This paper
develops a methodology for modeling consumer response that integrates
previous research in stochastic brand selection, diffusion of
innovation, test market analysis, and new product design. The
methodology makes it practical to extend brand selection models
to include diffusion phenomena such as awareness, trial, and
information flow. Purchase timing and brand selection are interdependent
and both phenomena depend jointly on managerial controls such
as advertising, coupons, price-off promotion, product positioning,
and consumer characteristics.
Within this
general structure, we provide practical estimation procedures
(at least squares approximation to the maximum likelihood estimates)
to determine the parameters which link managerial controls to
consumer response. Closed form solutions are derived for cumulative
awareness, cumulative trial, penetration, expected sales, and
purchases due to promotion - all as a function of time. We also
provide simplified expressions for equilibrium (t ® µ) market
share. Tradeoffs among complexity of the diffusion process,
number of managerial variables, nonstationarity, complexity
of purchase timing, consumer segmentation, and sample size are
made explicit so that the marketing scientist can customize
his analyses to the managerial problems that he faces.
The effects
of sample size, data interval frequency, and collinearity in
the explanatory variables are investigated with simulations
based on a five-state consumer response process which depends
on 8-10 marketing variables.
The paper
closes with a brief description of the application and predictive
test of a consumer response model based on the methodology.
