Description:
The primary emphasis of quantitative marketing is on the evaluation of firm policies for determination of marketing variables as well as the optimization of these policies. Thus, our methods of inference must apply to decision problems and to policy evaluation. In addition, the data we have access to in marketing shapes the demands on inference in a qualitatively different manner than in economics. In marketing, we typically have data at a much lower level of product aggregation. This poses the challenge of creating demand systems for large numbers of products with an element of discrete demand. Traditionally, market researchers have used data at a very high level of geographic aggregation which raises real concerns regarding valid causal inferences. The availability of data at lower level of geographic aggregation, including consumer panel data, holds out the prospect for improved causal inferences even in the absence of experimentation. In addition, the highly targeted nature of some forms of advertising pose significant challenges for ad effect measurement. At the same time, heterogeneity in the effects of marketing variables create a profusion of model parameters placing further demands on the method of inference. In this chapter, we will review methods of inference as they apply specifically to the challenges posed in marketing applications.