U.S.
Department of Energy; and the Census Bureau, Office of Policy Analysis and
Bureau of Economic Analysis in the U.S. Department of Commerce.

Because of the additional time required to clear the supplemental measures
report, Bureau officials responsible for the Income and Poverty Estimates
asked for a later date to issue their report. Consequently, the Bureau's
Associate Director for Communications, with the Director's consent,
scheduled Friday, September 26, 2003, as the release date for the Income
and Poverty Estimates, and both reports were issued on that date.10

Under the Bureau's guidance for dealing with the media, Census Bureau
analysts are to arrange their work schedules to be available for inquiries
for 2 to 3 days after a data release. This is why, prior to 2003, the
Bureau tended to release the Income and Poverty Estimates earlier in the
week: it obviated the need for analysts to work on the weekend.
Additionally, Bureau officials said that because of the Internet and cable
television, the news cycle is no longer viewed as a cycle and has instead
become a 24 hours a day, 7 days a week operation. Thus, many of the
media's inquiries occur the same day as the data are released. While it
seldom does so, the Bureau has released other reports on Fridays, such as
its 2001 health insurance report.11

For the 2004 and 2005 releases of Income and Poverty Estimates, the data
were released in August at the same time as data from the American
Community Survey. Bureau officials reported the Income and Poverty
Estimates (which come from the Bureau's Current Population Survey) are

10The Bureau cited several other factors for changing the date of the 2003
Income and Poverty Estimates release: (1) the September 2003 report was
the first income and poverty report to include tables and text based on
the new Census 2000-consistent race groups; (2) the September 2003 report
was also the first to reflect Census 2000-consistent industries and
occupations; and (3) new tables and analysis on alternative income
definitions were added to the income report released in September 2003.

11
http://www.census.gov/Press-Release/www/releases/archives/health_care_insurance/
000526.html (downloaded Apr. 20, 2006).

one of several sources of income and poverty information issued by the
Bureau. Starting in 2003, the Bureau began releasing income and poverty
information from the American Community Survey, which produces data
independent from the Current Population Survey. Bureau officials reported
that for methodological and other reasons, estimates from the Current
Population Survey, in some cases, did not match estimates from the
American Community Survey, causing confusing press coverage. In August
2004, when the Bureau released the two data sets at the same time, the
press release that accompanied the estimates explained why the two sets of
numbers might not match. (According to Bureau officials, the plans to move
the release date from September 2004 to August 2004 were in place well
before the actual release.) Going forward, the Bureau plans to continue
its practice of releasing the American Community Survey data and the
Income and Poverty Estimates simultaneously around the last Thursday in
August.

According to the Bureau, Several Factors Affected the Change in Location of the
2003 Release

According to a senior official we interviewed in the Bureau's Public
Information Office, the location of the 2003 Income and Poverty Estimates
news conference was changed from the National Press Club in Washington,
D.C., at the request of the Director of the Census Bureau, to help raise
awareness of the Bureau's new headquarters building, which was under
construction. The groundbreaking ceremony at the new site on the Bureau's
campus in Suitland, Maryland, had taken place several weeks earlier, and a
Bureau official reported the Director wanted the media to see the
improvements the Bureau was making at its headquarters location, as well
as to foster a spirit of good feeling, and highlight how Bureau officials
hoped that the new building would help improve the morale of Bureau
employees. The Bureau provided bus service for attendees from the original
location at the National Press Club in downtown D.C., to Bureau
headquarters in Suitland, a distance of around 8 miles.

Additionally, according to the Bureau's Associate Director for Strategic
Planning and Innovation, the location of the news conference is no longer
as relevant as it once was because of changes in technology. The 2003 news
conference was broadcast in real time via the Internet, and materials were
made available on the Bureau's Web site.12 The Associate Director for
Strategic Planning and Innovation noted that because of these advances and
accommodations, news media on-location attendance has declined over recent
years. Yet, overall media participation has increased via the availability
of Web casts, satellite-feed transmissions and telephone-audio access.
Consequently, the Suitland, Maryland headquarters is now the primary
location for this annual news conference.

12 http://www.census.gov/Press-Release/www/2003/IncomePoverty.html
(downloaded Apr. 20, 2006).

Bureau Officials Made Key Decisions on Releasing Income and Poverty Estimates

Because the Bureau did not maintain a written record of the release
decision, a precise list of the personnel involved and time line of events
is unavailable. However, according to the Bureau officials we interviewed,
the following Bureau employees were involved in the process for releasing
the Income and Poverty Estimates in 2003:

           o  Director of the Census Bureau;
           o  Deputy Director/Chief Operating Officer;
           o  Chief of the Bureau's Housing and Household Economic Statistics
           Division;
           o  Assistant Division Chief for Income, Poverty, and Health
           Statistics;
           o  Associate Director for Demographic Programs, now serving as the
           Associate Director for Strategic Planning and Innovation;
           o  Associate Director for Communications;
           o  Staff from the Bureau's Housing and Household Economic
           Statistics Division;
           o  Staff from the Bureau's Administrative and Customer Services
           Division; and
           o  Chief and Deputy Chief of the Bureau's Public Information
           Office.

