Executive Summary
Data fusion refers to the process by which different data sets are “fused” or merged so that they can be used jointly. For example, if one data set contains media exposure information and another data sets contains product purchase information, the process of data fusion would seek to produce one data set with both media exposure and purchase information for all observations. To fuse the data, all procedures rely on a set of common variables, such as demographic information.
In many marketing applications, data fusion is a means to the end of making joint inferences about media exposure and purchase behavior. For example, a firm would like to advertise its products on television shows whose audience has the greatest potential interest in the products. This means that the firm would like to infer the purchase behavior given viewership of all possible television shows. The authors develop a new set of data fusions methods for this specific problem.
The generic data fusion problem is to make inferences about the joint distribution of two sets of variables without any direct observations of the joint distribution. Instead, information is only available about each set separately along with some other set of common variables. The standard approach to data fusion creates a fused data set with the variables of interest and the common variables. The approach developed in this article directly estimates the joint distribution of just the variables of interest. For the case of either discrete or continuous variables, the approach yields a solution that can be implemented with standard statistical models and software.
In typical marketing applications, the common variables are psychographic or demographic variables, and the variables to be fused involve media viewing and product purchase. For this example, the approach directly estimates the joint distribution of media viewing and product purchase without including the common variables. This is the object required for marketing decisions. In marketing applications, fusion of discrete variables is required. The authors develop a method for relaxing the assumption of conditional independence for this case. They illustrate the approach with product purchase and media-viewing data from a large survey of British consumers.
Biography
Zvi Gilula is Professor (and Chairman) of Statistics at the Hebrew University, Jerusalem, where he received his doctoral degree, and he is Adjunct Professor of Statistics and Econometrics in the Graduate School of Business at the University of Chicago. His area of research and expertise is categorical data analysis, and he has been an associate editor for Journal of the American Statistical Association (Theory and Methods) since 1986. He was also an associate editor for the British Journal of the Marketing Research Society between 1999 and 2002. Zvi Gilula’s articles have been published mainly in Journal of the American Statistical Association, Biometrika, Journal of the Royal Statistical Society, and Annals of Statistics. He has been a consultant for a variety of companies and organizations, such as Hoffman LaRoche (Switzerland), Solvay (Germany), Navistar (the United States ), The Israeli National Lottery, the Israeli Advertising Association, TNS, BMRB, KMR (the United Kingdom), and The Morningside Group (China).
Robert E. McCulloch is Sigmund E. Edelstone Professor of Econometrics and Statistics in the Graduate School of Business at the University of Chicago. He obtained his BS from the University of Toronto in 1981 in Mathematics and Economics and his doctoral degree in Statistics from the University of Minnesota in 1985. His research interests include Bayesian statistics, graphical methods, model selection, and Bayesian data mining (CART).
Peter E. Rossi is Joseph T. and Bernice S. Lewis Professor of Marketing and Statistics in the Graduate School of Business at the University of Chicago. He received his doctoral degree from University of Chicago and his BA from Oberlin College. He has published a number of articles in peer-reviewed journals in marketing, economics, statistics, and econometrics, including Quantitative Marketing and Economics, Marketing Science, Journal of Marketing Research, American Economic Review, Journal of the American Statistical Association, Econometrica, Journal of Political Economy, Journal of Econometrics, Biometrika, Journal of Business and Economic Statistics, and Journal of Economic Theory. His areas of research interest include pricing and promotion, target marketing, direct marketing, micromarketing, limited dependent variable models, and Bayesian statistical methods. A fellow of the American Statistical Association and the Journal of Econometrics, he is the founding editor of Quantitative Marketing and Economics and past associate editor for Journal of the American Statistical Association, Journal of Econometrics, and Journal of Business and Economic Statistics. He is also the director of Kilts Center for Marketing in the Graduate School of Business at the University of Chicago.
J Marketing Research, Volume 43, Number 1, February 2006
View Table of Contents.