Experimental Design

The power of industrial experimentation can be staggering. Businesses sometimes have problems that do not seem to be solvable. Frequently, this is because of an interaction between factors. The human brain is just not able to detect interactions – you need statistics for that. This is why experimental design is able to save companies millions of dollars and can solve "insolvable" problems. (Ouellette, 2003)

Experimental design (DOE or EXD) in industry involves using sound principles (frequently dictated by statistical procedures) in designing a study to learn something about your business by changing a business process and analyzing the effect. By utilizing these principles, you can determine the nature of reality based on data with economical experiments. The reasons why we might conduct such a study might be:

• Eliminating a problem (defect, defective) that has resisted the "usual" fixes
• Reducing variability in a process to improve the output quality
• Moving a process average on target
• or all three

EXD is a tool used within the context of a strategy (for example, advanced problem solving or quality improvement) and as a tool can be used for many different purposes. However, in general an experimental design follows these steps:

• State the problem or research purpose
• State the research question
• State the dependent variable(s)
• Select the associated criterion measure(s)
• Identify and classify all independent variables
• Perform measurement systems studies as appropriate
• Select levels of the incorporated treatment variables
• Select an appropriate experimental design
• Develop the experimental plan
• Run the experiment
• Analyze the results
• Run confirmation studies as required
• Report the findings

The topic of EXD is sophisticated and requires a significant amount of training. ROI offers world-class training both on-site as well as on-line.