Bureau officials said that prerelease access to the Income and Poverty
Estimates is tightly controlled because of the possible economic impact of
the data. They stated its contents are shared with staff on a need-to-know
basis, where only those individuals who are involved with drafting the
report or the accompanying press release have access to the information.
They noted further that key steps in preparing and releasing the report
included the following:

           1. Program staff from the Bureau's Housing and Household Economic
           Statistics Division drafted the Income and Poverty Estimates
           report.
           2. A branch chief reviewed and approved the draft followed by the
           Associate Division Chief, the Division Chief, and ultimately the
           Associate Director for Demographic Programs, who reports to the
           Bureau Director. These senior officials reviewed the report for
           such things as clarity and presentation.
           3. When the content of the report was finalized, the Bureau's
           Public Information Office was sent a copy so it could draft a
           press release.
           4. The final draft was sent to the Bureau's Administrative and
           Customer Service Division, which designed the tables and figures,
           edited the text, and prepared a camera-ready version of the report
           for printing.

According to Bureau officials, the Department of Commerce had only limited
access to information from the Income and Poverty Estimates report before
it was issued, and Commerce officials played no role in the
decision-making process surrounding its release. For example, Commerce's
Office of Public Affairs reviewed the press release that accompanied the
report and thus had access to some of the numbers as well as the key
findings in the report. However, the office did not have access to any of
the tables that are placed on the Internet. (According to the Bureau,
Commerce usually provides a "hook" for the news media. In 2003, the press
release was issued Friday, September 26, and noted, on the first line,
that the nation's poverty rate rose from 11.7 percent in 2001 to 12.1
percent in 2002.)13 Moreover, the Bureau considers the press release part
of the report and holds it to the same standards for statistical quality
as the report itself.

Additionally, according to Bureau officials, the Division Chief and the
Assistant Division Chief briefed the Director of the Census Bureau on the
report about a week before the September 26, 2003, press conference.
Commerce's Under Secretary for Economic Affairs was briefed a day or two
before the press conference and the Under Secretary's staff were provided
with the final report at that time. (The report was also provided to the
Council of Economic Advisers the afternoon before the press conference.)

The Bureau Is Updating Its Practices for Releasing the Income and Poverty
Estimates

The then Chief of the Census Bureau's Housing and Household Economic
Statistics Division told us the Bureau is updating its practices for
releasing the Income and Poverty Estimates. The official stated that the
Bureau was prompted to revisit the 1985 memo by the fact the memo does not
include all of the Bureau's long-standing data dissemination practices;
that some of the practices in the 1985 memo are obsolete given the age of
the guidance; and the rise of the Internet and other technological
advances have had an effect. The official added that the process for
releasing Income and Poverty Estimates has become more formalized over
time.

13
http://www.census.gov/Press-Release/www/releases/archives/income_wealth/001371.html
(downloaded Apr. 20, 2006).

Bureau officials began drafting these revisions after the 2004 release. In
addition to updating the obsolete practices, Bureau officials stated they
planned to document the current practice of combining the Income and
Poverty Estimates release with the American Community Survey release. The
Bureau plans to issue its updated practices prior to the next release of
the Income and Poverty Estimates expected in August 2006.

    Governmentwide Guidance Is Being Developed That May Improve Statistical
                     Agencies' Data Dissemination Practices

Most of the 14 statistical agencies we reviewed reported general adherence
to NRC's guidance, important for (1) the wide dissemination of data, and
(2) maintaining a strong position of independence, although there were
some notable gaps. OMB, in concert with the statistical agencies, has
developed draft guidance on the release and dissemination of statistical
products that, according to OMB officials, parallel NRC's guidance. To the
extent it is comparable to NRC's guidance and other widely accepted
procedures for disseminating data, the proposed OMB directive could
promote more consistent adherence to practices that promote broader
dissemination of statistical data and enhance the data's credibility.

Most Agencies' Data Dissemination Practices Generally Adhered to NRC's Guidance
for Disseminating Information and Maintaining Their Independence

According to NRC, statistical agencies must have "vigorous and
well-planned dissemination programs to get information into the hands of
users who need it on a timely basis." Attributes of a good dissemination
program include using a variety of mechanisms to inform the widest
possible audience about available data products and how to acquire them.
Agencies should also have arrangements for archiving the information so
that it is available for future use, as well as a publications policy that
describes, among other things, the types of data products that will be
made available, the frequency of their release, and the audiences they
serve.

