Statistical Matching A Frequentist Theory, Practical Applications, and Alternative Bayesian Approaches

Statistically matching of separate survey samples - can this be efficient? When there is no single source file available about all the information of interest, techniques of matching different data sets are often applied. Then individual respondents on one survey are matched to those on another based on some common characteristics. The respondents in the resulting data set will have all the answers to all the questions in both original surveys. For example, government policy questions as well as media planning tasks may be answered by means of such a statistically matched data set. This book covers a wide range of different aspects concerning statistical matching that in Europe typically is called data fusion. A theoretical framework is derived to determine the advantages and disadvantages of statistical matching. Its history and practical applications are discussed, and alternative approaches are proposed and evaluated with real world marketing data. Answers to the question of efficiency are provided. A book about statistical matching will be of interest to researchers and practitioners in many data analysis areas, starting with data collection and the production of public use micro files, data banks, and data bases. People in the areas of database marketing, public health analysis, socioeconomic modeling, and also official statistics also will find it useful. Susanne Rässler is senior research assistant and lecturer at the Institute of Statistics and Econometrics at the University of Erlangen-Nürnberg in Germany. She received her Ph.D. in 1995, having written a book about survey sampling theory with the focus on sampling with unequal probabilities. Later she started research about statistical matching. This book is the result of her "habilitation" thesis according to German academic tradition.