Statistical Matching: Theory and Practice

Book Preface

Statistical matching is a relatively new area of research which has been receiving increasing attention in response to the flood of data which are now available. It has the practical objective of drawing information piecewise from different independent sample surveys.

The origins of statistical matching can be traced back to the mid-1960s, when a comprehensive data set with information on socio-demographic variables, income and tax returns by family was created by matching the 1966 Tax File and the 1967 Survey of Economic Opportunities; see Okner (1972). Interest in procedures for producing information from distinct sample surveys rose in the following years, although not without controversy. Is it possible to draw joint information on two variables never jointly observed but distinctly available in two independent sample surveys? Are standard statistical techniques able to solve this problem? As a matter of fact, there are two opposite aspects: the practical aspect that aims to produce a large amount of information rapidly and at low cost, and the theoretical aspect that needs to assess whether this production process is justifiable. This book is positioned at the boundary of these two aspects.