Efficient Design Testing With Statistical Concept of Degree of Freedom

Abstract

Efficient testing of design prototypes or early batches of products is essential to meeting time-to-market as well as field performance and other requirements such as those specified in environmental and safety standards. Most design engineers are concerned with the technical arrangements and execution of such tests but may not be making use of an approach based on information transformation that is better known to statisticians in the form of experimental design. In this paper, we suggest the use of strategies and tools of experimental design from the perspective of information utilization, and highlight ways in which specific performance characteristics of a new or improved product can be effectively and efficiency tested, quantified, and evaluated. The approach also bridges the usual gap that exists between design engineers and mathematical statisticians, as traditionally each profession tends to have its own parlance and mindset in empirical studies. The techniques illustrated in this paper tend to “fall through the crack” in various texts on empirical product testing; they help put in place a framework that can be fruitfully adopted by design engineers who do not necessarily have an extensive background in statistics. Knowledge of this framework is essential today when on one hand, time for product development, testing and evaluation is short in a globalized market, while on the other is the prevalence of computing hardware and software that has virtually eliminated the chores of data-crunching arising from product evaluation and proofing activities.