NRC also notes that a statistical agency needs to be politically
independent; that is, it "must be impartial and avoid even the appearance
that its collection, analysis, and reporting processes might be
manipulated for political purposes. . . ." Elements of this practice
include having the authority for decisions associated with the scope,
content, and publication of the data, as well as the authority for the
selection and promotion of professional, operational, and technical staff.

As shown in table 2, the data dissemination procedures of the 14
statistical agencies we reviewed included elements that were generally
aligned with NRC's guidance for the wide dissemination of data and
maintaining a strong position of independence.

Table 2: Most Statistical Agencies Reported That They Generally Adhered to
the National Research Council's Guidance for Releasing Information

                                                   Number of        Number of 
                                               Agencies That Agencies That Do 
                                                    Reported       Not Report 
                                               Following NRC    Following NRC 
                                                    Guidance         Guidance 
     Wide Dissemination of Data Elements                     
1 A variety of avenues for data                           
     dissemination, chosen to reach as broad a               
     public as reasonably possible                           
     a) Regular communication of major                    14                - 
     findings may be disseminated via an                     
     agency's Internet Web site, government                  
     depository libraries, conference exhibits               
     and programs, newsletters and journals,                 
     e-mail address lists, and the media for                 
     regular communication of major findings                 
     b) The broadest possible audience of                 14                - 
     potential users may be informed about                   
     available data products and how to obtain               
     them, and may include providing direct                  
     access to data on the Internet,                         
     depositing data products in libraries,                  
     establishing a network of data centers                  
     (such as the Census Bureau's state data                 
     centers), holding exhibits and making                   
     presentations at conferences, and                       
     maintaining lists of individuals and                    
     organizations to notify of new data                     
     c) Data releases are made available in               14                - 
     printed publications, on                                
     computer-readable media (e.g., CD-ROM),                 
     and on the Internet                                     
2 Release of data in a variety of formats              14                - 
     (e.g., printed reports, various kinds of                
     computer-readable data files with                       
     careful, complete documentation)                        
3 Policies for the preservation of data                14                - 
     that guide what data to retain and how                  
     they are to be archived for future                      
     secondary analysis                                      
4 An established publications policy that                 
     describes, for a data collection program,               
     the types of reports and other data                     
     releases to be made available, the                      
     audience to be served, and the frequency                
     of release                                              
     a) the types of reports to be made                    9                5 
     available                                               
     b) data releases to be made available                11                3 
     c) the audience to be served                          7                7 
     d) the frequency of release                           8                6 
5 Procedures for release of information                12                2 
     that preclude actual or perceived                       
     political interference. In particular,                  
     the timing of the public release of data                
     should be the responsibility of the                     
     statistical agency                                      

      Maintaining a Strong Position of Independence Elements               
6  Dissemination policies that foster regular, frequent release of    8  6 
      major findings from an agency's statistical programs to the          
      public via the media, the Internet, and other means                  
7  Adherence to predetermined schedules in the public release of        
      important statistical indicators to prevent even the appearance      
      of manipulation of release dates for political purposes              
      a) Predetermined schedules exist for the public release of data   12  2 
      (e.g., social and economic indicators and other statistical          
      information), and                                                    
      b) When an agency modifies a customary release schedule, this is   5  9 
      done for statistical purposes, and the agency announces and          
      explains the change in advance                                       
8  Recognition by policy officials outside the statistical agency of  4 10 
      the agency's authority to release statistical information without    
      prior clearance by department policy officials                       
9  Authority for statistical agency heads and qualified staff to     14  - 
      speak about the agency's statistics before Congress, with            
      congressional staff, and before public bodies                        
10 Maintenance of a clear distinction between statistical            12  2 
      information and policy interpretations of such information by the    
      President, the secretary of the department, or others in the         
      executive branch                                                     
11 Authority for professional decisions over the scope, content, and 12  2 
      frequency of data compiled, analyzed, or published                   
12 Authority for selection and promotion of professional, technical,  5  9 
      and operational staff                                                
13 Authority to ensure that information technology systems for data     
      processing and analysis securely maintain the integrity and          
      confidentiality of data and reliably support timely and accurate     
      production of key statistics                                         
      a) Statistical agency has authority to secure information          8  6 
      technology systems for data processing and analysis                  
      b) Statistical agency controls access to records of individual    12  2 
      respondents by policy, program, or regulatory agencies               

Source: GAO analysis.

