Inter-relating data sets for product development the reverse engineering approach

Product researchers often use a variety of data sets for the same set of products. They include formulations, consumer ratings, expert panel ratings, and instrumental measures. Although there are attempts to inter-relate these measures, for the most part, it is rare to include all sets of data in one general model, and then estimate the profile of one subset of variables using the profile of another subset of variables. This paper shows how the researcher can integrate formulations, consumer data, expert panel data, and instrumental measures using a set of equations created from the same set of independent variables. These variables may be systematically arrayed by experimental design, or created from principal components factor analysis. In order to inter-relate two profiles, the researcher sets one profile as the target or goal and then systematically explores the combinations of ingredients (or factor scores). The researcher looks for that combination that generates an expected profile lying `closeʹ to the goal profile.