This appendix describes the organization of the panel study and the procedures used by the panel to gather relevant background information, explore the issues, and develop its recommendations. This description covers two case study workshops that were held prior to the formal convening of the panel, but made important contributions to the panel's work.
PANEL ORGANIZATION AND ACTIVITIES
The panel was jointly convened in 1989 by the Committee on National Statistics (CNSTAT) of the National Academy of Sciences-National Research Council and the Social Science Research Council (SSRC). Its goal was to provide recommendations to federal agencies to aid them in their stewardship of data for public policy decisions and research. The panel was comprised of experts in the fields of ethics, privacy, respondent issues, public policy, legislation, the history of the federal statistical system, and statistics. Supporting the panel's work was a staff consisting of a study director, two consultants, a senior project assistant, and a staff liaison from the Social Science Research Council. (See Appendix B for biographical sketches of panel members and staff.)
Funding for the study was provided by the National Science Foundation, the Bureau of the Census of the U.S. Department of
Commerce, the Bureau of Labor Statistics of the U.S. Department of Labor, the Internal Revenue Service Statistics of Income Division of the U.S. Department of Treasury, the National Institute on Aging of the U.S. Department of Health and Human Services, the National Center for Education Statistics of the U.S. Department of Education, and, through their general contributions to the work of the Committee on National Statistics, several other federal agencies.
The panel accomplished its work in several different ways. Prior to the first meeting of the panel, and as a foundation for the panel's work, the Committee on National Statistics and the Social Science Research Council sponsored two workshops: the Longitudinal Retirement History Workshop, held in September 1987, and the Workshop on Confidentiality of and Access to Doctorate Records, held in November 1988. In January 1991 the panel sponsored a workshop on confidentiality of and access to National Center for Education Statistics data. A conference on disclosure limitation approaches and data access was held in March 1991. The panel held nine deliberative meetings from November 1989 through January 1992. The panel also met with representatives from the federal statistical community, privacy advocates, and other concerned individuals and organizations. Finally, the panel commissioned papers to inform its deliberations, including several papers that served as background for the Workshop on Confidentiality of and Access to National Center for Education Statistics Data and the Conference on Disclosure Limitation Approaches and Data Access (see below).
The panel held nine meetings in Washington, D.C., between November 1989 and January 1992. Through presentations by agency representatives at these meetings, the panel was informed about the complexities and realities of the missions of the agencies and the role that those missions and environment play in agencies' efforts to protect confidentiality while permitting access to data for research. Agency representatives identified important aspects of confidentiality and data access and provided information on the subtleties of some of the problems they face in dealing with these issues. In addition, the panel held lengthy discussions with researchers, privacy advocates, and others about data sharing and record linkage, statistical disclosure limitation, administrative policies of the agencies, legislative problems and solutions, the needs of researchers, and the needs of the federal statistical agencies. The length, duration, and number of meetings was intended to encourage debate and ultimately develop consensus on the issues. Throughout
these meetings, the panel was mindful of its goal to provide helpful recommendations to the federal statistical agencies.
At a June 1991 meeting, several privacy advocates were invited. The meeting focused on the public's interest, ethical issues, and cognitive research on informed consent and notification in relation to confidentiality and data access concerns.
LONGITUDINAL RETIREMENT HISTORY WORKSHOP
The Longitudinal Retirement History Workshop, chaired by Jerry A. Hausman and held on September 18–19, 1987, was conducted at the request of the National Institute on Aging (NIA) and the Census Bureau. The National Center for Health Statistics provided additional support for the workshop.
The National Institute on Aging requested the workshop to help answer a question of whether its proposed reinterview of surviving panel members and spouses of decedents from the Longitudinal Retirement History Survey was feasible. Additional goals of the workshop were to learn some of the problems and issues in protecting confidential data, discuss the disclosure limitation practices of the Census Bureau, develop suggestions for methods to access confidential data for research, and provide information to NIA on how to achieve its goals while maintaining adequate protection of the data. Workshop participants were particularly interested in the legal, ethical, and policy questions surrounding the issues of recontacting the panel members, linking the data from the reinterview with the old data, and making the linked data files available to researchers for analysis.