Twelve or more of the agencies reported having data dissemination
practices possessing four of the five elements related to the wide
dissemination of data. All 14 agencies reported their data dissemination
practices followed NRC's guidance for (1) having multiple avenues for
disseminating data, (2) releasing data in a variety of formats, and (3)
having policies to guide what data should be preserved and how it should
be archived. Similarly, 12 or more of the agencies' dissemination
practices had characteristics associated with five of the eight elements
corresponding to maintaining a strong position of independence. These
elements include (1) adherence to predetermined data release schedules,
and (2) authority to make decisions over the scope, content, and frequency
of the data compiled, analyzed, or published.

A greater number of agencies' data dissemination practices lacked certain
elements important for maintaining a strong position of independence. An
example of one of these elements is NRC's guidance suggesting statistical
agencies should have the "authority to release statistical information and
accompanying materials (including press releases) without prior clearance
by department policy officials" so there is "no opportunity for or
perception of political manipulation of any of the information."

However, 10 of the 14 selected agencies reported varying degrees of
clearance required by department officials. For example, at 2 agencies,
the department rather than the statistical agency releases statistical
information. Other agencies have the authority to release statistical
information, except for press releases, without departmental clearance,
although in some cases, the department's clearance process is limited to
reviewing the grammar, punctuation, and other editorial aspects. (Among
the agencies in our review, 11 agencies use press releases; 1 of these 11
agencies first publishes data from all of its major programs via a press
release; and the 3 remaining agencies reported they do not use press
releases as a vehicle to disseminate data.) With other agencies the
clearance process is more involved. For example, one agency said it
summarizes the data for the press release making sure it is fair and
complete, while officials at the departmental level might insert comments
from the cabinet secretary into the release. Further, 6 of the 14 agencies
lacked dissemination policies that promote the regular and frequent
release of major findings from an agency's statistical program.

As for the Bureau, officials reported that their agency generally adhered
to NRC's recommended guidelines. A notable gap was that Bureau officials
did not report adhering to announcing and explaining modifications to a
customary release schedule in advance (7b in table 2).14 Bureau officials
also lacked the authority to release statistical information and
accompanying materials (including press releases) without prior clearance
by department policy officials (8 in table 2). Also, while the Bureau's
established publications policy describes the frequency of release of data
collection programs, the Bureau reported the policy does not describe the
types of reports to be made available, the data releases to be made
available, or the audience to be served (4a-c in table 2).

14In its comments on a draft of this report, the Department of Commerce
noted that prior to the 2004 release of the Income and Poverty Estimates
the Census Bureau Director wrote a memorandum that described the rationale
for the change to the customary release schedule. Although the Bureau
announced the rationale for the schedule change in this instance, in our
survey of agencies' data dissemination procedures, Bureau officials told
us that the Bureau does not always adhere to the practice of announcing
and explaining modifications to a customary release schedule in advance.

OMB Is Developing Governmentwide Data Dissemination Guidance That Could Help
Strengthen the Credibility of Statistical Information

OMB has been working with the federal Interagency Council on Statistical
Policy to develop guidance for the release and dissemination of
statistical products. According to OMB officials, the guidance is intended
to help ensure statistical products are policy-neutral, timely, and
accurate. OMB officials told us their directive is similar to the NRC's
recommended practices, as well as to OMB's Statistical Policy Directive
Number 3, which applies only to the 38 market-sensitive principal economic
indicators produced by the Departments of Agriculture, Commerce, Labor,
and Treasury, as well as the Federal Reserve Board. However, OMB officials
told us this new directive will not be as stringent as Statistical Policy
Directive Number 3, because the data covered by the directive are released
less frequently than the principal economic indicators, and the data are
not considered to be market-sensitive. OMB expects to issue the directive
for public comment in the spring of 2006.

To the extent OMB's dissemination directive appropriately addresses the
principles underlying NRC's guidance and Statistical Policy Directive
Number 3, the directive could enhance the quality and credibility of
federal statistical data, in part, by replacing the patchwork of
agency-specific guidance with a more transparent, commonly accepted, and
consistently applied framework for disseminating data. For example, OMB's
directive could help promote more consistent adherence to key data release
practices such as the wide dissemination of data and maintaining an
agency's independent position.

As noted in the previous section, the dissemination procedures at several
statistical agencies we examined lacked elements important for these
practices, including (1) authority to release statistical information
without prior clearance by department policy officials, (2) data
dissemination policies that foster the frequent release of major findings
from an agency's statistical programs, and (3) an established publications
policy that describes the types of reports and other data releases to be
made available. As a result, their data products could be better
protected, with the directive, from the appearance of, or actual political
involvement. More specifically, OMB's new directive could address how best
to address the gaps that exist between agencies' data dissemination
practices on the one hand, and NRC's guidance on the other.

