The matching methodology: some statistical properties (Q1187644)

This monograph is concerned with both theoretical investigations and empirical evaluation of the quality of synthetic files created from multiple but incomplete data sources by using file merging methodology. In Chapter 1, the problem of merging micro-files and various steps involved in the existing methodology to solve this problem are described. Specifically, a general framework to categorize known models for matching files are proposed. Chapter 2 deals with the situation pertinent to merging files on the same individuals. A review of known results in this direction is given. New optimality properties of a maximum likelihood matching strategy are established. Some small-sample and large-sample properties of the number of correct matches, which shed some light on the reliability of the synthetic file arising from using the maximum likelihood strategy, are derived. In Chapter 3, the situation of merging data on similar individuals is treated. The bulk of the discussion in this chapter is confined to matching two files of data that are sampled from a trivariate normal population.