A Factor-Analytic Probit Model for Representing the Market Structure in Panel Data

Abstract

Internal market structure analysis infers both brand attributes and consumer preferences for those attributes from preference or choice data. The authors exploit a new method for estimating probit models from panel data to infer market structures that can be displayed in few dimensions, even though the model can represent every possible vector of purchase probabilities. The result outperforms each of several other models, including Choice Map, SCULPTRE, and Chintagunta's latent class model in terms of goodness of fit, predictive validity, and face validity for a detergent data set. Because theirs is the only market structure model to outperform the structureless Dirichlet-multinomial stochastic brand choice model, the other methods cannot claim to have recovered market structure for these data.

Managerial applicability of graphical formats to support positioning decisions

“…In fact, it is only possible for a company to reposition a brand apart from the competition if the perceptions of their own as well as competitor brands are known. The uniqueness of a brand is implied by a certain attribute (Elrod and Keane 1995;Erdem 1996) and can be emphasized by means of advertising or achieved by adapting the physical properties of the brand.…”

Using store level scanner data to improve category management decisions: Developing positioning maps

“…As proposed by Elrod (1988), Chintagunta (1994), and Elrod and Keane (1995), preference constants can be interpreted as results of the compensatory valuation of different brand attributes or dimensions. Accordingly, assume:…”

“…Elrod (1991), DeSarbo et al (1994, Keane (1995), andElrod et al (2002) demonstrate the multiple methodological traditions oriented toward this goal.…”

“…Both the parametric approach (Elrod, 1988;Elrod and Keane, 1995) and the semi-parametric approach (Chintagunta, 1994;Andrews and Manrai, 1999) have been considered. These models indicate the positioning of brands in a set of latent attributes or dimensions, as well as the distribution of consumers' preferences with regard to these dimensions.…”

Bayesian estimation of multivariate-normal models when dimensions are absent

“…Compare the conjoint situation with the existing applications of the MNP to scanner data, where alternatives, and hence the random components of utility, are usually associated with brands (McCulloch and Rossi, 1994;Elrod and Keane, 1995) or brand-attribute combinations (Chintagunta (1992) considers brand-size combinations of ketchup, and Rossi et al (1996), consider brand-liquid combinations of canned tuna). These correlations have the natural and compelling interpretation of unobserved pairwise similarities between products (SKUs) at the brand or brand-attribute level.…”