As OMB moves forward with its new directive, our interviews with OMB and
statistical agency officials, as well as our past work on data quality
guidance and internal control standards, identified the following
questions that will be important for OMB's dissemination directive to
consider:

Coverage: What will be covered by the directive?-principal statistical
agencies only?-the statistical functions of all agencies?-or only
statistical products? It will be important for OMB's directive to clearly
define what it does and does not cover so that both statistical agencies
and their parent organizations share the same understanding of their
respective authorities, and help ensure dissemination procedures are
consistently implemented. Certain roles, responsibilities, and processes
need to be clarified as well. Indeed, officials at two statistical
agencies we spoke with said there is ambiguity as to whether a statistical
press release is a statistical product and if so, whether statistical
agencies can issue them with or without first getting releases cleared at
the departmental level. Additionally, OMB has issued a number of
guidelines, directives, and standards on federal statistics. Are there any
gaps and overlaps among them, and can they be better integrated?

Documentation: To what extent, and how, should agencies document their
data dissemination procedures and policies, and how often should they be
reviewed and updated? The agencies we examined did not always document
their processes for disseminating statistical data, relying instead on
professional practice. However, as NRC points out: "Although a
long-standing culture of data quality contributes to professional
practice, an agency should also seek to develop and document standards
through an explicit process."

Moreover, documented guidance would lend more transparency to the data
dissemination process, and thus provide a basis for agencies to explain
their dissemination decisions to policy makers, news media, and the
public. Indeed, an OMB official told us that Statistical Policy Directive
Number 3 is a useful tool for explaining to high level policy officials
the procedures agencies must follow to maintain the integrity of the data,
and why the officials may not access principal economic data before it is
released to the public nor comment on it until after its release.
Documented guidance could also help ensure continuity in the face of
employee turnover.

The importance of documenting agencies' data dissemination practices can
be seen in the Bureau's experience in releasing data from the 2000 Census
on the homeless and others without conventional housing, when the Bureau
was criticized for shifting its position on reporting components of this
population. In our 2003 report, we noted that although the Bureau's
decision stemmed from its concerns over the reliability of the underlying
data, the Bureau's lack of documented, clear, transparent, and
consistently applied guidance governing the release of data from the 2000
Census hampered the Bureau in explaining its actions. Had such guidance
been in place, it could have helped the Bureau be more accountable and
consistent in its dealings with the public, and helped to ensure that the
Bureau's decisions both were, and appeared to be, totally objective.15

Flexibility: How much leeway should agencies have in implementing OMB's
directive? Agency officials we spoke with noted the different missions of
the various statistical agencies and cautioned against a one-size-fits-all
approach. As a result, it might not be practical to require all agencies
to meet predetermined release dates because it could lead to additional
workload burdens and staffing issues.

Monitoring: How will OMB ensure agencies comply with its directive?
Indeed, the effectiveness of the policies and procedures laid out in OMB's
directive will rest in large part on the extent to which agencies and
their parent departments adhere to them. Related questions include whether
there should be a regular assessment of agencies' compliance, and if so,
how often should it occur, and whether this should be done by OMB, or by
the agencies through a self-assessment.

Posting Data: Should agencies' dissemination policies include written
guidance for releasing information via specific channels? Indeed, although
NRC's guidance calls on agencies to disseminate data using a variety of
outlets so that the information reaches as wide an audience as possible,
should agencies also have a standard set of conduits where the public will
know an agency's data will always be available? Such conduits might
include, among others, an agency's Web site. Because all of an agency's
data products would be, at a minimum, available from a central point of
access, it could help strengthen an agency's credibility because the
public would always know where to find it.

15GAO, Decennial Census: Methods for Collecting and Reporting Data on the
Homeless and Others without Conventional Housing Need Refinement,
GAO-03-227 (Washington, D.C.: Jan. 17, 2003).

                                  Conclusions

A key lesson learned from the Bureau's experience is the importance of
fully documented, specific practices for maintaining the integrity of data
products, and by extension, the credibility of the agencies that release
them. Thus, as the Bureau updates its practices for releasing the Income
and Poverty Estimates, it will be important for the Bureau to more
thoroughly document its dissemination procedures so they are clear to the
public.

Further, OMB's efforts to develop governmentwide guidance on data
dissemination is a positive step toward enhancing the credibility of
federal statistical data, especially to the extent the directive mirrors
NRC's guidance and Statistical Policy Directive Number 3, as it would
replace each statistical agency's procedures with a more transparent,
commonly accepted, and consistently applied framework for disseminating
information. As OMB works to complete its directive, it will be important
for it to pay particular attention to those elements dealing with the wide
dissemination of data and maintaining a strong position of independence
that, our survey of statistical agencies suggests can be adhered to by a
greater number of agencies. Likewise, OMB should also consider other
aspects of agencies' data dissemination efforts that could make its
directive more comprehensive.