After considering the questions above, the workshop participants outlined three possible courses of action. Two were deemed useful in a broad context: allowing access of researchers to microdata through their appointments as special sworn employees of the Census Bureau and the establishment of data resource centers where Census Bureau employees could process researcher requests. A third proposed course of action, one that was thought to be the most feasible for the follow-up of the Longitudinal Retirement History Survey, involved obtaining consent from the survey respondents to transfer data in existing files to another agency whose data are not subject to the conditions of Title 13 (see Chapter 5). The workshop participants also developed some ideas to help guide
the Panel on Confidentiality and Data Access, which at the time had not yet been convened.
WORKSHOP ON CONFIDENTIALITY OF AND ACCESS TO DOCTORATE RECORDS
The Workshop on Confidentiality of and Access to Doctorate Records, chaired by George T. Duncan and held November 4–5, 1988, was convened by CNSTAT and SSRC. The purposes of the workshop were to determine if and how mechanisms for allowing greater researcher access to data from the Doctorate Records File and the Survey of Doctorate Records could be developed without compromising the confidentiality of the data and to identify issues that would be important for the proposed Panel on Confidentiality and Data Access to address.
Workshop participants discussed various approaches for providing increased access to the Doctorate Records File and the Survey of Doctorate Records. Topics discussed included the creation of public-use data files with all information that would allow identification of the respondent removed; modification of statements of use; providing respondents an opportunity to approve or disapprove of the proposed uses of the data; development of signed agreements that clearly define users responsibilities in maintaining the confidentiality of the data; and providing remote access to the data through a ''gatekeeper," who would monitor access requests and ensure adherence to any confidentiality provisions associated with the remote access. The general conclusion of participants was that the tension between confidentiality and data access could not be resolved by any one action.
WORKSHOP ON CONFIDENTIALITY OF AND ACCESS TO NATIONAL CENTER FOR EDUCATION STATISTICS DATA
The panel's Workshop on Confidentiality of and Access to National Center for Education Statistics Data, chaired by William M. Mason, was held in January 1991. The purposes of the workshop were to investigate confidentiality and access issues as they apply to the National Center for Education Statistics (NCES) and to obtain information for the panel's deliberations. Participants from the educational research community, federal government, education associations, and state departments of education were invited to attend the workshop. Panel members also attended the meeting.
During the workshop, participants discussed the evolution of confidentiality legislation covering the NCES particularly Section 3001(m) of the Augustus F. Hawkins-Robert T. Stafford Elementary and Secondary School Improvement Amendments of 1988 (the Hawkins-Stafford Act, P.L. 100–297) and Section 252 of the Excellence in Mathematics, Science and Engineering Act of 1990 (P.L. 101–589), which amended the confidentiality provisions of the General Education Provisions Act (P.L. 90–247). In a paper commissioned by the panel, Diana C. Pullin and Kenneth B. Newton discussed the impact of Section 252 on the policies and practices of NCES in permitting access to their data while respecting the confidentiality of respondents' identities. Discussion focused on the various data systems maintained by NCES, including the sources of the data and the type of data items in the data sets; NCES's confidentiality provisions; users' experiences before and after the Hawkins-Stafford Act; and NCES's statistical and administrative procedures to prevent the disclosure of individually identifiable information. Finally, participants assessed the issues and discussed possible changes in procedures, legislation, regulations, and interpretations that could increase access to data and protect the confidentiality of the data.
REQUEST FOR DOCUMENTATION
In fall 1990, the panel contacted approximately 40 federal agencies, requesting documentation of their policies and practices for ensuring the confidentiality of data that they collect and disseminate. The agencies were selected based on the number of their statistical activities and budgets for statistical programs. Information was requested on agencies' statutes, regulations, and internal guidelines; Privacy Act notices describing the nature of systems of records that it maintains (Title 5 U.S.C. § 552a(e)(4)); informed consent and notification procedures (for individuals and establishments); disclosure limitation techniques; access to nonpublic-use data, including the type of arrangements made for the access; instances of denied access; and other organizational arrangements and guidelines used to establish policies for protecting the confidentiality of data and reviewing requests for access to data. The panel also asked the agencies to provide information on their experiences as users of identifiable data.
The panel received extensive information on agencies' practices.
The documentation was invaluable to its work in assessing the agencies' confidentiality and access practices and their future needs. The panel and staff conducted many follow-up contacts with the agencies. In addition, representatives from the agencies attended panel meetings. The text of the request appears as Attachment 1.
INVITATION FOR COMMENTS
In fall 1990, the panel also published an "Invitation for Comments" in a variety of professional journals and newsletters. The panel requested comments from users of federal statistics on access problems, suggestions for improving access, and information on persons or businesses harmed by disclosure of identifiable data. The "Invitation" provided an opportunity for concerned individuals to submit their comments, suggestions, and experiences for the panel's consideration. Approximately 20 responses were received. The text of the "Invitation for Comments" is included as Attachment 2.
CONFERENCE ON DISCLOSURE LIMITATION APPROACHES AND DATA ACCESS
Identification and examination of methods to maximize access to federal statistical data while maintaining the confidentiality of the respondents' identities were the foci of the Conference on Disclosure Limitation Approaches and Data Access, held in March 1991. Researchers, federal government employees, and panel members attended the conference, chaired by George T. Duncan. To understand the underlying issues, participants reviewed data dissemination policies of federal agencies and the effect of increasingly powerful technologies on the agencies' efforts to provide data and maintain confidentiality. To facilitate the discussion of the issues, the conference was divided into four sessions: basic issues, statistical disclosure limitation, computer issues, and assessment of current legislation and restricted access procedures.
The panel commissioned several papers to serve as background information for the conference. These papers were presented at the conference and are discussed in Chapter 6 of the report. They are published in a special issue of the Journal of Official Statistics (1993(2)).
The panel commissioned 11 papers to help its deliberations.
Paper commissioned for the Workshop on Confidentiality of and Access to National Center for Education Statistics Data
Kenneth B. Newton and Diana C. Pullin, "The Impact of Section 252 of the Excellence in Mathematics, Science and Engineering Act of 1990 on the Policies and Practices of the National Center for Education Statistics"
Papers commissioned for the Conference on Disclosure Limitation Approaches and Data Access
Session 1: Basic Issues
Diane Lambert, "Measures of Disclosure Risk and Harm"
Paul D. Reynolds, "Privacy Advances in Social and Policy Science: Balancing Present Costs and Future Gains"
Discussants: William Butz, Thomas Plewes, Eleanor Singer
Session 2: Statistical Disclosure Limitation
Wayne A. Fuller, "Use of Masking Procedures for Disclosure Limitation"
Roderick J.A. Little, "Statistical Analysis of Masked Data"
Discussants: Brian V. Greenberg, Donald B. Rubin
Session 3: Computer Issues
Sallie Keller-McNulty and Elizabeth A. Unger, "Database Systems: Inferential Security"
Discussants: Gerald Gates, Teresa F. Lunt
Session 4: Creative Solutions to the Disclosure Problem Under the Status Quo
Joe S. Cecil, "Confidentiality Legislation and the Federal Statistical System"
Thomas B. Jabine, "Procedures for Restricted Data Access"
Discussants: John P. Fanning, F. Thomas Juster, Nancy J. Kirkendall
Thomas B. Jabine, "Statistical Disclosure Limitation: Current Federal Practices and Research"
Robert H. Mugge, "Informed Consent in U.S. Government Surveys"
Eleanor Singer, "Summary of Informed Consent Literature for National Research Council Panel on Confidentiality and Data Access"
Most of the papers and discussants' comments appear in revised form in a special issue of the Journal of Official Statistics (9(2)).
COMMITTEE ON NATIONAL STATISTICS
NATIONAL ACADEMY OF SCIENCES
2101 Constitution Avenue, N.W., Washington, D.C. 20418
Panel on Confidentiality and Data Access
CNSTAT: (202) 334–2550
SSRC: (212) 661–0280
Request for Documentation of Policies and Practices
FROM: George T. Duncan, Chair
DATE: October 31, 1990
Federal agencies differ with respect to their statutes, regulations, and organizational arrangements that pertain to the confidentiality of and the access to statistical data that they collect and disseminate. The review of policies and procedures related to confidentiality and data access is one of the most important activities of our panel study. Therefore, we are asking you to assist the Panel on Confidentiality and Data Access in obtaining documentation on your agency's policies and practices.
The Panel plans to collect information in two steps:
From a broad cross-section of federal statistical agencies and units, the Panel will request that each agency provide documentation related to its confidentiality and data access policies and practices; and
Following the review of these materials, the Panel and staff will request meetings with key officials of a few agencies. The meetings will serve the purpose of adding to or clarifying information provided in the documents. They will also provide the Panel an opportunity to obtain information on topics for which there was no documentation.
