The application of stochastic models to study buyer behavior in consumer durable product categories

Byunghoon Lim, Purdue University

Abstract

I. The purpose of this study is to compare the degree of consumer heterogeneity in brand choice probabilities across consumer non-durable and durable goods and industrial products. Statistical problems encountered in estimating polarization for durable goods with few purchases per sampling unit are addressed. Specifically, we develop a test for pooling data across durable goods such as different kitchen and video products. We also develop a measure of consumer heterogeneity by decomposing observed variance in purchase proportions data into binomial variance and that due to heterogeneity in choice probabilities. Finally, we report our empirical findings from analysis of selected packaged goods (regular coffee and saltine crackers) and durable goods (kitchen and video products) and an industrial product (choice of aviation fuel suppliers). II. The purpose of this paper is to present an application of the beta-logistic model in the analysis of the buying pattern of consumer durable goods. Even though the logit model has been extensively applied to describe the buying pattern of frequently bought product categories, the application of the logit model has been very limited in other product categories such as consumer durable goods. The beta-logistic model solves the above problems by estimating accurate parameters at the brand level and increasing the model fit in the absence of the loyalty variables. Mathematical forms and the properties of the beta-logistic model are explained for a bivariate and a multivariate case and compared to those of the logit model. Finally empirical findings and managerial implications are presented from the analysis of two types of consumer durable goods--color TV's and casual shoes. III. In this study, we present an application of selected brand switching data based methods to identify and test competitive structures within the French automobile market. We perform cluster analysis of brand interaction indices (National Purchase Diary 1975) and proximity measures (Rao and Sabavala 1981) to identify alternate market structure hypotheses. Next, we use confirmatory techniques based on the multinomial-Dirichlet model (Kalwani and Morrison 1977) and the "forced switching" model (Urban, Johnson and Hauser 1984) to test the hypothesized market structures. Finally, we examine the degree of conformance of the results obtained using these approaches and present a summary view of our understanding of inter-brand competition in the French automobile market. (Abstract shortened by UMI.)

Degree

Ph.D.

Advisors

Kalwani, Purdue University.

Subject Area

Marketing

COinS