Census Bureau

Record Linkage Workshop, Washington, D.C.

March 20-21, 1997


The Committee on Applied and Theoretical Statistics of the National Academy of Sciences, the Washington Statistical Society, the Bureau of the Census, the National Agricultural Statistical Service, and The George Washington University are sponsoring a two-day workshop on record linkage in Washington, D.C.

The only previous major U.S. record linkage workshop (in 1985) resulted in a widely cited proceedings. Much has been accomplished in recent years and new areas of related research have arisen. A sampling of these would include the increased privacy concerns due to record linkage, the growing importance of more efficient use of statistical resources by using record linkage, heightened interest in linkage technology due to debates such as health care reform, issues in physical security of data, and the measurement of nondisclosure and reidentification risks in public use microdata files.

Record linkage involves the merging and unduplication of lists that may be used as survey frames or in conjunction with administrative files. Duplicates are a major source of error, are often not accounted for in quantitative analyses, and are very expensive to deal with during or after data collection. Computerized record linkage concepts were introduced by Newcombe in the 1950's and mathematically formalized by Fellegi and Sunter in a classic 1969 paper in the Journal of the American Statistical Association. Modern record linkage is practical because of new methods and software that have extended standard algorithms or introduced new algorithms in statistics, computer science, and operations research.

Talks and software exhibits are planned. Talks will be divided into two groups. A few longer talks will be selected from the submitted abstracts as invited papers. The remainder will be considered contributed papers. Contributed paper sessions will consist of 15 to 20 minute talks followed by 5 minutes of discussion. Applications of linkage technology are particularly welcome -- for example in epidemiology, demography, establishment surveys, and education, among others.

Martha Fair (Statistics Canada) and Patricia Whitridge (Statistics Canada) will give a short course/tutorial on record linkage on Thursday, March 20 from 8:30 a.m. to 11:30 a.m. Martha was a developer of Statistics Canada's GRLS record linkage system, has developed epidemiological applications, and has taught very well received tutorials. Patricia has been involved in the application of record linkage to the Canadian Census of Agriculture.

A number of software vendors will be providing demonstrations on Friday, March 21.

A web page for the workshop is http://www.census.gov/srd/www/reclink/reclink.html. The web page has more current information, a conference brochure, abstracts of invited presentations, and a short annotated bibliography of background papers.

Abstracts of talks or proposals for software presentations, including a brief description of the presentation, the presenter's name, affiliation, address (mailing, fax, e-mail as applicable), and phone number may be submitted anytime prior to January 3, 1997. Early submissions are encouraged.

Submit abstracts to Fritz Scheuren, Conference Chair, 1402 Ruffner Road, Alexandria VA 22302. For questions or abstracts, you may use e-mail if you wish scheuren@aol.com. Questions may also be directed to Bill Winkler (301-457-4729 voice, 301-457-2299 fax, or bwinkler@census.gov or Fritz Scheuren (at 703-549-1120 voice, or 202-994-6917 fax).