Applying New Developments in Dimension Reduction Techniques in Statistics to Data-Rich Marketing Research

• Activity Number: 586
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics and Marketing
• Author(s): Neeraj Bharadwaj and Yuexiao Dong*+
• Companies: Temple University and Temple University
• Keywords: Factor analysis ; Parallel analysis ; Principal component analysis ; Sliced inverse regression ; Sufficient dimension reduction

Abstract

In the presence of large amounts of data, marketing researchers have relied upon dimension reduction methods (i.e., principal component analysis (PCA) and factor analysis) to reduce the number of variables effectively. Such traditional approaches extract a few significant factors that are linear combinations of the original predictor variables, but ignore the role of the response variables. To incorporate the information from both the predictor and the response variables, Naik et al. (2000) introduced sliced inverse regression (SIR), a sufficient dimension reduction technique, into the marketing research literature. SIR provides a viable alternative to PCA type dimension reduction methods as demonstrated by Monte Carlo simulations and empirical studies. Over the last decade, there have been significant advances in sufficient dimension reduction in statistics, yet few attempts to utilize them in marketing research. In this interdisciplinary research, we seek to address the call made by Naik at al. (2000) to marketing scholars and practitioners to "improve decision making by better utilizing vast amounts of marketing information" by applying recent advances in SIR from statistics.