Marketing Letters 3:4, (1992): 407-417 © 1992 Kluwer Academic Publishers, Manufactured in the Netherlands. A Decomposition of Repeat Buying V. KUMAR University of Houston, Department of Marketing, College of Business Administralion, Houston, TX 77204-6283 AMIT GHOSH Cleveland Stare University Departrnent of Marketing, College of Business, C!eveland, OH 44115 GERARD J. TELLIS* University of Southern California, School of Business Administration, Los Angeles, CA 90089-1421 Key words: Preference, Inertia, Coupon Proneness, Impulse Buying [November, 1991 Revised: May 1992] A bstract The authors decompose repeat buying for frequently purchased nondurables. The results are very similar for two categories each over a different city and time period. A factor anatysis of 18 mea- sures of repeat buying obtains four principal factors that explain 79-85% of the variance: Prefer- ence, Inertia, Coupon Proneness and Impulse Buying. A cluster analysis of factors on these di- mensions yields four segments, with distinct behavioral characteristics. Studies in many areas of marketing suggest that brand loyalty is an important dimension of repetitive buying of low-involvement, low-cost, frequently-pur- chased products. Thus, a better understanding of brand loyaIty would greatly pro- mote our understanding of consumer behavior for such products. This goal has led to at least 80 studies on brand loyalty (Jacoby and Chestnut 1978; Elrod 1987). Kim, Batra and Lehmann (1991) recently presented a study which attempts to review past measures of brand loyalty and test them for reliability and validity using a single-source scanner data. Nevertheless, several difficulties with past research binder a complete description of repetitive buying. First, no single defi- nition or measure of loyalty is completely satisfactory. In particular, attitudinal measures suffer from problems with reliability or self report bias, especially for single period surveys. Behavioral measures suffer from an inability to separate loyalty from response to marketing variables and either of them from random events. In addition, loyalty itself may be multi-dimensional consisting of true pref- erence for brands and inertia to switch brands. Similarly, brand switching needs *The authors thank IRI for the data.