Formulation a microdata dissemination policy
It is important for a data producing agency to publish a clear, transparent microdata dissemination policy. the policy should not be specific to a particular dataset. It should provide the overall information on accessibility to the data.
A good policy will at least contain:
- Information on whether and under what conditions the agency shares microdata
- A statement on the objectives of the policy
- Definition of the terms used in the policy
- References to the national legislation
- Identification on the responsibilities (who decides on the policy itself; who is in charge of its implementation)
- Pricing policy (not the price of each dataset; just an overal statement on the pricing policy)
- Information on the enforcement of the policy (penalties and others)
- The date (and version number) of the policy
The dissemination policy should be made available on the agency website. It should be accompanied by:
- Information specific to each dataset;
- Clear implementation procedures, which will define how, practically, users can request access to data (for licensed files and files available in data enclave), who will assess their requests, how and when they will be informed about the outcome of their requests, etc.
Example of a Microdata Release Policy
Preamble
The ABC Bureau of Statistics (ABCBOS) is committed to achieving excellence in the provision of timely, reliable and affordable official statistics for informed decision making in order to maximize the welfare of all ABCans. This is achieved through the collection and analysis of censuses, surveys and the use of administrative data as well as the dissemination a range of statistical products and providing assistance and services to users.
This policy addresses the conditions and the manner in which anonymized microdata files may be released to users for research purposes. It also strives to identify the different levels of anonymization for different categories of data use.
different approaches to providing access to microdata files.
The Policy for the Release of Anonymized Microdata Files
Objective: the objective of this policy is to define the nature of the anonymized microdata files that will be released, the intended use of these files and the conditions under which these files will be released.
Rationale: The ABCBOS aims to support research and teaching about the economic and social conditions of the ABCan population. In particular, this policy aims to support the needs of specialized researchers and students to by providing anonymized microdata files to be used strictly for research purposes. The output from this type of research forms an important the basis for decision-making about policies and programs in ABC.
Policy Statement: The ABCBOS will release microdata files for use by researchers for scientific research purposes when:
• The Director General is satisfied that all reasonable steps have been taken to prevent the identification of individual respondents
• The release of the data will substantially enhance the analytic value of the data that have been collected
• For all but purely public files, researchers disclose the nature and objectives of their intended research,
• It can be demonstrated that there are no credible alternative sources for these data, and
• The researchers have signed an appropriate undertaking and
• Legal provisions have been adhered to.
Definitions
For the purposes of this policy, microdata are defined as files of records pertaining to individual respondent units. Microdata files for dissemination purposes differ from those used within the ABCBOS in that all direct and indirect identifiers have been removed through various anonymisation procedures.
Anonymisation refers to the process of removing direct and indirect identifiers from the survey file to conceal the characteristics and the identity of individual respondents.
Direct identifiers include such information as names, addresses or other direct personal identifiers which must be removed from all files made available to users.
Indirect identifiers refer to characteristics which are shared with several other respondents and when combined with other information, can lead to compromising the identity of the respondent.
For the purposes of this policy, dissemination refers to the act of making microdata files and supporting metadata available for access and use with the data.
Types of Microdata files
There are three types of microdata files covered by this policy:
• Public Use Files (PUFs): Microdata files that are disseminated for general public use outside ABCBOS. They have been highly anonymised by removing the names and addresses and by collapsing geographic and respondent characteristic details to ensure that identification of individuals is highly unlikely. These files are made available for downloading from the ABCBOS site to individuals who identify themselves by name, provide their email addresses and agree to abide by the terms and conditions appropriate for a PUF. See Appendix B
• Licensed files: These files require that there be a signed agreement between the ABCBOS and major users, to permit them to access data files that are less highly anonymized and/or more sensitive than PUFs. For these files, all individual identifiers have been removed and some characteristic detail may be collapsed or removed. Licensing agreements are only entered into with bona fide users working for registered organizations. The primary and secondary researchers must be identified by name and a responsible officer of the organization must co-sign the license agreement. See Appendix C
• On-site Access files: The ABCBOS may, in the fullness of time and based on special needs, permit users to work on-site within the ABCBOS premises. This would permit access to sensitive data which could not be adequately anonymised for release outside of the ABCBOS. All data removed from the ABCBOS premises must be reviewed by an officer of ABCBOS to ensure that confidentiality is not breached. Researchers will be asked to outline their research objectives. See Appendix D. Please note that On-site access may not be planned by the ABCBOS at this time in which case this bullet and Appendix can be removed. It is included in this draft version for the sake of completeness
Classes of uses and users
The ABCBOS recognizes that microdata files are intended for specialized users with advanced quantitative skills. This policy distinguishes among the following classes of users and uses:
• Policymakers and researchers employed by line-ministries and planning departments who are engaged in the development of regional and national strategies and programs, including the monitoring and evaluation of these programs.
