Academic journal article Research-Technology Management

How to (Almost) Schedule Innovation: The Scientific Method and Statistical Design of Experiments Can Help

Academic journal article Research-Technology Management

How to (Almost) Schedule Innovation: The Scientific Method and Statistical Design of Experiments Can Help

Article excerpt

"You can't schedule innovation." This is a persistent and persuasive argument negating the relevance of project management in research and development. Since the purpose of research is to discover the unknown, the argument goes, it is impossible to predict whether it will even be successful, much less how long it will take. Consequently, attempts to force disciplines of project management onto a research effort will only add bureaucracy, frustration, and stifling of creativity, with no added benefit.

It is tempting to yield to this argument. In new product development and commercialization, however, an opposing school of thought also exists. Customers measure promises vs. delivery. Shareholders and directors (or owners for privately held firms) measure return on R&D investment. Competitors constantly threaten profitability. To the degree that a firm can commercialize new products on schedule, on budget and within specification, it will benefit financially.

So how do we reconcile the desire to define schedules and budgets with the inherent unpredictability of research and development? The purpose of this article is to discuss how application of the scientific method and statistical design of experiments can help in this regard.

The Two Tools

1. The Scientific Method.--The term "scientific method" is a broad one. The concepts under the umbrella of the method have evolved over hundreds of years, beginning with the ancient Greek philosophers. A nice summary of the important elements of the method as described by Kemeny is depicted in Figure 1 (1). It shows the area below the dotted line as the world of facts. The area above the line is the world of mathematics, the world that defines relationships among the facts. We use the term mathematics in the broadest sense. It is the ideal language for forming scientific theories. In some scientific disciplines, such as anthropology or the social sciences, numerical formulas and equations may not apply. Nevertheless, one of the numerous non-numerical branches of mathematics, such as topology, will allow the scientists to express theories regarding relationships among the facts.

[FIGURE 1 OMITTED]

A flowchart of activity depicting the application of the method as often taught in school science fairs is shown in Figure 2. It begins with induction and deduction: the student is asked to state hypotheses that can be tested by observation (Step A). The remaining steps comprise verification. Step B involves defining the experiments required to test the hypotheses. Step C comprises execution of those experiments and collection of the data. At step D, the original hypothesis or hypotheses are evaluated in light of the data. Interpretation includes answering such questions as:

[FIGURE 2 OMITTED]

* What is the repeatability of an individual experiment?

* Based on the data, is there more than one possible cause for any given observation (i.e., is it one cause or a combination of causes)?

* Do inexplicable observations imply there may be "lurking" variables? That is, are there experimental variables whose values I did not consider that could have drifted from one experiment to the next and caused large differences in observed results?

At step E, actions are defined based on the disciplined thinking from step D. One possibility is that new and refined hypotheses can be postulated (induction) and the steps repeated.

2. Statistical Design of Experiments!.--Some researchers have told me that they do not believe in experimental design. I contend that all experiments are designed. Some are designed by intuition and gut feel. Some say Thomas Edison followed this strategy. Other experiments are designed according to a rigorous statistical protocol of orthogonal combinations of variables. In either case, the experiments are designed. The researcher performs laboratory work to answer questions. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.