                      Recommendations for Executive Action

To help improve the Bureau's data dissemination practices and thus enhance
the agency's actual and perceived position of independence, we recommend
that the Secretary of Commerce direct the Bureau to, as part of its
efforts to update its practices for releasing the Income and Poverty
Estimates, fully document its key data dissemination practices for
releasing the Income and Poverty Estimates.

Further, to help improve governmentwide data dissemination practices that
would further safeguard the integrity of federal statistical data, we
recommend that the Director of OMB ensure his agency, in completing its
draft directive on the release of federal statistical products, considers
whether and how to address areas where our survey indicates there are gaps
between NRC's existing guidance and agencies' practices. These areas
include the extent to which agencies should have (1) full authority to
release statistical information without prior clearance by their
respective departments, (2) data dissemination policies that foster the
frequent release of major findings from agency's statistical programs, and
(3) an established publications policy describing the types of reports and
other releases an agency has available.

We are also recommending that the Director of OMB direct his agency to
include in its directive additional elements and characteristics important
for agencies' data dissemination practices, including (1) clear
definitions of what is, and what is not covered by the directive, (2) the
extent to which agencies should document their data dissemination guidance
and how often the guidance should be reviewed, (3) the amount of
flexibility agencies have in implementing OMB's guidance, (4) procedures
for monitoring agencies' adherence to its directive, and (5) the
feasibility of requiring agencies to distribute data products through a
standard set of channels as well as through other outlets as appropriate,
so that the public will always know at least one source it can turn to and
obtain agency data.

                       Agency Comments and Our Evaluation

In written comments on a draft of this report, Commerce neither agreed nor
disagreed with our recommendation for the Bureau to fully document its key
data dissemination practices for releasing the Income and Poverty
Estimates. However, Commerce reiterated the point we made in our report
that the Bureau is updating its practices for releasing the Income and
Poverty Estimates. Commerce noted that the updated document-which details
the dissemination practices for the Income and Poverty Estimates-is under
review. The Bureau plans to issue it prior to the next release of the
Income and Poverty Estimates expected in August 2006. Commerce also
provided some technical corrections and suggestions where additional
context might be needed, and we revised the report to reflect these
comments as appropriate. Commerce's comments are reprinted in their
entirety in appendix II.

The Director of OMB did not have comments on our recommendations to them.
However, OMB officials provided suggestions for technical corrections and
we revised the report to reflect these suggestions as appropriate.

As agreed with your offices, unless you publicly announce its contents
earlier, we plan no further distribution of this report until 30 days from
its issue date. At that time, we will send copies of the report to
interested congressional committees, the Director of the U.S. Census
Bureau, and the Director of the Office of Management and Budget. Copies
will be made available to others on request. This report will also be
available at no charge on GAO's home page at http://www.gao.gov .

If you or your staff have any questions about this report, please contact
me at (202) 512-6806 or [email protected]. Contact points for our Office of
Congressional Relations and Public Affairs can be found on the last page
of this report. GAO staff that made major contributions to this report are
listed in appendix III.

Brenda S. Farrell Acting Director Strategic Issues

Appendix I: Scope and Methodology

To address the extent to which the U.S. Bureau of the Census (Bureau)
adhered to its dissemination practices for the release of the 2003 annual
Income and Poverty Estimates and subsequent releases we asked Bureau
officials (in the Housing and Household Economic Statistics Division and
the Bureau's Public Information Office, among others) to identify the
Bureau and Department of Commerce officials who participated in the data
dissemination decisions, and interviewed the identified officials to
determine their role in the decision-making process, and whether they had
prerelease access to the information. We compared their actions to the
Bureau's data dissemination practices. The dissemination process includes
the steps from approval of the report content up to and including public
distribution of the report. Some of these practices are documented in the
Bureau's Policy and Procedures Manual, while others are undocumented
practices that we identified by interviewing cognizant Bureau officials.

Because written records of key activities related to the release (e.g.
e-mails, meeting agendas and notes) were either not retained or never
created, much of our reconstruction of the release was based on interviews
with the officials involved. We interviewed many of these officials both
as a group (by department) and individually to obtain as complete a
picture of the events as possible, and corroborated the information we
received from the various parties involved.

To assess the extent to which the Bureau and other federal statistical
agencies1 followed data dissemination practices that the National Academy
of Sciences' National Research Council (NRC) recommended in its 2005
report, Principles and Practices for a Federal Statistical Agency, we
surveyed officials at 14 federal statistical agencies. (NRC prepared the
report to assist statistical agencies in making their products as sound as
possible.)

