The Study on the Asymmetry of Inertia and Variety-Seeking State - Using Section-Aggregated Multinomial Logit Analysis

관성 및 다양성추구 상태의 비대칭성에 관한 연구 - 구간통합 다항로짓분석을 활용하여

  • Received : 2012.01.10
  • Accepted : 2013.02.28
  • Published : 2013.03.01

Abstract

Customer's purchase state consists of purchase inertia and variety-seeking. As the growing brand familiarity triggers the increase of brand attractiveness, customers purchase state will be of inertia. However the excessively growing brand familiarity ignites the decrease of brand attractiveness. Followingly the purchase state will be tend to plunge into the variety-seeking state. The main topic of this study is to validate the asymmetric formation of customer's purchase states between inertia and variety-seeking. In order to follow up the main topic, this article introduces a model to freely describe the velocity of value changes depending upon the purchase states. This model will help overcome the limitation of the past studies having been based on the symmetric value changes. Based on this approach marketer will be able to decide the timing of sales promotions. This research utilized local telecommunication carrier's database of smartphone application purchase/download records. This database was collected from two years (2009 and 2010) span, the time when the smartphones started commodifying in Korea whilst most of the past studies had used purchase data of maturity stage products. From this approach utilizing the introduction stage data in the product life cycle, the probability of brand choice depending upon the purchase state on the early-stage can be probed. Moreover, this study tries to expand the research methodology from the other areas of research by knowledge sharing. Here this study introduces the methodology of section-aggregated multinomial logit to simultaneously estimate the parameters that were included in the plural multinomial logit functions while the plural functions were inter-connected. This adoption of section-aggregated multinomial logit model procedures from the computerized statistics areas is expected to nourish the marketing research for more precise analysis and estimation of effects of marketing activities.