The purpose of this memorandum concerns the first step: a request from the Panel to provide documentation about your agency's policies and practices. The Panel is primarily interested in obtaining
documentation on policies and practices that pertain to the data your agency produces. In addition, the Panel would like to learn about your agency's experiences as a user of identifiable data from federal and non-federal sources. For instance, has your agency experienced difficulties or delays in accessing data produced by another federal agency? If so, what were the problems and how were they resolved? Examples of the relevant types of documentation are given below.
Please note that the Panel is asking your agency to provide copies of existing documents. We are not asking your agency to prepare additional written statements at this time. Since the documents collected from each federal agency will be used to develop the Panel's report and may be summarized in ways that can lead to the identification of an agency, they cannot be considered to be confidential.
To assist you in collecting this documentation, please consider the scope of this study: it includes publicly-supported statistical data collection activities such as censuses, surveys, administrative record data (when used for statistical purposes), and epidemiological studies. We are interested in statistical data for both individuals and establishments. Data from clinical trials, while very important, will not be considered in this study. There are some special issues associated with clinical trial data that would require a separate study focusing on the bioethical aspects of confidentiality and data access.
I. DOCUMENTATION FOR DATA PRODUCED BY YOUR AGENCY
For data produced by your agency, we ask that you provide relevant documentation of your agency's confidentiality and data access policies and practices. We use the word "documentation" to refer to a broad array of existing materials, such as published technical papers, standards manuals, copies of consent forms, memoranda, policy statements, statutes, interagency agreements, etc., that are relevant to each of the topics below. Please remember that the Panel is interested in data collected from both individuals (except in clinical trials) and establishments. Note that the Panel has copies of all documents listed in the appendices and does not require additional copies.
Statutes, regulations, agency guidelines: Statutes, regulations,
agency guidelines (ex., National Center for Health Statistics' Staff Manual on Confidentiality), etc., that pertain to the confidentiality of and access to your agency's data. Please note that we are not interested in government-wide statutes and guidelines pertaining to data used for statistical purposes, such as the Privacy Act—the panel will contact the Office of Management and Budget for government-wide documents. However, if your agency has its own regulations or guidelines for compliance with government-wide statutes, we are interested in those.
Privacy Act system of records notice: We would like to obtain system of records notices for two or three of your agency's major systems that are used for statistical purposes. In addition, notices from other systems that have features of special interest for the panel study would also be appreciated.
Respondent notification and consent procedures for both individuals and establishments: The Panel would like to obtain examples, from questionnaires or other survey forms, of the respondent notification and consent procedures (RNCPs) used by your agency to collect data from both individuals and establishments. Note that the term RNCP, as used in this memorandum, refers to a wide range of notices provided by federal agencies to individuals and businesses. It includes, but is not limited to, informed consent statements or agreements, confidentiality notification statements, and privacy act notifications. Some topics of interest follow:
How RNCPs differ for individuals and establishments, especially for agencies that collect information from both.
The RNCP(s) used by your agency when it plans to link survey data with administrative record data (for example, when your agency plans to use Social Security Number for the linkage).
In addition, the Panel also requests agency documents that contain instructions for staff on how to write RNCPs.
Disclosure limitation techniques: Statistical methods and procedures are sometimes used to limit disclosure risk in the release of tabulations and microdata to the public, for example,
not releasing separate data for geographic areas with small populations. Some topics of interest are:
The procedures that are used by your agency to limit disclosure risk.
If relevant, how procedures differ for individuals and establishments.
Facilitating access to nonpublic-use data: Some agencies have implemented special procedures to facilitate access to data that have not been released for public-use. Some topics of interest to the Panel follow:
Whether your agency has ever provided access to data that were not released for public-use.
If so, the type of arrangements used by your agency and its experience with such "special arrangements." Some examples include:
allowing controlled access to specially sworn employees who are required to relocate to the principal office of your agency;
allowing controlled access to specially sworn employees who work in regional offices of the agency; and
employing special licensing agreements.
Inability to facilitate access: Documents from your agency of cases where access has been denied and the reason for denial. Documentation might be contained in correspondence or in published papers, e.g., G. Gates, "Census Bureau microdata: Providing useful research data while protecting the anonymity of respondents" in the Proceedings of the Survey Research Section (American Statistical Association, 1988).
Organizational arrangements: Existing documentation that describes organizational arrangements in your agency that are used (1) to establish confidentiality guidelines and (2) to review requests for access (for example, the Census Bureau's Microdata Review Panel). Essentially, the Panel is interested
in tracing the "standard flow" of a request for access to unpublished data through your agency.