• International agencies involved in the conduct of special studies aimed at identifying development and support opportunities and the development programs and infrastructures within ABCD
• Research and academic institutes involved in social and economic research
• Students and professors mainly engaged in educational activities, and
• Other users who are involved in conducting scientific research (to be approved on a case by case basis
• (to be approved on a case by case basis).
This policy recognizes that different products will have different characteristics and will be of varying use to the users depending on their research objectives. The nature of the files produced will depend on the user needs that the survey/dissemination team have identified prior to producing the file.
Notes on degree of anonymization
There are three main types of anonymized files that may be produced under the terms of this policy. The major differences among these files are the levels of geographic and characteristic detail.
• Files that have less geographic detail and less characteristic detail (i.e., geography has been collapsed and variable detail has been reduced)
• Files that have less geographic detail and more variable detail and
• Files that have more or full geographic detail and more variable detail. (Such files are generally only available through data enclaves).
If all the survey variables are available in the file and the geography is detailed, then the greater the risk of identifying a respondent. Accordingly users requiring the maximum level of geographic detail may have to be prepared to work with less characteristic (variable) detail.
Files that have had both the variable detail and the geographic codes reduced are invaluable for students and professors for teaching and learning purposes so long as they are easily available to them. Such files can be made available as PUFs from the ABCBOS web site (see below).
All files will be reviewed by a Microdata Release Committee prior to release
Notes on Access Methods
The nature of the anonymized file will determine the manner in which it will be disseminated. The ABCBOS will release its dissemination microdata files on its ADA (ABCan Data Archive) server. PUFs are made available by having users complete an online form in which they provide their contact information. Having contact information enables ABCBOS to monitor the popularity of its products and enables contact with the users if there subsequent problems with the file or updates to the survey. We need to bring Dennis up to speed on NADA. It is not an archive in the sense of ICPSR but it a darned sight better than many NSO’s have. Now with AASDA on its way, the case is even stronger
Files requiring the user to sign a license are made available after the user completes and returns an access form which they can obtain from the ADA site. The ABCBOS may choose to provide these users with on-line access from a secure server, secure FTP or CD-ROM.
Responsibilities
Survey managers: will be responsible for
• identifying the needs of key stakeholders and ensuring the creation of an anonymized file that meets the needs to the user community to extent that possible under the Statistics Act.
• preparing an initial screening of the microdata file, identify any potential problem areas to be resolved and preparing a submission to the Microdata Release Committee.
The Microdata Release Committee: will be responsible for
• reviewing all requests by ABCBOS survey managers for the release of anonymized microdata files, using established criteria
• approving all files for release or providing guidance to the survey manger on how to improve the file before it may be released
• overseeing the licensing process and resolving issues dealing with possible breaches
• revising the guidelines that mangers use to create anonymized microdata files as necessary.
The Dissemination Group within the NSO: will be responsible for
• reviewing requests for access to licensed microdata files by researchers
•providing access to the data files for users who have been approvrevising the guidelines that mangers use to create anonymized microdata files as necessary.
• overseeing the licensing process and resolving issues dealing with possible breachesed for data ac.ess.
• responding to users’ requests for support and additional information
The Director General: will approve all releases of anonymized microdata files to users based on the advice and recommendation of the Microdata Release Committee.
Sponsors and Line-ministries: will be responsible for identifying their research needs and priorities to the survey managers prior to the design of the microdata files.
Timing and Release Policy
Recognizing the important needs of users and the importance of timely data, the ABCBOS will strive to release microdata files within 6-12 months after the first release of data from the survey. Surveys may need to be released in phases to ensure that the survey data have been reviewed and analyzed by the ABCBOS staff prior to their being anonymized for use outside the ABCBOS. This assures users that the ABCBOS will be able to answer questions about the survey.
Cost Recovery Policy
It is the policy of the ABCBOS to encourage broad use of its products by making them affordable for users. Accordingly, ABCBOS attempts to ensure that the costs of creating anonymized microdata files are built-in to the survey budget. At the same time, ABCBOS attempts to recover costs associated with the provisions of special services that benefit only a specific group.
The following cost-recovery principles will apply to the dissemination of microdata files.
• Public use files posted in the ABCBOS data server for broad download will be available without charge.
• Licensed data files will be made available to sponsors without charge providing that the cost of preparing the files is part of the survey budget.
• Line ministries are expected to ensure that the ABCBOS has sufficient funding to support the preparation of files for their use.
• Users demanding special services are expected to negotiate a cost-recovery agreement with the ABCBOS
Policy Update
This is version 1.3 of the policy, which was adopted on [... date ...] and became effective on [ ... date ...].
For Further Information
For further information about access to microdata, please contact: [put contact here]
