Brand Choice Inertia as One Aspect of the Notion of Brand Loyalty

A very parsimonious first-order multibrand model of the individual consumer for frequently purchased consumer goods is developed, generalizing the zero-order multinomial model. The model incorporates the notion of brand choice inertia, which is a form of short-term loyalty. Besides the equilibrium marginal choice probabilities of the multinomial model, a single additional parameter of choice inertia is defined. A maximum likelihood estimation procedure is derived and used on individual data from a subsample of the Secodip Consumer Panel, France: 2,401 purchases of cooking oil from eighty-nine of the consumers who were in the panel for the entire 1971-1972 period are used. Through the likelihood ratio test procedure, the model is shown to significantly improve on the multinomial model. An aggregate version of the model is also developed and estimated by a minimum chi-square procedure on four independent data sets. It is compared with the linear learning model which has been preferred over other probabilistic models of brand choice in most empirical studies in the past fifteen years. On the criterion of the p-level corresponding to the chi-square procedure, the model is superior for data set one, basically equivalent for data set two and its special case of the multinomial model being best for data sets three and four. The model operationalizes in a simple way the notion of carryover effect of marketing actions: a short-term perturbation from equilibrium--due to marketing mix actions, for example--has some residual effect on future purchase occasions. Moreover, in the context of new product modeling, the inertia model is in accord with the empirical observation by Parfitt and Collins of leveling-off of repeat-buying.