Specifically, we surveyed officials at the Bureau, and 13 additional
federal statistical agencies to collect information on the procedures they
followed when releasing data. These 14 agencies comprise the Interagency
Council on Statistical Policy, a body that coordinates federal statistical
work and advises Office of Management and Budget (OMB) on statistical
matters. The 14 agencies are:

1A federal statistical agency's principal function is the compilation and
analysis of data and the dissemination of information for statistical
purposes.

1. Bureau of Economic Analysis, U.S. Department of Commerce

2. Bureau of Justice Statistics, U.S. Department of Justice

3. Bureau of Labor Statistics, U.S. Department of Labor

4. Bureau of Transportation Statistics, U.S. Department of Transportation

5. Bureau of the Census, U.S. Department of Commerce

6. Economic Research Service, U.S. Department of Agriculture

7. Energy Information Administration, U.S. Department of Energy

8. National Agricultural Statistics Service, U.S. Department of
Agriculture

9. National Center for Education Statistics, U.S. Department of Education

10. National Center for Health Statistics, U.S. Department of Health and
Human Services

11. Office of Environmental Information, Environmental Protection Agency

12. Office of Research, Evaluation, and Statistics, Social Security
Administration

13. Science Resources Statistics Division, National Science Foundation

14. Statistics of Income Division, Internal Revenue Service, U.S.
Department of the Treasury

In surveying the agencies, we reviewed relevant documents such as agency
policy manuals, and interviewed key officials who included, depending on
the agency, top management officials, chief statisticians, as well as
management staff from program, communications, or public affairs offices.

We compared the information they provided us to certain practices that the
NRC has determined are important to federal statistical agencies in the
successful conduct of their missions. Specifically, we focused on two NRC
practices (1) wide dissemination of data, and (2) a strong position of
independence, because the 13 guidelines or elements associated with these
two practices are particularly important for data dissemination. The first
practice, the wide dissemination of data, is associated with the mechanics
of making the information available to the public, including the media for
releasing the information, as well as how it is formatted and archived.
The elements of the second practice, a strong position of independence,
are essential for maintaining the credibility of statistical agencies, as
well as for providing an unimpeded flow of information to data users.

To obtain a broader perspective on the governmentwide framework for
helping to ensure data quality, we also interviewed OMB officials about
OMB's role in coordinating and overseeing the data dissemination
activities and reviewed appropriate OMB documents such as Statistical
Policy Directive Number 2 and Number 3.

We conducted our work between March 2005 and April 2006 in accordance with
generally accepted government auditing standards.

Appendix II: Comments from the Department of Commerce

Appendix III: GAO Contact and Acknowledgments

                                  GAO Contact

Brenda S. Farrell, (202) 512-6806

                                Acknowledgments

In addition to the individual named above, Robert Goldenkoff, Assistant
Director, as well as Timothy Wexler, April Thompson, Robert Parker, Jay
Smale, Michael Volpe, Andrea Levine, and Amy Rosewarne made key
contributions to this report.

Related GAO Reports

Quality of Federal Data

Information Quality Act: National Agricultural Statistics Service
Implements First Steps, but Documentation of Census of Agriculture Could
Be Improved. GAO-05-644 . Washington, D.C.: September 23, 2005.

Data Mining: Agencies Have Taken Key Steps to Protect Privacy in Selected
Efforts, but Significant Compliance Issues Remain. GAO-05-866 .
Washington, D.C.: August 15, 2005.

Data Quality: Improvements to Count Correction Efforts Could Produce More
Accurate Census Data. GAO-05-463 . Washington, D.C.: June 20, 2005.

Data Quality: Census Bureau Needs to Accelerate Efforts to Develop and
Implement Data Quality Review Standards. GAO-05-86 . Washington, D.C.:
November 17, 2004.

Decennial Census: Methods for Collecting and Reporting Hispanic Subgroup
Data Need Refinement. GAO-03-228 . Washington, D.C.: January 17, 2003.

Decennial Census: Methods for Collecting and Reporting Data on the
Homeless and Others without Conventional Housing Need Refinement.
GAO-03-227 . Washington, D.C.: January 17, 2003.

2000 Census: Refinements to Full Count Review Program Could Improve Future
Data Quality. GAO-02-562 . Washington, D.C.: July 3, 2002.

(450394)

GAO's Mission

The Government Accountability Office, the audit, evaluation and
investigative arm of Congress, exists to support Congress in meeting its
constitutional responsibilities and to help improve the performance and
accountability of the federal government for the American people. GAO
examines the use of public funds; evaluates federal programs and policies;
and provides analyses, recommendations, and other assistance to help
Congress make informed oversight, policy, and funding decisions. GAO's
commitment to good government is reflected in its core values of
accountability, integrity, and reliability.