II. DOCUMENTATION FOR ACCESSING IDENTIFIABLE DATA FROM FEDERAL AND NON-FEDERAL SOURCES
The second area of interest to the Panel concerns your agency's experiences in accessing identifiable data from other federal agencies and non-federal sources—data to be used for statistical purposes. Specifically, we are interested in those experiences that concern access to microdata which would not meet public-use criteria.
Federal agencies may require identifiable data from other agencies to (1) use as a frame for their surveys; (2) link with their own survey data; (3) do analyses. The Panel is interested in the extent to which your agency had difficulty in obtaining such data (documentation might be letters or published papers). The Panel is also interested in knowing if special treatment was required for these data, above and beyond the standards applied to microdata produced by your own agency.
Data from other federal agencies: We are requesting information about your agency's experiences in accessing data from other federal agencies, especially the barriers encountered. Topics of interest include:
The extent to which your agency uses data for statistical purposes that are produced by other agencies.
Cases in which your agency experienced "difficulties" in accessing the federal data that it needed. For instance, barriers that a federal agency might experience are:
an agency sponsors surveys but is unable to get fully detailed files from the collecting agency; or
a federal agency or program experiences delays in getting administrative data or business lists from other agencies—data that are needed for use in its own statistical programs.
Confidentiality and access issues for non-federal data: A federal agency sometimes serves an important role as a repository
of non-federal survey and/or administrative data. These non-federal data provide valuable sources of statistical information that supplement federal data. Some issues of interest to the Panel include the following:
The extent to which your agency uses identifiable, non-federal records for statistical purposes (e.g., administrative data produced on the state-level).
The provisions that exist concerning confidentiality to and access of these data, especially where agency policies for data that it produces differ from its policies covering non-federal data.
Finally, agency papers and/or publications that describe the preceding items, but which might also include the history of access to unpublished data or the history of changes to relevant confidentiality laws and regulations, would be most helpful.
If there is someone else the Panel should contact in your agency or program to obtain the documentation that we are requesting, please provide Virginia de Wolf, Study Director, (202–334–2550) with the name and telephone number. If you have any questions, please direct them to Virginia. We would appreciate receiving your agency's documentation by December 3, 1990. Thank you very much for your efforts in helping the Panel.
Attachment 2 PANEL ON CONFIDENTIALITY AND DATA ACCESS
Invitation for Comments
Many users of Federal statistics are aware of the balance that must be struck between protecting the confidentiality of information provided by persons and businesses for statistical purposes and the need to make publicly-collected data widely available for legitimate research and statistical uses.
In search of new ways to deal with this issue, the Committee on National Statistics and the Social Science Research Council, with support from several Federal agencies, have convened a Panel on Confidentiality and Data Access. As part of its two-year study, the Panel, which had its first meeting in December of last year, will be compiling relevant information from both producers and users of Federal statistics.
The scope of this panel study includes publicly-supported statistical data collection activities on individuals and establishments, such as censuses, surveys, administrative record data (when used for statistical purposes), and epidemiological studies. Data from clinical trials, while very important, will not be considered in this study. There are some special issues associated with clinical trial data that would require a separate study focusing on the bioethical aspects of confidentiality and data access.
Readers of this notice are invited to submit short statements on any or all of the following subjects:
Access problems. Specific examples of instances where Federal agency confidentiality laws or policies have made it impossible for you or your colleagues to obtain data needed in your work or caused excessive delays in arranging for access to the data. Please indicate the sources and specific kinds of data desired and the purposes for which the data were needed.
Suggestions for improving access. Have you had any experience in obtaining access to data not disclosed for general public-use? How was this arranged? Do you have suggestions for improving data access with appropriate safeguards to maintain confidentiality
and without undue risk of adverse effects on public cooperation with censuses and surveys?
Persons or businesses harmed by disclosure. Do you know of any instances in which persons or businesses were harmed by unlawful or unintended disclosure of information they provided to the government under the condition that the information was to be used only for statistical purposes? How did this happen? What were the consequences? (This category differs from the first two in that statements need not be based on your own personal experience.)
Please submit your statements to me c/o Committee on National Statistics, National Research Council, 2101 Constitution Avenue, NW, Washington DC 20418. If you have any questions, please call Virginia de Wolf, Study Director, on 202/334–2550. We look forward to hearing from you.
George T. Duncan, Chair
Panel on Confidentiality and Data Access