Sensitivity Analysis of the Rating Scales and Worth Calculation Schemes Used in QFD Matrices

Abstract

Quality Function Deployment (QFD) is matrix method that identifies relative worth of product requirements from the customer requirements and their importance. Understanding the relative worth enables engineers to evaluate the potential of design concepts to achieve important requirements. In a QFD matrix called “House of Quality” or QFD I, engineers assess correlations between product requirements and customer requirements using a linear (e.g., 1–3–5) or an exponential (e.g., 1–3–9) rating scale. The exponential scale assigns product requirements that have large correlations with customer requirements a higher ratings of 9 instead of 5, and therefore, gives them larger relative worth. This paper studies how the choice of linear 1–3–5 and exponential 1–3–9 rating scales changes the relative worth of product requirements. To avoid being restricted to any specific pattern of a QFD matrix, this paper uses simulations and analytic approaches to obtain distributions of changes of relative worth, and to calculate the upper bounds of these changes. Finally, in an illustrative example, the authors integrate QFD and concept evaluation activities and provide a case in which the choice of rating scale in a QFD I matrix changes the optimal concept.