[Federal Register: May 5, 2003 (Volume 68, Number 86)]
[Notices]
[Page 23693-23694]
From the Federal Register Online via GPO Access [wais.access.gpo.gov]
[DOCID:fr05my03-54]
=======================================================================
-----------------------------------------------------------------------
DEPARTMENT OF COMMERCE
Bureau of the Census
[Docket Number 030328074-3074-01]
Revised Confidentiality Criteria for Bureau of the Census Public
Use Data Products
AGENCY: Bureau of the Census, Department of Commerce.
ACTION: General notice.
-----------------------------------------------------------------------
SUMMARY: The Bureau of the Census (Census Bureau) is issuing this
Notice to update the criteria used to assess and review public use data
products prior to their release. The criteria help ensure that
confidential information is not inadvertently disclosed.
DATES: This notice is effective May 5, 2003.
FOR FURTHER INFORMATION CONTACT: Requests for additional information on
this Notice should be directed to Laura Zayatz, Chair, Disclosure
Review Board, Statistical Research Division, U.S. Census Bureau, Room
3209, Federal Building 4, Washington, DC 20233, (301) 763-4955 or by
fax at (301) 457-2299.
SUPPLEMENTARY INFORMATION:
Background
Title 13, United States Code, Section 9, requires the Census Bureau
to protect the confidentiality of individual respondents. Title 13 also
requires the Census Bureau to release information from its data
collections to the public. In order to comply with the latter
requirement, while protecting respondents' confidentiality, the Census
Bureau has developed standard procedures for protecting its data
products from disclosure of identifying information. A Federal Register
Notice (46 FR 22017) published on April 15, 1981, described those
procedures.
Since that Notice appeared, some methodological improvements have
been introduced and additional formal review has been instituted.
Specific changes to the earlier criteria are the following items under
the ``Criteria for Releasing Public Use Microdata'' section of this
Notice.
[sbull] Paragraph (4). To ensure confidentiality protection, the
Census Bureau now requires the completion of a Checklist on Disclosure
Potential of Data.
[sbull] Paragraph (4)(e). To better protect the confidentiality of
data, the Census Bureau may require the use of data swapping. This
technique is discussed further in that paragraph.
[sbull] Paragraph (5). The name of the Microdata Review Panel was
changed to the Disclosure Review Board.
[sbull] ``Criteria for Releasing Public Use Tabular Data'' section.
This entire section is new. In 1995, the routine review performed by
the Disclosure Review Board was extended to include tabular data as
well as microdata.
Also, the scope of the Disclosure Review Board was enlarged to
include setting disclosure limitation rules, monitoring the Census
Bureau's adherence to its confidentiality policy requirements, and
resolving any problems, questions, and issues not covered by the
general criteria described in the earlier Federal Register Notice. This
Notice updates the information previously published and includes
criteria for releasing tabular data as well as microdata.
Confidentiality Criteria for Census Bureau Public Use Data Products
Data Subject to Disclosure Protection
All data released to the public are subject to disclosure
protection. These public use data products include both microdata and
tabular data. Microdata
[[Page 23694]]
are records containing information about individuals or households, or
about businesses, with all personal identifiers removed. Tabular data
may be frequency counts (e.g., number of Hispanic males) or magnitude
data presenting sums of a variable of interest (e.g., total value of
sales) from individuals, households, firms, or establishments.
Because some business firms and establishments are selected with
certainty, the disclosure risk for economic microdata files can be
quite high. For this reason, microdata files containing information on
establishments are rarely produced. However, any economic microdata
file produced for public release would be subject to disclosure
constraints comparable to those provided for individual and household
microdata files.
Criteria for avoiding disclosure in data released by the Census
Bureau for public use have been the subject of internal research,
consultations with stakeholders, and input from experts in disclosure
limitation techniques over the past few decades. Based on those
findings and subsequent assessments of data needs and disclosure risk,
the Census Bureau has established a release policy stating criteria for
release. The Census Bureau will apply these criteria to all data
products released to the public.
Criteria for Releasing Public Use Microdata
Files of records containing data about households and individuals,
or about businesses, can be made available for public use, provided the
appropriate disclosure avoidance requirements have been met. For public
use microdata about individuals and households, the following
conditions must be met:
(1) The records contain no names, addresses, or other unique
identifiers.
(2) The records include no geographic or related information that
would identify an area of fewer than 100,000 population.
(3) Once a file has been released with one set of geographic
identification, the same records cannot be released with different
identification if the two geographic schemes in combination identify
any area with fewer than 100,000 population.
(4) Specifications for each file (or groups of files) must be
reviewed to assure that confidentiality is protected. To do so, the
Census Bureau's Checklist on Disclosure Potential of Data must be
completed by the data producers and reviewed by disclosure experts at
the Census Bureau. (The Checklist is available on the Census Bureau Web
site at http://www.census.gov/srd/sdc/index.html.) This review may
result in:
(a) The removal or reduction in detail of any variable considered
likely to identify an especially small and visible population (e.g.,
persons with high income or rare demographic characteristics).
(b) The use of a minimum area population criterion that is higher
than 100,000 for that particular file(s) (e.g., for a file with
neighborhood characteristics).
(c) The introduction of ``noise'' (i.e., small amounts of random
variation) into selected data items.
(d) The subsampling of records so that public-use microdata do not
include all respondents included in a large survey.
(e) The use of data swapping (i.e., locating pairs of matching
households in the database, based on a set of predetermined variables,
and swapping those households across geographic areas to add
uncertainty for households with unique characteristics).
More information on these methods is provided on the Census
Bureau's Internet site at http://factfinder.census.gov/home/en/confidentiality.html#microdata
.
(5) The Disclosure Review Board concurs that the data meet
disclosure avoidance criteria for public release.
Comparable criteria will be developed and applied in the event that
microdata about businesses are being considered for public release.
Criteria for Releasing Public Use Tabular Data
Although tabular data products had routinely undergone disclosure
protection, it was not centralized. This Notice documents standardized
procedures now in use that have been adopted to protect the
confidentiality of information in frequency counts and magnitude data.
Tabulations containing data about households, individuals, firms, and
establishments can be made available for public use provided that:
(1) Specifications for each file (or groups of files) are reviewed
to assure that confidentiality is protected for any response provided
by an individual, a household, a firm, or an establishment. To do so,
the Census Bureau's Checklist on Disclosure Potential of Data is
completed by the data producers and reviewed by disclosure experts at
the Census Bureau. (The Checklist is available on the Census Bureau Web
site at http://www.census.gov/srd/sdc/index.html.) This review may
result in:
(a) The removal or reduction in detail of any variable in a
tabulation considered likely to identify an especially small and
visible population or group of establishments.
(b) The introduction of ``noise'' into selected data items.
(c) The use of data swapping for selected households.
(d) The use of rounding in selected tabulations.
(e) The use of thresholds in selected tabulations.
(f) The use of cell suppression (i.e., not displaying certain cell
values that would identify unique cases in a table or disclose an
individual response; often complementary cells also are blanked out to
prevent disclosures by subtracting the balance from the total for a
given row or column).
(2) The Disclosure Review Board concurs that the data meet
disclosure avoidance criteria for public release.
Dated: April 29, 2003.
Charles Louis Kincannon,
Director, Bureau of the Census.
[FR Doc. 03-10952 Filed 5-2-03; 8:45 am]
BILLING CODE 3510-07-P