Some Bayesian Approaches to Record Linkage

• Activity Number: 200
• Type: Topic Contributed
• Date/Time: Tuesday, August 13, 2002 : 10:30 AM to 12:20 PM
• Sponsor: Section on Survey Research Methods*
• Author(s): Brunero Liseo*+ and Marco Fortini and Alessandra Nuccitelli and Mauro Scanu
• Affiliation(s): University of Rome, La Sapienza and Istat, Rome Italy and Istat, Rome Italy and Istat, Rome Italy
• Keywords: False Match Rate ; MCMC ; Bayes factor

Abstract

Record Linkage (RL) refers to the use of an algorithmic technique to match records from different data sets that correspond to the same statistical unit. RL is ubiquitous in official statistics: estimation of population size via capture-recapture methods, testing of disclosure strategies, and coverage measurement surveys are only few examples.

A key difficult for any statistical analysis with RL, is the intensive computational burden. A Bayesian way to RL correctly clarifies the complexity of the problem, and it helps in finding adequate solutions. In this paper, we briefly discuss the validity of the more common statistical models for RL, and we propose a fully Bayesian approach to record linkage. We use standard Metropolis-Hastings and Simulated Annealing algorithms to derive the marginal posterior distribution of a matrix-valued parameter which indicates the "configuration" of matches between the two lists.