Obtaining Copies of GAO Reports and Testimony

The fastest and easiest way to obtain copies of GAO documents at no cost
is through GAO's Web site ( www.gao.gov ). Each weekday, GAO posts newly
released reports, testimony, and correspondence on its Web site. To have
GAO e-mail you a list of newly posted products every afternoon, go to
www.gao.gov and select "Subscribe to Updates."

Order by Mail or Phone

The first copy of each printed report is free. Additional copies are $2
each. A check or money order should be made out to the Superintendent of
Documents. GAO also accepts VISA and Mastercard. Orders for 100 or more
copies mailed to a single address are discounted 25 percent. Orders should
be sent to:

U.S. Government Accountability Office 441 G Street NW, Room LM Washington,
D.C. 20548

To order by Phone: Voice: (202) 512-6000 TDD: (202) 512-2537 Fax: (202)
512-6061

To Report Fraud, Waste, and Abuse in Federal Programs

Contact:

Web site: www.gao.gov/fraudnet/fraudnet.htm E-mail: [email protected]
Automated answering system: (800) 424-5454 or (202) 512-7470

Congressional Relations

Gloria Jarmon, Managing Director, [email protected] (202) 512-4400 U.S.
Government Accountability Office, 441 G Street NW, Room 7125 Washington,
D.C. 20548

Public Affairs

Paul Anderson, Managing Director, [email protected] (202) 512-4800 U.S.
Government Accountability Office, 441 G Street NW, Room 7149 Washington,
D.C. 20548

www.gao.gov/cgi-bin/getrpt? GAO-06-607 .

To view the full product, including the scope and methodology,
click on the link above.

For more information, contact Brenda S. Farrell at (202) 512-6806 or
[email protected].

Highlights of GAO-06-607 , a report to congressional requesters

May 2006

DATA QUALITY

Expanded Use of Key Dissemination Practices Would Further Safeguard the
Integrity of Federal Statistical Data

In 2003, the Bureau of the Census (Bureau) changed the day and location of
the release of its Income and Poverty Estimates. Some data users believed
the change was an effort to suppress unfavorable news and questioned the
Bureau's data dissemination practices. GAO was asked to assess whether (1)
the Bureau adhered to its dissemination practices for the 2003 and later
releases, and (2) the Bureau and 13 other federal statistical agencies
follow data release practices recommended by the National Research Council
(NRC). GAO reviewed the Bureau's dissemination process for the 2003 thru
2005 Income and Poverty Estimates.

What GAO Recommends

The Secretary of Commerce should direct the Bureau to fully document its
dissemination practices for the Income and Poverty Estimates. GAO also
recommends that OMB consider (1) how to address gaps identified between
agencies' dissemination practices and NRC's guidance, and (2) how OMB's
proposed guidance should address documentation, coverage, and other
questions noted in this report. In its comments on a draft of this report,
Commerce neither agreed nor disagreed with GAO's recommendation but
reiterated GAO's finding that the Bureau is updating its practices for
releasing the Income and Poverty Estimates. OMB did not have comments.

While not all of the Bureau's data dissemination practices are documented,
GAO was able to determine through discussions with Bureau officials and
review of available documentation, that the Bureau adhered to most of its
long-standing data release practices. However, the Bureau did depart from
the traditional day and location for the release of the Income and Poverty
Estimates report in 2003 and subsequent years. According to the Bureau,
the day of the 2003 release was changed because of a delay in producing a
companion report, and the location was changed from Washington, D.C., to
Suitland, Maryland, in part, because the Director of the Census Bureau
stated that he wanted to raise awareness that the construction of its new
headquarters had just started. Some of the Bureau's documented practices,
such as guidance on who has authority to choose the release date and
location, lacked specificity. Also, the Bureau's documented Income and
Poverty practices are outdated as they are contained in a 21-year-old
memo. The Bureau is updating it, to among other things, reflect current
technology.

Bureau Press Conference for Release of Income and Poverty Estimates

Most of the 14 statistical agencies in GAO's review generally adhered to
NRC's guidance, important for (1) the wide dissemination of data, and (2)
maintaining a strong position of independence. Still, there were some
notable gaps. For example, 6 of the 14 agencies lacked dissemination
policies (as recommended by NRC) that promote the regular and frequent
release of major findings from an agency's statistical program. The Office
of Management and Budget (OMB), in concert with other statistical
agencies, is developing governmentwide guidance on the release and
dissemination of statistical products that, according to OMB officials,
parallels NRC's and other generally accepted release practices. OMB's
guidance could foster more consistent adherence to practices that promote
broader dissemination of statistical data and enhance its credibility,
especially to the extent they address gaps GAO found between agencies'
data dissemination practices and NRC's guidance.
*** End of document. ***