Modeling count data: Applications to consumer and physician markets

Hui-ming Wang, Purdue University

Abstract

We investigate the extent of interrelation in consumers' store brand purchasing behavior across different product categories and explore how it differs based on the degree of similarity between categories. The effects of customer shopping behavior and their relationship with the store on the propensity to purchase store brand are examined. We use a multivariate mixed Poisson model and calibrate it on the data from a national warehouse club that sells a premium store brand. The results show strong support for the relationship of store brand purchasing across product categories. Some of the associations between dissimilar product categories are as strong as the one between similar categories. Price spread between the national and store brands and frequency of store visits are found to positively affect the propensity to purchase store brands. Prescription drug market is characterized by constant changes. New drug launches, drug benefit extensions, and continuous clinical findings affect physicians' efficacy perceptions over time. We develop a model of physicians' learning behavior and study the role of personal selling in the learning process. We find that physicians are characterized by differential learning tendencies. Detailing and sampling have positive informative effects on efficacy perceptions. Increased learning tendencies are associated with decreased detailing responsiveness. High volume prescribers have higher learning tendencies than others.

Degree

Ph.D.

Advisors

Kalwani, Purdue University.

Subject Area

Marketing

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS