Optimal Construction of Experimental Clusters

A critical issue in many marketing research studies is the creation of experimental clusters for the purpose of testing new products, advertising campaigns and other marketing decisions prior to widespread introduction in the marketplace. These experimental clusters are often called "test markets," "panels," and other such names, but regardless of the name, their central purpose is to allow controlled testing leading to inference to a population. Very often management wishes the experimental clusters to be "matched" on ratios reflecting change over time, comparative competitive position, etc. This paper examines Has problem of constructing such experimental clusters when cluster composition is measured by several ratios and the optimal construction would yield clusters with ratios as close as possible to those of the population. This paper demonstrates that it is possible to obtain optimal solutions through the solution of a mixed-integer linear programming problem which serves as a surrogate for the nonlinear formulation of the problem. An illustration of the successful application of this procedure is shown.